Skip to content
GitLab
Menu
Projects
Groups
Snippets
Loading...
Help
Help
Support
Community forum
Keyboard shortcuts
?
Submit feedback
Contribute to GitLab
Sign in / Register
Toggle navigation
Menu
Open sidebar
yangql
composable_kernel-1
Commits
03eaee6a
Unverified
Commit
03eaee6a
authored
Apr 15, 2023
by
Rostyslav Geyyer
Committed by
GitHub
Apr 15, 2023
Browse files
Add more macros to turn on/off denorm fix (#678)
Co-authored-by:
Rosty Geyyer
<
rosty.geyyer@amd.com
>
parent
e85178b4
Changes
2
Show whitespace changes
Inline
Side-by-side
Showing
2 changed files
with
2 additions
and
2 deletions
+2
-2
include/ck/tensor_operation/gpu/grid/gridwise_gemm_xdlops_bwd_weight.hpp
...or_operation/gpu/grid/gridwise_gemm_xdlops_bwd_weight.hpp
+1
-1
include/ck/tensor_operation/gpu/grid/gridwise_gemm_xdlops_v2r3.hpp
...k/tensor_operation/gpu/grid/gridwise_gemm_xdlops_v2r3.hpp
+1
-1
No files found.
include/ck/tensor_operation/gpu/grid/gridwise_gemm_xdlops_bwd_weight.hpp
View file @
03eaee6a
...
@@ -265,7 +265,7 @@ struct GridwiseGemm_bk0mk1_bk0nk1_mn_xdlops_bwd_weight
...
@@ -265,7 +265,7 @@ struct GridwiseGemm_bk0mk1_bk0nk1_mn_xdlops_bwd_weight
// we convert fp16->fp32->bf16 and execute bf16 mfma instruction
// we convert fp16->fp32->bf16 and execute bf16 mfma instruction
// when mfma if fixed, remove this section and update
// when mfma if fixed, remove this section and update
// FloatABAdjusted -> FloatAB throughout this file
// FloatABAdjusted -> FloatAB throughout this file
#if defined(__gfx90a__)
#if
CK_WORKAROUND_DENORM_FIX &&
defined(__gfx90a__)
using
FloatABAdjusted
=
conditional_t
<
is_same_v
<
FloatAB
,
ck
::
half_t
>
,
ck
::
bhalf_t
,
FloatAB
>
;
using
FloatABAdjusted
=
conditional_t
<
is_same_v
<
FloatAB
,
ck
::
half_t
>
,
ck
::
bhalf_t
,
FloatAB
>
;
#else
#else
using
FloatABAdjusted
=
FloatAB
;
using
FloatABAdjusted
=
FloatAB
;
...
...
include/ck/tensor_operation/gpu/grid/gridwise_gemm_xdlops_v2r3.hpp
View file @
03eaee6a
...
@@ -135,7 +135,7 @@ struct GridwiseGemm_k0mk1_k0nk1_mn_xdlops_v2r3
...
@@ -135,7 +135,7 @@ struct GridwiseGemm_k0mk1_k0nk1_mn_xdlops_v2r3
// we convert fp16->fp32->bf16 and execute bf16 mfma instruction
// we convert fp16->fp32->bf16 and execute bf16 mfma instruction
// when mfma if fixed, remove this section and update
// when mfma if fixed, remove this section and update
// FloatABAdjusted -> FloatAB throughout this file
// FloatABAdjusted -> FloatAB throughout this file
#if defined(__gfx90a__)
#if
CK_WORKAROUND_DENORM_FIX &&
defined(__gfx90a__)
using
FloatABAdjusted
=
conditional_t
<
is_same_v
<
FloatAB
,
ck
::
half_t
>
,
ck
::
bhalf_t
,
FloatAB
>
;
using
FloatABAdjusted
=
conditional_t
<
is_same_v
<
FloatAB
,
ck
::
half_t
>
,
ck
::
bhalf_t
,
FloatAB
>
;
#else
#else
using
FloatABAdjusted
=
FloatAB
;
using
FloatABAdjusted
=
FloatAB
;
...
...
Write
Preview
Markdown
is supported
0%
Try again
or
attach a new file
.
Attach a file
Cancel
You are about to add
0
people
to the discussion. Proceed with caution.
Finish editing this message first!
Cancel
Please
register
or
sign in
to comment