-
Notifications
You must be signed in to change notification settings - Fork 588
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
[Fused moe] add tuning fused configs for qwen2 57b and mixtral 8x7b (#…
- Loading branch information
Showing
2 changed files
with
292 additions
and
0 deletions.
There are no files selected for viewing
146 changes: 146 additions & 0 deletions
146
...sed_moe_triton/configs/E=64,N=640,device_name=NVIDIA_GeForce_RTX_4090,dtype=fp8_w8a8.json
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,146 @@ | ||
| { | ||
| "1": { | ||
| "BLOCK_SIZE_M": 16, | ||
| "BLOCK_SIZE_N": 32, | ||
| "BLOCK_SIZE_K": 128, | ||
| "GROUP_SIZE_M": 1, | ||
| "num_warps": 4, | ||
| "num_stages": 3 | ||
| }, | ||
| "2": { | ||
| "BLOCK_SIZE_M": 16, | ||
| "BLOCK_SIZE_N": 32, | ||
| "BLOCK_SIZE_K": 128, | ||
| "GROUP_SIZE_M": 32, | ||
| "num_warps": 4, | ||
| "num_stages": 2 | ||
| }, | ||
| "4": { | ||
| "BLOCK_SIZE_M": 16, | ||
| "BLOCK_SIZE_N": 64, | ||
| "BLOCK_SIZE_K": 256, | ||
| "GROUP_SIZE_M": 16, | ||
| "num_warps": 4, | ||
| "num_stages": 3 | ||
| }, | ||
| "8": { | ||
| "BLOCK_SIZE_M": 32, | ||
| "BLOCK_SIZE_N": 64, | ||
| "BLOCK_SIZE_K": 256, | ||
| "GROUP_SIZE_M": 32, | ||
| "num_warps": 8, | ||
| "num_stages": 3 | ||
| }, | ||
| "16": { | ||
| "BLOCK_SIZE_M": 16, | ||
| "BLOCK_SIZE_N": 32, | ||
| "BLOCK_SIZE_K": 256, | ||
| "GROUP_SIZE_M": 16, | ||
| "num_warps": 4, | ||
| "num_stages": 5 | ||
| }, | ||
| "24": { | ||
| "BLOCK_SIZE_M": 16, | ||
| "BLOCK_SIZE_N": 64, | ||
| "BLOCK_SIZE_K": 256, | ||
| "GROUP_SIZE_M": 64, | ||
| "num_warps": 4, | ||
| "num_stages": 2 | ||
| }, | ||
| "32": { | ||
| "BLOCK_SIZE_M": 16, | ||
| "BLOCK_SIZE_N": 32, | ||
| "BLOCK_SIZE_K": 256, | ||
| "GROUP_SIZE_M": 1, | ||
| "num_warps": 8, | ||
| "num_stages": 3 | ||
| }, | ||
| "48": { | ||
| "BLOCK_SIZE_M": 16, | ||
| "BLOCK_SIZE_N": 64, | ||
| "BLOCK_SIZE_K": 256, | ||
| "GROUP_SIZE_M": 64, | ||
| "num_warps": 8, | ||
| "num_stages": 2 | ||
| }, | ||
| "64": { | ||
| "BLOCK_SIZE_M": 32, | ||
| "BLOCK_SIZE_N": 32, | ||
| "BLOCK_SIZE_K": 256, | ||
| "GROUP_SIZE_M": 1, | ||
| "num_warps": 4, | ||
| "num_stages": 2 | ||
| }, | ||
| "96": { | ||
| "BLOCK_SIZE_M": 16, | ||
| "BLOCK_SIZE_N": 32, | ||
| "BLOCK_SIZE_K": 128, | ||
| "GROUP_SIZE_M": 16, | ||
| "num_warps": 4, | ||
| "num_stages": 3 | ||
| }, | ||
| "128": { | ||
| "BLOCK_SIZE_M": 32, | ||
| "BLOCK_SIZE_N": 64, | ||
| "BLOCK_SIZE_K": 256, | ||
| "GROUP_SIZE_M": 16, | ||
| "num_warps": 8, | ||
| "num_stages": 2 | ||
| }, | ||
| "256": { | ||
| "BLOCK_SIZE_M": 32, | ||
| "BLOCK_SIZE_N": 64, | ||
| "BLOCK_SIZE_K": 128, | ||
| "GROUP_SIZE_M": 16, | ||
| "num_warps": 8, | ||
| "num_stages": 3 | ||
| }, | ||
| "512": { | ||
| "BLOCK_SIZE_M": 32, | ||
| "BLOCK_SIZE_N": 64, | ||
| "BLOCK_SIZE_K": 128, | ||
| "GROUP_SIZE_M": 32, | ||
| "num_warps": 4, | ||
| "num_stages": 4 | ||
| }, | ||
| "1024": { | ||
| "BLOCK_SIZE_M": 64, | ||
| "BLOCK_SIZE_N": 128, | ||
| "BLOCK_SIZE_K": 128, | ||
| "GROUP_SIZE_M": 16, | ||
| "num_warps": 8, | ||
| "num_stages": 2 | ||
| }, | ||
| "1536": { | ||
| "BLOCK_SIZE_M": 128, | ||
| "BLOCK_SIZE_N": 128, | ||
| "BLOCK_SIZE_K": 64, | ||
| "GROUP_SIZE_M": 32, | ||
| "num_warps": 8, | ||
| "num_stages": 3 | ||
| }, | ||
| "2048": { | ||
| "BLOCK_SIZE_M": 128, | ||
| "BLOCK_SIZE_N": 128, | ||
| "BLOCK_SIZE_K": 64, | ||
| "GROUP_SIZE_M": 32, | ||
| "num_warps": 8, | ||
| "num_stages": 3 | ||
| }, | ||
| "3072": { | ||
| "BLOCK_SIZE_M": 128, | ||
| "BLOCK_SIZE_N": 128, | ||
| "BLOCK_SIZE_K": 64, | ||
| "GROUP_SIZE_M": 64, | ||
| "num_warps": 8, | ||
| "num_stages": 3 | ||
| }, | ||
| "4096": { | ||
| "BLOCK_SIZE_M": 128, | ||
| "BLOCK_SIZE_N": 128, | ||
| "BLOCK_SIZE_K": 64, | ||
| "GROUP_SIZE_M": 1, | ||
| "num_warps": 8, | ||
| "num_stages": 3 | ||
| } | ||
| } |
146 changes: 146 additions & 0 deletions
146
...sed_moe_triton/configs/E=8,N=3584,device_name=NVIDIA_GeForce_RTX_4090,dtype=fp8_w8a8.json
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,146 @@ | ||
| { | ||
| "1": { | ||
| "BLOCK_SIZE_M": 32, | ||
| "BLOCK_SIZE_N": 64, | ||
| "BLOCK_SIZE_K": 256, | ||
| "GROUP_SIZE_M": 1, | ||
| "num_warps": 8, | ||
| "num_stages": 3 | ||
| }, | ||
| "2": { | ||
| "BLOCK_SIZE_M": 16, | ||
| "BLOCK_SIZE_N": 64, | ||
| "BLOCK_SIZE_K": 256, | ||
| "GROUP_SIZE_M": 64, | ||
| "num_warps": 4, | ||
| "num_stages": 2 | ||
| }, | ||
| "4": { | ||
| "BLOCK_SIZE_M": 16, | ||
| "BLOCK_SIZE_N": 32, | ||
| "BLOCK_SIZE_K": 128, | ||
| "GROUP_SIZE_M": 1, | ||
| "num_warps": 4, | ||
| "num_stages": 4 | ||
| }, | ||
| "8": { | ||
| "BLOCK_SIZE_M": 32, | ||
| "BLOCK_SIZE_N": 32, | ||
| "BLOCK_SIZE_K": 128, | ||
| "GROUP_SIZE_M": 64, | ||
| "num_warps": 8, | ||
| "num_stages": 5 | ||
| }, | ||
| "16": { | ||
| "BLOCK_SIZE_M": 16, | ||
| "BLOCK_SIZE_N": 64, | ||
| "BLOCK_SIZE_K": 256, | ||
| "GROUP_SIZE_M": 1, | ||
| "num_warps": 4, | ||
| "num_stages": 2 | ||
| }, | ||
| "24": { | ||
| "BLOCK_SIZE_M": 16, | ||
| "BLOCK_SIZE_N": 32, | ||
| "BLOCK_SIZE_K": 256, | ||
| "GROUP_SIZE_M": 32, | ||
| "num_warps": 4, | ||
| "num_stages": 3 | ||
| }, | ||
| "32": { | ||
| "BLOCK_SIZE_M": 16, | ||
| "BLOCK_SIZE_N": 64, | ||
| "BLOCK_SIZE_K": 256, | ||
| "GROUP_SIZE_M": 1, | ||
| "num_warps": 4, | ||
| "num_stages": 2 | ||
| }, | ||
| "48": { | ||
| "BLOCK_SIZE_M": 32, | ||
| "BLOCK_SIZE_N": 32, | ||
| "BLOCK_SIZE_K": 256, | ||
| "GROUP_SIZE_M": 16, | ||
| "num_warps": 4, | ||
| "num_stages": 2 | ||
| }, | ||
| "64": { | ||
| "BLOCK_SIZE_M": 32, | ||
| "BLOCK_SIZE_N": 32, | ||
| "BLOCK_SIZE_K": 256, | ||
| "GROUP_SIZE_M": 32, | ||
| "num_warps": 4, | ||
| "num_stages": 2 | ||
| }, | ||
| "96": { | ||
| "BLOCK_SIZE_M": 32, | ||
| "BLOCK_SIZE_N": 32, | ||
| "BLOCK_SIZE_K": 256, | ||
| "GROUP_SIZE_M": 32, | ||
| "num_warps": 4, | ||
| "num_stages": 2 | ||
| }, | ||
| "128": { | ||
| "BLOCK_SIZE_M": 32, | ||
| "BLOCK_SIZE_N": 32, | ||
| "BLOCK_SIZE_K": 256, | ||
| "GROUP_SIZE_M": 32, | ||
| "num_warps": 4, | ||
| "num_stages": 2 | ||
| }, | ||
| "256": { | ||
| "BLOCK_SIZE_M": 32, | ||
| "BLOCK_SIZE_N": 64, | ||
| "BLOCK_SIZE_K": 128, | ||
| "GROUP_SIZE_M": 64, | ||
| "num_warps": 4, | ||
| "num_stages": 4 | ||
| }, | ||
| "512": { | ||
| "BLOCK_SIZE_M": 64, | ||
| "BLOCK_SIZE_N": 64, | ||
| "BLOCK_SIZE_K": 128, | ||
| "GROUP_SIZE_M": 32, | ||
| "num_warps": 8, | ||
| "num_stages": 2 | ||
| }, | ||
| "1024": { | ||
| "BLOCK_SIZE_M": 128, | ||
| "BLOCK_SIZE_N": 128, | ||
| "BLOCK_SIZE_K": 64, | ||
| "GROUP_SIZE_M": 32, | ||
| "num_warps": 8, | ||
| "num_stages": 3 | ||
| }, | ||
| "1536": { | ||
| "BLOCK_SIZE_M": 128, | ||
| "BLOCK_SIZE_N": 128, | ||
| "BLOCK_SIZE_K": 64, | ||
| "GROUP_SIZE_M": 32, | ||
| "num_warps": 8, | ||
| "num_stages": 3 | ||
| }, | ||
| "2048": { | ||
| "BLOCK_SIZE_M": 128, | ||
| "BLOCK_SIZE_N": 128, | ||
| "BLOCK_SIZE_K": 64, | ||
| "GROUP_SIZE_M": 64, | ||
| "num_warps": 8, | ||
| "num_stages": 3 | ||
| }, | ||
| "3072": { | ||
| "BLOCK_SIZE_M": 128, | ||
| "BLOCK_SIZE_N": 128, | ||
| "BLOCK_SIZE_K": 64, | ||
| "GROUP_SIZE_M": 64, | ||
| "num_warps": 8, | ||
| "num_stages": 3 | ||
| }, | ||
| "4096": { | ||
| "BLOCK_SIZE_M": 128, | ||
| "BLOCK_SIZE_N": 128, | ||
| "BLOCK_SIZE_K": 64, | ||
| "GROUP_SIZE_M": 1, | ||
| "num_warps": 8, | ||
| "num_stages": 3 | ||
| } | ||
| } |