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
change
sglang
Commits
e11ab79e
Unverified
Commit
e11ab79e
authored
Oct 10, 2024
by
HAI
Committed by
GitHub
Oct 10, 2024
Browse files
[Performance, hardware] MoE tuning update to AMD MI300x GPUs (#1619)
parent
01fdb2f3
Changes
2
Hide whitespace changes
Inline
Side-by-side
Showing
2 changed files
with
353 additions
and
57 deletions
+353
-57
python/sglang/srt/layers/fused_moe/configs/E=8,N=4096,device_name=AMD_Instinct_MI300X,dtype=float8.json
...,N=4096,device_name=AMD_Instinct_MI300X,dtype=float8.json
+178
-57
python/sglang/srt/layers/fused_moe/configs/E=8,N=8192,device_name=AMD_Instinct_MI300X,dtype=float8.json
...,N=8192,device_name=AMD_Instinct_MI300X,dtype=float8.json
+175
-0
No files found.
python/sglang/srt/layers/fused_moe/configs/E=8,N=4096,device_name=AMD_Instinct_MI300X,dtype=float8.json
View file @
e11ab79e
{
"8"
:
{
"BLOCK_SIZE_M"
:
16
,
"BLOCK_SIZE_N"
:
64
,
"BLOCK_SIZE_K"
:
256
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
4
,
"num_stages"
:
0
,
"waves_per_eu"
:
1
,
"matrix_instr_nonkdim"
:
16
,
"kpack"
:
1
},
"16"
:
{
"BLOCK_SIZE_M"
:
32
,
"BLOCK_SIZE_N"
:
64
,
"BLOCK_SIZE_K"
:
256
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
8
,
"num_stages"
:
0
,
"waves_per_eu"
:
2
,
"matrix_instr_nonkdim"
:
16
,
"kpack"
:
2
},
"32"
:
{
"BLOCK_SIZE_M"
:
32
,
"BLOCK_SIZE_N"
:
64
,
"BLOCK_SIZE_K"
:
256
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
8
,
"num_stages"
:
0
,
"waves_per_eu"
:
1
,
"matrix_instr_nonkdim"
:
16
,
"kpack"
:
2
},
"2048"
:
{
"BLOCK_SIZE_M"
:
128
,
"BLOCK_SIZE_N"
:
128
,
"BLOCK_SIZE_K"
:
128
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
8
,
"num_stages"
:
0
,
"waves_per_eu"
:
2
,
"matrix_instr_nonkdim"
:
16
,
"kpack"
:
2
},
"32768"
:
{
"BLOCK_SIZE_M"
:
256
,
"BLOCK_SIZE_N"
:
256
,
"BLOCK_SIZE_K"
:
64
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
8
,
"num_stages"
:
0
,
"waves_per_eu"
:
0
,
"matrix_instr_nonkdim"
:
16
,
"kpack"
:
1
}
}
{
"4"
:
{
"BLOCK_SIZE_M"
:
16
,
"BLOCK_SIZE_N"
:
64
,
"BLOCK_SIZE_K"
:
256
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
4
,
"num_stages"
:
0
,
"waves_per_eu"
:
4
,
"matrix_instr_nonkdim"
:
16
,
"kpack"
:
2
},
"8"
:
{
"BLOCK_SIZE_M"
:
16
,
"BLOCK_SIZE_N"
:
64
,
"BLOCK_SIZE_K"
:
256
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
4
,
"num_stages"
:
0
,
"waves_per_eu"
:
1
,
"matrix_instr_nonkdim"
:
16
,
"kpack"
:
1
},
"16"
:
{
"BLOCK_SIZE_M"
:
32
,
"BLOCK_SIZE_N"
:
64
,
"BLOCK_SIZE_K"
:
256
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
8
,
"num_stages"
:
0
,
"waves_per_eu"
:
2
,
"matrix_instr_nonkdim"
:
16
,
"kpack"
:
2
},
"32"
:
{
"BLOCK_SIZE_M"
:
32
,
"BLOCK_SIZE_N"
:
64
,
"BLOCK_SIZE_K"
:
256
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
8
,
"num_stages"
:
0
,
"waves_per_eu"
:
1
,
"matrix_instr_nonkdim"
:
16
,
"kpack"
:
2
},
"64"
:
{
"BLOCK_SIZE_M"
:
32
,
"BLOCK_SIZE_N"
:
64
,
"BLOCK_SIZE_K"
:
256
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
4
,
"num_stages"
:
0
,
"waves_per_eu"
:
2
,
"matrix_instr_nonkdim"
:
16
,
"kpack"
:
2
},
"128"
:
{
"BLOCK_SIZE_M"
:
16
,
"BLOCK_SIZE_N"
:
64
,
"BLOCK_SIZE_K"
:
256
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
4
,
"num_stages"
:
0
,
"waves_per_eu"
:
1
,
"matrix_instr_nonkdim"
:
16
,
"kpack"
:
1
},
"256"
:
{
"BLOCK_SIZE_M"
:
16
,
"BLOCK_SIZE_N"
:
64
,
"BLOCK_SIZE_K"
:
256
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
4
,
"num_stages"
:
0
,
"waves_per_eu"
:
1
,
"matrix_instr_nonkdim"
:
16
,
"kpack"
:
1
},
"512"
:
{
"BLOCK_SIZE_M"
:
64
,
"BLOCK_SIZE_N"
:
64
,
"BLOCK_SIZE_K"
:
256
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
4
,
"num_stages"
:
0
,
"waves_per_eu"
:
2
,
"matrix_instr_nonkdim"
:
16
,
"kpack"
:
2
},
"1024"
:
{
"BLOCK_SIZE_M"
:
128
,
"BLOCK_SIZE_N"
:
128
,
"BLOCK_SIZE_K"
:
128
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
8
,
"num_stages"
:
0
,
"waves_per_eu"
:
4
,
"matrix_instr_nonkdim"
:
16
,
"kpack"
:
2
},
"2048"
:
{
"BLOCK_SIZE_M"
:
128
,
"BLOCK_SIZE_N"
:
128
,
"BLOCK_SIZE_K"
:
128
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
8
,
"num_stages"
:
0
,
"waves_per_eu"
:
2
,
"matrix_instr_nonkdim"
:
16
,
"kpack"
:
2
},
"4096"
:
{
"BLOCK_SIZE_M"
:
128
,
"BLOCK_SIZE_N"
:
128
,
"BLOCK_SIZE_K"
:
128
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
8
,
"num_stages"
:
0
,
"waves_per_eu"
:
2
,
"matrix_instr_nonkdim"
:
16
,
"kpack"
:
2
},
"8192"
:
{
"BLOCK_SIZE_M"
:
256
,
"BLOCK_SIZE_N"
:
256
,
"BLOCK_SIZE_K"
:
64
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
8
,
"num_stages"
:
0
,
"waves_per_eu"
:
2
,
"matrix_instr_nonkdim"
:
16
,
"kpack"
:
1
},
"16384"
:
{
"BLOCK_SIZE_M"
:
256
,
"BLOCK_SIZE_N"
:
256
,
"BLOCK_SIZE_K"
:
64
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
8
,
"num_stages"
:
0
,
"waves_per_eu"
:
1
,
"matrix_instr_nonkdim"
:
16
,
"kpack"
:
1
},
"32768"
:
{
"BLOCK_SIZE_M"
:
256
,
"BLOCK_SIZE_N"
:
256
,
"BLOCK_SIZE_K"
:
64
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
8
,
"num_stages"
:
0
,
"waves_per_eu"
:
0
,
"matrix_instr_nonkdim"
:
16
,
"kpack"
:
1
},
"65536"
:
{
"BLOCK_SIZE_M"
:
256
,
"BLOCK_SIZE_N"
:
256
,
"BLOCK_SIZE_K"
:
64
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
8
,
"num_stages"
:
0
,
"waves_per_eu"
:
1
,
"matrix_instr_nonkdim"
:
16
,
"kpack"
:
1
},
"131072"
:
{
"BLOCK_SIZE_M"
:
256
,
"BLOCK_SIZE_N"
:
128
,
"BLOCK_SIZE_K"
:
64
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
4
,
"num_stages"
:
0
,
"waves_per_eu"
:
2
,
"matrix_instr_nonkdim"
:
16
,
"kpack"
:
2
}
}
python/sglang/srt/layers/fused_moe/configs/E=8,N=8192,device_name=AMD_Instinct_MI300X,dtype=float8.json
0 → 100644
View file @
e11ab79e
{
"4"
:
{
"BLOCK_SIZE_M"
:
16
,
"BLOCK_SIZE_N"
:
64
,
"BLOCK_SIZE_K"
:
256
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
4
,
"num_stages"
:
0
,
"waves_per_eu"
:
4
,
"matrix_instr_nonkdim"
:
16
,
"kpack"
:
2
},
"8"
:
{
"BLOCK_SIZE_M"
:
16
,
"BLOCK_SIZE_N"
:
64
,
"BLOCK_SIZE_K"
:
256
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
4
,
"num_stages"
:
0
,
"waves_per_eu"
:
1
,
"matrix_instr_nonkdim"
:
16
,
"kpack"
:
1
},
"16"
:
{
"BLOCK_SIZE_M"
:
32
,
"BLOCK_SIZE_N"
:
64
,
"BLOCK_SIZE_K"
:
256
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
8
,
"num_stages"
:
0
,
"waves_per_eu"
:
2
,
"matrix_instr_nonkdim"
:
16
,
"kpack"
:
2
},
"32"
:
{
"BLOCK_SIZE_M"
:
32
,
"BLOCK_SIZE_N"
:
64
,
"BLOCK_SIZE_K"
:
256
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
8
,
"num_stages"
:
0
,
"waves_per_eu"
:
1
,
"matrix_instr_nonkdim"
:
16
,
"kpack"
:
2
},
"64"
:
{
"BLOCK_SIZE_M"
:
32
,
"BLOCK_SIZE_N"
:
64
,
"BLOCK_SIZE_K"
:
256
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
4
,
"num_stages"
:
0
,
"waves_per_eu"
:
2
,
"matrix_instr_nonkdim"
:
16
,
"kpack"
:
2
},
"128"
:
{
"BLOCK_SIZE_M"
:
16
,
"BLOCK_SIZE_N"
:
64
,
"BLOCK_SIZE_K"
:
256
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
4
,
"num_stages"
:
0
,
"waves_per_eu"
:
1
,
"matrix_instr_nonkdim"
:
16
,
"kpack"
:
1
},
"256"
:
{
"BLOCK_SIZE_M"
:
128
,
"BLOCK_SIZE_N"
:
256
,
"BLOCK_SIZE_K"
:
128
,
"GROUP_SIZE_M"
:
32
,
"num_warps"
:
8
,
"num_stages"
:
4
},
"512"
:
{
"BLOCK_SIZE_M"
:
64
,
"BLOCK_SIZE_N"
:
64
,
"BLOCK_SIZE_K"
:
256
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
4
,
"num_stages"
:
0
,
"waves_per_eu"
:
2
,
"matrix_instr_nonkdim"
:
16
,
"kpack"
:
2
},
"1024"
:
{
"BLOCK_SIZE_M"
:
128
,
"BLOCK_SIZE_N"
:
128
,
"BLOCK_SIZE_K"
:
128
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
8
,
"num_stages"
:
0
,
"waves_per_eu"
:
4
,
"matrix_instr_nonkdim"
:
16
,
"kpack"
:
2
},
"2048"
:
{
"BLOCK_SIZE_M"
:
128
,
"BLOCK_SIZE_N"
:
128
,
"BLOCK_SIZE_K"
:
128
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
8
,
"num_stages"
:
0
,
"waves_per_eu"
:
2
,
"matrix_instr_nonkdim"
:
16
,
"kpack"
:
2
},
"4096"
:
{
"BLOCK_SIZE_M"
:
128
,
"BLOCK_SIZE_N"
:
128
,
"BLOCK_SIZE_K"
:
128
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
8
,
"num_stages"
:
0
,
"waves_per_eu"
:
2
,
"matrix_instr_nonkdim"
:
16
,
"kpack"
:
2
},
"8192"
:
{
"BLOCK_SIZE_M"
:
256
,
"BLOCK_SIZE_N"
:
256
,
"BLOCK_SIZE_K"
:
64
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
8
,
"num_stages"
:
0
,
"waves_per_eu"
:
2
,
"matrix_instr_nonkdim"
:
16
,
"kpack"
:
1
},
"16384"
:
{
"BLOCK_SIZE_M"
:
256
,
"BLOCK_SIZE_N"
:
256
,
"BLOCK_SIZE_K"
:
64
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
8
,
"num_stages"
:
0
,
"waves_per_eu"
:
1
,
"matrix_instr_nonkdim"
:
16
,
"kpack"
:
1
},
"32768"
:
{
"BLOCK_SIZE_M"
:
256
,
"BLOCK_SIZE_N"
:
256
,
"BLOCK_SIZE_K"
:
64
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
8
,
"num_stages"
:
0
,
"waves_per_eu"
:
0
,
"matrix_instr_nonkdim"
:
16
,
"kpack"
:
1
},
"65536"
:
{
"BLOCK_SIZE_M"
:
256
,
"BLOCK_SIZE_N"
:
256
,
"BLOCK_SIZE_K"
:
64
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
8
,
"num_stages"
:
0
,
"waves_per_eu"
:
1
,
"matrix_instr_nonkdim"
:
16
,
"kpack"
:
1
},
"131072"
:
{
"BLOCK_SIZE_M"
:
256
,
"BLOCK_SIZE_N"
:
128
,
"BLOCK_SIZE_K"
:
64
,
"GROUP_SIZE_M"
:
1
,
"num_warps"
:
4
,
"num_stages"
:
0
,
"waves_per_eu"
:
2
,
"matrix_instr_nonkdim"
:
16
,
"kpack"
:
2
}
}
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