Commit 177520a9 authored by zhuwenwen's avatar zhuwenwen
Browse files

add deepseek-v2 tp2 fused_moe_config

parent f0e7d72d
{
"1": {
"BLOCK_SIZE_M": 16,
"BLOCK_SIZE_N": 32,
"BLOCK_SIZE_K": 128,
"GROUP_SIZE_M": 1,
"num_warps": 1,
"num_stages": 2,
"waves_per_eu": 0
},
"2": {
"BLOCK_SIZE_M": 16,
"BLOCK_SIZE_N": 32,
"BLOCK_SIZE_K": 64,
"GROUP_SIZE_M": 1,
"num_warps": 1,
"num_stages": 2,
"waves_per_eu": 0
},
"4": {
"BLOCK_SIZE_M": 16,
"BLOCK_SIZE_N": 32,
"BLOCK_SIZE_K": 128,
"GROUP_SIZE_M": 1,
"num_warps": 2,
"num_stages": 2,
"waves_per_eu": 0
},
"8": {
"BLOCK_SIZE_M": 16,
"BLOCK_SIZE_N": 32,
"BLOCK_SIZE_K": 128,
"GROUP_SIZE_M": 1,
"num_warps": 2,
"num_stages": 2,
"waves_per_eu": 0
},
"16": {
"BLOCK_SIZE_M": 16,
"BLOCK_SIZE_N": 32,
"BLOCK_SIZE_K": 128,
"GROUP_SIZE_M": 1,
"num_warps": 2,
"num_stages": 2,
"waves_per_eu": 0
},
"24": {
"BLOCK_SIZE_M": 32,
"BLOCK_SIZE_N": 64,
"BLOCK_SIZE_K": 128,
"GROUP_SIZE_M": 1,
"num_warps": 8,
"num_stages": 2,
"waves_per_eu": 0
},
"32": {
"BLOCK_SIZE_M": 16,
"BLOCK_SIZE_N": 32,
"BLOCK_SIZE_K": 128,
"GROUP_SIZE_M": 1,
"num_warps": 4,
"num_stages": 2,
"waves_per_eu": 0
},
"48": {
"BLOCK_SIZE_M": 16,
"BLOCK_SIZE_N": 64,
"BLOCK_SIZE_K": 128,
"GROUP_SIZE_M": 1,
"num_warps": 4,
"num_stages": 2,
"waves_per_eu": 0
},
"64": {
"BLOCK_SIZE_M": 16,
"BLOCK_SIZE_N": 64,
"BLOCK_SIZE_K": 128,
"GROUP_SIZE_M": 4,
"num_warps": 4,
"num_stages": 2,
"waves_per_eu": 0
},
"96": {
"BLOCK_SIZE_M": 16,
"BLOCK_SIZE_N": 64,
"BLOCK_SIZE_K": 128,
"GROUP_SIZE_M": 1,
"num_warps": 4,
"num_stages": 2,
"waves_per_eu": 0
},
"128": {
"BLOCK_SIZE_M": 32,
"BLOCK_SIZE_N": 32,
"BLOCK_SIZE_K": 128,
"GROUP_SIZE_M": 1,
"num_warps": 4,
"num_stages": 2,
"waves_per_eu": 0
},
"256": {
"BLOCK_SIZE_M": 32,
"BLOCK_SIZE_N": 32,
"BLOCK_SIZE_K": 128,
"GROUP_SIZE_M": 1,
"num_warps": 4,
"num_stages": 2,
"waves_per_eu": 0
},
"512": {
"BLOCK_SIZE_M": 64,
"BLOCK_SIZE_N": 64,
"BLOCK_SIZE_K": 128,
"GROUP_SIZE_M": 1,
"num_warps": 4,
"num_stages": 2,
"waves_per_eu": 0
},
"1024": {
"BLOCK_SIZE_M": 128,
"BLOCK_SIZE_N": 128,
"BLOCK_SIZE_K": 64,
"GROUP_SIZE_M": 1,
"num_warps": 4,
"num_stages": 2,
"waves_per_eu": 0
},
"1536": {
"BLOCK_SIZE_M": 128,
"BLOCK_SIZE_N": 128,
"BLOCK_SIZE_K": 64,
"GROUP_SIZE_M": 1,
"num_warps": 4,
"num_stages": 2,
"waves_per_eu": 0
},
"2048": {
"BLOCK_SIZE_M": 128,
"BLOCK_SIZE_N": 128,
"BLOCK_SIZE_K": 64,
"GROUP_SIZE_M": 16,
"num_warps": 4,
"num_stages": 2,
"waves_per_eu": 0
},
"3072": {
"BLOCK_SIZE_M": 128,
"BLOCK_SIZE_N": 128,
"BLOCK_SIZE_K": 64,
"GROUP_SIZE_M": 1,
"num_warps": 4,
"num_stages": 2,
"waves_per_eu": 0
},
"4096": {
"BLOCK_SIZE_M": 128,
"BLOCK_SIZE_N": 128,
"BLOCK_SIZE_K": 64,
"GROUP_SIZE_M": 1,
"num_warps": 4,
"num_stages": 2,
"waves_per_eu": 0
}
}
{
"1": {
"BLOCK_SIZE_M": 16,
"BLOCK_SIZE_N": 64,
"BLOCK_SIZE_K": 64,
"BLOCK_SIZE_N": 32,
"BLOCK_SIZE_K": 32,
"GROUP_SIZE_M": 1,
"num_warps": 4,
"num_warps": 2,
"num_stages": 2,
"num_ldmatrixes": 1
"num_ldmatrixes": 1,
"waves_per_eu": 0
},
"2": {
"BLOCK_SIZE_M": 32,
"BLOCK_SIZE_M": 16,
"BLOCK_SIZE_N": 128,
"BLOCK_SIZE_K": 64,
"BLOCK_SIZE_K": 16,
"GROUP_SIZE_M": 1,
"num_warps": 8,
"num_warps": 1,
"num_stages": 2,
"num_ldmatrixes": 1
"num_ldmatrixes": 1,
"waves_per_eu": 0
},
"4": {
"BLOCK_SIZE_M": 32,
"BLOCK_SIZE_N": 128,
"BLOCK_SIZE_K": 64,
"BLOCK_SIZE_M": 16,
"BLOCK_SIZE_N": 256,
"BLOCK_SIZE_K": 32,
"GROUP_SIZE_M": 1,
"num_warps": 4,
"num_stages": 2,
"num_ldmatrixes": 1
"num_ldmatrixes": 1,
"waves_per_eu": 0
},
"8": {
"BLOCK_SIZE_M": 16,
"BLOCK_SIZE_N": 256,
"BLOCK_SIZE_M": 32,
"BLOCK_SIZE_N": 128,
"BLOCK_SIZE_K": 64,
"GROUP_SIZE_M": 1,
"num_warps": 8,
"num_warps": 4,
"num_stages": 2,
"num_ldmatrixes": 1
"num_ldmatrixes": 1,
"waves_per_eu": 0
},
"16": {
"BLOCK_SIZE_M": 16,
"BLOCK_SIZE_N": 256,
"BLOCK_SIZE_K": 64,
"GROUP_SIZE_M": 1,
"num_warps": 8,
"num_warps": 4,
"num_stages": 2,
"num_ldmatrixes": 1
"num_ldmatrixes": 1,
"waves_per_eu": 0
},
"24": {
"BLOCK_SIZE_M": 32,
"BLOCK_SIZE_M": 16,
"BLOCK_SIZE_N": 128,
"BLOCK_SIZE_K": 64,
"GROUP_SIZE_M": 1,
"num_warps": 4,
"num_stages": 2,
"num_ldmatrixes": 1
"num_ldmatrixes": 1,
"waves_per_eu": 0
},
"32": {
"BLOCK_SIZE_M": 16,
......@@ -60,7 +66,8 @@
"GROUP_SIZE_M": 1,
"num_warps": 8,
"num_stages": 2,
"num_ldmatrixes": 1
"num_ldmatrixes": 1,
"waves_per_eu": 0
},
"48": {
"BLOCK_SIZE_M": 16,
......@@ -69,7 +76,8 @@
"GROUP_SIZE_M": 1,
"num_warps": 8,
"num_stages": 2,
"num_ldmatrixes": 1
"num_ldmatrixes": 1,
"waves_per_eu": 0
},
"64": {
"BLOCK_SIZE_M": 16,
......@@ -78,7 +86,8 @@
"GROUP_SIZE_M": 1,
"num_warps": 8,
"num_stages": 2,
"num_ldmatrixes": 1
"num_ldmatrixes": 1,
"waves_per_eu": 0
},
"96": {
"BLOCK_SIZE_M": 16,
......@@ -87,25 +96,28 @@
"GROUP_SIZE_M": 1,
"num_warps": 8,
"num_stages": 2,
"num_ldmatrixes": 1
"num_ldmatrixes": 1,
"waves_per_eu": 0
},
"128": {
"BLOCK_SIZE_M": 32,
"BLOCK_SIZE_N": 256,
"BLOCK_SIZE_N": 128,
"BLOCK_SIZE_K": 64,
"GROUP_SIZE_M": 1,
"num_warps": 8,
"num_warps": 4,
"num_stages": 2,
"num_ldmatrixes": 1
"num_ldmatrixes": 1,
"waves_per_eu": 0
},
"256": {
"BLOCK_SIZE_M": 64,
"BLOCK_SIZE_N": 128,
"BLOCK_SIZE_K": 64,
"GROUP_SIZE_M": 1,
"num_warps": 8,
"num_warps": 4,
"num_stages": 2,
"num_ldmatrixes": 1
"num_ldmatrixes": 1,
"waves_per_eu": 0
},
"512": {
"BLOCK_SIZE_M": 64,
......@@ -114,7 +126,8 @@
"GROUP_SIZE_M": 1,
"num_warps": 8,
"num_stages": 2,
"num_ldmatrixes": 1
"num_ldmatrixes": 1,
"waves_per_eu": 0
},
"1024": {
"BLOCK_SIZE_M": 128,
......@@ -123,42 +136,47 @@
"GROUP_SIZE_M": 1,
"num_warps": 4,
"num_stages": 2,
"num_ldmatrixes": 1
"num_ldmatrixes": 1,
"waves_per_eu": 0
},
"1536": {
"BLOCK_SIZE_M": 128,
"BLOCK_SIZE_N": 128,
"BLOCK_SIZE_K": 64,
"GROUP_SIZE_M": 1,
"BLOCK_SIZE_M": 64,
"BLOCK_SIZE_N": 256,
"BLOCK_SIZE_K": 32,
"GROUP_SIZE_M": 4,
"num_warps": 4,
"num_stages": 2,
"num_ldmatrixes": 1
"num_ldmatrixes": 1,
"waves_per_eu": 0
},
"2048": {
"BLOCK_SIZE_M": 128,
"BLOCK_SIZE_N": 128,
"BLOCK_SIZE_K": 64,
"GROUP_SIZE_M": 1,
"BLOCK_SIZE_N": 256,
"BLOCK_SIZE_K": 32,
"GROUP_SIZE_M": 4,
"num_warps": 4,
"num_stages": 2,
"num_ldmatrixes": 1
"num_ldmatrixes": 1,
"waves_per_eu": 0
},
"3072": {
"BLOCK_SIZE_M": 128,
"BLOCK_SIZE_N": 128,
"BLOCK_SIZE_K": 64,
"BLOCK_SIZE_M": 64,
"BLOCK_SIZE_N": 256,
"BLOCK_SIZE_K": 32,
"GROUP_SIZE_M": 1,
"num_warps": 4,
"num_stages": 2,
"num_ldmatrixes": 1
"num_ldmatrixes": 1,
"waves_per_eu": 0
},
"4096": {
"BLOCK_SIZE_M": 128,
"BLOCK_SIZE_N": 128,
"BLOCK_SIZE_K": 64,
"BLOCK_SIZE_N": 256,
"BLOCK_SIZE_K": 32,
"GROUP_SIZE_M": 1,
"num_warps": 4,
"num_stages": 2,
"num_ldmatrixes": 1
"num_ldmatrixes": 1,
"waves_per_eu": 0
}
}
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment