Skip to content

Add float4_e2m1fn_x2 support for concat #2315

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 2 commits into from
Nov 18, 2025
Merged

Add float4_e2m1fn_x2 support for concat #2315

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
42c644e
Add float4_e2m1fn_x2 support for concat
CuiYifeng Nov 10, 2025
d3f73a9
Add FP4 UT
CuiYifeng Nov 10, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
3 changes: 2 additions & 1 deletion src/ATen/native/xpu/sycl/Shape.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -395,7 +395,8 @@ void cat_out_kernel(
kBool,
kBFloat16,
AT_EXPAND(AT_FLOAT8_TYPES),
AT_EXPAND(AT_BAREBONES_UNSIGNED_TYPES));
AT_EXPAND(AT_BAREBONES_UNSIGNED_TYPES),
kFloat4_e2m1fn_x2);
} else {
offset = 0;
for (j = 0; j < numInputs; j++) {
Expand Down
18 changes: 18 additions & 0 deletions test/regressions/test_cat.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -7,6 +7,9 @@
from torch.testing._internal.common_dtype import float8_types_and
from torch.testing._internal.common_utils import run_tests, TestCase

cpu_device = torch.device("cpu")
xpu_device = torch.device("xpu")


class TestTorchMethod(TestCase):
def _create_input_tensors(self, shape, dtype, memory_format=None):
Expand Down Expand Up @@ -61,6 +64,21 @@ def test_cat_simple(self, dtype):

self._test_cat_float8_core(tensors, dim, dtype)

def _float4_dummy_tensor(self, shape, device):
data = torch.ones(shape, dtype=torch.uint8, device=device)
return data.view(torch.float4_e2m1fn_x2)

def test_cat_float4_simple(self):
input_cpu1 = self._float4_dummy_tensor([2, 2, 6], device=cpu_device)
input_cpu2 = self._float4_dummy_tensor([2, 2, 6], device=cpu_device)
output_cpu = torch.stack([input_cpu1, input_cpu2]).view(torch.uint8)

input_xpu1 = self._float4_dummy_tensor([2, 2, 6], device=xpu_device)
input_xpu2 = self._float4_dummy_tensor([2, 2, 6], device=xpu_device)
output_xpu = torch.stack([input_xpu1, input_xpu2]).view(torch.uint8)

self.assertEqual(output_xpu, output_cpu)

def test_cat_8d(self, dtype=torch.float):
input1 = torch.randn([256, 8, 8, 3, 3, 3, 3], dtype=dtype)
input2 = torch.randn([256, 8, 8, 3, 3, 3, 3], dtype=dtype)
Expand Down