Skip to content

Commit e0bb430

Browse files
WANDY666WANDY666
committed
add tune configs
1 parent af0a080 commit e0bb430

10 files changed

+10
-0
lines changed

‎lightllm/common/all_kernel_configs/_fwd_kernel_flash_decode_diverse_stage1:v1/{block_seq=256,gqa_group_size=16,out_dtype=torch.bfloat16,q_head_dim=128}_NVIDIA_GeForce_RTX_5090.json‎

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1 @@
1+
{"4096": {"8": {"BLOCK_N": 64, "num_warps": 8, "num_stages": 4}, "32": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 1}, "128": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 1}, "256": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 1}}, "8192": {"8": {"BLOCK_N": 16, "num_warps": 4, "num_stages": 1}, "32": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 1}, "128": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 1}, "256": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 1}}}

‎lightllm/common/all_kernel_configs/_fwd_kernel_flash_decode_diverse_stage1:v1/{block_seq=256,gqa_group_size=16,out_dtype=torch.float16,q_head_dim=128}_NVIDIA_GeForce_RTX_5090.json‎

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1 @@
1+
{"4096": {"8": {"BLOCK_N": 64, "num_warps": 8, "num_stages": 4}, "32": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 1}, "128": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 1}, "256": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 1}}, "8192": {"8": {"BLOCK_N": 16, "num_warps": 4, "num_stages": 1}, "32": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 1}, "128": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 1}, "256": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 1}}}

‎lightllm/common/all_kernel_configs/_fwd_kernel_flash_decode_diverse_stage1:v1/{block_seq=256,gqa_group_size=2,out_dtype=torch.bfloat16,q_head_dim=128}_NVIDIA_GeForce_RTX_5090.json‎

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1 @@
1+
{"4096": {"8": {"BLOCK_N": 64, "num_warps": 8, "num_stages": 3}, "32": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 2}, "128": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 2}, "256": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 2}}, "8192": {"8": {"BLOCK_N": 16, "num_warps": 4, "num_stages": 9}, "32": {"BLOCK_N": 16, "num_warps": 4, "num_stages": 1}, "128": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 2}, "256": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 2}}}

‎lightllm/common/all_kernel_configs/_fwd_kernel_flash_decode_diverse_stage1:v1/{block_seq=256,gqa_group_size=2,out_dtype=torch.float16,q_head_dim=128}_NVIDIA_GeForce_RTX_5090.json‎

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1 @@
1+
{"4096": {"8": {"BLOCK_N": 64, "num_warps": 8, "num_stages": 3}, "32": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 2}, "128": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 2}, "256": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 2}}, "8192": {"8": {"BLOCK_N": 16, "num_warps": 4, "num_stages": 9}, "32": {"BLOCK_N": 16, "num_warps": 4, "num_stages": 1}, "128": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 2}, "256": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 2}}}

‎lightllm/common/all_kernel_configs/_fwd_kernel_flash_decode_diverse_stage1:v1/{block_seq=256,gqa_group_size=4,out_dtype=torch.bfloat16,q_head_dim=128}_NVIDIA_GeForce_RTX_5090.json‎

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1 @@
1+
{"4096": {"8": {"BLOCK_N": 64, "num_warps": 8, "num_stages": 7}, "32": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 3}, "128": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 4}, "256": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 3}}, "8192": {"8": {"BLOCK_N": 16, "num_warps": 4, "num_stages": 9}, "32": {"BLOCK_N": 16, "num_warps": 4, "num_stages": 2}, "128": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 3}, "256": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 3}}}

‎lightllm/common/all_kernel_configs/_fwd_kernel_flash_decode_diverse_stage1:v1/{block_seq=256,gqa_group_size=4,out_dtype=torch.float16,q_head_dim=128}_NVIDIA_GeForce_RTX_5090.json‎

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1 @@
1+
{"4096": {"8": {"BLOCK_N": 64, "num_warps": 8, "num_stages": 7}, "32": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 3}, "128": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 4}, "256": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 3}}, "8192": {"8": {"BLOCK_N": 16, "num_warps": 4, "num_stages": 9}, "32": {"BLOCK_N": 16, "num_warps": 4, "num_stages": 2}, "128": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 3}, "256": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 3}}}

‎lightllm/common/all_kernel_configs/_fwd_kernel_flash_decode_diverse_stage1:v1/{block_seq=256,gqa_group_size=5,out_dtype=torch.bfloat16,q_head_dim=128}_NVIDIA_GeForce_RTX_5090.json‎

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1 @@
1+
{"4096": {"8": {"BLOCK_N": 64, "num_warps": 16, "num_stages": 4}, "32": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 5}, "128": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 5}, "256": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 5}}, "8192": {"8": {"BLOCK_N": 16, "num_warps": 4, "num_stages": 2}, "32": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 4}, "128": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 5}, "256": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 5}}}

‎lightllm/common/all_kernel_configs/_fwd_kernel_flash_decode_diverse_stage1:v1/{block_seq=256,gqa_group_size=5,out_dtype=torch.float16,q_head_dim=128}_NVIDIA_GeForce_RTX_5090.json‎

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1 @@
1+
{"4096": {"8": {"BLOCK_N": 64, "num_warps": 16, "num_stages": 4}, "32": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 5}, "128": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 5}, "256": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 5}}, "8192": {"8": {"BLOCK_N": 16, "num_warps": 4, "num_stages": 2}, "32": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 4}, "128": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 5}, "256": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 5}}}

‎lightllm/common/all_kernel_configs/_fwd_kernel_flash_decode_diverse_stage1:v1/{block_seq=256,gqa_group_size=8,out_dtype=torch.bfloat16,q_head_dim=128}_NVIDIA_GeForce_RTX_5090.json‎

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1 @@
1+
{"4096": {"8": {"BLOCK_N": 64, "num_warps": 8, "num_stages": 1}, "32": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 2}, "128": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 5}, "256": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 5}}, "8192": {"8": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 1}, "32": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 3}, "128": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 5}, "256": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 5}}}

‎lightllm/common/all_kernel_configs/_fwd_kernel_flash_decode_diverse_stage1:v1/{block_seq=256,gqa_group_size=8,out_dtype=torch.float16,q_head_dim=128}_NVIDIA_GeForce_RTX_5090.json‎

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1 @@
1+
{"4096": {"8": {"BLOCK_N": 64, "num_warps": 8, "num_stages": 1}, "32": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 2}, "128": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 5}, "256": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 5}}, "8192": {"8": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 1}, "32": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 3}, "128": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 5}, "256": {"BLOCK_N": 16, "num_warps": 2, "num_stages": 5}}}

0 commit comments

Comments
 (0)