leavelet/cutlass
1
0
Fork 0
Code Issues Pull Requests Actions 26 Packages Projects Releases Wiki Activity

Add couple configs into generator.py for mixed input MM (#1350)

Browse Source 
* Add couple configs into generator.py for mixed input MM

* change one unit test name; reenable 128x32 in the profiler

* Added U8/BF16 tests.

---------

Co-authored-by: Haicheng Wu <haichengw@nvidia.com>
Co-authored-by: Haicheng Wu <57973641+hwu36@users.noreply.github.com>
This commit is contained in:
Aleksandar Samardžić
2024-08-16 00:59:29 -04:00
committed by GitHub
co-authored by Haicheng Wu Haicheng Wu
parent 865be73a97
commit 3f084f7f3c
21 changed files with 1506 additions and 80 deletions
Download Patch File Download Diff File Expand all files Collapse all files
test/unit/gemm/device/testbed_universal.h
+7 -2
View File
@@ -112,8 +112,13 @@ struct TestbedUniversal {
scope_max = is_unsigned_int ? 2 : 1;
scope_min = is_unsigned_int ? 0 : -1;
} else if (bits_output == 16) {
scope_max = is_unsigned_int ? 10 : 5;
scope_min = is_unsigned_int ? 0 : -5;
constexpr auto u8_bf16 =
(cutlass::platform::is_same<ElementA, uint8_t>::value &&
cutlass::platform::is_same<ElementB, cutlass::bfloat16_t>::value) ||
(cutlass::platform::is_same<ElementA, cutlass::bfloat16_t>::value &&
cutlass::platform::is_same<ElementB, uint8_t>::value);
scope_max = is_unsigned_int ? 10 : (u8_bf16 ? 3 : 5);
scope_min = is_unsigned_int ? 0 : (u8_bf16 ? -3 : -5);
} else {
scope_max = 8;
scope_min = -8;