source
stringclasses
13 values
gpu
stringclasses
4 values
model
stringclasses
13 values
kernel_family
stringclasses
7 values
kernel_name
stringclasses
328 values
dtype
stringclasses
2 values
held_out
bool
2 classes
M
float64
1
16.4k
N
float64
8
248k
K
float64
2.05k
29.6k
bs
float64
1
1
seq
float64
1
16.4k
n_heads
float64
8
64
head_dim
float64
64
256
kv_heads
float64
1
64
numel
float64
64
531M
op_type
stringclasses
177 values
gpu_time_duration_ms
float64
0
703
launch_block_size
float64
16
512
launch_grid_size
float64
1
1.04M
dram_bytes_sum
float64
1.27
94.3B
launch_registers_per_thread
float64
16
255
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_256x128_ldg8_relu_f2f_stages_64x3_tn
bf16
true
128
8,192
8,192
1
128
null
null
null
null
q_proj
0.1465
256
96
136,360,000
238
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_128x64_ldg8_f2f_stages_64x4_tn
bf16
true
128
1,024
8,192
1
128
null
null
null
null
k_proj
0.02522
128
64
18,890,000
158
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_128x64_ldg8_f2f_stages_64x4_tn
bf16
true
128
1,024
8,192
1
128
null
null
null
null
v_proj
0.02493
128
64
18,890,000
158
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_256x128_ldg8_f2f_stages_64x3_tn
bf16
true
128
8,192
8,192
1
128
null
null
null
null
o_proj
0.13802
256
96
136,340,000
238
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_256x128_ldg8_f2f_stages_64x3_tn
bf16
true
128
29,568
8,192
1
128
null
null
null
null
gate
0.59981
256
116
495,180,000
238
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_256x128_ldg8_f2f_stages_64x3_tn
bf16
true
128
29,568
8,192
1
128
null
null
null
null
up
0.58589
256
116
495,320,000
238
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_256x128_ldg8_f2f_stages_64x3_tn
bf16
true
128
8,192
29,568
1
128
null
null
null
null
down_proj
0.40381
256
96
492,040,000
238
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_256x128_ldg8_relu_f2f_stages_64x3_tn
bf16
true
128
8,192
8,192
1
128
null
null
null
null
q_proj
0.14227
256
96
136,360,000
238
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_128x64_ldg8_f2f_stages_64x4_tn
bf16
true
128
1,024
8,192
1
128
null
null
null
null
k_proj
0.02611
128
64
18,900,000
158
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_128x64_ldg8_f2f_stages_64x4_tn
bf16
true
128
1,024
8,192
1
128
null
null
null
null
v_proj
0.02547
128
64
18,890,000
158
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_256x128_ldg8_f2f_stages_64x3_tn
bf16
true
128
8,192
8,192
1
128
null
null
null
null
o_proj
0.13427
256
96
136,350,000
238
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_256x128_ldg8_f2f_stages_64x3_tn
bf16
true
128
29,568
8,192
1
128
null
null
null
null
gate
0.59018
256
116
492,400,000
238
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_256x128_ldg8_f2f_stages_64x3_tn
bf16
true
128
29,568
8,192
1
128
null
null
null
null
up
0.58454
256
116
495,670,000
238
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_256x128_ldg8_f2f_stages_64x3_tn
bf16
true
128
8,192
29,568
1
128
null
null
null
null
down_proj
0.40048
256
96
492,040,000
238
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_256x128_ldg8_relu_f2f_stages_64x3_tn
bf16
true
128
8,192
8,192
1
128
null
null
null
null
q_proj
0.1448
256
96
136,360,000
238
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_128x64_ldg8_f2f_stages_64x4_tn
bf16
true
128
1,024
8,192
1
128
null
null
null
null
k_proj
0.02547
128
64
18,890,000
158
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_128x64_ldg8_f2f_stages_64x4_tn
bf16
true
128
1,024
8,192
1
128
null
null
null
null
v_proj
0.02499
128
64
18,890,000
158
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_256x128_ldg8_f2f_stages_64x3_tn
bf16
true
128
8,192
8,192
1
128
null
null
null
null
o_proj
0.13741
256
96
136,340,000
238
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_256x128_ldg8_f2f_stages_64x3_tn
bf16
true
128
29,568
8,192
1
128
null
null
null
null
gate
0.5913
256
116
495,580,000
238
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_256x128_ldg8_f2f_stages_64x3_tn
bf16
true
128
29,568
8,192
1
128
null
null
null
null
up
0.58816
256
116
495,200,000
238
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_256x128_ldg8_f2f_stages_64x3_tn
bf16
true
128
8,192
29,568
1
128
null
null
null
null
down_proj
0.40064
256
96
492,040,000
238
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_256x128_ldg8_relu_f2f_stages_64x3_tn
bf16
true
128
8,192
8,192
1
128
null
null
null
null
q_proj
0.1425
256
96
136,360,000
238
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_128x64_ldg8_f2f_stages_64x4_tn
bf16
true
128
1,024
8,192
1
128
null
null
null
null
k_proj
0.02621
128
64
18,890,000
158
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_128x64_ldg8_f2f_stages_64x4_tn
bf16
true
128
1,024
8,192
1
128
null
null
null
null
v_proj
0.02509
128
64
18,890,000
158
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_256x128_ldg8_f2f_stages_64x3_tn
bf16
true
128
8,192
8,192
1
128
null
null
null
null
o_proj
0.13731
256
96
136,340,000
238
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_256x128_ldg8_f2f_stages_64x3_tn
bf16
true
128
29,568
8,192
1
128
null
null
null
null
gate
0.58605
256
116
493,910,000
238
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_256x128_ldg8_f2f_stages_64x3_tn
bf16
true
128
29,568
8,192
1
128
null
null
null
null
up
0.58115
256
116
495,510,000
238
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_256x128_ldg8_f2f_stages_64x3_tn
bf16
true
128
8,192
29,568
1
128
null
null
null
null
down_proj
0.39818
256
96
492,080,000
238
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_256x128_ldg8_relu_f2f_stages_64x3_tn
bf16
true
128
8,192
8,192
1
128
null
null
null
null
q_proj
0.1416
256
96
136,360,000
238
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_128x64_ldg8_f2f_stages_64x4_tn
bf16
true
128
1,024
8,192
1
128
null
null
null
null
k_proj
0.02509
128
64
18,890,000
158
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_128x64_ldg8_f2f_stages_64x4_tn
bf16
true
128
1,024
8,192
1
128
null
null
null
null
v_proj
0.02486
128
64
18,890,000
158
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_256x128_ldg8_f2f_stages_64x3_tn
bf16
true
128
8,192
8,192
1
128
null
null
null
null
o_proj
0.13206
256
96
136,340,000
238
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_256x128_ldg8_f2f_stages_64x3_tn
bf16
true
128
29,568
8,192
1
128
null
null
null
null
gate
0.59645
256
116
492,670,000
238
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_256x128_ldg8_f2f_stages_64x3_tn
bf16
true
128
29,568
8,192
1
128
null
null
null
null
up
0.58925
256
116
494,280,000
238
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_256x128_ldg8_f2f_stages_64x3_tn
bf16
true
128
8,192
29,568
1
128
null
null
null
null
down_proj
0.40934
256
96
492,040,000
238
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_256x128_ldg8_relu_f2f_stages_64x3_tn
bf16
true
128
8,192
8,192
1
128
null
null
null
null
q_proj
0.14454
256
96
136,360,000
238
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_128x64_ldg8_f2f_stages_64x4_tn
bf16
true
128
1,024
8,192
1
128
null
null
null
null
k_proj
0.02547
128
64
18,900,000
158
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_128x64_ldg8_f2f_stages_64x4_tn
bf16
true
128
1,024
8,192
1
128
null
null
null
null
v_proj
0.02506
128
64
18,890,000
158
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_256x128_ldg8_f2f_stages_64x3_tn
bf16
true
128
8,192
8,192
1
128
null
null
null
null
o_proj
0.13296
256
96
136,340,000
238
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_256x128_ldg8_f2f_stages_64x3_tn
bf16
true
128
29,568
8,192
1
128
null
null
null
null
gate
0.58883
256
116
495,570,000
238
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_256x128_ldg8_f2f_stages_64x3_tn
bf16
true
128
29,568
8,192
1
128
null
null
null
null
up
0.59098
256
116
492,420,000
238
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_256x128_ldg8_f2f_stages_64x3_tn
bf16
true
128
8,192
29,568
1
128
null
null
null
null
down_proj
0.40794
256
96
492,040,000
238
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_256x128_ldg8_relu_f2f_stages_64x3_tn
bf16
true
128
8,192
8,192
1
128
null
null
null
null
q_proj
0.14368
256
96
136,360,000
238
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_128x64_ldg8_f2f_stages_64x4_tn
bf16
true
128
1,024
8,192
1
128
null
null
null
null
k_proj
0.02525
128
64
18,890,000
158
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_128x64_ldg8_f2f_stages_64x4_tn
bf16
true
128
1,024
8,192
1
128
null
null
null
null
v_proj
0.02515
128
64
18,890,000
158
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_256x128_ldg8_f2f_stages_64x3_tn
bf16
true
128
8,192
8,192
1
128
null
null
null
null
o_proj
0.13699
256
96
136,340,000
238
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_256x128_ldg8_f2f_stages_64x3_tn
bf16
true
128
29,568
8,192
1
128
null
null
null
null
gate
0.59722
256
116
494,130,000
238
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_256x128_ldg8_f2f_stages_64x3_tn
bf16
true
128
29,568
8,192
1
128
null
null
null
null
up
0.59411
256
116
492,750,000
238
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_256x128_ldg8_f2f_stages_64x3_tn
bf16
true
128
8,192
29,568
1
128
null
null
null
null
down_proj
0.40934
256
96
492,040,000
238
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
gemm
ampere_bf16_s16816gemm_bf16_256x128_ldg8_f2f_stages_64x3_tn
bf16
true
128
152,064
8,192
1
128
null
null
null
null
lm_head
1.98
256
594
2,540,000,000
238
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.00736
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.0063
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.0063
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.0065
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.00618
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.00627
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.0063
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.0063
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.00627
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.0064
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.00643
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.00618
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.00627
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.00637
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.00634
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.00653
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.00685
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.0063
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.00621
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.00624
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.00614
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.00634
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.00621
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.00605
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.00621
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.00621
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.00618
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.00634
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.00627
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.00618
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.00621
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.00614
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.00621
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.00608
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.00624
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.00634
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.00624
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.00621
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.00614
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.00618
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.00621
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.00627
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.0063
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.00627
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.00624
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.00627
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.00608
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.00614
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.00618
512
256
1,060,000
32
Qwen2.5-72B-Instruct_prefill
A100
Qwen-72B
splitk_reduce
void cublasLt::splitKreduce_kernel<32, 16, int, __nv_bfloat16, __nv_bfloat16, float, __nv_bfloat16, 0, __nv_bfloat16, __nv_bfloat16, __nv_bfloat16, 1, 1, 0>(cublasLt::cublasSplitKParams<T6>, const T4 *, const T10 *, T9 *, T5 *, const T6 *, const T6 *, const T11 *, const T4 *, T11 *, void *, long, T6 *, int *, T6 *, T6 ...
bf16
true
128
1,024
8,192
null
null
null
null
null
null
null
0.00627
512
256
1,060,000
32