Skip to content
GitLab
Menu
Projects
Groups
Snippets
Loading...
Help
Help
Support
Community forum
Keyboard shortcuts
?
Submit feedback
Contribute to GitLab
Sign in
Toggle navigation
Menu
Open sidebar
OpenDAS
vllm_cscc
Commits
abe855d6
Unverified
Commit
abe855d6
authored
Jun 06, 2024
by
Philipp Moritz
Committed by
GitHub
Jun 06, 2024
Browse files
[Kernel] Retune Mixtral 8x22b configs for FP8 on H100 (#5294)
parent
4efff036
Changes
2
Show whitespace changes
Inline
Side-by-side
Showing
2 changed files
with
91 additions
and
91 deletions
+91
-91
vllm/model_executor/layers/fused_moe/configs/E=8,N=2048,device_name=NVIDIA_H100_80GB_HBM3,dtype=float8.json
...=2048,device_name=NVIDIA_H100_80GB_HBM3,dtype=float8.json
+55
-55
vllm/model_executor/layers/fused_moe/configs/E=8,N=4096,device_name=NVIDIA_H100_80GB_HBM3,dtype=float8.json
...=4096,device_name=NVIDIA_H100_80GB_HBM3,dtype=float8.json
+36
-36
No files found.
vllm/model_executor/layers/fused_moe/configs/E=8,N=2048,device_name=NVIDIA_H100_80GB_HBM3,dtype=float8.json
View file @
abe855d6
{
"1"
:
{
"BLOCK_SIZE_M"
:
1
6
,
"BLOCK_SIZE_N"
:
32
,
"BLOCK_SIZE_K"
:
64
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
8
,
"num_stages"
:
3
"BLOCK_SIZE_M"
:
6
4
,
"BLOCK_SIZE_N"
:
128
,
"BLOCK_SIZE_K"
:
128
,
"GROUP_SIZE_M"
:
1
6
,
"num_warps"
:
4
,
"num_stages"
:
5
},
"2"
:
{
"BLOCK_SIZE_M"
:
1
6
,
"BLOCK_SIZE_N"
:
6
4
,
"BLOCK_SIZE_K"
:
64
,
"BLOCK_SIZE_M"
:
6
4
,
"BLOCK_SIZE_N"
:
25
6
,
"BLOCK_SIZE_K"
:
128
,
"GROUP_SIZE_M"
:
16
,
"num_warps"
:
4
,
"num_stages"
:
5
},
"4"
:
{
"BLOCK_SIZE_M"
:
64
,
"BLOCK_SIZE_N"
:
3
2
,
"BLOCK_SIZE_K"
:
256
,
"GROUP_SIZE_M"
:
64
,
"num_warps"
:
8
,
"num_stages"
:
5
"BLOCK_SIZE_N"
:
2
56
,
"BLOCK_SIZE_K"
:
128
,
"GROUP_SIZE_M"
:
32
,
"num_warps"
:
4
,
"num_stages"
:
4
},
"8"
:
{
"BLOCK_SIZE_M"
:
1
6
,
"BLOCK_SIZE_N"
:
6
4
,
"BLOCK_SIZE_K"
:
256
,
"BLOCK_SIZE_M"
:
6
4
,
"BLOCK_SIZE_N"
:
25
6
,
"BLOCK_SIZE_K"
:
128
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
4
,
"num_stages"
:
3
"num_stages"
:
5
},
"16"
:
{
"BLOCK_SIZE_M"
:
128
,
"BLOCK_SIZE_M"
:
64
,
"BLOCK_SIZE_N"
:
64
,
"BLOCK_SIZE_K"
:
6
4
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
8
,
"BLOCK_SIZE_K"
:
25
6
,
"GROUP_SIZE_M"
:
32
,
"num_warps"
:
4
,
"num_stages"
:
3
},
"24"
:
{
"BLOCK_SIZE_M"
:
64
,
"BLOCK_SIZE_N"
:
25
6
,
"BLOCK_SIZE_N"
:
6
4
,
"BLOCK_SIZE_K"
:
128
,
"GROUP_SIZE_M"
:
64
,
"num_warps"
:
8
,
"num_stages"
:
2
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
4
,
"num_stages"
:
3
},
"32"
:
{
"BLOCK_SIZE_M"
:
64
,
"BLOCK_SIZE_N"
:
25
6
,
"BLOCK_SIZE_K"
:
64
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
8
,
"BLOCK_SIZE_N"
:
6
4
,
"BLOCK_SIZE_K"
:
128
,
"GROUP_SIZE_M"
:
1
6
,
"num_warps"
:
4
,
"num_stages"
:
3
},
"48"
:
{
"BLOCK_SIZE_M"
:
64
,
"BLOCK_SIZE_N"
:
25
6
,
"BLOCK_SIZE_N"
:
6
4
,
"BLOCK_SIZE_K"
:
128
,
"GROUP_SIZE_M"
:
64
,
"num_warps"
:
8
,
"num_stages"
:
4
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
4
,
"num_stages"
:
3
},
"64"
:
{
"BLOCK_SIZE_M"
:
128
,
"BLOCK_SIZE_N"
:
25
6
,
"BLOCK_SIZE_M"
:
64
,
"BLOCK_SIZE_N"
:
6
4
,
"BLOCK_SIZE_K"
:
128
,
"GROUP_SIZE_M"
:
6
4
,
"num_warps"
:
8
,
"num_stages"
:
2
"GROUP_SIZE_M"
:
1
6
,
"num_warps"
:
4
,
"num_stages"
:
3
},
"96"
:
{
"BLOCK_SIZE_M"
:
64
,
"BLOCK_SIZE_N"
:
25
6
,
"BLOCK_SIZE_N"
:
6
4
,
"BLOCK_SIZE_K"
:
128
,
"GROUP_SIZE_M"
:
16
,
"num_warps"
:
4
,
"num_stages"
:
4
"num_stages"
:
3
},
"128"
:
{
"BLOCK_SIZE_M"
:
64
,
"BLOCK_SIZE_N"
:
25
6
,
"BLOCK_SIZE_K"
:
64
,
"GROUP_SIZE_M"
:
32
,
"num_warps"
:
8
,
"BLOCK_SIZE_N"
:
6
4
,
"BLOCK_SIZE_K"
:
128
,
"GROUP_SIZE_M"
:
16
,
"num_warps"
:
4
,
"num_stages"
:
3
},
"256"
:
{
"BLOCK_SIZE_M"
:
128
,
"BLOCK_SIZE_N"
:
64
,
"BLOCK_SIZE_M"
:
64
,
"BLOCK_SIZE_N"
:
128
,
"BLOCK_SIZE_K"
:
128
,
"GROUP_SIZE_M"
:
64
,
"num_warps"
:
8
,
"num_stages"
:
4
"num_warps"
:
4
,
"num_stages"
:
3
},
"512"
:
{
"BLOCK_SIZE_M"
:
64
,
"BLOCK_SIZE_N"
:
256
,
"BLOCK_SIZE_M"
:
128
,
"BLOCK_SIZE_N"
:
128
,
"BLOCK_SIZE_K"
:
128
,
"GROUP_SIZE_M"
:
64
,
"num_warps"
:
4
,
"num_stages"
:
3
"num_warps"
:
8
,
"num_stages"
:
5
},
"1024"
:
{
"BLOCK_SIZE_M"
:
128
,
"BLOCK_SIZE_N"
:
256
,
"BLOCK_SIZE_K"
:
128
,
"GROUP_SIZE_M"
:
32
,
"GROUP_SIZE_M"
:
64
,
"num_warps"
:
8
,
"num_stages"
:
4
},
...
...
@@ -139,7 +139,7 @@
"BLOCK_SIZE_M"
:
128
,
"BLOCK_SIZE_N"
:
256
,
"BLOCK_SIZE_K"
:
128
,
"GROUP_SIZE_M"
:
32
,
"GROUP_SIZE_M"
:
16
,
"num_warps"
:
8
,
"num_stages"
:
4
}
...
...
vllm/model_executor/layers/fused_moe/configs/E=8,N=4096,device_name=NVIDIA_H100_80GB_HBM3,dtype=float8.json
View file @
abe855d6
{
"1"
:
{
"BLOCK_SIZE_M"
:
1
6
,
"BLOCK_SIZE_N"
:
32
,
"BLOCK_SIZE_K"
:
64
,
"BLOCK_SIZE_M"
:
6
4
,
"BLOCK_SIZE_N"
:
128
,
"BLOCK_SIZE_K"
:
128
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
4
,
"num_stages"
:
5
},
"2"
:
{
"BLOCK_SIZE_M"
:
64
,
"BLOCK_SIZE_N"
:
64
,
"BLOCK_SIZE_K"
:
64
,
"GROUP_SIZE_M"
:
1
6
,
"BLOCK_SIZE_N"
:
128
,
"BLOCK_SIZE_K"
:
128
,
"GROUP_SIZE_M"
:
6
4
,
"num_warps"
:
4
,
"num_stages"
:
4
},
...
...
@@ -20,59 +20,59 @@
"BLOCK_SIZE_N"
:
128
,
"BLOCK_SIZE_K"
:
128
,
"GROUP_SIZE_M"
:
64
,
"num_warps"
:
8
,
"num_stages"
:
2
"num_warps"
:
4
,
"num_stages"
:
3
},
"8"
:
{
"BLOCK_SIZE_M"
:
64
,
"BLOCK_SIZE_N"
:
25
6
,
"BLOCK_SIZE_N"
:
6
4
,
"BLOCK_SIZE_K"
:
128
,
"GROUP_SIZE_M"
:
32
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
4
,
"num_stages"
:
3
"num_stages"
:
5
},
"16"
:
{
"BLOCK_SIZE_M"
:
64
,
"BLOCK_SIZE_N"
:
128
,
"BLOCK_SIZE_K"
:
256
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
8
,
"num_stages"
:
2
"BLOCK_SIZE_N"
:
64
,
"BLOCK_SIZE_K"
:
128
,
"GROUP_SIZE_M"
:
1
6
,
"num_warps"
:
4
,
"num_stages"
:
5
},
"24"
:
{
"BLOCK_SIZE_M"
:
64
,
"BLOCK_SIZE_N"
:
128
,
"BLOCK_SIZE_K"
:
256
,
"GROUP_SIZE_M"
:
1
,
"BLOCK_SIZE_N"
:
64
,
"BLOCK_SIZE_K"
:
128
,
"GROUP_SIZE_M"
:
1
6
,
"num_warps"
:
4
,
"num_stages"
:
3
"num_stages"
:
5
},
"32"
:
{
"BLOCK_SIZE_M"
:
64
,
"BLOCK_SIZE_N"
:
64
,
"BLOCK_SIZE_K"
:
128
,
"BLOCK_SIZE_N"
:
128
,
"BLOCK_SIZE_K"
:
256
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
8
,
"num_stages"
:
3
"num_warps"
:
4
,
"num_stages"
:
4
},
"48"
:
{
"BLOCK_SIZE_M"
:
64
,
"BLOCK_SIZE_N"
:
128
,
"BLOCK_SIZE_N"
:
64
,
"BLOCK_SIZE_K"
:
128
,
"GROUP_SIZE_M"
:
64
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
4
,
"num_stages"
:
2
"num_stages"
:
3
},
"64"
:
{
"BLOCK_SIZE_M"
:
64
,
"BLOCK_SIZE_N"
:
64
,
"BLOCK_SIZE_K"
:
256
,
"GROUP_SIZE_M"
:
1
6
,
"BLOCK_SIZE_K"
:
128
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
4
,
"num_stages"
:
2
"num_stages"
:
3
},
"96"
:
{
"BLOCK_SIZE_M"
:
128
,
"BLOCK_SIZE_M"
:
64
,
"BLOCK_SIZE_N"
:
64
,
"BLOCK_SIZE_K"
:
128
,
"GROUP_SIZE_M"
:
1
,
...
...
@@ -81,11 +81,11 @@
},
"128"
:
{
"BLOCK_SIZE_M"
:
64
,
"BLOCK_SIZE_N"
:
128
,
"BLOCK_SIZE_K"
:
256
,
"BLOCK_SIZE_N"
:
64
,
"BLOCK_SIZE_K"
:
128
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
8
,
"num_stages"
:
2
"num_warps"
:
4
,
"num_stages"
:
3
},
"256"
:
{
"BLOCK_SIZE_M"
:
128
,
...
...
@@ -93,7 +93,7 @@
"BLOCK_SIZE_K"
:
128
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
8
,
"num_stages"
:
5
"num_stages"
:
3
},
"512"
:
{
"BLOCK_SIZE_M"
:
128
,
...
...
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