github/tinygrad
1
1
Fork 0
mirror of https://github.com/tinygrad/tinygrad.git synced 2026-04-07 03:00:26 -04:00
Code Issues Packages Projects Releases Wiki Activity

Fix amx shape [run_process_replay] (#6524)

Browse Source 
* fix amx shape (sz,sz,sz) -> (sz,sz,1)

* revert check
This commit is contained in:
ignaciosica
2024-09-15 22:49:55 -03:00
committed by GitHub
parent 1683b274b6
commit c447ec2190
3 changed files with 13 additions and 11 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
test/test_linearizer.py
View File

@@ -1040,7 +1040,8 @@ class TestLinearizer(unittest.TestCase):
def test_tensor_cores(self):
for tc in Device[Device.DEFAULT].renderer.tensor_cores:
if (getenv("EMULATE_CUDA") or getenv("EMULATE_INTEL")) and (tc.dtype_in == dtypes.bfloat16 or tc.dtype_out == dtypes.bfloat16): continue
helper_tc_allclose(tc.dims[0], tc.dims[1], tc.dims[2], tc.dtype_in, tc.dtype_out, axis=0, tc_opt=0)
# 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_padded(self):
@@ -1061,7 +1062,8 @@ class TestLinearizer(unittest.TestCase):
# 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)
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)
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)
# check correctness
helper_tc_allclose(tc.dims[0]+pad, tc.dims[1]+pad, tc.dims[2]+pad, tc.dtype_in, tc.dtype_out, tc_opt=2)