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
norm
vllm
Commits
dd823f7f
Commit
dd823f7f
authored
Mar 25, 2024
by
zhuwenwen
Browse files
add bf16
parent
9fed1f5d
Changes
2
Hide whitespace changes
Inline
Side-by-side
Showing
2 changed files
with
8 additions
and
8 deletions
+8
-8
csrc/cache_kernels.cu
csrc/cache_kernels.cu
+4
-4
csrc/dispatch_utils.h
csrc/dispatch_utils.h
+4
-4
No files found.
csrc/cache_kernels.cu
View file @
dd823f7f
...
...
@@ -251,8 +251,8 @@ void reshape_and_cache(
CALL_RESHAPE_AND_CACHE
(
float
,
float
,
false
);
}
else
if
(
key
.
dtype
()
==
at
::
ScalarType
::
Half
)
{
CALL_RESHAPE_AND_CACHE
(
uint16_t
,
uint16_t
,
false
);
//
} else if (key.dtype() == at::ScalarType::BFloat16) {
//
CALL_RESHAPE_AND_CACHE(__nv_bfloat16, __nv_bfloat16, false);
}
else
if
(
key
.
dtype
()
==
at
::
ScalarType
::
BFloat16
)
{
CALL_RESHAPE_AND_CACHE
(
__nv_bfloat16
,
__nv_bfloat16
,
false
);
}
// } else if (kv_cache_dtype == "fp8_e5m2") {
// if (key.dtype() == at::ScalarType::Float) {
...
...
@@ -308,8 +308,8 @@ void convert_fp8_e5m2(
CALL_CONVERT_FP8_E5M2
(
uint8_t
,
float
);
}
else
if
(
src_cache
.
dtype
()
==
at
::
ScalarType
::
Half
)
{
CALL_CONVERT_FP8_E5M2
(
uint8_t
,
uint16_t
);
//
} else if (src_cache.dtype() == at::ScalarType::BFloat16) {
//
CALL_CONVERT_FP8_E5M2(uint8_t, __nv_bfloat16);
}
else
if
(
src_cache
.
dtype
()
==
at
::
ScalarType
::
BFloat16
)
{
CALL_CONVERT_FP8_E5M2
(
uint8_t
,
__nv_bfloat16
);
}
else
if
(
dst_cache
.
dtype
()
==
at
::
ScalarType
::
Float
)
{
CALL_CONVERT_FP8_E5M2
(
float
,
uint8_t
);
}
else
if
(
dst_cache
.
dtype
()
==
at
::
ScalarType
::
Half
)
{
...
...
csrc/dispatch_utils.h
View file @
dd823f7f
...
...
@@ -8,8 +8,8 @@
#define VLLM_DISPATCH_CASE_FLOATING_TYPES(...) \
AT_DISPATCH_CASE(at::ScalarType::Float, __VA_ARGS__) \
AT_DISPATCH_CASE(at::ScalarType::Half, __VA_ARGS__)
//
AT_DISPATCH_CASE(at::ScalarType::BFloat16, __VA_ARGS__)
AT_DISPATCH_CASE(at::ScalarType::Half, __VA_ARGS__)
\
AT_DISPATCH_CASE(at::ScalarType::BFloat16, __VA_ARGS__)
#define VLLM_DISPATCH_FLOATING_TYPES(TYPE, NAME, ...) \
AT_DISPATCH_SWITCH( \
...
...
@@ -17,8 +17,8 @@
#define VLLM_DISPATCH_CASE_FLOATING_AND_BYTE_TYPES(...) \
AT_DISPATCH_CASE(at::ScalarType::Float, __VA_ARGS__) \
AT_DISPATCH_CASE(at::ScalarType::Half, __VA_ARGS__)
//
AT_DISPATCH_CASE(at::ScalarType::BFloat16, __VA_ARGS__)
\
AT_DISPATCH_CASE(at::ScalarType::Half, __VA_ARGS__)
\
AT_DISPATCH_CASE(at::ScalarType::BFloat16, __VA_ARGS__)
// AT_DISPATCH_CASE(at::ScalarType::Byte, __VA_ARGS__)
#define VLLM_DISPATCH_FLOATING_AND_BYTE_TYPES(TYPE, NAME, ...) \
...
...
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