from typing import Union
import numpy as np
import unittest
from dataclasses import replace
from test.helpers import ast_const
from tinygrad.opt.kernel import Opt, OptOps, KernelOptError, Kernel
from tinygrad.codegen.lowerer import get_grouped_dims
from tinygrad.uop.ops import UOp, Ops, GroupOp
from tinygrad.device import Device, Buffer, is_dtype_supported
from tinygrad.shape.shapetracker import ShapeTracker
from tinygrad.shape.view import View
from tinygrad.tensor import Tensor, _to_np_dtype
from tinygrad.engine.realize import run_schedule, lower_schedule, CompiledRunner
from tinygrad.opt.heuristic import hand_coded_optimizations
from tinygrad.helpers import prod, Context, getenv, CI, flatten, dedup, AMX
from tinygrad.dtype import DType, dtypes
def helper_realized_ast(r:Union[Tensor, list[Tensor]]) -> tuple[UOp, list[Buffer]]:
if isinstance(r, Tensor): r = [r]
s = Tensor.schedule(*r)
run_schedule(s[:-1]) # run all kernels except the last one
assert s[-1].ast.op is Ops.SINK, f"helper_realized_ast expects a SINK {s[-1]}"
# now all input buffers in s[-1] should be realized
# create fresh buffers for the outputs
bufs = [Buffer((x).device, x.size, x.dtype).allocate() if i < len(s[-1].ast.src) else x for i,x in enumerate(s[-1].bufs)]
return s[-1].ast, bufs
def helper_tc_allclose(N:int, M:int, K:int, dtype_in:DType, dtype_out:DType, axis:int=0, tc_select:int=-1, tc_opt:int=0, use_tensor_cores:int=1):
a, b = Tensor.rand(M, K, dtype=dtype_in), Tensor.rand(K, N, dtype=dtype_in)
np_a, np_b = a.numpy(), b.numpy()
r = a.matmul(b, dtype=dtype_out)
if dtype_in == dtypes.bfloat16: r = r.float()
realized_ast, bufs = helper_realized_ast(r)
k = Kernel(realized_ast)
k.apply_tensor_cores(use_tensor_cores, axis=axis, tc_select=tc_select, tc_opt=tc_opt)
prg = CompiledRunner(replace(k.to_program(), device=Device.DEFAULT))
if use_tensor_cores == 1: assert len([uop for uop in k.uops if uop.op is Ops.WMMA]) > 0, "wmma not triggered"
elif use_tensor_cores == 3: assert len([uop for uop in k.uops if uop.op is Ops.DEFINE_LOCAL]) == 2, "local buffers not triggered"
assert len([x for x in k.applied_opts if x.op is OptOps.TC]) == 1, "tensor core opt not included"
prg.exec(bufs)
if dtype_in == dtypes.half: tc_atol, tc_rtol = 1e-2, 1e-3
elif dtype_in == dtypes.bfloat16: tc_atol, tc_rtol = 1e-2, 1e-2
else: tc_atol, tc_rtol = 5e-3, 1e-4
c = bufs[0].numpy().reshape((M,N))
np.testing.assert_allclose(c, np_a @ np_b, atol=tc_atol, rtol=tc_rtol)
def helper_tc_ensure_uops_and_opts_count(N: int, M:int, K:int, dtype_in:DType, dtype_out:DType, axis:int=0, tc_select:int=-1, tc_opt:int=0,
ensure_triggered:bool=True):
a, b = Tensor.rand(M, K, dtype=dtype_in), Tensor.rand(K, N, dtype=dtype_in)
r = a.matmul(b, dtype=dtype_out)
sched = r.schedule()
realized_ast = sched[-1].ast
k = Kernel(realized_ast)
k.apply_tensor_cores(1, axis=axis, tc_select=tc_select, tc_opt=tc_opt)
k.linearize()
wmmas = len([uop for uop in k.uops if uop.op is Ops.WMMA])
tcs = len([x for x in k.applied_opts if x.op is OptOps.TC])
if ensure_triggered:
assert wmmas > 0, "tensor core not triggered"
assert tcs == 1, "tensor core opt not included"
else:
assert wmmas == 0, "tensor core is incorrectly triggered"
assert tcs == 0, "tensor core opt is incorrectly included"
class TestLinearizer(unittest.TestCase):
def test_arg_dedup(self):
# NOTE: this realize exists because Tensor.numpy calls .contiguous() internally
# without contiguous folding, rand.to("CPU") and rand.contiguous().to("CPU") are different UOps.
# this test asserts they are the identical Buffer
# having different buffers is fine for correctness, because the outputs match.
a, b = Tensor.randn(4).realize(), Tensor.randn(4).realize()
np_a, np_b = a.numpy(), b.numpy()
c = ((a.shrink(((0, 2),)) - a.shrink(((2, 4),))) - (b.shrink(((0, 2),)) - b.shrink(((2, 4),))))
lowered = [x[1] for x in lower_schedule(c.schedule())]
for ei in lowered: ei.run()
rawbufs = lowered[-1].bufs
assert len(rawbufs) == 3 and set(rawbufs[1:]) == {a.uop.base.realized, b.uop.base.realized}
np_c = (np_a[:2] - np_a[2:]) - (np_b[:2] - np_b[2:])
np.testing.assert_allclose(np_c, c.numpy(), atol=1e-4, rtol=1e-4)
def test_load_removed(self):
a = Tensor.rand(1).realize()
b = Tensor.rand(1).realize()
ta = Tensor.where(Tensor(True), a, b).numpy()
tb = Tensor.where(Tensor(False), a, b).numpy()
np.testing.assert_equal(a.numpy(), ta)
np.testing.assert_equal(b.numpy(), tb)
def test_multioutput(self):
dtype, st = dtypes.int, ShapeTracker.from_shape((8,))
g0, g1, g2, g3 = [UOp(Ops.DEFINE_GLOBAL, dtype.ptr(), arg=i) for i in range(4)]
a = UOp(Ops.LOAD, dtype, src=(g2.view(st),))
b = UOp(Ops.LOAD, dtype, src=(g3.view(st),))
out0 = UOp(Ops.STORE, dtypes.void, src=(g0.view(st), a + b))
out1 = UOp(Ops.STORE, dtypes.void, src=(g1.view(st), a * b))
sink = UOp(Ops.SINK, src=(out0, out1))
a_t = Tensor.full(st.shape, 2).contiguous().realize()
b_t = Tensor.full(st.shape, 3).contiguous().realize()
lin = helper_linearizer_ast(sink, [a_t, b_t], wanna_output=[a_t.numpy()+b_t.numpy(), a_t.numpy()*b_t.numpy()])[0]
stores = [u for u in lin.uops if u.op is Ops.STORE]
mutable_bufs = dedup(flatten([[x for x in u.src[0].toposort() if x.op is Ops.DEFINE_GLOBAL] for u in stores]))
assert len(mutable_bufs) == len(stores) == 2
self.assertSetEqual(set([u.arg for u in mutable_bufs]), set([0,1]))
def _test_no_nested_ranges(self, lins, skip=None):
for l in lins:
range_in_acc = flatten([[x for x in u.src if x.op is Ops.RANGE] for u in l.uops if u.op is Ops.DEFINE_ACC])
ranges = [u.op for u in l.uops if (u.op is Ops.RANGE and u in range_in_acc) or (u.op is Ops.ENDRANGE and u.src[0] in range_in_acc)]
for i,u in enumerate(ranges):
if skip and i in skip: continue
assert ranges[i-1] != u, f"multireduce nested the ranges! {ranges[i-1], {u}}"
@unittest.expectedFailure
def test_const_alu_indexing(self):
st = ShapeTracker.from_shape((4,)).to_uop()
load = UOp.load(UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(), arg=1, src=()), st, dtype=dtypes.float)
op = load+UOp.const(dtypes.float, 1.0)*UOp.const(dtypes.float, -1)
store = UOp.store(UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(), arg=0, src=()), st, op)
Tensor.manual_seed(0)
x = Tensor.randn(4,).realize()
helper_linearizer_ast(store.sink(), [x], wanna_output=[x.numpy()+1*-1], opts=[])
# shapeless CONST in AST is not supported
@unittest.expectedFailure
def test_const_alu_indexing_one_const_fine(self):
st = ShapeTracker.from_shape((4,)).to_uop()
load = UOp.load(UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(), arg=1, src=()), st, dtype=dtypes.float)
op = load+UOp.const(dtypes.float, 1.0)
store = UOp.store(UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(), arg=0, src=()), st, op)
Tensor.manual_seed(0)
x = Tensor.randn(4,).realize()
helper_linearizer_ast(store.sink(), [x], wanna_output=[x.numpy()+1], opts=[])
@unittest.skipIf(CI and Device.DEFAULT in {"AMD"}, "AMD CI doesn't support multiple sync threads yet")
@unittest.skipUnless(Device[Device.DEFAULT].renderer.has_local, "test requires locals")
@unittest.skipUnless(Device[Device.DEFAULT].renderer.has_shared, "test requires shared")
def test_multireduce(self):
Tensor.manual_seed(0)
x = Tensor.randn(32, dtype=dtypes.float).realize()
st_x = x.uop.st
g0, g1 = [UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(), arg=i) for i in range(2)]
first_x = UOp(Ops.LOAD, dtypes.float, (g1.view(st_x.reshape((1, 32)).expand((32, 32))),))
first_reduce = UOp(Ops.REDUCE_AXIS, dtypes.float, (first_x,), (Ops.ADD, (1,)))
second_x = UOp(Ops.LOAD, dtypes.float, (g1.view(st_x.reshape((32, 1))),))
diff = second_x + first_reduce*ast_const(dtypes.float, -1, (32, 1))
second_reduce = UOp(Ops.REDUCE_AXIS, dtypes.float, (diff,), (Ops.ADD, (0,)))
store = UOp(Ops.STORE, dtypes.void, (g0.view(ShapeTracker.from_shape((1, 1))), second_reduce))
sink = UOp(Ops.SINK, src=(store,))
opts = [
[Opt(OptOps.GROUPTOP, 0, 2), Opt(OptOps.GROUPTOP, 1, 2)], # grouping
[Opt(OptOps.GROUPTOP, 0, 8), Opt(OptOps.GROUPTOP, 1, 8)],
[Opt(OptOps.GROUPTOP, 0, 16), Opt(OptOps.GROUPTOP, 1, 16)],
[Opt(OptOps.GROUPTOP, 0, 32), Opt(OptOps.GROUPTOP, 0, 32)],
[Opt(OptOps.UNROLL, 0, 2), Opt(OptOps.UNROLL, 1, 2)], # unroll reduce
[Opt(OptOps.UNROLL, 0, 4), Opt(OptOps.UNROLL, 1, 4)],
[Opt(OptOps.UNROLL, 0, 8), Opt(OptOps.UNROLL, 1, 8)] if Device.DEFAULT not in {"NV", "METAL"} else [], # can't do float8,
[Opt(OptOps.GROUPTOP, 0, 2), Opt(OptOps.GROUPTOP, 1, 2), Opt(OptOps.UNROLL, 2, 2), Opt(OptOps.UNROLL, 3, 2)], # grouping + unrolling
[Opt(OptOps.UNROLL, 0, 2), Opt(OptOps.UNROLL, 1, 2), Opt(OptOps.GROUPTOP, 0, 2), Opt(OptOps.GROUPTOP, 1, 2)],
[Opt(OptOps.GROUPTOP, 0, 4), Opt(OptOps.GROUPTOP, 1, 4), Opt(OptOps.UNROLL, 2, 8), Opt(OptOps.UNROLL, 2, 8)],
[Opt(OptOps.UNROLL, 0, 4), Opt(OptOps.UNROLL, 1, 4), Opt(OptOps.GROUPTOP, 0, 8), Opt(OptOps.GROUPTOP, 0, 8)],
]
wanna_output = (x.numpy()-x.numpy().sum(-1, keepdims=True)).sum(-1).reshape(1,1)
lins = helper_linearizer_ast(sink, [x], wanna_output=[wanna_output], opts=opts)
self._test_no_nested_ranges(lins, [0])
@unittest.skipIf(CI and Device.DEFAULT in {"AMD"}, "AMD CI doesn't support multiple sync threads yet")
@unittest.skipUnless(Device[Device.DEFAULT].renderer.has_local, "test requires locals")
@unittest.skipUnless(Device[Device.DEFAULT].renderer.has_shared, "test requires shared")
def test_mid_dim_multireduce(self):
Tensor.manual_seed(0)
x = Tensor.randn(27, 32, 5, dtype=dtypes.float).realize()
st_x = x.uop.st
g0, g1 = [UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(), arg=i) for i in range(2)]
first_x = UOp(Ops.LOAD, dtypes.float, (g1.view(st_x.reshape((27, 1, 32, 5)).expand((27, 32, 32, 5))),))
first_reduce = UOp(Ops.REDUCE_AXIS, dtypes.float, (first_x,), (Ops.ADD, (2,)))
second_x = UOp(Ops.LOAD, dtypes.float, (g1.view(st_x.reshape((27, 32, 1, 5))),))
diff = second_x + first_reduce*ast_const(dtypes.float, -1, (27, 32, 1, 5))
second_reduce = UOp(Ops.REDUCE_AXIS, dtypes.float, (diff,), (Ops.ADD, (1,)))
store = UOp(Ops.STORE, src=(g0.view(ShapeTracker.from_shape((27, 1, 1, 5))), second_reduce))
sink = UOp(Ops.SINK, src=(store,))
opts = [
# locals
[Opt(OptOps.LOCAL, 0, 3)],
[Opt(OptOps.LOCAL, 0, 9)],
[Opt(OptOps.LOCAL, 0, 27)],
# grouping
[Opt(OptOps.GROUPTOP, 0, 2), Opt(OptOps.GROUPTOP, 1, 2)],
[Opt(OptOps.GROUPTOP, 0, 8), Opt(OptOps.GROUPTOP, 1, 8)],
[Opt(OptOps.GROUPTOP, 0, 16), Opt(OptOps.GROUPTOP, 1, 16)],
[Opt(OptOps.GROUPTOP, 0, 32), Opt(OptOps.GROUPTOP, 0, 32)],
# # unroll
[Opt(OptOps.UNROLL, 0, 2), Opt(OptOps.UNROLL, 1, 2)],
[Opt(OptOps.UNROLL, 0, 4), Opt(OptOps.UNROLL, 1, 4)],
[Opt(OptOps.UNROLL, 0, 8), Opt(OptOps.UNROLL, 1, 8)] if Device.DEFAULT not in {"NV", "METAL"} else [],
# # upcasting
[Opt(OptOps.UPCAST, 0, 3)],
[Opt(OptOps.UPCAST, 0, 9)],
# locals with grouping
[Opt(OptOps.LOCAL, 0, 3), Opt(OptOps.GROUPTOP, 0, 2), Opt(OptOps.GROUPTOP, 1, 2)],
# locals with unroll
[Opt(OptOps.LOCAL, 0, 3), Opt(OptOps.UNROLL, 0, 2), Opt(OptOps.UNROLL, 1, 2)],
# locals with upcasting
[Opt(OptOps.LOCAL, 0, 3), Opt(OptOps.UPCAST, 0, 9)],
# grouping with unrolling
[Opt(OptOps.GROUPTOP, 0, 2), Opt(OptOps.GROUPTOP, 1, 2), Opt(OptOps.UNROLL, 2, 2), Opt(OptOps.UNROLL, 3, 2)],
[Opt(OptOps.GROUPTOP, 0, 4), Opt(OptOps.GROUPTOP, 1, 4), Opt(OptOps.UNROLL, 2, 8), Opt(OptOps.UNROLL, 2, 8)],
# grouping with upcasting
[Opt(OptOps.GROUPTOP, 0, 2), Opt(OptOps.GROUPTOP, 1, 2), Opt(OptOps.UPCAST, 0, 3)],
# locals with grouping with unroll
[Opt(OptOps.LOCAL, 0, 3), Opt(OptOps.GROUPTOP, 0, 2), Opt(OptOps.GROUPTOP, 1, 2), Opt(OptOps.UNROLL, 2, 2), Opt(OptOps.UNROLL, 3, 2)],
[Opt(OptOps.LOCAL, 0, 3), Opt(OptOps.GROUPTOP, 0, 4), Opt(OptOps.GROUPTOP, 1, 4), Opt(OptOps.UNROLL, 2, 8), Opt(OptOps.UNROLL, 2, 8)],
# locals with grouping with upcasting
[Opt(OptOps.LOCAL, 0, 3), Opt(OptOps.UPCAST, 0, 3), Opt(OptOps.GROUPTOP, 0, 2), Opt(OptOps.GROUPTOP, 1, 2)],
[Opt(OptOps.LOCAL, 0, 9), Opt(OptOps.UPCAST, 0, 3), Opt(OptOps.GROUPTOP, 0, 2), Opt(OptOps.GROUPTOP, 1, 2)],
# grouping with unrolling and upcasting
[Opt(OptOps.UPCAST, 0, 3), Opt(OptOps.GROUPTOP, 0, 2), Opt(OptOps.GROUPTOP, 1, 2), Opt(OptOps.UNROLL, 2, 2), Opt(OptOps.UNROLL, 3, 2)],
[Opt(OptOps.UPCAST, 0, 3), Opt(OptOps.GROUPTOP, 0, 4), Opt(OptOps.GROUPTOP, 1, 4), Opt(OptOps.UNROLL, 2, 8), Opt(OptOps.UNROLL, 2, 8)],
# locals + grouping + unrolling + upcasting
[Opt(OptOps.LOCAL, 0, 3), Opt(OptOps.UPCAST, 0, 3), Opt(OptOps.GROUPTOP, 0, 2), Opt(OptOps.GROUPTOP, 1, 2),
Opt(OptOps.UNROLL, 2, 2), Opt(OptOps.UNROLL, 3, 2)],
]
wanna_output = (x.numpy()-x.numpy().sum(axis=1, keepdims=True)).sum(axis=1).reshape(27,1,1,5)
lins = helper_linearizer_ast(sink, [x], wanna_output=[wanna_output], opts=opts)
self._test_no_nested_ranges(lins, [0])
def test_triple_multireduce(self):
Tensor.manual_seed(0)
x0 = Tensor.randn(27, 32, 5, dtype=dtypes.float).realize()
x1 = Tensor.randn(27, 32, 5, dtype=dtypes.float).realize()
x2 = Tensor.randn(27, 32, 5, dtype=dtypes.float).realize()
g0, g1, g2, g3 = [UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(), arg=i) for i in range(4)]
first_x = UOp(Ops.LOAD, dtypes.float, (g1.view(x0.uop.st.reshape((27, 1, 1, 32, 5)).expand((27, 32, 32, 32, 5))),))
first_reduce = UOp(Ops.REDUCE_AXIS, dtypes.float, (first_x,), (Ops.ADD, (3,)))
second_x = UOp(Ops.LOAD, dtypes.float, (g2.view(x1.uop.st.reshape((27, 1, 32, 1, 5)).expand((27, 32, 32, 1, 5))),))
diff = (second_x+first_reduce*ast_const(dtypes.float, -1, (27, 32, 32, 1, 5)))
second_reduce = UOp(Ops.REDUCE_AXIS, dtypes.float, (diff,), (Ops.ADD, (2,)))
third_x = UOp(Ops.LOAD, dtypes.float, (g3.view(x2.uop.st.reshape((27, 32, 1, 1, 5))),))
mul = (third_x*second_reduce)
third_reduce = UOp(Ops.REDUCE_AXIS, dtypes.float, (mul,), (Ops.ADD, (1,)))
store = UOp(Ops.STORE, src=(g0.view(ShapeTracker.from_shape((27, 1, 1, 1, 5))), third_reduce))
sink = UOp(Ops.SINK, src=(store,))
wanna_output = (x2.numpy()*(x1.numpy()-x0.numpy().sum(axis=1, keepdims=True)).sum(axis=1, keepdims=True)).sum(axis=1).reshape(27,1,1,1,5)
lins = helper_linearizer_ast(sink, [x0,x1,x2], wanna_output=[wanna_output])
self._test_no_nested_ranges(lins, [0])
@unittest.skipIf(CI and Device.DEFAULT in {"AMD"}, "AMD CI doesn't support multiple sync threads yet")
@unittest.skipUnless(Device[Device.DEFAULT].renderer.has_local, "test requires locals")
@unittest.skipUnless(Device[Device.DEFAULT].renderer.has_shared, "test requires shared")
@unittest.skip("this is not supported, it worked by luck")
def test_double_reduce_multireduce(self):
Tensor.manual_seed(0)
x = Tensor.randn(8, 32, 8, 16, dtype=dtypes.float).realize()
st = x.uop.st
g0, g1 = [UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(), arg=i) for i in range(2)]
first_x = UOp(Ops.LOAD, dtypes.float, (g1, st.reshape((8, 1, 32, 8, 1, 16)).expand((8, 32, 32, 8, 16, 16)).to_uop()))
first_reduce = UOp(Ops.REDUCE_AXIS, dtypes.float, (first_x,), (Ops.ADD, (2, 5)))
second_x = UOp(Ops.LOAD, dtypes.float, (g1, st.reshape((8, 32, 1, 8, 16, 1)).to_uop()))
neg_first_reduce = first_reduce * ast_const(dtypes.float, -1, (8, 32, 1, 8, 16, 1))
squares = (second_x+neg_first_reduce)
squares_sum = UOp(Ops.REDUCE_AXIS, dtypes.float, (squares,), (Ops.ADD, (1, 4)))
store = UOp(Ops.STORE, src=(g0, ShapeTracker.from_shape((8, 1, 1, 8, 1, 1)).to_uop(), squares_sum,))
sink = UOp(Ops.SINK, src=(store,))
wanna_output = (x.numpy()-x.numpy().sum(axis=(1,3), keepdims=True)).sum(axis=(1,3)).reshape((8,1,1,8,1,1))
opts = [
# openCL / GPU=1 is 256 max threads
# grouping
[Opt(OptOps.GROUPTOP, 0, 2), Opt(OptOps.GROUPTOP, 1, 2)], # first dim of both reduces
[Opt(OptOps.GROUPTOP, 0, 2), Opt(OptOps.GROUPTOP, 3, 2)], # both dims of the second reduce
[Opt(OptOps.GROUPTOP, 2, 2), Opt(OptOps.GROUPTOP, 3, 2)], # second dim of both reduces
[Opt(OptOps.GROUPTOP, 1, 2), Opt(OptOps.GROUPTOP, 3, 2)], # both dims of the first reduce
# group all reduce dims
[Opt(OptOps.GROUPTOP, 0, 2), Opt(OptOps.GROUPTOP, 1, 2), Opt(OptOps.GROUPTOP, 2, 2), Opt(OptOps.GROUPTOP, 3, 2)],
# checking how it works with 2 grouped reduces + unrolling
[Opt(OptOps.GROUPTOP, 0, 4), Opt(OptOps.GROUPTOP, 1, 4), Opt(OptOps.GROUPTOP, 2, 4), Opt(OptOps.GROUPTOP, 3, 4),
Opt(OptOps.UNROLL, 0, 2), Opt(OptOps.UNROLL, 1, 2), Opt(OptOps.UNROLL, 2, 2), Opt(OptOps.UNROLL, 3, 2)],
# Checking how it works with 2 grouped reduces + locals.
[Opt(OptOps.LOCAL, 0, 2), Opt(OptOps.LOCAL, 0, 4),
Opt(OptOps.GROUPTOP, 0, 2), Opt(OptOps.GROUPTOP, 1, 2), Opt(OptOps.GROUPTOP, 2, 2), Opt(OptOps.GROUPTOP, 3, 2)],
# Checking how it works with 2 grouped reduces + locals + unroll.
[Opt(OptOps.LOCAL, 0, 2),
Opt(OptOps.GROUPTOP, 0, 4), Opt(OptOps.GROUPTOP, 1, 4), Opt(OptOps.GROUPTOP, 2, 4), Opt(OptOps.GROUPTOP, 3, 4),
Opt(OptOps.UNROLL, 0, 2), Opt(OptOps.UNROLL, 1, 2), Opt(OptOps.UNROLL, 2, 2), Opt(OptOps.UNROLL, 3, 2)],
# Checking how it works with 2 grouped reduces + locals + upcast.
[Opt(OptOps.LOCAL, 0, 2), Opt(OptOps.UPCAST, 0, 2),
Opt(OptOps.GROUPTOP, 0, 2), Opt(OptOps.GROUPTOP, 1, 2), Opt(OptOps.GROUPTOP, 2, 2), Opt(OptOps.GROUPTOP, 3, 2)],
# Checking how it works with 2 grouped reduces + locals + upcast + unroll.
[Opt(OptOps.LOCAL, 0, 2), Opt(OptOps.UPCAST, 0, 2),
Opt(OptOps.GROUPTOP, 0, 4), Opt(OptOps.GROUPTOP, 1, 4), Opt(OptOps.GROUPTOP, 2, 4), Opt(OptOps.GROUPTOP, 3, 4),
Opt(OptOps.UNROLL, 0, 2), Opt(OptOps.UNROLL, 1, 2), Opt(OptOps.UNROLL, 2, 2), Opt(OptOps.UNROLL, 3, 2)],
]
lins = helper_linearizer_ast(sink, [x], wanna_output=[wanna_output], opts=opts)
self._test_no_nested_ranges(lins, [0, 1])
@unittest.skipIf(CI and Device.DEFAULT in {"AMD"}, "AMD CI doesn't support multiple sync threads yet")
@unittest.skipUnless(Device[Device.DEFAULT].renderer.has_local, "test requires locals")
@unittest.skipUnless(Device[Device.DEFAULT].renderer.has_shared, "test requires shared")
def test_partial_opt_multireduce(self):
# check how it works with one reduce optimized and one unoptimized
Tensor.manual_seed(0)
x = Tensor.randn(27, 15, 5, dtype=dtypes.float).softmax(1).realize()
g0, g1 = [UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(), arg=i) for i in range(2)]
first_x = UOp(Ops.LOAD, dtypes.float, (g1.view(x.uop.st.reshape((27, 1, 15, 5)).expand((27, 15, 15, 5))),))
first_reduce = UOp(Ops.REDUCE_AXIS, dtypes.float, (first_x,), (Ops.ADD, (2,)))
second_x = UOp(Ops.LOAD, dtypes.float, (g1.view(x.uop.st.reshape((27, 15, 1, 5))),))
diff = (second_x+first_reduce*ast_const(dtypes.float, -1, (27, 15, 1, 5)))
second_reduce = UOp(Ops.REDUCE_AXIS, dtypes.float, (diff,), (Ops.ADD, (1,)))
store = UOp(Ops.STORE, src=(g0.view(ShapeTracker.from_shape((27, 1, 1, 5))), second_reduce))
sink = UOp(Ops.SINK, src=(store,))
opts = [
[Opt(OptOps.GROUPTOP, 0, 3)], # grouping
[Opt(OptOps.GROUPTOP, 1, 3)],
[Opt(OptOps.GROUPTOP, 0, 15)],
[Opt(OptOps.GROUPTOP, 1, 15)],
[Opt(OptOps.UNROLL, 0, 3)],
[Opt(OptOps.UNROLL, 1, 3)],
]
wanna_output = (x.numpy()-x.numpy().sum(axis=1, keepdims=True)).sum(axis=1).reshape(27,1,1,5)
lins = helper_linearizer_ast(sink, [x], wanna_output=[wanna_output], opts=opts)
self._test_no_nested_ranges(lins, [0])
@unittest.skipIf(CI and Device.DEFAULT in {"AMD"}, "AMD CI doesn't support multiple sync threads yet")
@unittest.skipUnless(Device[Device.DEFAULT].renderer.has_local, "test requires locals")
@unittest.skipUnless(Device[Device.DEFAULT].renderer.has_shared, "test requires shared")
def test_multireduce_with_parallel(self):
Tensor.manual_seed(0)
x = Tensor.randn(4, 32, dtype=dtypes.float).realize()
x_p = Tensor.randn(4, 32, dtype=dtypes.float).realize()
g0, g1, g2 = [UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(), arg=i) for i in range(3)]
first_x = UOp(Ops.LOAD, dtypes.float, (g1.view(x.uop.st.reshape((4, 1, 32)).expand((4, 32, 32))),))
first_x_p = UOp(Ops.LOAD, dtypes.float, (g2.view(x_p.uop.st.reshape((4, 1, 32)).expand((4, 32, 32))),))
first_reduce = UOp(Ops.REDUCE_AXIS, dtypes.float, (first_x,), (Ops.ADD, (2,)))
first_reduce_p = UOp(Ops.REDUCE_AXIS, dtypes.float, (first_x_p.alu(Ops.EXP2),), (Ops.ADD, (2,)))
second_x = UOp(Ops.LOAD, dtypes.float, (g1.view(x.uop.st.reshape((4, 32, 1))),))
diff = (second_x+(first_reduce + first_reduce_p)*ast_const(dtypes.float, -1, (4, 32, 1)))
second_reduce = UOp(Ops.REDUCE_AXIS, dtypes.float, (diff,), (Ops.ADD, (1,)))
store = UOp(Ops.STORE, src=(g0.view(ShapeTracker.from_shape((4, 1, 1))), second_reduce))
sink = UOp(Ops.SINK, src=(store,))
opts = [
# [Opt(OptOps.GROUPTOP, 0, 2), Opt(OptOps.GROUPTOP, 1, 2)], # grouping
# [Opt(OptOps.GROUPTOP, 0, 8), Opt(OptOps.GROUPTOP, 1, 8)],
# [Opt(OptOps.GROUPTOP, 0, 16), Opt(OptOps.GROUPTOP, 1, 16)],
# [Opt(OptOps.GROUPTOP, 0, 32), Opt(OptOps.GROUPTOP, 0, 32)],
[Opt(OptOps.UNROLL, 0, 2), Opt(OptOps.UNROLL, 1, 2)], # unroll reduce
[Opt(OptOps.UNROLL, 0, 4), Opt(OptOps.UNROLL, 1, 4)],
[Opt(OptOps.UNROLL, 0, 8), Opt(OptOps.UNROLL, 1, 8)] if Device.DEFAULT not in {"NV", "METAL"} else [], # can't do float8,
# [Opt(OptOps.GROUPTOP, 0, 2), Opt(OptOps.GROUPTOP, 1, 2), Opt(OptOps.UNROLL, 2, 2), Opt(OptOps.UNROLL, 3, 2)], # grouping + unrolling
# [Opt(OptOps.UNROLL, 0, 2), Opt(OptOps.UNROLL, 1, 2), Opt(OptOps.GROUPTOP, 0, 2), Opt(OptOps.GROUPTOP, 1, 2)],
# [Opt(OptOps.GROUPTOP, 0, 4), Opt(OptOps.GROUPTOP, 1, 4), Opt(OptOps.UNROLL, 2, 8), Opt(OptOps.UNROLL, 2, 8)],
# [Opt(OptOps.UNROLL, 0, 4), Opt(OptOps.UNROLL, 1, 4), Opt(OptOps.GROUPTOP, 0, 8), Opt(OptOps.GROUPTOP, 0, 8)],
]
wanna_output = (x.numpy()-(x.numpy().sum(-1, keepdims=True)+np.exp2(x_p.numpy()).sum(-1, keepdims=True))).sum(-1).reshape(4, 1,1)
lins = helper_linearizer_ast(sink, [x,x_p], wanna_output=[wanna_output], opts=opts)
self._test_no_nested_ranges(lins, [0])
@unittest.skipIf(CI and Device.DEFAULT in {"AMD"}, "AMD CI doesn't support multiple sync threads yet")
def test_multiout_multireduce(self):
# check how multireduce works with multioutput
Tensor.manual_seed(0)
x = Tensor.randn(27, 15, 5, dtype=dtypes.float).realize()
g0, g1, g2 = [UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(), arg=i) for i in range(3)]
first_x = UOp(Ops.LOAD, dtypes.float, (g2.view(x.uop.st.reshape((27, 1, 15, 5)).expand((27, 15, 15, 5))),))
first_reduce = UOp(Ops.REDUCE_AXIS, dtypes.float, (first_x,), (Ops.ADD, (2,)))
second_x = UOp(Ops.LOAD, dtypes.float, (g2.view(x.uop.st.reshape((27, 15, 1, 5))),))
diff = (second_x+first_reduce*ast_const(dtypes.float, -1, (27, 15, 1, 5)))
second_reduce = UOp(Ops.REDUCE_AXIS, dtypes.float, (diff,), (Ops.ADD, (1,)))
store0 = UOp(Ops.STORE, src=(g0.view(ShapeTracker.from_shape((27, 1, 1, 5))), second_reduce))
second_out = second_reduce * ast_const(dtypes.float, 1/15, (27, 1, 1, 5))
store1 = UOp(Ops.STORE, src=(g1.view(ShapeTracker.from_shape((27, 1, 1, 5))), second_out))
sink = UOp(Ops.SINK, src=(store0, store1))
wanna_output = (x.numpy()-x.numpy().sum(axis=1, keepdims=True)).sum(axis=1).reshape(27,1,1,5)
helper_linearizer_ast(sink, [x], wanna_output=[wanna_output, wanna_output/15])
@unittest.skipIf(CI and Device.DEFAULT in {"AMD"}, "AMD CI doesn't support multiple sync threads yet")
def test_multiout_intermediate_multireduce(self):
# check how it outputing at different stages of the multireduce works
# TODO: Fails because the stores shapes do not match: store1.shape = (27,15,1,5) != store0.shape = (27,1,1,5)
# so the output shapes are different (FAIL!),
# if we change the shape of store1 to be contiguous, it will match store0 but not the value it's storing (FAIL!)
Tensor.manual_seed(0)
x = Tensor.randn(27, 15, 5, dtype=dtypes.float).realize()
g0, g1, g2 = [UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(), arg=i) for i in range(3)]
first_x = UOp(Ops.LOAD, dtypes.float, src=(g2.view(x.uop.st.reshape((27, 1, 15, 5)).expand((27, 15, 15, 5))),))
first_reduce = UOp(Ops.REDUCE_AXIS, dtypes.float, (first_x,), (Ops.ADD, (2,)))
second_x = UOp(Ops.LOAD, dtypes.float, src=(g2.view(x.uop.st.reshape((27, 15, 1, 5))),))
diff = (second_x+first_reduce*ast_const(dtypes.float, -1, (27, 15, 1, 5)))
second_reduce = UOp(Ops.REDUCE_AXIS, dtypes.float, (diff,), (Ops.ADD, (1,)))
store0 = UOp(Ops.STORE, src=(g0.view(ShapeTracker.from_shape((27, 1, 1, 5))), second_reduce))
store1 = UOp(Ops.STORE, src=(g1.view(ShapeTracker(views=(View(shape=(27,15,1,5), strides=(5,0,1,1), offset=0, mask=None, contiguous=False),))), first_reduce)) # noqa: E501
wanna_output0 = (x.numpy()-x.numpy().sum(axis=1, keepdims=True)).sum(axis=1).reshape(27,1,1,5)
wanna_output1 = x.numpy().sum(axis=1).reshape(27,1,1,5)
sink = UOp(Ops.SINK, src=(store0, store1))
with self.assertRaises(RuntimeError): # AST is invalid
helper_linearizer_ast(sink, [x], wanna_output=[wanna_output0, wanna_output1])
@unittest.skipIf(CI and Device.DEFAULT in {"AMD"}, "AMD CI doesn't support multiple sync threads yet")
def test_complete_unroll_multireduce(self):
Tensor.manual_seed(0)
x = Tensor.randn(27, 3, 5, dtype=dtypes.float).realize()
g0, g1 = [UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(), arg=i) for i in range(2)]
first_x = UOp(Ops.LOAD, dtypes.float, src=(g1.view(x.uop.st.reshape((27, 1, 3, 5)).expand((27, 3, 3, 5))),))
first_reduce = UOp(Ops.REDUCE_AXIS, dtypes.float, (first_x,), (Ops.ADD, (2,)))
second_x = UOp(Ops.LOAD, dtypes.float, src=(g1.view(x.uop.st.reshape((27, 3, 1, 5))),))
diff = (second_x+first_reduce*ast_const(dtypes.float, -1, (27, 3, 1, 5)))
second_reduce = UOp(Ops.REDUCE_AXIS, dtypes.float, (diff,), (Ops.ADD, (1,)))
store = UOp(Ops.STORE, src=(g0.view(ShapeTracker.from_shape((27, 1, 1, 5))), second_reduce))
sink = UOp(Ops.SINK, src=(store,))
opts = [[Opt(OptOps.UNROLL, 0, 3), Opt(OptOps.UNROLL, 0, 3)]]
wanna_output = (x.numpy()-x.numpy().sum(axis=1, keepdims=True)).sum(axis=1).reshape(27,1,1,5)
helper_linearizer_ast(sink, [x], wanna_output=[wanna_output], opts=opts)
@unittest.skipIf(CI and Device.DEFAULT in {"AMD"}, "AMD CI doesn't support multiple sync threads yet")
def test_upcast_multireduce(self):
Tensor.manual_seed(0)
x = Tensor.randn(27, 3, 5, dtype=dtypes.float).realize()
g0, g1 = [UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(), arg=i) for i in range(2)]
first_x = UOp(Ops.LOAD, dtypes.float, src=(g1.view(x.uop.st.reshape((27, 1, 3, 5)).expand((27, 3, 3, 5))),))
first_reduce = UOp(Ops.REDUCE_AXIS, dtypes.float, (first_x,), (Ops.ADD, (2,)))
second_x = UOp(Ops.LOAD, dtypes.float, src=(g1.view(x.uop.st.reshape((27, 3, 1, 5))),))
diff = (second_x+first_reduce*ast_const(dtypes.float, -1, (27, 3, 1, 5)))
second_reduce = UOp(Ops.REDUCE_AXIS, dtypes.float, (diff,), (Ops.ADD, (1,)))
store = UOp(Ops.STORE, src=(g0.view(ShapeTracker.from_shape((27, 1, 1, 5))), second_reduce))
sink = UOp(Ops.SINK, src=(store,))
opts = [[Opt(OptOps.UPCAST, 0, 3)]]
wanna_output = (x.numpy()-x.numpy().sum(axis=1, keepdims=True)).sum(axis=1).reshape(27,1,1,5)
helper_linearizer_ast(sink, [x], wanna_output=[wanna_output], opts=opts)
@unittest.skipIf(CI and Device.DEFAULT in {"AMD"}, "AMD CI doesn't support multiple sync threads yet")
@unittest.skipUnless(Device[Device.DEFAULT].renderer.has_local, "test requires locals")
@unittest.skip("can't group with multiple reduces yet")
def test_early_endif(self):
# make sure the if block of a grouped reduce can be closed early and the result loaded back in
Tensor.manual_seed(0)
x = Tensor.randn(27, 12, 5, dtype=dtypes.float).realize()
g0, g1 = [UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(), arg=i) for i in range(2)]
first_x = UOp(Ops.LOAD, dtypes.float, (g1, x.uop.st.reshape((27, 1, 12, 5)).expand((27, 12, 12, 5)).to_uop()))
first_reduce = UOp(Ops.REDUCE_AXIS, dtypes.float, (first_x,), (Ops.ADD, (2,)))
second_x = UOp(Ops.LOAD, dtypes.float, (g1, x.uop.st.reshape((27, 12, 1, 5)).to_uop()))
diff = (second_x+first_reduce*ast_const(dtypes.float, -1, (27, 12, 1, 5)))
second_reduce = UOp(Ops.REDUCE_AXIS, dtypes.float, (diff,), (Ops.ADD, (1,)))
store = UOp(Ops.STORE, src=(g0, ShapeTracker.from_shape((27, 1, 1, 5)).to_uop(), second_reduce))
sink = UOp(Ops.SINK, src=(store,))
opts = [[Opt(OptOps.GROUPTOP, 0, 3), Opt(OptOps.GROUPTOP, 1, 3)]]
wanna_output = (x.numpy()-x.numpy().sum(axis=1, keepdims=True)).sum(axis=1).reshape(27,1,1,5)
helper_linearizer_ast(sink, [x], wanna_output=[wanna_output], opts=opts)
@unittest.skipIf(CI and Device.DEFAULT in {"AMD"}, "AMD CI doesn't support multiple sync threads yet")
def test_mean_std_multireduce(self):
Tensor.manual_seed(0)
x = Tensor.randn(15, 25, 35, dtype=dtypes.float).realize()
g0, g1 = [UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(), arg=i) for i in range(2)]
first_x = UOp(Ops.LOAD, dtypes.float, src=(g1.view(x.uop.st.reshape((15, 25, 1, 35)).expand((15, 25, 35, 35))),))
first_reduce = UOp(Ops.REDUCE_AXIS, dtypes.float, (first_x,), (Ops.ADD, (3,)))
neg_mean = first_reduce * ast_const(dtypes.float, -1/35, (15, 25, 35, 1))
second_x = UOp(Ops.LOAD, dtypes.float, src=(g1.view(x.uop.st.reshape((15, 25, 35, 1))),))
squares = (second_x+neg_mean)*(second_x+neg_mean)
squares_sum = UOp(Ops.REDUCE_AXIS, dtypes.float, (squares,), (Ops.ADD, (2,)))
variance = squares_sum * ast_const(dtypes.float, 1/35, (15, 25, 1, 1))
std = variance.alu(Ops.SQRT)
store = UOp(Ops.STORE, src=(g0.view(ShapeTracker.from_shape((15, 25, 1, 1))), std))
sink = UOp(Ops.SINK, src=(store,))
wanna_output = x.numpy().std(axis=2, ddof=0).reshape((15,25,1,1))
helper_linearizer_ast(sink, [x], wanna_output=[wanna_output])
@unittest.skipIf(CI and Device.DEFAULT in {"AMD"}, "AMD CI doesn't support multiple sync threads yet")
def test_mean_std_multireduce_mid_dim(self):
Tensor.manual_seed(0)
x = Tensor.randn(15, 25, 35, dtype=dtypes.float).realize()
g0, g1 = [UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(), arg=i) for i in range(2)]
first_x = UOp(Ops.LOAD, dtypes.float, src=(g1.view(x.uop.st.reshape((15, 1, 25, 35)).expand((15, 25, 25, 35))),))
first_reduce = UOp(Ops.REDUCE_AXIS, dtypes.float, (first_x,), (Ops.ADD, (2,)))
neg_mean = first_reduce * ast_const(dtypes.float, -0.04, (15, 25, 1, 35))
second_x = UOp(Ops.LOAD, dtypes.float, src=(g1.view(x.uop.st.reshape((15, 25, 1, 35))),))
squares = (second_x+neg_mean)*(second_x+neg_mean)
squares_sum = UOp(Ops.REDUCE_AXIS, dtypes.float, (squares,), (Ops.ADD, (1,)))
variance = squares_sum * ast_const(dtypes.float, 0.04, (15, 1, 1, 35))
std = variance.alu(Ops.SQRT)
store = UOp(Ops.STORE, src=(g0.view(ShapeTracker.from_shape((15, 1, 1, 35))), std))
sink = UOp(Ops.SINK, src=(store,))
wanna_output = x.numpy().std(axis=1, ddof=0).reshape((15,1,1,35))
helper_linearizer_ast(sink, [x], wanna_output=[wanna_output])
@unittest.skipIf(CI and Device.DEFAULT in {"AMD"}, "AMD CI doesn't support multiple sync threads yet")
@unittest.expectedFailure
def test_mean_std_multireduce_multiout(self):
# TODO: Similar error to test_multiout_intermediate_multireduce (implicit expand vs shape mismatch)
Tensor.manual_seed(0)
x = Tensor.randn(15, 25, 35, dtype=dtypes.float).realize()
g0, g1, g2 = [UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(), arg=i) for i in range(3)]
first_x = UOp(Ops.LOAD, dtypes.float, (g2, x.uop.st.reshape((15, 25, 1, 35)).expand((15, 25, 35, 35)).to_uop()))
first_reduce = UOp(Ops.REDUCE_AXIS, dtypes.float, (first_x,), (Ops.ADD, (3,)))
neg_mean = first_reduce * ast_const(dtypes.float, -1/35, (15, 25, 35, 1))
second_x = UOp(Ops.LOAD, dtypes.float, (g2, x.uop.st.reshape((15, 25, 35, 1)).to_uop()))
squares = (second_x+neg_mean)*(second_x+neg_mean)
squares_sum = UOp(Ops.REDUCE_AXIS, dtypes.float, (squares,), (Ops.ADD, (2,)))
variance = squares_sum * ast_const(dtypes.float, 1/35, (15, 25, 1, 1))
std = variance.alu(Ops.SQRT)
store_mean = UOp(Ops.STORE, src=(g1, ShapeTracker.from_shape((15, 25, 1, 1)).to_uop(), neg_mean))
store_std = UOp(Ops.STORE, src=(g0, ShapeTracker.from_shape((15, 25, 1, 1)).to_uop(), std))
sink = UOp(Ops.SINK, src=(store_std, store_mean))
wanna_output = [x.numpy().std(axis=2, ddof=0).reshape(15,25,1,1), x.numpy().mean(axis=2).reshape(15,25,1,1)]
lins = helper_linearizer_ast(sink, [x], wanna_output=wanna_output)
for k in lins:
assert len([u for u in k.uops if u.op is Ops.DEFINE_ACC]) == 2, "got more than two accs (implies the kernel didn't reuse the mean reduce)"
@unittest.skipIf(CI and Device.DEFAULT in {"PTX", "AMD", "NV"}, "ocelot/remu doesn't have multiple wave syncs yet")
def test_var_multireduce(self):
Tensor.manual_seed(0)
x = Tensor.randn(3, 27, 32, dtype=dtypes.float).realize()
g0, g1 = [UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(), arg=i) for i in range(2)]
# push reduce (3, 27, 32) -> (3, 27, 1) -> (3, 27, 32) expand to LOAD
first_x = UOp(Ops.LOAD, dtypes.float, src=(g1.view(x.uop.st.reshape((3, 27, 1, 32)).expand((3, 27, 32, 32))),))
first_reduce = UOp(Ops.REDUCE_AXIS, dtypes.float, (first_x,), (Ops.ADD, (3,)))
neg_mean = first_reduce * ast_const(dtypes.float, -0.03125, (3, 27, 32, 1))
# store = UOp(Ops.STORE, src=(g0, ShapeTracker.from_shape((3, 27, 32, 1)).to_uop(), mean))
# verify_lazyop(store)
second_x = UOp(Ops.LOAD, dtypes.float, src=(g1.view(x.uop.st.reshape((3, 27, 32, 1))),))
squares = (second_x+neg_mean)*(second_x+neg_mean)
squares_sum = UOp(Ops.REDUCE_AXIS, dtypes.float, (squares,), (Ops.ADD, (2,)))
variance = squares_sum * ast_const(dtypes.float, 0.03125, (3, 27, 1, 1))
store = UOp(Ops.STORE, src=(g0.view(ShapeTracker.from_shape((3, 27, 1, 1))), variance))
sink = UOp(Ops.SINK, src=(store,))
wanna_output = x.numpy().var(axis=2, ddof=0).reshape((3,27,1,1))
helper_linearizer_ast(sink, [x], wanna_output=[wanna_output])
# tinygrad ref
y_tiny = x.var(axis=2, correction=0).reshape(3,27,1,1)
np.testing.assert_allclose(y_tiny.numpy(), wanna_output, atol=1e-4, rtol=1e-4)
@unittest.skipIf(CI and Device.DEFAULT in {"AMD"}, "AMD CI doesn't support multiple sync threads yet")
def test_softmax_multireduce(self):
x = Tensor.rand(4, 32).realize()
g0, g1 = [UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(), arg=i) for i in range(2)]
first_x = UOp(Ops.LOAD, dtypes.float, src=(g1.view(x.uop.st.reshape((4, 1, 32,)).expand((4, 32, 32))),))
max_x = UOp(Ops.REDUCE_AXIS, dtypes.float, (first_x,), (Ops.MAX, (2,)))
second_x = UOp(Ops.LOAD, dtypes.float, src=(g1.view(x.uop.st.reshape((4, 32, 1,))),))
centered_x = second_x+max_x*ast_const(dtypes.float, -1, (4, 32, 1))
exp_x = centered_x.alu(Ops.EXP2)
sum_exp_x = UOp(Ops.REDUCE_AXIS, dtypes.float, (exp_x,), (Ops.ADD, (1,)))
# y = exp_x * sum_exp_x.alu(Ops.RECIP) # kernels cannot do a return to full shape
recip_sum_exp_x = sum_exp_x.alu(Ops.RECIP)
store = UOp(Ops.STORE, src=(g0.view(ShapeTracker.from_shape((4,1,1))), recip_sum_exp_x))
sink = UOp(Ops.SINK, src=(store,))
expected = 1/np.exp2(x.numpy() - x.numpy().max(axis=-1, keepdims=True)).sum(axis=-1, keepdims=True).reshape(4,1,1)
helper_linearizer_ast(sink, [x], wanna_output=[expected])
@unittest.skipIf(CI and Device.DEFAULT in {"PTX", "AMD", "NV"}, "very slow")
def test_indexing_multireduce(self):
dataset = Tensor.rand(16384, 256).realize()
idxs = Tensor([0,3,5,6]).realize()
with Context(FUSE_ARANGE=1):
sink = dataset[idxs].contiguous().kernelize().uop.base.src[1].arg.ast
real_index = dataset.numpy()[idxs.numpy()].reshape(4, 1, 256, 1)
helper_linearizer_ast(sink, [dataset, idxs], wanna_output=[real_index])
# AssertionError: repeated stores in uops
def test_argmax_multireduce_axis0(self):
t = Tensor.randn(10, 20).realize()
t_max = t.max((0,)).realize()
real_argmax = np.argmax(t.numpy(), axis=0, keepdims=False).reshape(1, 20, 1)
ast = UOp(Ops.SINK, dtypes.void, arg=None, src=(
UOp(Ops.STORE, dtypes.void, arg=None, src=(
UOp(Ops.VIEW, dtypes.int.ptr(20), arg=ShapeTracker(views=(View(shape=(1, 20, 1), strides=(0, 1, 0), offset=0, mask=None, contiguous=True),)), src=( # noqa: E501
UOp(Ops.DEFINE_GLOBAL, dtypes.int.ptr(-1), arg=0, src=()),)),
UOp(Ops.ADD, dtypes.int, arg=None, src=(
UOp(Ops.ADD, dtypes.int, arg=None, src=(
UOp(Ops.CONST, dtypes.int, arg=10, src=(
x6:=UOp(Ops.VIEW, dtypes.void, arg=ShapeTracker(views=(View(shape=(1, 20, 1), strides=(0, 0, 0), offset=0, mask=None, contiguous=False),)), src=()),)), # noqa: E501
UOp(Ops.MUL, dtypes.int, arg=None, src=(
x8:=UOp(Ops.CONST, dtypes.int, arg=-1, src=(
x6,)),
UOp(Ops.REDUCE_AXIS, dtypes.int, arg=(Ops.MAX, (0,)), src=(
UOp(Ops.MUL, dtypes.int, arg=None, src=(
UOp(Ops.CAST, dtypes.int, arg=None, src=(
UOp(Ops.CMPNE, dtypes.bool, arg=None, src=(
UOp(Ops.CMPNE, dtypes.bool, arg=None, src=(
UOp(Ops.LOAD, dtypes.float, arg=None, src=(
UOp(Ops.VIEW, dtypes.float.ptr(200), arg=ShapeTracker(views=(View(shape=(10, 20, 1), strides=(20, 1, 0), offset=0, mask=None, contiguous=True),)), src=( # noqa: E501
UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(-1), arg=1, src=()),)),)),
UOp(Ops.LOAD, dtypes.float, arg=None, src=(
UOp(Ops.VIEW, dtypes.float.ptr(20), arg=ShapeTracker(views=(View(shape=(10, 20, 1), strides=(0, 1, 0), offset=0, mask=None, contiguous=False),)), src=( # noqa: E501
UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(-1), arg=2, src=()),)),)),)),
UOp(Ops.CONST, dtypes.bool, arg=True, src=(
x21:=UOp(Ops.VIEW, dtypes.void, arg=ShapeTracker(views=(View(shape=(10, 20, 1), strides=(0, 0, 0), offset=0, mask=None, contiguous=False),)), src=()),)),)),)), # noqa: E501
UOp(Ops.ADD, dtypes.int, arg=None, src=(
UOp(Ops.REDUCE_AXIS, dtypes.int, arg=(Ops.ADD, (2,)), src=(
UOp(Ops.WHERE, dtypes.int, arg=None, src=(
UOp(Ops.VALID, dtypes.bool, arg=None, src=(
UOp(Ops.VIEW, dtypes.void, arg=ShapeTracker(views=(View(shape=(11, 19), strides=(0, 0), offset=0, mask=((0, 11), (9, 19)), contiguous=False), View(shape=(10, 20, 10), strides=(1, 0, 20), offset=0, mask=None, contiguous=False))), src=()),)), # noqa: E501
UOp(Ops.CONST, dtypes.int, arg=-1, src=(
x28:=UOp(Ops.VIEW, dtypes.void, arg=ShapeTracker(views=(View(shape=(10, 20, 10), strides=(0, 0, 0), offset=0, mask=None, contiguous=False),)), src=()),)), # noqa: E501
UOp(Ops.CONST, dtypes.int, arg=0, src=(
x28,)),)),)),
UOp(Ops.CONST, dtypes.int, arg=10, src=(
x21,)),)),)),)),)),)),
x8,)),)),))
helper_linearizer_ast(ast, [t, t_max], wanna_output=[real_argmax])
def test_argmax_multireduce_flat(self):
t = Tensor.randn(10, 20).realize()
t_max = t.max().realize()
real_argmax = np.argmax(t.numpy())
ast = UOp(Ops.SINK, dtypes.void, arg=None, src=(
UOp(Ops.STORE, dtypes.void, arg=None, src=(
UOp(Ops.VIEW, dtypes.int.ptr(1), arg=ShapeTracker(views=(View(shape=(1, 1), strides=(0, 0), offset=0, mask=None, contiguous=True),)), src=(
UOp(Ops.DEFINE_GLOBAL, dtypes.int.ptr(-1), arg=0, src=()),)),
UOp(Ops.ADD, dtypes.int, arg=None, src=(
UOp(Ops.ADD, dtypes.int, arg=None, src=(
UOp(Ops.CONST, dtypes.int, arg=200, src=(
x6:=UOp(Ops.VIEW, dtypes.void, arg=ShapeTracker(views=(View(shape=(1, 1), strides=(0, 0), offset=0, mask=None, contiguous=True),)), src=()),)), # noqa: E501
UOp(Ops.MUL, dtypes.int, arg=None, src=(
x8:=UOp(Ops.CONST, dtypes.int, arg=-1, src=(
x6,)),
UOp(Ops.REDUCE_AXIS, dtypes.int, arg=(Ops.MAX, (0,)), src=(
UOp(Ops.MUL, dtypes.int, arg=None, src=(
UOp(Ops.CAST, dtypes.int, arg=None, src=(
UOp(Ops.CMPNE, dtypes.bool, arg=None, src=(
UOp(Ops.CMPNE, dtypes.bool, arg=None, src=(
UOp(Ops.LOAD, dtypes.float, arg=None, src=(
UOp(Ops.VIEW, dtypes.float.ptr(200), arg=ShapeTracker(views=(View(shape=(200, 1), strides=(1, 0), offset=0, mask=None, contiguous=True),)), src=( # noqa: E501
UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(-1), arg=1, src=()),)),)),
UOp(Ops.LOAD, dtypes.float, arg=None, src=(
UOp(Ops.VIEW, dtypes.float.ptr(1), arg=ShapeTracker(views=(View(shape=(200, 1), strides=(0, 0), offset=0, mask=None, contiguous=False),)), src=( # noqa: E501
UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(-1), arg=2, src=()),)),)),)),
UOp(Ops.CONST, dtypes.bool, arg=True, src=(
x21:=UOp(Ops.VIEW, dtypes.void, arg=ShapeTracker(views=(View(shape=(200, 1), strides=(0, 0), offset=0, mask=None, contiguous=False),)), src=()),)),)),)), # noqa: E501
UOp(Ops.ADD, dtypes.int, arg=None, src=(
UOp(Ops.REDUCE_AXIS, dtypes.int, arg=(Ops.ADD, (1,)), src=(
UOp(Ops.WHERE, dtypes.int, arg=None, src=(
UOp(Ops.VALID, dtypes.bool, arg=None, src=(
UOp(Ops.VIEW, dtypes.void, arg=ShapeTracker(views=(View(shape=(201, 399), strides=(0, 0), offset=0, mask=((0, 201), (199, 399)), contiguous=False), View(shape=(200, 200), strides=(1, 400), offset=0, mask=None, contiguous=False))), src=()),)), # noqa: E501
UOp(Ops.CONST, dtypes.int, arg=-1, src=(
x28:=UOp(Ops.VIEW, dtypes.void, arg=ShapeTracker(views=(View(shape=(200, 200), strides=(0, 0), offset=0, mask=None, contiguous=False),)), src=()),)), # noqa: E501
UOp(Ops.CONST, dtypes.int, arg=0, src=(
x28,)),)),)),
UOp(Ops.CONST, dtypes.int, arg=200, src=(
x21,)),)),)),)),)),)),
x8,)),)),))
helper_linearizer_ast(ast, [t, t_max], wanna_output=[real_argmax])
@unittest.skipIf(CI and Device.DEFAULT in {"AMD"}, "AMD CI doesn't support multiple sync threads yet")
def test_padto_sum_multireduce(self):
Tensor.manual_seed(0)
N = 17
x = Tensor.rand(N, N).realize()
opts = [
[Opt(OptOps.PADTO, 0, 32)],
[Opt(OptOps.PADTO, 0, 32), Opt(OptOps.UPCAST, 0, 8),],
# TODO: multireduce pads
# causes an issue because the acc won't be masked in the second reduce
# [Opt(OptOps.PADTO, 1, 32), Opt(OptOps.PADTO, 2, 32)]
]
g0, g1 = [UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(), arg=i) for i in range(2)]
x_ld0 = UOp(Ops.LOAD, dtypes.float, src=(g1.view(x.uop.st.reshape((1, N, N)).expand((N,N,N))),))
x_ld1 = UOp(Ops.LOAD, dtypes.float, src=(g1.view(x.uop.st.reshape((N, 1, N))),))
r0 = UOp(Ops.REDUCE_AXIS, dtypes.float, (x_ld0,), (Ops.ADD, (1,)))
r1 = UOp(Ops.REDUCE_AXIS, dtypes.float, (x_ld1+r0*ast_const(dtypes.float, -1, (N, 1, N)),),(Ops.ADD, (0,)))
store = UOp(Ops.STORE, src=(g0.view(ShapeTracker.from_shape((1,1,N))), r1))
sink = UOp(Ops.SINK, src=(store,))
helper_linearizer_ast(sink, [x], wanna_output=[(x.numpy()-x.numpy().sum(axis=0, keepdims=True)).sum(axis=0).reshape(1,1,N)], opts=opts)
x_ld0 = UOp(Ops.LOAD, dtypes.float, src=(g1.view(x.uop.st.reshape((N, 1, N)).expand((N,N,N))),))
x_ld1 = UOp(Ops.LOAD, dtypes.float, src=(g1.view(x.uop.st.reshape((N, N, 1))),))
r0 = UOp(Ops.REDUCE_AXIS, dtypes.float, (x_ld0,), (Ops.ADD, (2,)))
r1 = UOp(Ops.REDUCE_AXIS, dtypes.float, (x_ld1+r0*ast_const(dtypes.float, -1, (N, N, 1)),), (Ops.ADD, (1,)))
store = UOp(Ops.STORE, src=(g0.view(ShapeTracker.from_shape((N,1,1))), r1))
sink = UOp(Ops.SINK, src=(store,))
helper_linearizer_ast(sink, [x], wanna_output=[(x.numpy()-x.numpy().sum(axis=1, keepdims=True)).sum(axis=1).reshape(N,1,1)], opts=opts)
@unittest.skipIf(CI and Device.DEFAULT in {"AMD"}, "AMD CI doesn't support multiple sync threads yet")
def test_padto_max_multireduce(self):
Tensor.manual_seed(0)
N = 17
x = Tensor.rand(N, N).realize()
opts = [
[Opt(OptOps.PADTO, 0, 32)],
[Opt(OptOps.PADTO, 0, 32), Opt(OptOps.UPCAST, 0, 8),]
]
g0, g1 = [UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(), arg=i) for i in range(2)]
x_ld0 = UOp(Ops.LOAD, dtypes.float, src=(g1.view(x.uop.st.reshape((1, N, N)).expand((N,N,N))),))
x_ld1 = UOp(Ops.LOAD, dtypes.float, src=(g1.view(x.uop.st.reshape((N, 1, N))),))
r0 = UOp(Ops.REDUCE_AXIS, dtypes.float, (x_ld0,), (Ops.MAX, (1,)))
r1 = UOp(Ops.REDUCE_AXIS, dtypes.float, (x_ld1+r0*ast_const(dtypes.float, -1, (N, 1, N)),), (Ops.MAX, (0,)))
store = UOp(Ops.STORE, src=(g0.view(ShapeTracker.from_shape((1,1,N))), r1))
sink = UOp(Ops.SINK, src=(store,))
helper_linearizer_ast(sink, [x], wanna_output=[(x.numpy()-x.numpy().max(axis=0, keepdims=True)).max(axis=0).reshape(1,1,N)], opts=opts)
x_ld0 = UOp(Ops.LOAD, dtypes.float, src=(g1.view(x.uop.st.reshape((N, 1, N)).expand((N,N,N))),))
x_ld1 = UOp(Ops.LOAD, dtypes.float, src=(g1.view(x.uop.st.reshape((N, N, 1))),))
r0 = UOp(Ops.REDUCE_AXIS, dtypes.float, (x_ld0,), (Ops.MAX, (2,)))
r1 = UOp(Ops.REDUCE_AXIS, dtypes.float, (x_ld1+r0*ast_const(dtypes.float, -1, (N, N, 1)),), (Ops.MAX, (1,)))
store = UOp(Ops.STORE, src=(g0.view(ShapeTracker.from_shape((N,1,1))), r1))
sink = UOp(Ops.SINK, src=(store,))
helper_linearizer_ast(sink, [x], wanna_output=[(x.numpy()-x.numpy().max(axis=1, keepdims=True)).max(axis=1).reshape(N,1,1)], opts=opts)
@unittest.skipIf(CI and Device.DEFAULT in {"AMD"}, "AMD CI doesn't support multiple sync threads yet")
def test_padto_where_multireduce(self):
# ternary operators try to use both ridxs
# we need to make sure the ternary operators nest properly
N = 17
x = Tensor.rand(N, N).realize()
a = Tensor.rand(1, 1).realize()
b = Tensor.rand(1, 1).realize()
opts = [[Opt(OptOps.PADTO, 0, 32)],[Opt(OptOps.PADTO, 0, 32), Opt(OptOps.UPCAST, 0, 8),],]
wanna_output = np.where(0.5*17 < (x.numpy()+np.where(0.75*17 < x.numpy().sum(axis=1,keepdims=True), a.numpy(), b.numpy())).sum(axis=1),0.0,1.0).reshape((N,1,1)) # noqa: E501
ld0 = x.uop.st.reshape((N, 1, N)).expand((N,N,N))
ld1 = x.uop.st.reshape((N, N, 1))
ast = UOp(Ops.SINK, src=(
UOp(Ops.STORE, src=(
UOp(Ops.VIEW, dtypes.float.ptr(), arg=ShapeTracker(views=(View(shape=(N, 1, 1), strides=(1, 0, 0), offset=0, mask=None, contiguous=True),)), src=( # noqa: E501
UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(), arg=0),)),
UOp(Ops.WHERE, dtypes.float, arg=None, src=(
UOp(Ops.CMPLT, dtypes.bool, arg=None, src=(
ast_const(dtypes.float, 0.5*N, (N, 1, 1)),
UOp(Ops.REDUCE_AXIS, dtypes.float, arg=(Ops.ADD, (1,)), src=(
UOp(Ops.ADD, dtypes.float, arg=None, src=(
UOp(Ops.LOAD, dtypes.float, src=(
UOp(Ops.VIEW, dtypes.float.ptr(), arg=ld1, src=(
UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(), arg=1),)),)),
UOp(Ops.WHERE, dtypes.float, arg=None, src=(
UOp(Ops.CMPLT, dtypes.bool, arg=None, src=(
ast_const(dtypes.float, 0.75*N, (N, N, 1)),
UOp(Ops.REDUCE_AXIS, dtypes.float, arg=(Ops.ADD, (2,)), src=(
UOp(Ops.LOAD, dtypes.float, src=(
UOp(Ops.VIEW, dtypes.float.ptr(), arg=ld0, src=(
UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(), arg=1),)),)),)),)),
UOp(Ops.LOAD, dtypes.float, src=(
UOp(Ops.VIEW, dtypes.float.ptr(), arg=ShapeTracker(views=(View(shape=(N, N, 1), strides=(0, 0, 0), offset=0, mask=None, contiguous=False),)), src=( # noqa: E501
UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(), arg=2),)),)),
UOp(Ops.LOAD, dtypes.float, src=(
UOp(Ops.VIEW, dtypes.float.ptr(), arg=ShapeTracker(views=(View(shape=(N, N, 1), strides=(0, 0, 0), offset=0, mask=None, contiguous=False),)), src=( # noqa: E501
UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(), arg=3),)),)),)),)),)),)),
ast_const(dtypes.float, 0.0, (N, 1, 1)),
ast_const(dtypes.float, 1.0, (N, 1, 1)),)),)),))
helper_linearizer_ast(ast, [x,a,b], opts=opts, wanna_output=[wanna_output])
ld0 = x.uop.st.reshape((1, N, N)).expand((N,N,N))
ld1 = x.uop.st.reshape((N, 1, N))
wanna_output = np.where(0.5*17 < (x.numpy()+np.where(0.75*17 < x.numpy().sum(axis=0,keepdims=True), a.numpy(), b.numpy())).sum(axis=0),0.0,1.0).reshape(1,1,N) # noqa: E501
ast = UOp(Ops.SINK, src=(
UOp(Ops.STORE, src=(
UOp(Ops.VIEW, dtypes.float.ptr(), arg=ShapeTracker(views=(View(shape=(1, 1, N), strides=(0, 0, 1), offset=0, mask=None, contiguous=True),)), src=( # noqa: E501
UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(), arg=0, src=()),)),
UOp(Ops.WHERE, dtypes.float, arg=None, src=(
UOp(Ops.CMPLT, dtypes.bool, arg=None, src=(
ast_const(dtypes.float, 0.5*N, (1, 1, N)),
UOp(Ops.REDUCE_AXIS, dtypes.float, arg=(Ops.ADD, (0,)), src=(
UOp(Ops.ADD, dtypes.float, arg=None, src=(
UOp(Ops.LOAD, dtypes.float, src=(
UOp(Ops.VIEW, dtypes.float.ptr(), arg=ld1, src=(
UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(), arg=1, src=()),)),)),
UOp(Ops.WHERE, dtypes.float, arg=None, src=(
UOp(Ops.CMPLT, dtypes.bool, arg=None, src=(
ast_const(dtypes.float, 0.75*N, (N, 1, N)),
UOp(Ops.REDUCE_AXIS, dtypes.float, arg=(Ops.ADD, (1,)), src=(
UOp(Ops.LOAD, dtypes.float, src=(
UOp(Ops.VIEW, dtypes.float.ptr(), arg=ld0, src=(
UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(), arg=1, src=()),)),)),)),)),
UOp(Ops.LOAD, dtypes.float, src=(
UOp(Ops.VIEW, dtypes.float.ptr(), arg=ShapeTracker(views=(View(shape=(N, 1, N), strides=(0, 0, 0), offset=0, mask=None, contiguous=False),)), src=( # noqa: E501
UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(), arg=2, src=()),)),)),
UOp(Ops.LOAD, dtypes.float, src=(
UOp(Ops.VIEW, dtypes.float.ptr(), arg=ShapeTracker(views=(View(shape=(N, 1, N), strides=(0, 0, 0), offset=0, mask=None, contiguous=False),)), src=( # noqa: E501
UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(), arg=3, src=()),)),)),)),)),)),)),
ast_const(dtypes.float, 0.0, (1, 1, N)),
ast_const(dtypes.float, 1.0, (1, 1, N)),)),)),))
helper_linearizer_ast(ast, [x,a,b], opts=opts, wanna_output=[wanna_output])
# pad reduce axis
helper_linearizer_ast(ast, [x,a,b], opts=[[Opt(OptOps.PADTO, 1, 32)],], wanna_output=[wanna_output])
ld0 = x.uop.st.reshape((1,1,N,N)).expand((N,N,N,N))
ld1 = x.uop.st.reshape((N,N,1,1))
wanna_output = np.where(0.5*17 < (x.numpy()+np.where(0.75*17 < x.numpy().sum(keepdims=True), a.numpy(), b.numpy())).sum(keepdims=True),0.0,1.0).reshape((1,1,1,1))# noqa: E501
ast = UOp(Ops.SINK, src=(
UOp(Ops.STORE, src=(
UOp(Ops.VIEW, dtypes.float.ptr(), arg=ShapeTracker(views=(View(shape=(1, 1, 1, 1), strides=(0, 0, 0, 0), offset=0, mask=None, contiguous=True),)), src=( # noqa: E501
UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(), arg=0, src=()),)),
UOp(Ops.WHERE, dtypes.float, arg=None, src=(
UOp(Ops.CMPLT, dtypes.bool, arg=None, src=(
ast_const(dtypes.float, 0.5*N, (1, 1, 1, 1)),
UOp(Ops.REDUCE_AXIS, dtypes.float, arg=(Ops.ADD, (0, 1)), src=(
UOp(Ops.ADD, dtypes.float, arg=None, src=(
UOp(Ops.LOAD, dtypes.float, src=(
UOp(Ops.VIEW, dtypes.float.ptr(), arg=ShapeTracker(views=(View(shape=(N, N, 1, 1), strides=(N, 1, 0, 0), offset=0, mask=None, contiguous=True),)), src=( # noqa: E501
UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(), arg=1),)),)),
UOp(Ops.WHERE, dtypes.float, arg=None, src=(
UOp(Ops.CMPLT, dtypes.bool, arg=None, src=(
ast_const(dtypes.float, 0.75*N, (N, N, 1, 1)),
UOp(Ops.REDUCE_AXIS, dtypes.float, arg=(Ops.ADD, (2, 3)), src=(
UOp(Ops.LOAD, dtypes.float, src=(
UOp(Ops.VIEW, dtypes.float.ptr(), arg=ShapeTracker(views=(View(shape=(N, N, N, N), strides=(0, 0, N, 1), offset=0, mask=None, contiguous=False),)), src=( # noqa: E501
UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(), arg=1),)),)),)),)),
UOp(Ops.LOAD, dtypes.float, src=(
UOp(Ops.VIEW, dtypes.float.ptr(), arg=ShapeTracker(views=(View(shape=(N, N, 1, 1), strides=(0, 0, 0, 0), offset=0, mask=None, contiguous=False),)), src=( # noqa: E501
UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(), arg=2),)),)),
UOp(Ops.LOAD, dtypes.float, src=(
UOp(Ops.VIEW, dtypes.float.ptr(), arg=ShapeTracker(views=(View(shape=(N, N, 1, 1), strides=(0, 0, 0, 0), offset=0, mask=None, contiguous=False),)), src=( # noqa: E501
UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(), arg=3),)),)),)),)),)),)),
ast_const(dtypes.float, 0.0, (1, 1, 1, 1)),
ast_const(dtypes.float, 1.0, (1, 1, 1, 1)),)),)),))
helper_linearizer_ast(ast, [x,a,b], opts=[[Opt(OptOps.PADTO, 0, 32)],], wanna_output=[wanna_output])
@unittest.skipUnless(Device[Device.DEFAULT].renderer.has_local, "test requires locals")
@unittest.skipUnless(Device[Device.DEFAULT].renderer.has_shared, "test requires shared")
def test_end_local(self):
g0, g1 = [UOp(Ops.DEFINE_GLOBAL, dtypes.int.ptr(), arg=i) for i in range(2)]
load = UOp(Ops.LOAD, dtypes.int, (g1.view(ShapeTracker.from_shape((32,))),))
reduce = UOp(Ops.REDUCE_AXIS, dtypes.int, (load,), (Ops.ADD, (0,)))
store = UOp(Ops.STORE, src=(g0.view(ShapeTracker.from_shape((1,))), reduce))
sink = UOp(Ops.SINK, src=(store,))
load_t = Tensor.full(load.st_arg.shape, 1).contiguous().realize()
k = helper_linearizer_ast(sink, [load_t], wanna_output=[load_t.numpy().sum()])[1]
self.assertEqual(k.uops[-2].op, Ops.ENDIF)
self.assertEqual(k.uops[-1].op, Ops.SINK)
self.assertLess(k.uops.index([x for x in k.uops if x.op is Ops.STORE][-1]), k.uops.index(k.uops[-1]))
def test_two_nested_range(self):
a = Tensor.randn(2, ).realize()
out = a.reshape(2, 1).expand(2, 3).sum()
lin = helper_linearizer_opt(out, wanna_output=[np.broadcast_to(a.numpy().reshape(2, 1), (2, 3)).sum()])[0]
ranges = [i for i,u in enumerate(lin.uops) if u.op is Ops.RANGE]
assert len(ranges) == 1 # NOTE: it collapses now
# RANGE -> LOAD -> RANGE -> ASSIGN
#assert any(x.op is Ops.LOAD for x in lin.uops[ranges[0]:ranges[1]])
def test_three_nested_range(self):
a = Tensor.randn(2, ).realize()
out = a.reshape(2, 1).expand(2, 3).expand(2, 2, 3).sum()
lin = helper_linearizer_opt(out, wanna_output=[np.broadcast_to(np.broadcast_to(a.numpy().reshape(2, 1), (2, 3)), (2, 2, 3)).sum()])[0]
ranges = [i for i,u in enumerate(lin.uops) if u.op is Ops.RANGE]
assert len(ranges) == 1 # NOTE: it collapses now
# RANGE -> RANGE -> LOAD -> RANGE -> ASSIGN
# NOTE: nothing should toposort between the first two ranges
#assert ranges[0]+1 == ranges[1]
#assert any(x.op is Ops.LOAD for x in lin.uops[ranges[1]:ranges[2]])
def test_two_nested_range_alt_indexing(self):
a = Tensor([2, 2]).realize()
out = a.reshape(2, 1).pad(((1, 1), (1, 1)), value=2).sum()
lin = helper_linearizer_opt(out, wanna_output=[24])[0]
ranges = [i for i,u in enumerate(lin.uops) if u.op is Ops.RANGE]
# RANGE -> ALU -> RANGE -> ALU + LOAD -> ASSIGN
assert any(x.op in GroupOp.ALU for x in lin.uops[ranges[0]:ranges[1]])
assert not any(x.op is Ops.LOAD for x in lin.uops[ranges[0]:ranges[1]])
assert any(x.op in {*GroupOp.ALU, Ops.LOAD} for x in lin.uops[ranges[1]:])
def test_range_outer_op_before_phi(self):
a = Tensor.randn(4, 1).realize()
b = Tensor.randn(1, 1).realize()
out = (a + b[0]).sum() + b[0]
lin = helper_linearizer_opt(out, wanna_output=[(a.numpy()+b.numpy()[0]).sum()+b.numpy()])[0]
ranges = [i for i,u in enumerate(lin.uops) if u.op is Ops.RANGE]
# LOAD -> RANGE -> LOAD -> ASSIGN
assert len([x for x in lin.uops[:ranges[0]] if x.op is Ops.LOAD]) == 1
def test_range_outer_op_before_phi_nested_range(self):
a = Tensor.randn(2, ).realize()
b = Tensor.randn(1, 1).realize()
out = (a.reshape(2, 1).expand(2, 3) + b[0]).sum() + b[0]
lin = helper_linearizer_opt(out, wanna_output=[(np.broadcast_to(a.numpy().reshape(2, 1), (2, 3)) + b.numpy()[0]).sum() + b.numpy()])[0]
ranges = [i for i,u in enumerate(lin.uops) if u.op is Ops.RANGE]
assert len(ranges) == 1 # NOTE: it collapses now
#if getenv("PTX"):
# LOAD -> RANGE -> CAST -> ALU -> ALU -> LOAD -> ALU -> RANGE -> ALU -> ASSIGN
# assert lin.uops[ranges[0]-2].op is Ops.LOAD
# assert ranges[1] == ranges[0]+6
# assert [x.op for x in lin.uops[ranges[1]-2:ranges[1]]] == [Ops.LOAD, Ops.ALU]
# LOAD -> RANGE -> LOAD -> ALU -> RANGE -> ASSIGN
#else:
# assert lin.uops[ranges[0]-2].op is Ops.LOAD
# assert ranges[1] == ranges[0]+3
# assert [x.op for x in lin.uops[ranges[1]-2:ranges[1]]] == [Ops.LOAD, Ops.ALU]
def test_range_outer_op_after_phi(self):
a = Tensor.randn(4, 1).realize()
out = a.sum() * a.sum()
lin = helper_linearizer_opt(out, wanna_output=[a.numpy().sum()*a.numpy().sum()])[0]
# RANGE -> LOAD -> ASSIGN -> ALU
end = max(i for i,u in enumerate(lin.uops) if u.op is Ops.ENDRANGE)
# the INDEX can be first
assert lin.uops[end+1].op in GroupOp.ALU or lin.uops[end+2].op in GroupOp.ALU
def test_range_outer_op_after_phi_nested_range(self):
a = Tensor.randn(2, ).realize()
out = a.reshape(2, 1).expand(2, 3).sum() + a.reshape(2, 1).expand(2, 3).sum()
lin = helper_linearizer_opt(out, wanna_output=[(np.broadcast_to(a.numpy().reshape(2, 1), (2, 3))).sum()*2])[0]
# RANGE -> LOAD -> ASSIGN -> ALU
end = max(i for i,u in enumerate(lin.uops) if u.op is Ops.ENDRANGE)
# the INDEX can be first
assert lin.uops[end+1].op in GroupOp.ALU or lin.uops[end+2].op in GroupOp.ALU
def test_load_dedup(self):
# for different leaves in the AST, the same loads may occur.
a = Tensor.randn(4).realize()
# these are of size 3 to avoid float4 coalesce
r = a[:-1] + a[1:]
k = Kernel(r.schedule()[-1].ast)
k.upcast()
k.linearize()
num_loads = len([uop for uop in k.uops if uop.op is Ops.LOAD])
assert num_loads <= 4, "more load uops than needed"
assert num_loads >= 4, "unexpected number of uops, maybe this test needs updating?"
@unittest.skipIf(getenv("PTX"), "broken on ptx for some reason")
def test_load_cache_const_bufs(self):
# make sure const buffers are differentiated from local and mem buffers
ST, DT = ShapeTracker(views=(View(shape=((1,)), strides=(0, 0), offset=0, mask=None, contiguous=False),)).to_uop(), dtypes.int
VAL = ast_const(DT, 2, ST.arg.shape)
g0, g1 = [UOp(Ops.DEFINE_GLOBAL, DT.ptr(), arg=i) for i in range(2)]
# data1[0] + VAL
a = UOp(Ops.LOAD, DT, (g1.view(ST.arg),)) + VAL
# (literal const 1) + VAL
b = ast_const(DT, 1, ST.arg.shape) + VAL
store = UOp(Ops.STORE, src=(g0.view(ST.arg), (a+b)))
sink = UOp(Ops.SINK, src=(store,))
lin = Kernel(sink)
lin.linearize()
assert len(lin.uops) <= 10, "too many uops"
def test_upcast_cse(self):
# when upcasting, within a subtree, there may be common expressions.
a, b = Tensor.randn(1).realize(), Tensor.randn(1).realize()
r = a.expand([2]) + b.expand([2])
k = Kernel(r.schedule()[-1].ast)
k.upcast()
k.linearize()
num_ops = len([uop for uop in k.uops if uop.op in GroupOp.ALU])
assert num_ops <= 1, "more alu uops than needed"
@unittest.skipUnless(Device[Device.DEFAULT].renderer.supports_float4, "test requires float4")
def test_reduce_upcast(self):
x, w = Tensor.randn((1,1,3)).realize(), Tensor.randn((1,1,2)).realize()
r = Tensor.conv2d(x,w,padding=1).relu()
k = Kernel(r.schedule()[-1].ast)
k.upcast()
k.upcast()
k.linearize()
accs = [u for u in k.uops if u.op is Ops.DEFINE_ACC]
stores = [u for u in k.uops if u.op is Ops.STORE]
assert len(accs) == 0 # it's removed now
assert len(stores) == 1
assert stores[0].src[-1].dtype == dtypes.float.vec(4)
# NOTE: can reenable, it does work. it just makes BEAM slow
@unittest.expectedFailure
@unittest.skipUnless(Device.DEFAULT == "CPU", "test only for CPU")
def test_upcast_with_locals_cpu(self):
out = Tensor.ones(64,64).contiguous() @ Tensor.ones(64,64).contiguous()
k = Kernel(out.schedule()[-1].ast)
k.apply_opt(Opt(OptOps.LOCAL, axis=0, arg=4))
prg = k.to_program()
self.assertEqual(len(prg.src.split("for")), 5)
@unittest.skipUnless(Device[Device.DEFAULT].renderer.has_local, "test requires locals")
@unittest.skipUnless(Device[Device.DEFAULT].renderer.has_shared, "test requires shared")
@unittest.skipUnless(Device[Device.DEFAULT].renderer.supports_float4, "test requires float4")
@unittest.skipIf(getenv("PTX"), "broken on ptx for some reason")
def test_upcast_with_locals(self):
x, y = Tensor.rand(1,128), Tensor.rand(128, 128)
r = (x@y).relu()
k = Kernel(r.schedule()[-1].ast)
k.apply_opts([Opt(op=OptOps.GROUP, axis=0, arg=8), Opt(op=OptOps.LOCAL, axis=0, arg=4), Opt(op=OptOps.UPCAST, axis=0, arg=4)])
k.linearize()
stores = [u for u in k.uops if u.op is Ops.STORE]
# the first store is to lds and can be upcasted
assert stores[0].src[-1].dtype == dtypes.float.vec(4)
assert any(x.op is Ops.DEFINE_LOCAL for x in stores[0].toposort())
# the second store is to gds with no upcasts
assert stores[1].src[-1].dtype == dtypes.float
assert any(x.op is Ops.DEFINE_GLOBAL for x in stores[1].toposort())
def test_zero_fold(self):
a, b = Tensor.randn(1).realize(), Tensor.randn(1).realize()
r = Tensor.stack(a, b)
k = Kernel(r.schedule()[-1].ast)
k.upcast()
k.linearize()
num_ops = len([uop for uop in k.uops if uop.op in GroupOp.ALU])
assert num_ops == 0, "more alu uops than needed"
def test_sum_acc_dtype(self):
for tensor_dtype, acc_dtype in (
(dtypes.bool, dtypes.int), (dtypes.int16, dtypes.int), (dtypes.float16, dtypes.float), (dtypes.bfloat16, dtypes.float)):
if is_dtype_supported(tensor_dtype) and is_dtype_supported(acc_dtype):
a = Tensor([1, 2, 3], dtype=tensor_dtype).sum()
k = Kernel(a.schedule()[-1].ast)
k.linearize()
local = [uop for uop in k.uops if uop.op is Ops.DEFINE_ACC]
assert local[0].dtype == acc_dtype
def test_arg_acc_dtype(self):
def helper_arg_acc_dtype(c: Tensor, expected_dtype:DType):
k = Kernel(c.schedule()[-1].ast)
k.linearize()
local = [uop for uop in k.uops if uop.op is Ops.DEFINE_ACC]
assert local[0].dtype == expected_dtype
tests = (
(dtypes.float16, None, dtypes.float),
(dtypes.bfloat16, None, dtypes.float),
(dtypes.float, None, dtypes.float),
(dtypes.float16, dtypes.float16, dtypes.float16),
(dtypes.bfloat16, dtypes.bfloat16, dtypes.bfloat16),
(dtypes.float, dtypes.float16, dtypes.float16),
)
for tensor_dtype, acc_dtype, expected_dtype in tests:
if is_dtype_supported(tensor_dtype) and is_dtype_supported(acc_dtype) and is_dtype_supported(expected_dtype):
a, b = Tensor.rand(8, 8, dtype=tensor_dtype), Tensor.rand(8, 8, dtype=tensor_dtype)
helper_arg_acc_dtype(a.sum(dtype=acc_dtype), expected_dtype)
helper_arg_acc_dtype(a.matmul(b, dtype=acc_dtype), expected_dtype)
helper_arg_acc_dtype(Tensor.einsum("ki,ij->kj", a, b, dtype=acc_dtype), expected_dtype)
d, w = Tensor.rand(4, 8, 8, 8, dtype=tensor_dtype), Tensor.rand(8, 8, 2, 2, dtype=tensor_dtype)
helper_arg_acc_dtype(d.conv2d(w, dtype=acc_dtype), expected_dtype)
# TODO: don't skip bf16 for real device (METAL, AMD)
@unittest.skipUnless(Device[Device.DEFAULT].renderer.tensor_cores, "test requires tensor cores")
def test_tensor_cores(self):
for tc in Device[Device.DEFAULT].renderer.tensor_cores:
if not is_dtype_supported(tc.dtype_in) or not is_dtype_supported(tc.dtype_out): continue
# for AMX, tc.dims[2] == 1 so reduceop is None thus tensor_cores are not triggered
helper_tc_allclose(tc.dims[0], tc.dims[1], 2 if AMX else tc.dims[2], tc.dtype_in, tc.dtype_out, axis=0, tc_opt=0)
@unittest.skipUnless(Device[Device.DEFAULT].renderer.tensor_cores, "test requires tensor cores")
def test_tensor_cores_emulation(self):
for tc in Device[Device.DEFAULT].renderer.tensor_cores:
if not is_dtype_supported(tc.dtype_in) or not is_dtype_supported(tc.dtype_out): continue
# for AMX, tc.dims[2] == 1 so reduceop is None thus tensor_cores are not triggered
helper_tc_allclose(tc.dims[0], tc.dims[1], 2 if AMX else tc.dims[2], tc.dtype_in, tc.dtype_out, axis=0, tc_opt=0, use_tensor_cores=3)
@unittest.skipUnless(Device[Device.DEFAULT].renderer.tensor_cores, "test requires tensor cores")
def test_tensor_cores_codegen(self):
for tc in Device[Device.DEFAULT].renderer.tensor_cores:
if not is_dtype_supported(tc.dtype_in) or not is_dtype_supported(tc.dtype_out): continue
n, m, k = tc.dims[0], tc.dims[1], 2 if AMX else tc.dims[2]
a, b = Tensor.rand(m, k, dtype=tc.dtype_in), Tensor.rand(k, n, dtype=tc.dtype_in)
r = a.matmul(b, dtype=tc.dtype_out)
sched = r.schedule()
realized_ast = sched[-1].ast
kernel = Kernel(realized_ast)
kernel.apply_tensor_cores(1, axis=0, tc_select=-1, tc_opt=2)
kernel.linearize()
prg = kernel.to_program()
if Device.DEFAULT == "LLVM":
assert "0x201000" in prg.src
elif Device.DEFAULT == "AMD" and getenv("AMD_LLVM", 0):
assert "@llvm.amdgcn.wmma" in prg.src
elif Device[Device.DEFAULT].renderer.suffix == "PTX":
assert "mma.sync.aligned" in prg.src
else:
assert "__WMMA_" in prg.src
@unittest.skipIf(Device.DEFAULT in ("AMD", "AMD_LLVM") or (Device.DEFAULT == "PYTHON" and getenv("EMULATE_AMD")), "broken for AMD")
@unittest.skipUnless(Device[Device.DEFAULT].renderer.tensor_cores, "test requires tensor cores")
def test_tensor_cores_padded(self):
for tc in Device[Device.DEFAULT].renderer.tensor_cores:
if not is_dtype_supported(tc.dtype_in) or not is_dtype_supported(tc.dtype_out): continue
helper_tc_allclose(tc.dims[0]+(pad:=1), tc.dims[1]+pad, tc.dims[2]+pad, tc.dtype_in, tc.dtype_out, tc_opt=2)
# AMD compiler bug: AMD miscompiles non-zero padded tc kernels with -O3, producing wrong results, nans or hang (see #9606)
# Internal bug: zero-stride dimensions combined with a mask may produce wrong index/valid for pad == 1 on AMD
@unittest.skipUnless(Device.DEFAULT in ("AMD", "AMD_LLVM") or (Device.DEFAULT == "PYTHON" and getenv("EMULATE_AMD")), "test for AMD's tc")
@unittest.skipUnless(Device[Device.DEFAULT].renderer.tensor_cores, "test requires tensor cores")
@unittest.expectedFailure
def test_tensor_cores_padded_amd(self):
for tc in Device[Device.DEFAULT].renderer.tensor_cores:
if not is_dtype_supported(tc.dtype_in) or not is_dtype_supported(tc.dtype_out): continue
helper_tc_allclose(tc.dims[0]+(pad:=1), tc.dims[1]+pad, tc.dims[2]+pad, tc.dtype_in, tc.dtype_out, tc_opt=2)
@unittest.skipUnless(Device[Device.DEFAULT].renderer.tensor_cores, "test requires tensor cores")
def test_tensor_cores_padded_uops(self):
for tc in Device[Device.DEFAULT].renderer.tensor_cores:
pad = 1
# check that TC is triggered for TC_OPT=2
helper_tc_ensure_uops_and_opts_count(tc.dims[0]+pad, tc.dims[1]+pad, tc.dims[2]+pad,
tc.dtype_in, tc.dtype_out, tc_opt=2, ensure_triggered=True)
# check that TC is not triggered for TC_OPT<2
helper_tc_ensure_uops_and_opts_count(tc.dims[0]+pad, tc.dims[1]+pad, tc.dims[2]+pad,
tc.dtype_in, tc.dtype_out, tc_opt=1, ensure_triggered=False)
helper_tc_ensure_uops_and_opts_count(tc.dims[0]+pad, tc.dims[1]+pad, tc.dims[2]+pad,
tc.dtype_in, tc.dtype_out, tc_opt=0, ensure_triggered=False)
# check excessive padding doesn't trigger padded TC in TC_OPT=2
helper_tc_ensure_uops_and_opts_count(tc.dims[0]//4, tc.dims[1], tc.dims[2], tc.dtype_in, tc.dtype_out, tc_opt=2, ensure_triggered=False)
helper_tc_ensure_uops_and_opts_count(tc.dims[0], tc.dims[1]//4, tc.dims[2], tc.dtype_in, tc.dtype_out, tc_opt=2, ensure_triggered=False)
if not AMX: # AMX tc.dims[2] == 1
helper_tc_ensure_uops_and_opts_count(tc.dims[0], tc.dims[1], tc.dims[2]//4, tc.dtype_in, tc.dtype_out, tc_opt=2, ensure_triggered=False)
@unittest.skipIf(CI and Device.DEFAULT in {"AMD"}, "AMD CI is really slow here")
@unittest.skipUnless(Device[Device.DEFAULT].renderer.tensor_cores, "test requires tensor cores")
def test_tensor_cores_multi_reduce(self):
for tc in Device[Device.DEFAULT].renderer.tensor_cores:
if not is_dtype_supported(tc.dtype_in) or not is_dtype_supported(tc.dtype_out): continue
# this will be a M=G16, N=G32, M=G16, M=G16, K=R16, K=R16, K=R16 with 9 choices of TC MNK axes
golden_result = None
for axis in range(9):
a = Tensor.rand(16, 16, 29, 29, dtype=tc.dtype_in).realize()
b = Tensor.rand(32, 16, 16, 16, dtype=tc.dtype_in).realize()
c = a.conv2d(b, padding=1, dtype=tc.dtype_out)
realized_ast, real_bufs = helper_realized_ast(c)
k = Kernel(realized_ast)
k.apply_tensor_cores(1, axis=axis, tc_opt=2)
k.linearize()
assert len([uop for uop in k.uops if uop.op is Ops.WMMA]) > 0, "tensor core not triggered"
assert len([x for x in k.applied_opts if x.op is OptOps.TC]) == 1, "tensor core opt not included"
prg = CompiledRunner(k.to_program())
# TODO: support this even if numpy doesn't
if _to_np_dtype(real_bufs[0].dtype) is None: continue
real_bufs[0].copyin(np.zeros((real_bufs[0].size, ), dtype=_to_np_dtype(real_bufs[0].dtype)).data) # Zero to check that all values are filled
prg.exec(real_bufs)
result = np.frombuffer(real_bufs[0].as_buffer(), _to_np_dtype(real_bufs[0].dtype))
# ensure the results for each choice of axis matches
if golden_result is None: golden_result = np.frombuffer(real_bufs[0].as_buffer(), _to_np_dtype(real_bufs[0].dtype))
np.testing.assert_allclose(result, golden_result, atol=0.1, rtol=0.2)
# check that get_kernel_actions produces all 9 options
from tinygrad.opt.search import get_kernel_actions
tc_actions = [k for i, k in get_kernel_actions(Kernel(realized_ast), False).items() if k.applied_opts[0].op == OptOps.TC]
available_tc = len([x for x in Device[Device.DEFAULT].renderer.tensor_cores if x.dtype_in == tc.dtype_in and x.dtype_out == tc.dtype_out])
assert len(tc_actions) == 9 * available_tc, f"should contain 9 possible TC actions for every available TC, got {len(tc_actions)}"
@unittest.skipUnless(Device[Device.DEFAULT].renderer.tensor_cores, "test requires tensor cores")
def test_tensor_cores_unroll_phi(self):
tc = Device[Device.DEFAULT].renderer.tensor_cores[0]
x, y = Tensor.rand(128, 128, dtype=tc.dtype_in), Tensor.rand(128, 128, dtype=tc.dtype_in)
r = x.matmul(y, dtype=tc.dtype_out)
k = helper_linearizer_opt(r, [[Opt(OptOps.UNROLL, 0, 4)]], apply_tc=True, atol=3e-2, rtol=1e-3)[-1]
for u in k.uops:
if u.op is Ops.WMMA:
assert u.src[-1].src[0].op != Ops.ASSIGN
@unittest.skipUnless(Device[Device.DEFAULT].renderer.tensor_cores, "test requires tensor cores")
@unittest.skipIf(Device.DEFAULT in {"CPU", "LLVM"}, "CPU does not support using a different type for accumulation")
def test_tensor_cores_unroll_casted_phi(self):
tc = [tc for tc in Device[Device.DEFAULT].renderer.tensor_cores if tc.dtype_in != tc.dtype_out][0]
x, y = Tensor.rand(128, 128, dtype=tc.dtype_in), Tensor.rand(128, 128, dtype=tc.dtype_in)
r = x.matmul(y, dtype=tc.dtype_out)
k = helper_linearizer_opt(r, [[Opt(OptOps.UNROLL, 0, 4)]], apply_tc=True, atol=3e-2, rtol=1e-3)[-1]
for u in k.uops:
if u.op is Ops.WMMA:
#assert u.src[-1].dtype == dtypes.float.vec(prod(tc.thread_local_sizes[2]))
assert u.src[-1].src[0].op != Ops.ASSIGN
@unittest.skipUnless(Device[Device.DEFAULT].renderer.tensor_cores, "test requires tensor cores")
@unittest.skipIf(Device.DEFAULT in {"CPU", "LLVM"}, "CPU does not support using a different type for accumulation")
def test_tensor_cores_unroll_casted_phi_with_children(self):
# all ASSIGN children are outside the loop
tc = [tc for tc in Device[Device.DEFAULT].renderer.tensor_cores if tc.dtype_in != tc.dtype_out][0]
x, y = Tensor.rand(128, 128, dtype=tc.dtype_in), Tensor.rand(128, 128, dtype=tc.dtype_in)
r = x.matmul(y, dtype=tc.dtype_out).relu()
k = helper_linearizer_opt(r, [[Opt(OptOps.UNROLL, 0, 4)]], apply_tc=True, atol=3e-2, rtol=1e-3)[-1]
for u in k.uops:
if u.op is Ops.WMMA:
#assert u.src[-1].dtype == dtypes.float.vec(prod(tc.thread_local_sizes[2]))
assert u.src[-1].src[0].op != Ops.ASSIGN
@unittest.skipUnless(Device[Device.DEFAULT].renderer.supports_float4, "test requires float4")
def test_simple_unroll_no_between_phi_dependencies(self):
x, y = Tensor.rand(128, 128), Tensor.rand(128, 128)
r = (x@y).relu()
k = helper_linearizer_opt(r, [[Opt(OptOps.UNROLL, 0, 4), Opt(OptOps.UPCAST, 0, 4)]])[-1]
# the uops graph is RANGE -> DEFINE_ACC -> 4x ALU -> 4x ASSIGN -> ENDRANGE
for u in k.uops:
if u.op is Ops.ASSIGN:
assert u.src[1].op in GroupOp.ALU
# children of ASSIGN are placed after ENDRANGE
if any(x.op is Ops.ASSIGN for x in u.src):
end_range = [i for i, x in enumerate(k.uops) if x.op is Ops.ENDRANGE][0]
assert end_range < k.uops.index(u)
def test_grouped_dims(self):
def _assert_grouped_dims(prefix, dims, max_sizes, reverse_dims, expected_sizes, assert_same_length = True):
idxs = get_grouped_dims(prefix, dims, max_sizes, reverse_dims)
loop_idxs = dedup(flatten([[y for y in x.toposort() if y.op is Ops.SPECIAL] for x in idxs]))
loop_idxs = sorted(loop_idxs, key=lambda uop: uop.arg[0])
sizes = [x.arg[1] for x in loop_idxs]
assert len(idxs) == len(dims), f"expected idxs to have same length as dims {len(dims)}, got {len(idxs)}"
if assert_same_length:
assert len(loop_idxs) == min(len(sizes), len(dims)), f"expected idxs to have length {min(len(sizes), len(dims))}, got {len(loop_idxs)}"
assert sizes == expected_sizes, f"expected sizes={expected_sizes}, got {sizes=}"
# TODO: add these back after uop symbolic
# for i in range(len(dims)):
# assert idxs[i].max+1 == dims[i], f"idxs[{i}] should have max {dims[i]-1}"
# for i in range(len(loop_idxs)):
# assert loop_idxs[i].expr.startswith(prefix), f"loop_idxs[{i}] must start with {prefix}"
# assert loop_idxs[i].max+1 == sizes[i], f"loop_idxs[{i}] should have max {sizes[i]-1}"
# no-op
_assert_grouped_dims("gidx", (2,), (16,16,16), False, [2])
_assert_grouped_dims("gidx", (2,3), (16,16,16), False, [2,3])
# check reverse dims
_assert_grouped_dims("gidx", (2,3), (16,16,16), True, [3,2])
_assert_grouped_dims("gidx", (2,3,4), (16,16,16), False, [2,3,4])
# test splitting globals: len(dims) == len(max)
_assert_grouped_dims("gidx", (64,3,4), (16,16,16), False, [16,12,4])
_assert_grouped_dims("gidx", (64,3,4), (16,4,16), False, [16,3,16])
_assert_grouped_dims("gidx", (64,3,4), (16,16,16), True, [16,3,16])
_assert_grouped_dims("gidx", (128,3,4), (16,4,256), False, [16,3,32])
_assert_grouped_dims("gidx", (4,4,512), (16,4,256), False, [8,4,256])
# prefer group_dim strategy when possible
_assert_grouped_dims("gidx", (512,4,2), (8192,2,2), False, [2048,2])
# test splitting globals: len(dims) < len(max)
# len(dim) -> len(limited)
# 1 -> 2
_assert_grouped_dims("gidx", (128,), (16,16,256), False, [16,8], False)
# 1 -> 3
_assert_grouped_dims("gidx", (65536,), (16,16,256), False, [16,16,256], False)
# 2 -> 3
_assert_grouped_dims("gidx", (128,128), (16,16,256), False, [16,16,64], False)
# test when the only divisor is the square root of dim
_assert_grouped_dims("gidx", (121,), (12,12,12), False, [11,11], False)
# collapse on onto the left most axis
_assert_grouped_dims("gidx", (2,3,4,5), (16,16,16), False, [6,4,5])
_assert_grouped_dims("gidx", (2,3,4,5), (32,16,16), True, [20,3,2])
# _assert_grouped_dims("gidx", (Variable("start_pos",1,2),3,4,5), (32,16,16), True, [20,3,Variable("start_pos",1,2)])
# collapse on left-most available axis (the left most is too small)
_assert_grouped_dims("gidx", (2,3,4,5), (4,16,16), False, [2,12,5])
_assert_grouped_dims("gidx", (2,3,4,5), (16,16,16), True, [5,12,2])
# _assert_grouped_dims("gidx", (Variable("start_pos",1,2),3,4,5), (16,16,16), False, [Variable("start_pos",1,2)*3,4,5])
# dim too large and not factorable
with self.assertRaises(RuntimeError):
get_grouped_dims("gidx", (23,), (16,16,16), False,)
with self.assertRaises(RuntimeError):
get_grouped_dims("gidx", (128,3,4), (16,2,2), False,)
# too large for sizes
with self.assertRaises(RuntimeError):
get_grouped_dims("gidx", (2,3,4,5,6), (16,16,16))
# # variable too large
# with self.assertRaises(AssertionError):
# get_grouped_dims("gidx", (Variable("start_pos",0,16),3,4), (16,16,16), False,)
@unittest.skipUnless(Device[Device.DEFAULT].renderer.has_local, "test requires locals")
def test_default_global_reversed(self):
# shrink so that the dims do not collapse
t = Tensor.ones(5, 6, 7).contiguous().realize().shrink(((0, 4), (0, 5), (0, 6)))
k = helper_linearizer_opt(t+1)[0]
idxs = dedup([uop for uop in k.uops if uop.op is Ops.SPECIAL])
idxs = sorted(idxs, key=lambda uop: uop.arg[0])
assert idxs[0].arg == ('gidx0', 6), idxs[0].arg
assert idxs[1].arg == ('gidx1', 5), idxs[1].arg
assert idxs[2].arg == ('gidx2', 4), idxs[2].arg
def test_sum_collapse(self):
t = Tensor([2]).reshape(1, 1).expand(256, 256).sum()
sched = [si for si in t.schedule() if si.ast.op is Ops.SINK]
# sum_collapse is a full collapse now
assert len(sched) == 1
assert not any(u.op is Ops.REDUCE_AXIS for u in sched[0].ast.toposort()), "found reduce in sum collapse"
#lin = Kernel(sched[0].ast)
#assert not any(u.op is Ops.RANGE for u in lin.linearize().uops), "found loop in sum collapse"
def test_assign_fold(self):
a = Tensor.ones(4, 4).contiguous().realize()
m = Tensor.ones(4, 4).shrink(((1, 2), None)).pad(((1, 2), None))
a.assign(a+m)
a.realize()
np.testing.assert_equal(a.flatten().numpy(), [1.,1.,1.,1.,2.,2.,2.,2.,1.,1.,1.,1.,1.,1.,1.,1.])
def test_where_fold(self):
a = Tensor.ones(4, 4).contiguous().realize()
b = a.shrink(((1, 2), None)).pad(((1, 2), None))
a.assign(b.where(2, a))
sched = a.schedule()
assert len(sched) == 1
sched_copy = sched[:]
run_schedule(sched)
np.testing.assert_equal(a.flatten().numpy(), [1.,1.,1.,1.,2.,2.,2.,2.,1.,1.,1.,1.,1.,1.,1.,1.])
lin = Kernel(sched_copy[-1].ast)
lin.linearize()
assert not any(u.op == Ops.WHERE for u in lin.uops), "found where where where should be folded"
def test_phi_simplification(self):
def helper(t, max_ops=0):
k = helper_linearizer_opt(t)[-1]
uops = list(k.linearize().uops)
# ignore kernel optimized IF statements for now
if if_op:=next((u for u in uops if u.op is Ops.IF), None):
uops = uops[:uops.index(if_op)]
assert len(set([u.op for u in uops if u.op in {Ops.RANGE, Ops.SPECIAL}])) == 1, "has either specials or ranges, not both"
assert len([u for u in uops if u.op is Ops.ASSIGN]) == 0, "ASSIGN should have been simplified"
# TODO: once uops track min/max this will be fixed
#assert len([u for u in uops if u.op is Ops.MAX]) <= max_ops, "no unnecessary MAX ops"
helper(Tensor.arange(5.5, (3.5*300), 3.5), max_ops=2)
helper(Tensor.arange(-1, -100, -5), max_ops=2)
# NOTE: both of these split the reduce (this just wasn't tracked before)
#helper(Tensor.arange(-3.2, 6.7, 0.64), max_ops=2)
#helper(Tensor.arange(256), max_ops=2)
helper(Tensor.arange(255), max_ops=2)
@unittest.skipUnless(Device[Device.DEFAULT].renderer.supports_float4, "test requires float4")
def test_grouped_store_phis(self):
"""
float4 acc0 = float4(0.0,0.0,0.0,0.0);
{
acc0 = // ...
}
*((device float4*)(data0+alu2)) = float4(acc0.x,acc0.y,acc0.z,acc0.w);
simplifies to:
*((device float4*)(data0+alu2)) = acc0;
"""
x, y = Tensor.randn(64,64), Tensor.randn(64,64)
out = x.matmul(y)
k = helper_linearizer_opt(out)[-1]
# check that the float4 cast collapses
store_vals = [u.src[-1] for u in k.uops if u.op is Ops.STORE]
for val in store_vals:
assert val.dtype == dtypes.float.vec(4) # and val.op is not Ops.VECTORIZE
@unittest.skipUnless(Device[Device.DEFAULT].renderer.has_local, "test requires locals")
@unittest.skipUnless(Device[Device.DEFAULT].renderer.has_shared, "test requires shared")
@unittest.skipUnless(Device[Device.DEFAULT].renderer.supports_float4, "test requires float4")
def test_arange_opts(self):
a = Tensor.arange(128)
helper_linearizer_opt(a, [
[Opt(OptOps.GROUP, 0, 32)],
[Opt(OptOps.GROUPTOP, 0, 32)],
[Opt(op=OptOps.LOCAL, axis=0, arg=8)],
[Opt(op=OptOps.LOCAL, axis=0, arg=8), Opt(op=OptOps.UPCAST, axis=0, arg=0)],
[Opt(op=OptOps.LOCAL, axis=0, arg=8), Opt(op=OptOps.UPCAST, axis=0, arg=0), Opt(op=OptOps.GROUP, axis=0, arg=8)],
[Opt(op=OptOps.LOCAL, axis=0, arg=8), Opt(op=OptOps.UPCAST, axis=0, arg=0), Opt(op=OptOps.GROUP, axis=0, arg=8), Opt(op=OptOps.UNROLL, axis=1, arg=4)], # noqa: E501
])
@unittest.skipUnless(Device[Device.DEFAULT].renderer.supports_float4, "test requires float4")
def test_grouped_store_values(self):
x = Tensor.randn((4,3,6,6)).realize()
out = x.flip((0,1)).contiguous()
k = helper_linearizer_opt(out)[-1]
store_val = [u.src[-1] for u in k.uops if u.op is Ops.STORE][0]
assert store_val.dtype == dtypes.float.vec(4) and store_val.op is not Ops.VECTORIZE
@unittest.skipUnless(Device[Device.DEFAULT].renderer.has_local, "test requires locals")
@unittest.skipUnless(Device[Device.DEFAULT].renderer.has_shared, "test requires shared")
@unittest.skipUnless(Device[Device.DEFAULT].renderer.supports_float4, "test requires float4")
def test_grouped_store_locals_and_globals(self):
x, y = Tensor.rand(128, 128), Tensor.rand(128, 128)
out = x@y
opt = [Opt(OptOps.LOCAL, 0, 4), Opt(OptOps.GROUPTOP, 0, 8),
Opt(OptOps.UNROLL, 0, 4), Opt(OptOps.UPCAST, 0, 4), Opt(OptOps.UPCAST, 1, 2)] # upcast accs in both reduces
k = helper_linearizer_opt(out, opts=[opt])[-1]
def get_recursive(uop): return set.union(set(uop.src), [uop], *[get_recursive(v) for v in uop.src])
local_stores = [u for u in k.uops if u.op is Ops.STORE and any(x.op is Ops.DEFINE_LOCAL for x in get_recursive(u.src[0]))]
global_stores = [u for u in k.uops if u.op is Ops.STORE and any(x.op is Ops.DEFINE_GLOBAL for x in get_recursive(u.src[0]))]
barrier = [u for u in k.uops if u.op is Ops.BARRIER][0]
# check that the float4 cast collapses for all stores
for store in local_stores+global_stores:
assert store.src[-1].dtype.count > 1 # and store.src[2].op is not Ops.VECTORIZE
# # check the children's vins
# TODO: src ALU are not the same, should it?
# assert barrier.src == tuple(local_stores)
assert len([u for u in k.uops if u.op is Ops.IF and u.src[-1] == barrier]) == 1
@unittest.skipUnless(Device[Device.DEFAULT].renderer.has_local, "test requires locals")
@unittest.skipUnless(Device[Device.DEFAULT].renderer.has_shared, "test requires shared")
@unittest.skipUnless(Device[Device.DEFAULT].renderer.supports_float4, "test requires float4")
def test_grouped_store_local_only(self):
x, y = Tensor.rand(1,128), Tensor.rand(128, 128)
r = (x@y).relu()
k = helper_linearizer_opt(r)[-1]
stores = [u for u in k.uops if u.op is Ops.STORE]
# the float4 value stores directly in lds and we skip upcast
self.assertEqual(stores[0].src[-1].dtype, dtypes.float.vec(4))
#assert stores[0].src[-1].op is not Ops.VECTORIZE
# the global store doesn't change
assert stores[1].src[-1].dtype == dtypes.float
@unittest.skipUnless(Device[Device.DEFAULT].renderer.has_local, "test requires locals")
@unittest.skipUnless(Device[Device.DEFAULT].renderer.supports_float4, "test requires float4")
def test_skip_unmatching_upcasts(self):
Tensor.manual_seed(0)
ast = UOp(Ops.SINK, dtypes.void, arg=None, src=(
UOp(Ops.STORE, dtypes.void, arg=None, src=(
UOp(Ops.VIEW, dtypes.float.ptr(9600), arg=ShapeTracker(views=(View(shape=(240, 40, 1, 1), strides=(40, 1, 0, 0), offset=0, mask=None, contiguous=True),)), src=( # noqa: E501
UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(9600), arg=0, src=()),)),
UOp(Ops.LOAD, dtypes.float, arg=None, src=(
UOp(Ops.VIEW, dtypes.float.ptr(9600), arg=ShapeTracker(views=(View(shape=(240, 40, 1, 1), strides=(1, 240, 0, 0), offset=0, mask=None, contiguous=False),)), src=( # noqa: E501
UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(9600), arg=1, src=()),)),)),)),))
opt = [
Opt(op=OptOps.UPCAST, axis=1, arg=4), Opt(op=OptOps.LOCAL, axis=0, arg=16),
Opt(op=OptOps.LOCAL, axis=1, arg=2), Opt(op=OptOps.UPCAST, axis=3, arg=2)
]
k = helper_linearizer_ast(ast, [Tensor.randn(240*40).realize()], opts=[opt])[-1]
out = [u for u in k.uops if u.op is Ops.STORE][0]
assert out.src[-1].op is Ops.VECTORIZE and out.src[-1].dtype == dtypes.float.vec(4)
@unittest.skipUnless(Device[Device.DEFAULT].renderer.has_local, "test requires locals")
@unittest.skipUnless(Device[Device.DEFAULT].renderer.supports_float4, "test requires float4")
def test_skip_unmatching_upcasts_with_gep(self):
Tensor.manual_seed(0)
ast = UOp(Ops.SINK, dtypes.void, arg=None, src=(
UOp(Ops.STORE, dtypes.void, arg=None, src=(
UOp(Ops.VIEW, dtypes.float.ptr(256), arg=ShapeTracker(views=(View(shape=(8, 32, 1, 1), strides=(32, 1, 0, 0), offset=0, mask=None, contiguous=True),)), src=( # noqa: E501
UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(256), arg=0, src=()),)),
UOp(Ops.LOAD, dtypes.float, arg=None, src=(
UOp(Ops.VIEW, dtypes.float.ptr(256), arg=ShapeTracker(views=(View(shape=(8, 32, 1, 1), strides=(1, 8, 0, 0), offset=0, mask=None, contiguous=False),)), src=( # noqa: E501
UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(256), arg=1, src=()),)),)),)),))
opt = [Opt(op=OptOps.LOCAL, axis=1, arg=4), Opt(op=OptOps.UPCAST, axis=2, arg=2), Opt(op=OptOps.LOCAL, axis=1, arg=8),
Opt(op=OptOps.UPCAST, axis=1, arg=0), Opt(op=OptOps.UPCAST, axis=1, arg=4), Opt(op=OptOps.LOCAL, axis=0, arg=8),
Opt(op=OptOps.UPCAST, axis=1, arg=0), Opt(op=OptOps.UPCAST, axis=0, arg=2)]
k = helper_linearizer_ast(ast, [Tensor.randn(8*32).realize()], opts=[opt])[-1]
out = [u for u in k.uops if u.op is Ops.STORE][0]
assert out.src[-1].op is Ops.VECTORIZE and out.src[-1].dtype.count != 1
@unittest.skipUnless(Device[Device.DEFAULT].renderer.supports_float4, "need backends that support float4")
class TestFloat4(unittest.TestCase):
@staticmethod
def count_float4(k, n=4):
return (len([uop for uop in k.uops if uop.op is Ops.LOAD and uop.dtype == dtypes.float.vec(n)]),
len([uop for uop in k.uops if uop.op is Ops.STORE and uop.src[-1].dtype == dtypes.float.vec(n)]))
@staticmethod
def count_half4(k):
return (len([uop for uop in k.uops if uop.op is Ops.LOAD and uop.dtype == dtypes.half.vec(4)]),
len([uop for uop in k.uops if uop.op is Ops.STORE and uop.src[-1].dtype == dtypes.half.vec(4)]))
def test_float4_basic(self):
a = Tensor.empty(2, 8).realize()
b = Tensor.empty(2, 8).realize()
c = a + b
s = c.schedule()[0]
k = Kernel(s.ast)
k.apply_opts([Opt(op=OptOps.UPCAST, axis=0, arg=4)])
k.linearize()
assert TestFloat4.count_float4(k) == (2, 1)
@unittest.skipIf(Device.DEFAULT in {"CPU", "LLVM"} and AMX, "CPU with AMX upcasts float up to size 16")
def test_float4_multidim(self):
a = Tensor.empty(2, 8).realize()
b = Tensor.empty(2, 8).realize()
c = a + b
s = c.schedule()[0]
k = Kernel(s.ast)
k.shift_to(0, 4) # float4 dimension
k.shift_to(0, 2, insert_before=k.shape_len-1)
k.upcast()
k.upcast()
k.local_dims += 1
k.linearize()
assert TestFloat4.count_float4(k) == (4, 2)
@unittest.skipUnless(Device.DEFAULT in {"CPU", "LLVM"} and AMX, "Only CPU with AMX upcasts float up to size 16")
def test_float4_multidim_amx(self):
def kernel_for_shape(size, shift):
a = Tensor.empty(2, size).realize()
b = Tensor.empty(2, size).realize()
c = a + b
s = c.schedule()[0]
k = Kernel(s.ast)
k.shift_to(0, 4)
k.shift_to(0, shift, insert_before=k.shape_len-1)
k.upcast()
k.upcast()
k.local_dims += 1
k.linearize()
return k
sizes = [12, 8, 16]
shifts = [3, 2, 4]
excepted_upcast_size = [4, 8, 16]
expected_output = [(6,3), (2,1), (2,1)]
for i in range(len(sizes)):
assert TestFloat4.count_float4(kernel_for_shape(sizes[i], shifts[i]), excepted_upcast_size[i]) == expected_output[i]
def test_float4_unaligned_load(self):
a = Tensor.empty(9).realize().shrink(((1, 9),))
b = Tensor.empty(9).realize().shrink(((1, 9),))
c = a + b
s = c.schedule()[0]
k = Kernel(s.ast)
k.apply_opts([Opt(op=OptOps.UPCAST, axis=0, arg=4)])
k.linearize()
assert TestFloat4.count_float4(k) == (0, 1)
@unittest.skipIf(Device.DEFAULT in {"CPU", "LLVM"} and AMX, "CPU with AMX upcasts float up to size 16")
def test_float4_multidim_unaligned_load(self):
a = Tensor.empty(2, 9).realize().shrink(((0, 2), (1, 9),))
b = Tensor.empty(2, 9).realize().shrink(((0, 2), (1, 9),))
c = a + b
s = c.schedule()[0]
k = Kernel(s.ast)
k.shift_to(len(k.full_unupcasted_shape)-1, 4) # manual trigger float4 dim
k.upcast()
k.shift_to(len(k.full_unupcasted_shape)-1, 2, insert_before=k.shape_len-1)
k.upcast()
k.local_dims += 1
k.linearize()
assert TestFloat4.count_float4(k) == (0, 2)
@unittest.skipUnless(Device.DEFAULT in {"CPU", "LLVM"} and AMX, "Only CPU with AMX upcasts float up to size 16")
def test_float4_multidim_unaligned_load_amx(self):
def kernel_for_shape(size, shift):
a = Tensor.empty(2, size).realize().shrink(((0, 2), (1, size),))
b = Tensor.empty(2, size).realize().shrink(((0, 2), (1, size),))
c = a + b
s = c.schedule()[0]
k = Kernel(s.ast)
k.shift_to(len(k.full_unupcasted_shape)-1, 4) # manual trigger float4 dim
k.upcast()
k.shift_to(len(k.full_unupcasted_shape)-1, shift, insert_before=k.shape_len-1)
k.upcast()
k.local_dims += 1
k.linearize()
return k
sizes = [13, 9, 17]
shifts = [3, 2, 4]
excepted_upcast_size = [4, 8, 16]
expected_output = [(0,3), (0,1), (0,1)]
for i in range(len(sizes)):
assert TestFloat4.count_float4(kernel_for_shape(sizes[i], shifts[i]), excepted_upcast_size[i]) == expected_output[i]
def test_float4_sometimes_unaligned(self):
a = Tensor.empty(1, 1, 8).realize()
b = Tensor.empty(1, 1, 5).realize().shrink(((0, 1), (0, 1), (1, 5)))
c = a.conv2d(b)
# only the first and last conv dot products are aligned in a, and b is never aligned, so no
# float4 should be emitted (the reduce axis of size 4 is the float4 axis here)
s = c.schedule()[0]
k = Kernel(s.ast)
k.upcast()
k.linearize()
assert TestFloat4.count_float4(k) == (0, 0)
def test_float4_multidim_sometimes_unaligned(self):
a = Tensor.empty(1, 1, 7).realize()
b = Tensor.empty(1, 1, 5).realize().shrink(((0, 1), (0, 1), (1, 5)))
c = a.conv2d(b)
# the first conv dot product is aligned in a. If we upcast the output and reduce
# dimension, then we could do float4 for only that one set of loads, but we currently
# don't.
# UPDATE: now we do this fusion
s = c.schedule()[0]
k = Kernel(s.ast)
k.upcast()
k.upcast()
k.linearize()
assert TestFloat4.count_float4(k) in {(0,1), (1,1)}
def test_float4_noncontiguous(self):
a = Tensor.empty(4, 2).realize()
b = Tensor.empty(4, 2).realize()
c = a + b
# we will upcast the top axis of sz 4. they should not be coalesced into float4,
# since the top axis is not contiguous.
s = c.schedule()[0]
k = Kernel(s.ast)
k.shift_to(0, 4, top=True) # top axes are float4 axes
k.upcast()
k.linearize()
assert TestFloat4.count_float4(k) == (0, 0)
def test_float4_expand(self):
a = Tensor.empty(9).realize().shrink(((1, 9),))
b = Tensor.empty(2).realize().reshape((2, 1)).expand((2,4)).reshape((8,))
c = a + b
# we will upcast the top axis of sz 4. they should not be coalesced into float4,
# since the top axis is not contiguous.
s = c.schedule()[0]
k = Kernel(s.ast)
k.shift_to(0, 4) # float4 axis
k.upcast()
k.linearize()
assert TestFloat4.count_float4(k) == (0, 1)
def test_float4_heterogeneous(self):
a = Tensor.empty(8).realize()
b = Tensor.empty(9).realize().shrink(((1, 9),))
c = a + b
# should float4 b but not a
s = c.schedule()[0]
k = Kernel(s.ast)
k.shift_to(0, 4) # float4 axis
k.upcast()
k.linearize()
assert TestFloat4.count_float4(k) == (1, 1)
def test_half4_load_unrolled(self):
# from llama 7B shard 4 gpus
ast = UOp(Ops.SINK, dtypes.void, arg=None, src=(
UOp(Ops.STORE, dtypes.void, arg=None, src=(
UOp(Ops.VIEW, dtypes.float.ptr(96000), arg=ShapeTracker(views=(View(shape=(1, 3, 32000, 1), strides=(0, 32000, 1, 0), offset=0, mask=None, contiguous=True),)), src=( # noqa: E501
UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(96000), arg=0, src=()),)),
UOp(Ops.REDUCE_AXIS, dtypes.float, arg=(Ops.ADD, (3,)), src=(
UOp(Ops.CAST, dtypes.float, arg=None, src=(
UOp(Ops.MUL, dtypes.half, arg=None, src=(
UOp(Ops.LOAD, dtypes.half, arg=None, src=(
UOp(Ops.VIEW, dtypes.half.ptr(9216), arg=ShapeTracker(views=(View(shape=(1, 3, 32000, 1024), strides=(0, 4096, 0, 1), offset=0, mask=None, contiguous=False),)), src=( # noqa: E501
UOp(Ops.DEFINE_GLOBAL, dtypes.half.ptr(9216), arg=1, src=()),)),)),
UOp(Ops.LOAD, dtypes.half, arg=None, src=(
UOp(Ops.VIEW, dtypes.half.ptr(32768000), arg=ShapeTracker(views=(View(shape=(1, 3, 32000, 1024), strides=(0, 0, 1024, 1), offset=0, mask=None, contiguous=False),)), src=( # noqa: E501
UOp(Ops.DEFINE_GLOBAL, dtypes.half.ptr(32768000), arg=2, src=()),)),)),)),)),)),)),))
# TODO: fix this, expected might change but should be positive
for expected, opts in [
((7, 0), [Opt(op=OptOps.UPCAST, axis=1, arg=4), Opt(op=OptOps.UPCAST, axis=0, arg=3), Opt(op=OptOps.UNROLL, axis=0, arg=4)]),
((5, 0), [Opt(op=OptOps.UPCAST, axis=1, arg=4), Opt(op=OptOps.UNROLL, axis=0, arg=4)]),
((2, 0), [Opt(op=OptOps.UNROLL, axis=0, arg=4)]),
]:
k = Kernel(ast)
k.apply_opts(opts)
k.linearize()
count = TestFloat4.count_half4(k)
assert count == expected, f"{count=}, {expected=}"
@unittest.skip("this doesn't happen anymore")
def test_float4_acc(self):
# from float32 stable diffusion red tinybox
ast = UOp(Ops.SINK, dtypes.void, arg=None, src=(
UOp(Ops.STORE, dtypes.void, arg=None, src=(
UOp(Ops.VIEW, dtypes.float.ptr(33554432), arg=ShapeTracker(views=(View(shape=(1, 1, 128, 512, 512, 1, 1, 1), strides=(0, 0, 262144, 512, 1, 0, 0, 0), offset=0, mask=None, contiguous=True),)), src=( # noqa: E501
UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(33554432), arg=0, src=()),)),
UOp(Ops.ADD, dtypes.float, arg=None, src=(
UOp(Ops.REDUCE_AXIS, dtypes.float, arg=(Ops.ADD, (5, 6, 7)), src=(
UOp(Ops.MUL, dtypes.float, arg=None, src=(
UOp(Ops.LOAD, dtypes.float, arg=None, src=(
UOp(Ops.VIEW, dtypes.float.ptr(67108864), arg=ShapeTracker(views=(View(shape=(1, 1, 1, 256, 4, 514, 4, 514), strides=(0, 0, 0, 262144, 0, 512, 0, 1), offset=-513, mask=((0, 1), (0, 1), (0, 1), (0, 256), (0, 4), (1, 513), (0, 4), (1, 513)), contiguous=False), View(shape=(1, 1, 128, 512, 512, 256, 3, 3), strides=(0, 0, 0, 2056, 1, 4227136, 1058840, 515), offset=0, mask=None, contiguous=False))), src=( # noqa: E501
UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(67108864), arg=1, src=()),)),)),
UOp(Ops.LOAD, dtypes.float, arg=None, src=(
UOp(Ops.VIEW, dtypes.float.ptr(294912), arg=ShapeTracker(views=(View(shape=(1, 1, 128, 512, 512, 256, 3, 3), strides=(0, 0, 2304, 0, 0, 9, 3, 1), offset=0, mask=None, contiguous=False),)), src=( # noqa: E501
UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(294912), arg=2, src=()),)),)),)),)),
UOp(Ops.LOAD, dtypes.float, arg=None, src=(
UOp(Ops.VIEW, dtypes.float.ptr(128), arg=ShapeTracker(views=(View(shape=(1, 1, 128, 512, 512, 1, 1, 1), strides=(0, 0, 1, 0, 0, 0, 0, 0), offset=0, mask=None, contiguous=False),)), src=( # noqa: E501
UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(128), arg=3, src=()),)),)),)),)),))
for expected, opts in [
(1, [Opt(op=OptOps.UPCAST, axis=2, arg=4)]),
(4, [Opt(op=OptOps.UPCAST, axis=2, arg=4), Opt(op=OptOps.UPCAST, axis=0, arg=4)]),
]:
k = Kernel(ast)
k.apply_opts(opts)
k.linearize()
count = len([uop for uop in k.uops if uop.op is Ops.DEFINE_ACC and uop.dtype == dtypes.float.vec(4)])
assert count == expected, f"{count=}, {expected=}"
@unittest.skip("this doesn't happen anymore")
def test_float2_acc(self):
# from resnet
ast = UOp(Ops.SINK, dtypes.void, arg=None, src=(
UOp(Ops.STORE, dtypes.void, arg=None, src=(
UOp(Ops.VIEW, dtypes.half.ptr(212926464), arg=ShapeTracker(views=(View(shape=(1, 256, 1, 64, 1, 114, 1, 114), strides=(0, 831744, 0, 12996, 0, 114, 0, 1), offset=0, mask=None, contiguous=True),)), src=( # noqa: E501
UOp(Ops.DEFINE_GLOBAL, dtypes.half.ptr(212926464), arg=0, src=()),)),
UOp(Ops.CAST, dtypes.half, arg=None, src=(
UOp(Ops.REDUCE_AXIS, dtypes.float, arg=(Ops.ADD, (4, 6)), src=(
UOp(Ops.CAST, dtypes.float, arg=None, src=(
UOp(Ops.LOAD, dtypes.half, arg=None, src=(
UOp(Ops.VIEW, dtypes.half.ptr(462422016), arg=ShapeTracker(views=(View(shape=(256, 64, 3, 56, 2, 3, 56, 2), strides=(1806336, 28224, 3, 504, 0, 1, 9, 0), offset=0, mask=((0, 256), (0, 64), (0, 3), (0, 56), (0, 1), (0, 3), (0, 56), (0, 1)), contiguous=False), View(shape=(256, 64, 3, 115, 3, 115), strides=(7225344, 112896, 37632, 336, 112, 1), offset=0, mask=((0, 256), (0, 64), (0, 3), (0, 112), (0, 3), (0, 112)), contiguous=False), View(shape=(256, 64, 456, 456), strides=(7617600, 119025, 345, 1), offset=0, mask=((0, 256), (0, 64), (0, 345), (0, 345)), contiguous=False), View(shape=(1, 256, 1, 64, 4, 114, 4, 114), strides=(0, 13307904, 0, 207936, 51984, 456, 114, 1), offset=0, mask=None, contiguous=True))), src=( # noqa: E501
UOp(Ops.DEFINE_GLOBAL, dtypes.half.ptr(462422016), arg=1, src=()),)),)),)),)),)),)),))
for expected, opts in [
(16, [Opt(op=OptOps.LOCAL, axis=1, arg=16), Opt(op=OptOps.UPCAST, axis=1, arg=0), Opt(op=OptOps.UPCAST, axis=2, arg=2), Opt(op=OptOps.LOCAL, axis=2, arg=3), Opt(op=OptOps.UPCAST, axis=3, arg=4)]), # noqa: E501
(4, [Opt(op=OptOps.LOCAL, axis=1, arg=16), Opt(op=OptOps.UPCAST, axis=1, arg=0), Opt(op=OptOps.UPCAST, axis=2, arg=2)]),
]:
k = Kernel(ast)
k.apply_opts(opts)
k.linearize()
count = len([uop for uop in k.uops if uop.op is Ops.DEFINE_ACC and uop.dtype == dtypes.float.vec(2)])
assert count == expected, f"{count=}, {expected=}"
class TestHandCodedOpts(unittest.TestCase):
def test_masked_upcast(self):
layer_1 = Tensor.cat(*[Tensor.empty(5) for _ in range(4)])
layer_2 = Tensor.cat(layer_1.unsqueeze(0), Tensor.empty(6, 20))
s = layer_2.schedule()[-1]
k = Kernel(s.ast)
k.apply_opts(hand_coded_optimizations(k))
assert len(k.bufs) == 6 # make sure all ops are done in one kernel
# masked upcast should upcast masked axis of size 7
# masked upcast should not upcast large (20) last axis
# float4/other hcopt shouldn't upcast last axis, since we already have 7 upcast, and the last axis is not very contiguous
assert k.upcasted == 1 and k.full_shape[-1] == 7
@unittest.skipIf(Device.DEFAULT in {"METAL", "WEBGPU"}, "METAL/WEBGPU split this kernel since it has 37 buffers")
def test_masked_upcast_wino(self):
monster = Tensor.stack(*[Tensor.stack(*[Tensor.empty(16) for _ in range(6)]) for _ in range(6)])
s = monster.schedule()[-1]
k = Kernel(s.ast)
k.apply_opts(hand_coded_optimizations(k))
assert len(k.bufs) == 37 # make sure all ops are done in one kernel
# should upcast the two Tensor.stacks
assert k.upcasted >= 2 and k.full_shape[k.shape_len-k.upcasted:k.shape_len].count(6) == 2
def test_masked_upcast_wino_full(self):
with Context(WINO=1):
x,w = Tensor.rand(1,4,8,8, requires_grad=True).realize(), Tensor.rand(4,4,3,3, requires_grad=True).realize()
out = Tensor.conv2d(x,w, padding=1)
out.mean().backward()
upcasts = []
wino_schedule = out.schedule()
# collect upcasts of tile transform kernels
for i, si in enumerate(wino_schedule):
k = Kernel(si.ast)
k.apply_opts(hand_coded_optimizations(k))
if k.reduceop is not None: continue # not a tile transform kernel (there is a gemm reduce kernel)
if len(k.bufs) < 22: continue # not a tile transform kernel (there's a permute kernel at the end)
upcasts.append(tuple(k.full_shape[k.shape_len - k.upcasted:k.shape_len]))
assert len(upcasts) == 3 # 3 transformation matrices
assert len(wino_schedule) <= 4 # 4 kernels
# this test case's inputs are too small, so one of the 4-stacks became a local, which is fine i guess
assert upcasts.count((6, 6)) == 2 #and upcasts.count((4, 4)) == 1
backward_schedule = Tensor.schedule(x.grad, w.grad)
for si in backward_schedule:
k = Kernel(si.ast)
k.apply_opts(hand_coded_optimizations(k))
k.linearize()
if len(k.bufs) < 20: continue # not a tile transform kernel
# heuristic number to make sure that at least some upcasts but not too many upcasts are being done
assert 6 <= prod(k.full_shape[k.shape_len - k.upcasted:k.shape_len]) <= 216
assert len(backward_schedule) <= 13 # just the current number, but it could be better
def test_masked_upcast_many(self):
layer_1 = Tensor.cat(Tensor.rand(3, 4), Tensor.rand(4, 4))
layer_2 = Tensor.cat(layer_1.unsqueeze(0), Tensor.rand(6, 7, 4))
layer_3 = Tensor.cat(layer_2.unsqueeze(0), Tensor.rand(6, 7, 7, 4))
k = helper_linearizer_opt(layer_3)[-1]
assert len(k.bufs) == 5 # make sure all ops are done in one kernel
# check that we don't do too many upcasts
assert prod(k.full_shape[k.shape_len-k.upcasted:k.shape_len]) <= 49
@unittest.skipUnless(Device[Device.DEFAULT].renderer.has_local, "test requires locals")
def test_matvec(self):
N = 128
a = Tensor.rand(1, N).realize()
b = Tensor.rand(N, N).realize()
c = a @ b
k = helper_linearizer_opt(c)[-1]
assert k.group_for_reduces == 1
assert k.local_dims == 1
assert k.upcasted == 1
def helper_linearizer_ast(ast:UOp, inputs:list[Tensor], *args, **kwargs):
assert isinstance(ast, UOp), "ast must be UOp"
inbufs = [x.uop.base.buffer for x in inputs]
outbufs = [Buffer(inbufs[-1].device if inbufs else Device.DEFAULT, out.st_arg.size, out.src[1].dtype).allocate() \
for out in ast.src]
return _helper_linearizer_opt_ast(ast, outbufs+inbufs, *args, **kwargs)
def helper_linearizer_opt(r:Union[Tensor, list[Tensor]], *args, **kwargs):
realized_ast, real_bufs = helper_realized_ast(r)
return _helper_linearizer_opt_ast(realized_ast, real_bufs, *args, **kwargs)
def copyout_outputs(lin:Kernel, outbufs:list[Buffer]) -> list[np.ndarray]:
ret = []
for i,x in enumerate(outbufs):
shape: tuple[int, ...] = lin.ast.src[i].st_arg.shape
ret.append(np.frombuffer(x.as_buffer(), _to_np_dtype(x.dtype)).reshape(shape))
return ret
def reset_bufs(bufs:list[Buffer]):
for buf in bufs: buf.copyin(np.zeros((buf.size, ), dtype=_to_np_dtype(buf.dtype)).data) # Zero to check that all values are filled
def _helper_linearizer_opt_ast(realized_ast:UOp, real_bufs:list[Buffer], opts=[],
apply_tc=False, atol=1e-4, rtol=1e-4, color_sizes=[], wanna_output=[]) -> list[Kernel]:
lins: list[Kernel] = []
outbufs = [real_bufs[x.src[0].base.arg] for x in realized_ast.src]
device = real_bufs[0].device
def get_prg(k:Kernel): return CompiledRunner(replace(k.to_program(), device=device))
def check_opt(opts, create_k, expected_color_size):
k = create_k()
lins.append(k)
if apply_tc:
assert k.apply_tensor_cores(1, extra_opts=opts), "no tensor core triggered"
else:
k.apply_opts(opts)
if expected_color_size is not None:
cs = list(zip(k.colors(), k.full_shape))
assert cs == expected_color_size, f"expected={expected_color_size} got={cs}"
prg = get_prg(k)
reset_bufs(outbufs)
prg.exec(real_bufs)
for x,want in zip(copyout_outputs(k, outbufs), wanna_output): np.testing.assert_allclose(x, want, atol=atol, rtol=rtol)
# Get baseline if it is not provided, which is not optimized at all.
k = Kernel(realized_ast)
lins.append(k)
prg = get_prg(k)
prg.exec(real_bufs)
if len(wanna_output) == 0: wanna_output = copyout_outputs(k, outbufs)
else:
for buf,want in zip(copyout_outputs(k, outbufs), wanna_output): np.testing.assert_allclose(buf, want, atol=atol, rtol=rtol)
# Check correctness of handcoded optimiztions.
k = Kernel(realized_ast)
k.apply_opts(hand_coded_optimizations(k))
lins.append(k)
prg = get_prg(k)
reset_bufs(outbufs)
prg.exec(real_bufs)
for buf,want in zip(copyout_outputs(k, outbufs), wanna_output): np.testing.assert_allclose(buf, want, atol=atol, rtol=rtol)
for i,x in enumerate(opts): # Check custom transformations if any.
check_opt(x, lambda: Kernel(realized_ast), color_sizes[i] if i < len(color_sizes) else None)
return lins
class TestKernelOpts(unittest.TestCase):
@unittest.skipUnless(Device[Device.DEFAULT].renderer.has_local, "test requires locals")
@unittest.skipUnless(Device[Device.DEFAULT].renderer.has_shared, "test requires shared")
def test_local_and_grouped_reduce(self):
N = 128
Tensor.manual_seed(1882)
a = Tensor.rand(4, 4, N, N)
b = Tensor.rand(4, 4, N)
r = (b.sqrt() + ((a+1).sum(axis=3).exp()))
helper_linearizer_opt(r, [
[Opt(OptOps.LOCAL, 0, 2)],
[Opt(OptOps.LOCAL, 0, 8)],
[Opt(OptOps.LOCAL, 0, 16)], # Checking how it works with locals
[Opt(OptOps.GROUPTOP, 0, 2)],
[Opt(OptOps.GROUPTOP, 0, 32)],
[Opt(OptOps.GROUPTOP, 0, 64)], # Checking how it works with grouped reduce
[Opt(OptOps.LOCAL, 0, 2), Opt(OptOps.GROUPTOP, 0, 2)],
[Opt(OptOps.LOCAL, 0, 16), Opt(OptOps.GROUPTOP, 0, 16)],
[Opt(OptOps.LOCAL, 0, 32), Opt(OptOps.GROUPTOP, 0, 2)],
# Checking how it works with locals + grouped reduce
[Opt(OptOps.LOCAL, 0, 2), Opt(OptOps.GROUPTOP, 0, 64)],
# Checking how it works with locals + grouped reduce + upcasts
[Opt(OptOps.LOCAL, 0, 2), Opt(OptOps.GROUPTOP, 0, 2), Opt(OptOps.UPCAST, 0, 8), Opt(OptOps.UNROLL, 1, 4)],
# many local + many group
[Opt(OptOps.GROUP, 0, 2)] * 4,
[Opt(OptOps.LOCAL, 0, 2)] * 4,
[Opt(OptOps.LOCAL, 0, 2), Opt(OptOps.GROUP, 0, 2)] * 4,
])
def test_upcasts(self):
N = 16
Tensor.manual_seed(1772)
a = Tensor.rand(N, N)
b = Tensor.rand(N, N)
r = (a+b).sqrt() * ((a+1).exp())
helper_linearizer_opt(r, [
[Opt(OptOps.UPCAST, 0, 2)],
[Opt(OptOps.UPCAST, 0, 4)],
[Opt(OptOps.UPCAST, 0, 8)], # Checking how it works with upcasts
])
def test_full_upcast(self):
Tensor.manual_seed(1772)
a = Tensor.rand(4)
b = Tensor.rand(4)
r = (a+b).sqrt() * ((a+1).exp())
helper_linearizer_opt(r, [
[Opt(OptOps.UPCAST, 0, 4)], # Checking how it works with upcasts
])
@unittest.skipUnless(Device[Device.DEFAULT].renderer.has_local, "test requires locals")
@unittest.skipUnless(Device[Device.DEFAULT].renderer.has_shared, "test requires shared")
def test_matmul(self):
N = 128
Tensor.manual_seed(1552)
a = Tensor.rand(N, N)
b = Tensor.rand(N, N)
r = a@b
helper_linearizer_opt(r, [
[Opt(OptOps.UPCAST, 0, 2)],
[Opt(OptOps.UPCAST, 0, 4), Opt(OptOps.UPCAST, 1, 4)], # Checking how it works with upcasts
[Opt(OptOps.LOCAL, 0, 2)],
[Opt(OptOps.LOCAL, 1, 32)],
[Opt(OptOps.LOCAL, 0, 4), Opt(OptOps.LOCAL, 1, 4)],
[Opt(OptOps.LOCAL, 0, 4), Opt(OptOps.LOCAL, 1, 32)],
[Opt(OptOps.LOCAL, 0, 16), Opt(OptOps.LOCAL, 1, 8)], # Checking how it works with locals
[Opt(OptOps.GROUPTOP, 0, 2)],
[Opt(OptOps.GROUPTOP, 0, 32)],
[Opt(OptOps.GROUPTOP, 0, 32), Opt(OptOps.UNROLL, 0, 4)], # Checking how it works with grouped_reduce
[Opt(OptOps.LOCAL, 0, 2), Opt(OptOps.LOCAL, 1, 2), Opt(OptOps.GROUPTOP, 0, 32)],
[Opt(OptOps.LOCAL, 0, 8), Opt(OptOps.GROUPTOP, 0, 32)],
[Opt(OptOps.LOCAL, 0, 4), Opt(OptOps.LOCAL, 0, 8), Opt(OptOps.GROUPTOP, 0, 4)], # Checking how it works with local+grouped_reduce
# Checking all together
[Opt(OptOps.LOCAL, 0, 4), Opt(OptOps.LOCAL, 0, 4), Opt(OptOps.GROUPTOP, 0, 8), Opt(OptOps.UNROLL, 0, 4), Opt(OptOps.UPCAST, 0, 4),
Opt(OptOps.UPCAST, 1, 2)],
# Full global upcast + local
[Opt(OptOps.LOCAL, 0, 4), Opt(OptOps.LOCAL, 0, 4), Opt(OptOps.GROUPTOP, 0, 8), Opt(OptOps.UNROLL, 0, 4), Opt(OptOps.UPCAST, 0, 8)],
])
@unittest.skipUnless(Device[Device.DEFAULT].renderer.has_local, "test requires locals")
@unittest.skipUnless(Device[Device.DEFAULT].renderer.has_shared, "test requires shared")
def test_double_reduce(self):
N = 128
Tensor.manual_seed(1552)
a = Tensor.rand(8, N, 8, N)
r = a.sum(axis=(1,3))
helper_linearizer_opt(r, [
# openCL / GPU=1 is 256 max threads
[Opt(OptOps.GROUPTOP, 0, 2)], [Opt(OptOps.GROUPTOP, 0, 32)],
[Opt(OptOps.GROUPTOP, 1, 2)], [Opt(OptOps.GROUPTOP, 1, 32)], # Checking how it works with 1 grouped_reduce.
[Opt(OptOps.GROUPTOP, 0, 2), Opt(OptOps.GROUPTOP, 1, 2)],
[Opt(OptOps.GROUPTOP, 0, 16), Opt(OptOps.GROUPTOP, 1, 2)],
[Opt(OptOps.GROUPTOP, 0, 4), Opt(OptOps.GROUPTOP, 1, 64)], # Checking how it works with 2 grouped_reduces.
[Opt(OptOps.GROUPTOP, 0, 16), Opt(OptOps.GROUPTOP, 1, 2), Opt(OptOps.UNROLL, 0, 4)],
[Opt(OptOps.GROUPTOP, 0, 2), Opt(OptOps.GROUPTOP, 1, 32), Opt(OptOps.UNROLL, 2, 4)], # Checking how it works with 2 grouped_reduces + upcasts.
[Opt(OptOps.LOCAL, 0, 4), Opt(OptOps.LOCAL, 1, 4), Opt(OptOps.GROUPTOP, 0, 4), Opt(OptOps.GROUPTOP, 1, 4)],
# Checking how it works with 2 grouped_reduces + upcasts + locals.
[Opt(OptOps.LOCAL, 0, 4), Opt(OptOps.LOCAL, 1, 4), Opt(OptOps.GROUPTOP, 0, 2), Opt(OptOps.GROUPTOP, 1, 32), Opt(OptOps.UNROLL, 1, 4)],
[Opt(OptOps.LOCAL, 0, 2), Opt(OptOps.LOCAL, 1, 2), Opt(OptOps.GROUPTOP, 0, 8), Opt(OptOps.GROUPTOP, 1, 4), Opt(OptOps.UPCAST, 0, 2)],
[Opt(OptOps.LOCAL, 0, 2), Opt(OptOps.LOCAL, 1, 2), Opt(OptOps.GROUPTOP, 0, 8), Opt(OptOps.GROUPTOP, 1, 4), Opt(OptOps.UPCAST, 0, 2),
Opt(OptOps.UNROLL, 0, 4), Opt(OptOps.UNROLL, 1, 4)], # Checking how it works with 2 grouped_reduces + upcasts + locals.
[Opt(OptOps.LOCAL, 0, 4), Opt(OptOps.LOCAL, 1, 4), Opt(OptOps.GROUPTOP, 0, 4), Opt(OptOps.GROUPTOP, 1, 4), Opt(OptOps.UPCAST, 0, 2),
Opt(OptOps.UPCAST, 0, 2)], # No globals
])
@unittest.skipUnless(Device[Device.DEFAULT].renderer.tensor_cores, "test requires tensor cores")
@unittest.skipUnless(Device[Device.DEFAULT].renderer.has_local, "test requires locals")
def test_invalid_tensor_core_extra_opts(self):
N = 128
Tensor.manual_seed(1552)
a = Tensor.rand(N, N)
b = Tensor.rand(N, N)
realized_ast, _ = helper_realized_ast(a@b)
invalid_opts = [
[Opt(OptOps.LOCAL, 2, 2)],
[Opt(OptOps.UPCAST, 2, 2)],
[Opt(OptOps.LOCAL, 0, 2), Opt(OptOps.LOCAL, 2, 2)],
]
for x in invalid_opts:
k = Kernel(realized_ast)
with self.assertRaises(AssertionError):
assert k.apply_tensor_cores(use_tensor_cores=1, extra_opts=x), "no valid tensor core" # for METAL in runners
@unittest.skipUnless(Device[Device.DEFAULT].renderer.tensor_cores, "test requires tensor cores")
def test_buf_index_not_found_tensor_core(self):
ast = UOp(Ops.SINK, dtypes.void, arg=None, src=(
UOp(Ops.STORE, dtypes.void, arg=None, src=(
UOp(Ops.VIEW, dtypes.float.ptr(256), arg=ShapeTracker(views=(View(shape=(1, 256), strides=(0, 1), offset=0, mask=None, contiguous=True),)), src=( # noqa: E501
UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(256), arg=0, src=()),)),
UOp(Ops.REDUCE_AXIS, dtypes.float, arg=(Ops.ADD, (0,)), src=(
UOp(Ops.MUL, dtypes.float, arg=None, src=(
UOp(Ops.CAST, dtypes.float, arg=None, src=(
UOp(Ops.CMPNE, dtypes.bool, arg=None, src=(
UOp(Ops.LOAD, dtypes.int, arg=None, src=(
UOp(Ops.VIEW, dtypes.int.ptr(256), arg=ShapeTracker(views=(View(shape=(1243, 256), strides=(0, 1), offset=0, mask=None, contiguous=False),)), src=( # noqa: E501
UOp(Ops.DEFINE_GLOBAL, dtypes.int.ptr(256), arg=1, src=()),)),)),
UOp(Ops.LOAD, dtypes.int, arg=None, src=(
UOp(Ops.VIEW, dtypes.int.ptr(1243), arg=ShapeTracker(views=(View(shape=(1243, 256), strides=(1, 0), offset=0, mask=None, contiguous=False),)), src=( # noqa: E501
UOp(Ops.DEFINE_GLOBAL, dtypes.int.ptr(1243), arg=2, src=()),)),)),)),)),
UOp(Ops.LOAD, dtypes.float, arg=None, src=(
UOp(Ops.VIEW, dtypes.float.ptr(1243), arg=ShapeTracker(views=(View(shape=(1243, 256), strides=(1, 0), offset=0, mask=None, contiguous=False),)), src=( # noqa: E501
UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(1243), arg=3, src=()),)),)),)),)),)),))
k = Kernel(ast, opts=Device[Device.DEFAULT].renderer)
with self.assertRaises(KernelOptError):
k.apply_opt(Opt(OptOps.TC, 0, (-1, 1, 1)))
@unittest.skipUnless(Device[Device.DEFAULT].renderer.tensor_cores, "test requires tensor cores")
def test_tensor_core_opts(self):
N = 128
Tensor.manual_seed(1552)
for tc in Device[Device.DEFAULT].renderer.tensor_cores:
# bf16 buffer returns float32 numpy outputs so test would fail. testing opt with half suffices.
if tc.dtype_in != dtypes.half and tc.dtype_out != dtypes.half: continue
a, b = Tensor.rand(N, N, dtype=tc.dtype_in), Tensor.rand(N, N, dtype=tc.dtype_in)
r = a.matmul(b, dtype=tc.dtype_out)
(atol, rtol) = ((0.25, 0.01) if tc.dtype_out == dtypes.half else (3e-2, 1e-3)) if tc.dtype_in == dtypes.half else (1e-4, 1e-4)
helper_linearizer_opt(r, [
[],
[Opt(OptOps.UPCAST, 0, 4)],
[Opt(OptOps.UPCAST, 1, 4)],
[Opt(OptOps.UPCAST, 0, 4), Opt(OptOps.UPCAST, 1, 4)], # check upcasts
[Opt(OptOps.UNROLL, 0, 2)], # check unroll
[Opt(OptOps.UPCAST, 0, 4), Opt(OptOps.UNROLL, 0, 2)], # check combo of unroll and local
[Opt(OptOps.UPCAST, 0, 4), Opt(OptOps.UPCAST, 1, 4), Opt(OptOps.UNROLL, 0, 2)],
[Opt(OptOps.UPCAST, 0, 4), Opt(OptOps.UPCAST, 1, 4), Opt(OptOps.UNROLL, 0, 4)],
[Opt(OptOps.UPCAST, 1, 4), Opt(OptOps.UPCAST, 0, 4)], # check permutations
[Opt(OptOps.UNROLL, 0, 2), Opt(OptOps.UPCAST, 0, 4)],
[Opt(OptOps.UPCAST, 0, 4), Opt(OptOps.UNROLL, 0, 2), Opt(OptOps.UPCAST, 1, 4)],
[Opt(OptOps.UNROLL, 0, 2), Opt(OptOps.UPCAST, 1, 4), Opt(OptOps.UPCAST, 0, 4), Opt(OptOps.UNROLL, 0, 4)],
# [Opt(OptOps.GROUP, 0, 2)] # doesn't work because group_for_reduce dims become early locals (conflicting with TC)
], apply_tc=True, atol=atol, rtol=rtol)
@unittest.skipUnless(Device[Device.DEFAULT].renderer.tensor_cores, "test requires tensor cores")
@unittest.skipUnless(Device[Device.DEFAULT].renderer.has_local, "test requires locals")
def test_tensor_core_opts_locals(self):
N = 128
Tensor.manual_seed(1552)
for tc in Device[Device.DEFAULT].renderer.tensor_cores:
# bf16 buffer returns float32 numpy outputs so test would fail. testing opt with half suffices.
if tc.dtype_in != dtypes.half and tc.dtype_out != dtypes.half: continue
a, b = Tensor.rand(N, N, dtype=tc.dtype_in), Tensor.rand(N, N, dtype=tc.dtype_in)
r = a.matmul(b, dtype=tc.dtype_out)
(atol, rtol) = ((0.25, 0.01) if tc.dtype_out == dtypes.half else (3e-2, 1e-3)) if tc.dtype_in == dtypes.half else (1e-4, 1e-4)
helper_linearizer_opt(r, [
[Opt(OptOps.UNROLL, 0, 0)], # check full unroll of reduce with locals
[Opt(OptOps.LOCAL, 0, 4)], # check local
[Opt(OptOps.UPCAST, 0, 4), Opt(OptOps.UPCAST, 1, 4), Opt(OptOps.UNROLL, 0, 4), Opt(OptOps.LOCAL, 0, 2)],
[Opt(OptOps.LOCAL, 0, 2), Opt(OptOps.UPCAST, 1, 4), Opt(OptOps.UNROLL, 0, 2), Opt(OptOps.UPCAST, 0, 4)],
], apply_tc=True, atol=atol, rtol=rtol)
def test_padto_matmul(self):
if (CI and Device.DEFAULT in ["AMD", "NV", "CUDA"]):
self.skipTest("super slow on CUDA and AMD because of the big grid dims")
N = 17 * 17
Tensor.manual_seed(289)
a = Tensor.rand(N, N)
b = Tensor.rand(N, N)
helper_linearizer_opt(a@b, [
[Opt(OptOps.PADTO, 0, 32)],
[Opt(OptOps.PADTO, 1, 32)],
[Opt(OptOps.PADTO, 2, 32)],
[Opt(OptOps.PADTO, 0, 32), Opt(OptOps.PADTO, 1, 32)],
[Opt(OptOps.PADTO, 0, 32), Opt(OptOps.PADTO, 1, 32), Opt(OptOps.PADTO, 2, 32)],
# can optimize further post PADTO
[Opt(OptOps.PADTO, 0, 32), Opt(OptOps.PADTO, 1, 32), Opt(OptOps.UPCAST, 0, 2), Opt(OptOps.UPCAST, 1, 2),],
])
def test_padto_upcasted_not_ok(self):
N = 4
a = Tensor.rand(N, N)
b = Tensor.rand(N, N)
helper_linearizer_opt(a@b, [
[Opt(OptOps.UPCAST, 0, 0)],
[Opt(OptOps.UPCAST, 1, 0)],
[Opt(OptOps.UNROLL, 0, 0)],
[Opt(OptOps.PADTO, 0, 8)],
[Opt(OptOps.PADTO, 1, 8)],
[Opt(OptOps.PADTO, 2, 8)],
])
with self.assertRaises(KernelOptError):
helper_linearizer_opt(a@b, [[Opt(OptOps.UPCAST, 0, 0), Opt(OptOps.PADTO, 2, 8)]])
with self.assertRaises(KernelOptError):
helper_linearizer_opt(a@b, [[Opt(OptOps.UPCAST, 1, 0), Opt(OptOps.PADTO, 2, 8)]])
with self.assertRaises(KernelOptError):
helper_linearizer_opt(a@b, [[Opt(OptOps.UNROLL, 0, 0), Opt(OptOps.PADTO, 2, 8)]])
def test_padto_sum_ok(self):
N = 18 * 18
# NOTE: this setup prevents 17 * 17 contiguous merged into one dimension
a = Tensor.rand(N, N).realize().shrink(((0, 17), (0, 17))) * 100
b = (Tensor.rand(N, N) < 0.5).realize().shrink(((0, 17), (0, 17)))
helper_linearizer_opt(a.sum(0), [
[Opt(OptOps.PADTO, 0, 32)],
[Opt(OptOps.PADTO, 0, 32), Opt(OptOps.UPCAST, 0, 8),],
])
helper_linearizer_opt(a.sum(1), [
[Opt(OptOps.PADTO, 0, 32)],
[Opt(OptOps.PADTO, 0, 32), Opt(OptOps.UPCAST, 0, 8),],
])
# can pad sum reduce axis if there's no unsafe ops prior to sum
for axis in (0, 1):
helper_linearizer_opt(a.sum(), [[Opt(OptOps.PADTO, axis, 32)],])
helper_linearizer_opt(a.sum(0), [[Opt(OptOps.PADTO, axis, 32)],])
helper_linearizer_opt(b.sum(), [[Opt(OptOps.PADTO, axis, 32)],])
helper_linearizer_opt(b.sum(0), [[Opt(OptOps.PADTO, axis, 32)],])
helper_linearizer_opt(b.sum(dtype=dtypes.bool), [[Opt(OptOps.PADTO, axis, 32)],])
# TODO: why?
if Device.DEFAULT != "WEBGPU":
helper_linearizer_opt(b.sum(0, dtype=dtypes.bool), [[Opt(OptOps.PADTO, axis, 32)],])
helper_linearizer_opt(b.sum(1, dtype=dtypes.bool), [[Opt(OptOps.PADTO, axis, 32)],])
# having unsafe ops after sum is fine
helper_linearizer_opt(a.sum().exp(), [[Opt(OptOps.PADTO, 0, 32)],])
helper_linearizer_opt(a.sum(0).exp(), [[Opt(OptOps.PADTO, 1, 32)],])
def test_padto_sum_not_ok(self):
N = 18 * 18
# NOTE: this setup prevents 17 * 17 contiguous merged into one dimension
a = Tensor.rand(N, N).shrink(((0, 17), (0, 17))).exp()
# exp is not safe to pad
with self.assertRaises(KernelOptError):
helper_linearizer_opt(a.exp().sum(), [[Opt(OptOps.PADTO, 0, 32)],])
with self.assertRaises(KernelOptError):
helper_linearizer_opt(a.exp().sum(0), [[Opt(OptOps.PADTO, 1, 32)],])
b = a < 1
# lt is not safe to pad
with self.assertRaises(KernelOptError):
helper_linearizer_opt(b.sum(), [[Opt(OptOps.PADTO, 0, 32)],])
with self.assertRaises(KernelOptError):
helper_linearizer_opt(b.sum(0), [[Opt(OptOps.PADTO, 1, 32)],])
def test_padto_max(self):
N = 18 * 18
# NOTE: this setup prevents 17 * 17 contiguous merged into one axis
a = -Tensor.rand(N, N).shrink(((0, 17), (0, 17))) * 100
helper_linearizer_opt(a.max(0), [
[Opt(OptOps.PADTO, 0, 32)],
[Opt(OptOps.PADTO, 0, 32), Opt(OptOps.UPCAST, 0, 8),],
])
helper_linearizer_opt(a.max(1), [
[Opt(OptOps.PADTO, 0, 32)],
[Opt(OptOps.PADTO, 0, 32), Opt(OptOps.UPCAST, 0, 8),],
])
# cannot pad max kernel on reduce
with self.assertRaises(KernelOptError):
helper_linearizer_opt(a.max(), [[Opt(OptOps.PADTO, 0, 32)],])
with self.assertRaises(KernelOptError):
helper_linearizer_opt(a.max(0), [[Opt(OptOps.PADTO, 1, 32)],])
def test_padto_where(self):
Tensor.manual_seed(0)
N = 17 * 17
a = (Tensor.randn(N, N).realize().max(axis=0, keepdim=True) > 1).where(1, 0)
helper_linearizer_opt(a.max(0), [
[Opt(OptOps.PADTO, 0, 32)],
[Opt(OptOps.PADTO, 0, 32), Opt(OptOps.UPCAST, 0, 8),],
])
def test_padto_where_multioutput(self):
Tensor.manual_seed(0)
N = 17 * 17
r = Tensor.randn(N, N).realize().max(axis=0, keepdim=True) > 1
a0 = r.where(1, 0)
a1 = r.where(2, 0)
helper_linearizer_opt([a0.max(0), a1.max(0)], [
[Opt(OptOps.PADTO, 0, 32)],
[Opt(OptOps.PADTO, 0, 32), Opt(OptOps.UPCAST, 0, 8),],
])
@unittest.skipUnless(Device[Device.DEFAULT].renderer.has_local, "test requires locals")
@unittest.skipUnless(Device[Device.DEFAULT].renderer.has_shared, "test requires shared")
def test_padto_group(self):
Tensor.manual_seed(0)
g0, g1, g2 = [UOp(Ops.DEFINE_GLOBAL, dtypes.float.ptr(), arg=i) for i in range(3)]
ld0 = UOp(Ops.LOAD, dtypes.float, src=(g1.view(ShapeTracker(views=(View(shape=(2, 1, 4, 1, 3, 4, 2, 6, 1, 3), strides=(0, 0, 0, 0, 0, 18, 0, 3, 0, 1), offset=0, mask=None, contiguous=False),))),)) # noqa: E501
ld1 = UOp(Ops.LOAD, dtypes.float, src=(g2.view(ShapeTracker(views=(View(shape=(2, 1, 4, 1, 3, 4, 2, 6, 1, 3), strides=(0, 0, 0, 0, 0, 0, 0, 0, 0, 0), offset=0, mask=None, contiguous=False),))),)) # noqa: E501
store = UOp(Ops.STORE, src=(g0.view(ShapeTracker(views=(View(shape=(1, 1, 1, 1, 1, 4, 1, 6, 1, 3), strides=(0, 0, 0, 0, 0, 18, 0, 3, 0, 1), offset=0, mask=None, contiguous=True),))), UOp(Ops.REDUCE_AXIS, dtypes.float, (ld0*ld1,), (Ops.ADD, (0, 2, 4, 6)),))) # noqa: E501
sink = UOp(Ops.SINK, src=(store,))
data1 = Tensor.randn(2, 1, 4, 1, 3, 4, 2, 6, 1, 3).realize()
data2 = Tensor.randn(2, 1, 4, 1, 3, 4, 2, 6, 1, 3).realize()
helper_linearizer_ast(sink, [data1, data2], opts=[
#[Opt(OptOps.PADTO, 0, 32), Opt(OptOps.GROUP, 0, 4)],
#[Opt(OptOps.PADTO, 0, 32), Opt(OptOps.UPCAST, 0, 8)],
#[Opt(OptOps.PADTO, 0, 32), Opt(OptOps.UPCAST, 0, 8), Opt(OptOps.GROUP, 0, 4)]
])
@unittest.skipUnless(Device[Device.DEFAULT].renderer.has_local, "test requires locals")
@unittest.skipUnless(Device[Device.DEFAULT].renderer.has_shared, "test requires shared")
def test_color_shapes_with_local(self):
N = 32
Tensor.manual_seed(1552)
a = Tensor.rand(N, N)
b = Tensor.rand(N, N)
r = a@b
opts_shapes = [
([Opt(OptOps.LOCAL, 0, 2)], [("blue",16),("blue",32),("cyan",2),("red",32)]),
([Opt(OptOps.LOCAL, 0, 2),Opt(OptOps.GROUP, 0, 2)], [("blue",16),("blue",32),("cyan",2),("green",2),("red",16)]),
# check to ensure local_dims are stable for full UNROLL of first_reduce
([Opt(OptOps.LOCAL, 0, 2),Opt(OptOps.UNROLL, 0, 0)], [("blue",16),("blue",32),("cyan",2),("magenta",32)]),
([Opt(OptOps.UNROLL, 0, 0),Opt(OptOps.LOCAL, 0, 2)], [("blue",16),("blue",32),("cyan",2),("magenta",32)]),
# check behavior for full UNROLL on an existing GROUP
([Opt(OptOps.LOCAL, 0, 2),Opt(OptOps.GROUP, 0, 0),Opt(OptOps.UNROLL, 0, 2)], [("blue",16),("blue",32),("cyan",2),("green",16),("magenta",2)]),
([Opt(OptOps.LOCAL, 0, 2),Opt(OptOps.GROUP, 0, 0),Opt(OptOps.UNROLL, 0, 0)], [("blue",16),("blue",32),("cyan",2),("magenta",32)]),
([Opt(OptOps.GROUP, 0, 0),Opt(OptOps.LOCAL, 0, 2),Opt(OptOps.UNROLL, 0, 0)], [("blue",16),("blue",32),("cyan",2),("magenta",32)]),
([Opt(OptOps.GROUP, 0, 2),Opt(OptOps.UNROLL, 0, 0)], [("blue",32),("blue",32),("red",16),("magenta",2)]),
]
helper_linearizer_opt(r, [x[0] for x in opts_shapes], color_sizes=[x[1] for x in opts_shapes])
if __name__ == '__main__':
unittest.main()