code large_stringlengths 193 899 | workload_type large_stringclasses 15
values | gpu_name large_stringclasses 12
values | gpu_features large_stringclasses 12
values | runtime_ms float64 0.01 4.35k | flops float64 30 21,045B | memory_bytes int64 160 13.2B | arithmetic_intensity float64 0 6.37k | dtype_bytes int64 2 4 | workload_params large_stringlengths 11 79 |
|---|---|---|---|---|---|---|---|---|---|
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 256) x (256, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 256, dtype=torch.float16, device='cuda')
B = torch.randn(256, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchronize... | matmul | NVIDIA A100 80GB | {"gpu_name": "NVIDIA A100 80GB", "cuda_cores": 6912, "tensor_cores": 432, "memory_gb": 80, "memory_bandwidth_gbps": 2039, "base_clock_mhz": 765, "boost_clock_mhz": 1410, "sm_count": 108, "fp32_tflops": 19.5, "fp16_tflops": 312, "tdp_watts": 400, "compute_capability": 8.0, "l2_cache_mb": 40} | 0.040719 | 134,217,728 | 1,179,648 | 113.777778 | 2 | {"M": 256, "N": 1024, "K": 256, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 256) x (256, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 256, dtype=torch.float16, device='cuda')
B = torch.randn(256, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchronize... | matmul | NVIDIA L4 | {"gpu_name": "NVIDIA L4", "cuda_cores": 7424, "tensor_cores": 232, "memory_gb": 24, "memory_bandwidth_gbps": 300, "base_clock_mhz": 795, "boost_clock_mhz": 2040, "sm_count": 58, "fp32_tflops": 30.3, "fp16_tflops": 121, "tdp_watts": 72, "compute_capability": 8.9, "l2_cache_mb": 48} | 0.045118 | 134,217,728 | 1,179,648 | 113.777778 | 2 | {"M": 256, "N": 1024, "K": 256, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 256) x (256, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 256, dtype=torch.float16, device='cuda')
B = torch.randn(256, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchronize... | matmul | NVIDIA L40S | {"gpu_name": "NVIDIA L40S", "cuda_cores": 18176, "tensor_cores": 568, "memory_gb": 48, "memory_bandwidth_gbps": 864, "base_clock_mhz": 1110, "boost_clock_mhz": 2520, "sm_count": 142, "fp32_tflops": 91.6, "fp16_tflops": 183.2, "tdp_watts": 350, "compute_capability": 8.9, "l2_cache_mb": 96} | 0.040346 | 134,217,728 | 1,179,648 | 113.777778 | 2 | {"M": 256, "N": 1024, "K": 256, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 256) x (256, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 256, dtype=torch.float16, device='cuda')
B = torch.randn(256, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchronize... | matmul | NVIDIA RTX 3090 | {"gpu_name": "NVIDIA RTX 3090", "cuda_cores": 10496, "tensor_cores": 328, "memory_gb": 24, "memory_bandwidth_gbps": 936, "base_clock_mhz": 1395, "boost_clock_mhz": 1695, "sm_count": 82, "fp32_tflops": 35.6, "fp16_tflops": 71, "tdp_watts": 350, "compute_capability": 8.6, "l2_cache_mb": 6} | 0.040029 | 134,217,728 | 1,179,648 | 113.777778 | 2 | {"M": 256, "N": 1024, "K": 256, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 256) x (256, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 256, dtype=torch.float16, device='cuda')
B = torch.randn(256, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchronize... | matmul | NVIDIA RTX 4090 | {"gpu_name": "NVIDIA RTX 4090", "cuda_cores": 16384, "tensor_cores": 512, "memory_gb": 24, "memory_bandwidth_gbps": 1008, "base_clock_mhz": 2235, "boost_clock_mhz": 2520, "sm_count": 128, "fp32_tflops": 82.6, "fp16_tflops": 165.2, "tdp_watts": 450, "compute_capability": 8.9, "l2_cache_mb": 72} | 0.055593 | 134,217,728 | 1,179,648 | 113.777778 | 2 | {"M": 256, "N": 1024, "K": 256, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 256) x (256, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 256, dtype=torch.float16, device='cuda')
B = torch.randn(256, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchronize... | matmul | NVIDIA H100 SXM | {"gpu_name": "NVIDIA H100 SXM", "cuda_cores": 16896, "tensor_cores": 528, "memory_gb": 80, "memory_bandwidth_gbps": 3350, "base_clock_mhz": 1095, "boost_clock_mhz": 1830, "sm_count": 132, "fp32_tflops": 67, "fp16_tflops": 989, "tdp_watts": 700, "compute_capability": 9.0, "l2_cache_mb": 50} | 0.028358 | 134,217,728 | 1,179,648 | 113.777778 | 2 | {"M": 256, "N": 1024, "K": 256, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 256) x (256, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 256, dtype=torch.float16, device='cuda')
B = torch.randn(256, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchronize... | matmul | NVIDIA H100 PCIe | {"gpu_name": "NVIDIA H100 PCIe", "cuda_cores": 14592, "tensor_cores": 456, "memory_gb": 80, "memory_bandwidth_gbps": 2039, "base_clock_mhz": 1095, "boost_clock_mhz": 1620, "sm_count": 114, "fp32_tflops": 48, "fp16_tflops": 756, "tdp_watts": 350, "compute_capability": 9.0, "l2_cache_mb": 50} | 0.029486 | 134,217,728 | 1,179,648 | 113.777778 | 2 | {"M": 256, "N": 1024, "K": 256, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 256) x (256, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 256, dtype=torch.float16, device='cuda')
B = torch.randn(256, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchronize... | matmul | NVIDIA RTX A6000 | {"gpu_name": "NVIDIA RTX A6000", "cuda_cores": 10752, "tensor_cores": 336, "memory_gb": 48, "memory_bandwidth_gbps": 768, "base_clock_mhz": 1410, "boost_clock_mhz": 1860, "sm_count": 84, "fp32_tflops": 38.7, "fp16_tflops": 77.4, "tdp_watts": 300, "compute_capability": 8.6, "l2_cache_mb": 6} | 0.05609 | 134,217,728 | 1,179,648 | 113.777778 | 2 | {"M": 256, "N": 1024, "K": 256, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 512) x (512, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 512, dtype=torch.float32, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchronize... | matmul | NVIDIA T4 | {"gpu_name": "NVIDIA T4", "cuda_cores": 2560, "tensor_cores": 320, "memory_gb": 16, "memory_bandwidth_gbps": 320, "base_clock_mhz": 585, "boost_clock_mhz": 1590, "sm_count": 40, "fp32_tflops": 8.1, "fp16_tflops": 65, "tdp_watts": 70, "compute_capability": 7.5, "l2_cache_mb": 4} | 0.089934 | 268,435,456 | 3,670,016 | 73.142857 | 4 | {"M": 256, "N": 1024, "K": 512, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 512) x (512, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 512, dtype=torch.float32, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchronize... | matmul | NVIDIA V100 | {"gpu_name": "NVIDIA V100", "cuda_cores": 5120, "tensor_cores": 640, "memory_gb": 32, "memory_bandwidth_gbps": 900, "base_clock_mhz": 1230, "boost_clock_mhz": 1530, "sm_count": 80, "fp32_tflops": 15.7, "fp16_tflops": 125, "tdp_watts": 300, "compute_capability": 7.0, "l2_cache_mb": 6} | 0.058603 | 268,435,456 | 3,670,016 | 73.142857 | 4 | {"M": 256, "N": 1024, "K": 512, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 512) x (512, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 512, dtype=torch.float32, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchronize... | matmul | NVIDIA A10G | {"gpu_name": "NVIDIA A10G", "cuda_cores": 9216, "tensor_cores": 288, "memory_gb": 24, "memory_bandwidth_gbps": 600, "base_clock_mhz": 885, "boost_clock_mhz": 1695, "sm_count": 80, "fp32_tflops": 31.2, "fp16_tflops": 62.5, "tdp_watts": 150, "compute_capability": 8.6, "l2_cache_mb": 6} | 0.046767 | 268,435,456 | 3,670,016 | 73.142857 | 4 | {"M": 256, "N": 1024, "K": 512, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 512) x (512, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 512, dtype=torch.float32, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchronize... | matmul | NVIDIA A100 40GB | {"gpu_name": "NVIDIA A100 40GB", "cuda_cores": 6912, "tensor_cores": 432, "memory_gb": 40, "memory_bandwidth_gbps": 1555, "base_clock_mhz": 765, "boost_clock_mhz": 1410, "sm_count": 108, "fp32_tflops": 19.5, "fp16_tflops": 312, "tdp_watts": 400, "compute_capability": 8.0, "l2_cache_mb": 40} | 0.075072 | 268,435,456 | 3,670,016 | 73.142857 | 4 | {"M": 256, "N": 1024, "K": 512, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 512) x (512, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 512, dtype=torch.float32, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchronize... | matmul | NVIDIA A100 80GB | {"gpu_name": "NVIDIA A100 80GB", "cuda_cores": 6912, "tensor_cores": 432, "memory_gb": 80, "memory_bandwidth_gbps": 2039, "base_clock_mhz": 765, "boost_clock_mhz": 1410, "sm_count": 108, "fp32_tflops": 19.5, "fp16_tflops": 312, "tdp_watts": 400, "compute_capability": 8.0, "l2_cache_mb": 40} | 0.070958 | 268,435,456 | 3,670,016 | 73.142857 | 4 | {"M": 256, "N": 1024, "K": 512, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 512) x (512, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 512, dtype=torch.float32, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchronize... | matmul | NVIDIA L4 | {"gpu_name": "NVIDIA L4", "cuda_cores": 7424, "tensor_cores": 232, "memory_gb": 24, "memory_bandwidth_gbps": 300, "base_clock_mhz": 795, "boost_clock_mhz": 2040, "sm_count": 58, "fp32_tflops": 30.3, "fp16_tflops": 121, "tdp_watts": 72, "compute_capability": 8.9, "l2_cache_mb": 48} | 0.053747 | 268,435,456 | 3,670,016 | 73.142857 | 4 | {"M": 256, "N": 1024, "K": 512, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 512) x (512, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 512, dtype=torch.float32, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchronize... | matmul | NVIDIA L40S | {"gpu_name": "NVIDIA L40S", "cuda_cores": 18176, "tensor_cores": 568, "memory_gb": 48, "memory_bandwidth_gbps": 864, "base_clock_mhz": 1110, "boost_clock_mhz": 2520, "sm_count": 142, "fp32_tflops": 91.6, "fp16_tflops": 183.2, "tdp_watts": 350, "compute_capability": 8.9, "l2_cache_mb": 96} | 0.032955 | 268,435,456 | 3,670,016 | 73.142857 | 4 | {"M": 256, "N": 1024, "K": 512, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 512) x (512, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 512, dtype=torch.float32, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchronize... | matmul | NVIDIA RTX 3090 | {"gpu_name": "NVIDIA RTX 3090", "cuda_cores": 10496, "tensor_cores": 328, "memory_gb": 24, "memory_bandwidth_gbps": 936, "base_clock_mhz": 1395, "boost_clock_mhz": 1695, "sm_count": 82, "fp32_tflops": 35.6, "fp16_tflops": 71, "tdp_watts": 350, "compute_capability": 8.6, "l2_cache_mb": 6} | 0.062028 | 268,435,456 | 3,670,016 | 73.142857 | 4 | {"M": 256, "N": 1024, "K": 512, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 512) x (512, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 512, dtype=torch.float32, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchronize... | matmul | NVIDIA RTX 4090 | {"gpu_name": "NVIDIA RTX 4090", "cuda_cores": 16384, "tensor_cores": 512, "memory_gb": 24, "memory_bandwidth_gbps": 1008, "base_clock_mhz": 2235, "boost_clock_mhz": 2520, "sm_count": 128, "fp32_tflops": 82.6, "fp16_tflops": 165.2, "tdp_watts": 450, "compute_capability": 8.9, "l2_cache_mb": 72} | 0.062147 | 268,435,456 | 3,670,016 | 73.142857 | 4 | {"M": 256, "N": 1024, "K": 512, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 512) x (512, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 512, dtype=torch.float32, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchronize... | matmul | NVIDIA H100 SXM | {"gpu_name": "NVIDIA H100 SXM", "cuda_cores": 16896, "tensor_cores": 528, "memory_gb": 80, "memory_bandwidth_gbps": 3350, "base_clock_mhz": 1095, "boost_clock_mhz": 1830, "sm_count": 132, "fp32_tflops": 67, "fp16_tflops": 989, "tdp_watts": 700, "compute_capability": 9.0, "l2_cache_mb": 50} | 0.039325 | 268,435,456 | 3,670,016 | 73.142857 | 4 | {"M": 256, "N": 1024, "K": 512, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 512) x (512, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 512, dtype=torch.float32, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchronize... | matmul | NVIDIA H100 PCIe | {"gpu_name": "NVIDIA H100 PCIe", "cuda_cores": 14592, "tensor_cores": 456, "memory_gb": 80, "memory_bandwidth_gbps": 2039, "base_clock_mhz": 1095, "boost_clock_mhz": 1620, "sm_count": 114, "fp32_tflops": 48, "fp16_tflops": 756, "tdp_watts": 350, "compute_capability": 9.0, "l2_cache_mb": 50} | 0.056055 | 268,435,456 | 3,670,016 | 73.142857 | 4 | {"M": 256, "N": 1024, "K": 512, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 512) x (512, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 512, dtype=torch.float32, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchronize... | matmul | NVIDIA RTX A6000 | {"gpu_name": "NVIDIA RTX A6000", "cuda_cores": 10752, "tensor_cores": 336, "memory_gb": 48, "memory_bandwidth_gbps": 768, "base_clock_mhz": 1410, "boost_clock_mhz": 1860, "sm_count": 84, "fp32_tflops": 38.7, "fp16_tflops": 77.4, "tdp_watts": 300, "compute_capability": 8.6, "l2_cache_mb": 6} | 0.046951 | 268,435,456 | 3,670,016 | 73.142857 | 4 | {"M": 256, "N": 1024, "K": 512, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 512) x (512, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 512, dtype=torch.float16, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchronize... | matmul | NVIDIA T4 | {"gpu_name": "NVIDIA T4", "cuda_cores": 2560, "tensor_cores": 320, "memory_gb": 16, "memory_bandwidth_gbps": 320, "base_clock_mhz": 585, "boost_clock_mhz": 1590, "sm_count": 40, "fp32_tflops": 8.1, "fp16_tflops": 65, "tdp_watts": 70, "compute_capability": 7.5, "l2_cache_mb": 4} | 0.075171 | 268,435,456 | 1,835,008 | 146.285714 | 2 | {"M": 256, "N": 1024, "K": 512, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 512) x (512, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 512, dtype=torch.float16, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchronize... | matmul | NVIDIA V100 | {"gpu_name": "NVIDIA V100", "cuda_cores": 5120, "tensor_cores": 640, "memory_gb": 32, "memory_bandwidth_gbps": 900, "base_clock_mhz": 1230, "boost_clock_mhz": 1530, "sm_count": 80, "fp32_tflops": 15.7, "fp16_tflops": 125, "tdp_watts": 300, "compute_capability": 7.0, "l2_cache_mb": 6} | 0.04967 | 268,435,456 | 1,835,008 | 146.285714 | 2 | {"M": 256, "N": 1024, "K": 512, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 512) x (512, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 512, dtype=torch.float16, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchronize... | matmul | NVIDIA A10G | {"gpu_name": "NVIDIA A10G", "cuda_cores": 9216, "tensor_cores": 288, "memory_gb": 24, "memory_bandwidth_gbps": 600, "base_clock_mhz": 885, "boost_clock_mhz": 1695, "sm_count": 80, "fp32_tflops": 31.2, "fp16_tflops": 62.5, "tdp_watts": 150, "compute_capability": 8.6, "l2_cache_mb": 6} | 0.03307 | 268,435,456 | 1,835,008 | 146.285714 | 2 | {"M": 256, "N": 1024, "K": 512, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 512) x (512, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 512, dtype=torch.float16, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchronize... | matmul | NVIDIA A100 40GB | {"gpu_name": "NVIDIA A100 40GB", "cuda_cores": 6912, "tensor_cores": 432, "memory_gb": 40, "memory_bandwidth_gbps": 1555, "base_clock_mhz": 765, "boost_clock_mhz": 1410, "sm_count": 108, "fp32_tflops": 19.5, "fp16_tflops": 312, "tdp_watts": 400, "compute_capability": 8.0, "l2_cache_mb": 40} | 0.032398 | 268,435,456 | 1,835,008 | 146.285714 | 2 | {"M": 256, "N": 1024, "K": 512, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 512) x (512, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 512, dtype=torch.float16, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchronize... | matmul | NVIDIA A100 80GB | {"gpu_name": "NVIDIA A100 80GB", "cuda_cores": 6912, "tensor_cores": 432, "memory_gb": 80, "memory_bandwidth_gbps": 2039, "base_clock_mhz": 765, "boost_clock_mhz": 1410, "sm_count": 108, "fp32_tflops": 19.5, "fp16_tflops": 312, "tdp_watts": 400, "compute_capability": 8.0, "l2_cache_mb": 40} | 0.048197 | 268,435,456 | 1,835,008 | 146.285714 | 2 | {"M": 256, "N": 1024, "K": 512, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 512) x (512, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 512, dtype=torch.float16, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchronize... | matmul | NVIDIA L4 | {"gpu_name": "NVIDIA L4", "cuda_cores": 7424, "tensor_cores": 232, "memory_gb": 24, "memory_bandwidth_gbps": 300, "base_clock_mhz": 795, "boost_clock_mhz": 2040, "sm_count": 58, "fp32_tflops": 30.3, "fp16_tflops": 121, "tdp_watts": 72, "compute_capability": 8.9, "l2_cache_mb": 48} | 0.051018 | 268,435,456 | 1,835,008 | 146.285714 | 2 | {"M": 256, "N": 1024, "K": 512, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 512) x (512, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 512, dtype=torch.float16, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchronize... | matmul | NVIDIA L40S | {"gpu_name": "NVIDIA L40S", "cuda_cores": 18176, "tensor_cores": 568, "memory_gb": 48, "memory_bandwidth_gbps": 864, "base_clock_mhz": 1110, "boost_clock_mhz": 2520, "sm_count": 142, "fp32_tflops": 91.6, "fp16_tflops": 183.2, "tdp_watts": 350, "compute_capability": 8.9, "l2_cache_mb": 96} | 0.049564 | 268,435,456 | 1,835,008 | 146.285714 | 2 | {"M": 256, "N": 1024, "K": 512, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 512) x (512, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 512, dtype=torch.float16, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchronize... | matmul | NVIDIA RTX 3090 | {"gpu_name": "NVIDIA RTX 3090", "cuda_cores": 10496, "tensor_cores": 328, "memory_gb": 24, "memory_bandwidth_gbps": 936, "base_clock_mhz": 1395, "boost_clock_mhz": 1695, "sm_count": 82, "fp32_tflops": 35.6, "fp16_tflops": 71, "tdp_watts": 350, "compute_capability": 8.6, "l2_cache_mb": 6} | 0.023979 | 268,435,456 | 1,835,008 | 146.285714 | 2 | {"M": 256, "N": 1024, "K": 512, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 512) x (512, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 512, dtype=torch.float16, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchronize... | matmul | NVIDIA RTX 4090 | {"gpu_name": "NVIDIA RTX 4090", "cuda_cores": 16384, "tensor_cores": 512, "memory_gb": 24, "memory_bandwidth_gbps": 1008, "base_clock_mhz": 2235, "boost_clock_mhz": 2520, "sm_count": 128, "fp32_tflops": 82.6, "fp16_tflops": 165.2, "tdp_watts": 450, "compute_capability": 8.9, "l2_cache_mb": 72} | 0.038958 | 268,435,456 | 1,835,008 | 146.285714 | 2 | {"M": 256, "N": 1024, "K": 512, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 512) x (512, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 512, dtype=torch.float16, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchronize... | matmul | NVIDIA H100 SXM | {"gpu_name": "NVIDIA H100 SXM", "cuda_cores": 16896, "tensor_cores": 528, "memory_gb": 80, "memory_bandwidth_gbps": 3350, "base_clock_mhz": 1095, "boost_clock_mhz": 1830, "sm_count": 132, "fp32_tflops": 67, "fp16_tflops": 989, "tdp_watts": 700, "compute_capability": 9.0, "l2_cache_mb": 50} | 0.034982 | 268,435,456 | 1,835,008 | 146.285714 | 2 | {"M": 256, "N": 1024, "K": 512, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 512) x (512, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 512, dtype=torch.float16, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchronize... | matmul | NVIDIA H100 PCIe | {"gpu_name": "NVIDIA H100 PCIe", "cuda_cores": 14592, "tensor_cores": 456, "memory_gb": 80, "memory_bandwidth_gbps": 2039, "base_clock_mhz": 1095, "boost_clock_mhz": 1620, "sm_count": 114, "fp32_tflops": 48, "fp16_tflops": 756, "tdp_watts": 350, "compute_capability": 9.0, "l2_cache_mb": 50} | 0.022941 | 268,435,456 | 1,835,008 | 146.285714 | 2 | {"M": 256, "N": 1024, "K": 512, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 512) x (512, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 512, dtype=torch.float16, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchronize... | matmul | NVIDIA RTX A6000 | {"gpu_name": "NVIDIA RTX A6000", "cuda_cores": 10752, "tensor_cores": 336, "memory_gb": 48, "memory_bandwidth_gbps": 768, "base_clock_mhz": 1410, "boost_clock_mhz": 1860, "sm_count": 84, "fp32_tflops": 38.7, "fp16_tflops": 77.4, "tdp_watts": 300, "compute_capability": 8.6, "l2_cache_mb": 6} | 0.054966 | 268,435,456 | 1,835,008 | 146.285714 | 2 | {"M": 256, "N": 1024, "K": 512, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 1024) x (1024, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 1024, dtype=torch.float32, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA T4 | {"gpu_name": "NVIDIA T4", "cuda_cores": 2560, "tensor_cores": 320, "memory_gb": 16, "memory_bandwidth_gbps": 320, "base_clock_mhz": 585, "boost_clock_mhz": 1590, "sm_count": 40, "fp32_tflops": 8.1, "fp16_tflops": 65, "tdp_watts": 70, "compute_capability": 7.5, "l2_cache_mb": 4} | 0.16735 | 536,870,912 | 6,291,456 | 85.333333 | 4 | {"M": 256, "N": 1024, "K": 1024, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 1024) x (1024, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 1024, dtype=torch.float32, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA V100 | {"gpu_name": "NVIDIA V100", "cuda_cores": 5120, "tensor_cores": 640, "memory_gb": 32, "memory_bandwidth_gbps": 900, "base_clock_mhz": 1230, "boost_clock_mhz": 1530, "sm_count": 80, "fp32_tflops": 15.7, "fp16_tflops": 125, "tdp_watts": 300, "compute_capability": 7.0, "l2_cache_mb": 6} | 0.094447 | 536,870,912 | 6,291,456 | 85.333333 | 4 | {"M": 256, "N": 1024, "K": 1024, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 1024) x (1024, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 1024, dtype=torch.float32, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA A10G | {"gpu_name": "NVIDIA A10G", "cuda_cores": 9216, "tensor_cores": 288, "memory_gb": 24, "memory_bandwidth_gbps": 600, "base_clock_mhz": 885, "boost_clock_mhz": 1695, "sm_count": 80, "fp32_tflops": 31.2, "fp16_tflops": 62.5, "tdp_watts": 150, "compute_capability": 8.6, "l2_cache_mb": 6} | 0.065818 | 536,870,912 | 6,291,456 | 85.333333 | 4 | {"M": 256, "N": 1024, "K": 1024, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 1024) x (1024, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 1024, dtype=torch.float32, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA A100 40GB | {"gpu_name": "NVIDIA A100 40GB", "cuda_cores": 6912, "tensor_cores": 432, "memory_gb": 40, "memory_bandwidth_gbps": 1555, "base_clock_mhz": 765, "boost_clock_mhz": 1410, "sm_count": 108, "fp32_tflops": 19.5, "fp16_tflops": 312, "tdp_watts": 400, "compute_capability": 8.0, "l2_cache_mb": 40} | 0.092278 | 536,870,912 | 6,291,456 | 85.333333 | 4 | {"M": 256, "N": 1024, "K": 1024, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 1024) x (1024, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 1024, dtype=torch.float32, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA A100 80GB | {"gpu_name": "NVIDIA A100 80GB", "cuda_cores": 6912, "tensor_cores": 432, "memory_gb": 80, "memory_bandwidth_gbps": 2039, "base_clock_mhz": 765, "boost_clock_mhz": 1410, "sm_count": 108, "fp32_tflops": 19.5, "fp16_tflops": 312, "tdp_watts": 400, "compute_capability": 8.0, "l2_cache_mb": 40} | 0.108837 | 536,870,912 | 6,291,456 | 85.333333 | 4 | {"M": 256, "N": 1024, "K": 1024, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 1024) x (1024, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 1024, dtype=torch.float32, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA L4 | {"gpu_name": "NVIDIA L4", "cuda_cores": 7424, "tensor_cores": 232, "memory_gb": 24, "memory_bandwidth_gbps": 300, "base_clock_mhz": 795, "boost_clock_mhz": 2040, "sm_count": 58, "fp32_tflops": 30.3, "fp16_tflops": 121, "tdp_watts": 72, "compute_capability": 8.9, "l2_cache_mb": 48} | 0.059855 | 536,870,912 | 6,291,456 | 85.333333 | 4 | {"M": 256, "N": 1024, "K": 1024, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 1024) x (1024, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 1024, dtype=torch.float32, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA L40S | {"gpu_name": "NVIDIA L40S", "cuda_cores": 18176, "tensor_cores": 568, "memory_gb": 48, "memory_bandwidth_gbps": 864, "base_clock_mhz": 1110, "boost_clock_mhz": 2520, "sm_count": 142, "fp32_tflops": 91.6, "fp16_tflops": 183.2, "tdp_watts": 350, "compute_capability": 8.9, "l2_cache_mb": 96} | 0.061968 | 536,870,912 | 6,291,456 | 85.333333 | 4 | {"M": 256, "N": 1024, "K": 1024, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 1024) x (1024, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 1024, dtype=torch.float32, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA RTX 3090 | {"gpu_name": "NVIDIA RTX 3090", "cuda_cores": 10496, "tensor_cores": 328, "memory_gb": 24, "memory_bandwidth_gbps": 936, "base_clock_mhz": 1395, "boost_clock_mhz": 1695, "sm_count": 82, "fp32_tflops": 35.6, "fp16_tflops": 71, "tdp_watts": 350, "compute_capability": 8.6, "l2_cache_mb": 6} | 0.047676 | 536,870,912 | 6,291,456 | 85.333333 | 4 | {"M": 256, "N": 1024, "K": 1024, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 1024) x (1024, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 1024, dtype=torch.float32, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA RTX 4090 | {"gpu_name": "NVIDIA RTX 4090", "cuda_cores": 16384, "tensor_cores": 512, "memory_gb": 24, "memory_bandwidth_gbps": 1008, "base_clock_mhz": 2235, "boost_clock_mhz": 2520, "sm_count": 128, "fp32_tflops": 82.6, "fp16_tflops": 165.2, "tdp_watts": 450, "compute_capability": 8.9, "l2_cache_mb": 72} | 0.049112 | 536,870,912 | 6,291,456 | 85.333333 | 4 | {"M": 256, "N": 1024, "K": 1024, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 1024) x (1024, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 1024, dtype=torch.float32, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA H100 SXM | {"gpu_name": "NVIDIA H100 SXM", "cuda_cores": 16896, "tensor_cores": 528, "memory_gb": 80, "memory_bandwidth_gbps": 3350, "base_clock_mhz": 1095, "boost_clock_mhz": 1830, "sm_count": 132, "fp32_tflops": 67, "fp16_tflops": 989, "tdp_watts": 700, "compute_capability": 9.0, "l2_cache_mb": 50} | 0.045354 | 536,870,912 | 6,291,456 | 85.333333 | 4 | {"M": 256, "N": 1024, "K": 1024, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 1024) x (1024, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 1024, dtype=torch.float32, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA H100 PCIe | {"gpu_name": "NVIDIA H100 PCIe", "cuda_cores": 14592, "tensor_cores": 456, "memory_gb": 80, "memory_bandwidth_gbps": 2039, "base_clock_mhz": 1095, "boost_clock_mhz": 1620, "sm_count": 114, "fp32_tflops": 48, "fp16_tflops": 756, "tdp_watts": 350, "compute_capability": 9.0, "l2_cache_mb": 50} | 0.043741 | 536,870,912 | 6,291,456 | 85.333333 | 4 | {"M": 256, "N": 1024, "K": 1024, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 1024) x (1024, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 1024, dtype=torch.float32, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA RTX A6000 | {"gpu_name": "NVIDIA RTX A6000", "cuda_cores": 10752, "tensor_cores": 336, "memory_gb": 48, "memory_bandwidth_gbps": 768, "base_clock_mhz": 1410, "boost_clock_mhz": 1860, "sm_count": 84, "fp32_tflops": 38.7, "fp16_tflops": 77.4, "tdp_watts": 300, "compute_capability": 8.6, "l2_cache_mb": 6} | 0.052164 | 536,870,912 | 6,291,456 | 85.333333 | 4 | {"M": 256, "N": 1024, "K": 1024, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 1024) x (1024, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 1024, dtype=torch.float16, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA T4 | {"gpu_name": "NVIDIA T4", "cuda_cores": 2560, "tensor_cores": 320, "memory_gb": 16, "memory_bandwidth_gbps": 320, "base_clock_mhz": 585, "boost_clock_mhz": 1590, "sm_count": 40, "fp32_tflops": 8.1, "fp16_tflops": 65, "tdp_watts": 70, "compute_capability": 7.5, "l2_cache_mb": 4} | 0.039874 | 536,870,912 | 3,145,728 | 170.666667 | 2 | {"M": 256, "N": 1024, "K": 1024, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 1024) x (1024, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 1024, dtype=torch.float16, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA V100 | {"gpu_name": "NVIDIA V100", "cuda_cores": 5120, "tensor_cores": 640, "memory_gb": 32, "memory_bandwidth_gbps": 900, "base_clock_mhz": 1230, "boost_clock_mhz": 1530, "sm_count": 80, "fp32_tflops": 15.7, "fp16_tflops": 125, "tdp_watts": 300, "compute_capability": 7.0, "l2_cache_mb": 6} | 0.046894 | 536,870,912 | 3,145,728 | 170.666667 | 2 | {"M": 256, "N": 1024, "K": 1024, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 1024) x (1024, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 1024, dtype=torch.float16, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA A10G | {"gpu_name": "NVIDIA A10G", "cuda_cores": 9216, "tensor_cores": 288, "memory_gb": 24, "memory_bandwidth_gbps": 600, "base_clock_mhz": 885, "boost_clock_mhz": 1695, "sm_count": 80, "fp32_tflops": 31.2, "fp16_tflops": 62.5, "tdp_watts": 150, "compute_capability": 8.6, "l2_cache_mb": 6} | 0.074475 | 536,870,912 | 3,145,728 | 170.666667 | 2 | {"M": 256, "N": 1024, "K": 1024, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 1024) x (1024, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 1024, dtype=torch.float16, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA A100 40GB | {"gpu_name": "NVIDIA A100 40GB", "cuda_cores": 6912, "tensor_cores": 432, "memory_gb": 40, "memory_bandwidth_gbps": 1555, "base_clock_mhz": 765, "boost_clock_mhz": 1410, "sm_count": 108, "fp32_tflops": 19.5, "fp16_tflops": 312, "tdp_watts": 400, "compute_capability": 8.0, "l2_cache_mb": 40} | 0.060209 | 536,870,912 | 3,145,728 | 170.666667 | 2 | {"M": 256, "N": 1024, "K": 1024, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 1024) x (1024, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 1024, dtype=torch.float16, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA A100 80GB | {"gpu_name": "NVIDIA A100 80GB", "cuda_cores": 6912, "tensor_cores": 432, "memory_gb": 80, "memory_bandwidth_gbps": 2039, "base_clock_mhz": 765, "boost_clock_mhz": 1410, "sm_count": 108, "fp32_tflops": 19.5, "fp16_tflops": 312, "tdp_watts": 400, "compute_capability": 8.0, "l2_cache_mb": 40} | 0.020833 | 536,870,912 | 3,145,728 | 170.666667 | 2 | {"M": 256, "N": 1024, "K": 1024, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 1024) x (1024, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 1024, dtype=torch.float16, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA L4 | {"gpu_name": "NVIDIA L4", "cuda_cores": 7424, "tensor_cores": 232, "memory_gb": 24, "memory_bandwidth_gbps": 300, "base_clock_mhz": 795, "boost_clock_mhz": 2040, "sm_count": 58, "fp32_tflops": 30.3, "fp16_tflops": 121, "tdp_watts": 72, "compute_capability": 8.9, "l2_cache_mb": 48} | 0.050702 | 536,870,912 | 3,145,728 | 170.666667 | 2 | {"M": 256, "N": 1024, "K": 1024, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 1024) x (1024, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 1024, dtype=torch.float16, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA L40S | {"gpu_name": "NVIDIA L40S", "cuda_cores": 18176, "tensor_cores": 568, "memory_gb": 48, "memory_bandwidth_gbps": 864, "base_clock_mhz": 1110, "boost_clock_mhz": 2520, "sm_count": 142, "fp32_tflops": 91.6, "fp16_tflops": 183.2, "tdp_watts": 350, "compute_capability": 8.9, "l2_cache_mb": 96} | 0.04023 | 536,870,912 | 3,145,728 | 170.666667 | 2 | {"M": 256, "N": 1024, "K": 1024, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 1024) x (1024, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 1024, dtype=torch.float16, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA RTX 3090 | {"gpu_name": "NVIDIA RTX 3090", "cuda_cores": 10496, "tensor_cores": 328, "memory_gb": 24, "memory_bandwidth_gbps": 936, "base_clock_mhz": 1395, "boost_clock_mhz": 1695, "sm_count": 82, "fp32_tflops": 35.6, "fp16_tflops": 71, "tdp_watts": 350, "compute_capability": 8.6, "l2_cache_mb": 6} | 0.073997 | 536,870,912 | 3,145,728 | 170.666667 | 2 | {"M": 256, "N": 1024, "K": 1024, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 1024) x (1024, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 1024, dtype=torch.float16, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA RTX 4090 | {"gpu_name": "NVIDIA RTX 4090", "cuda_cores": 16384, "tensor_cores": 512, "memory_gb": 24, "memory_bandwidth_gbps": 1008, "base_clock_mhz": 2235, "boost_clock_mhz": 2520, "sm_count": 128, "fp32_tflops": 82.6, "fp16_tflops": 165.2, "tdp_watts": 450, "compute_capability": 8.9, "l2_cache_mb": 72} | 0.028929 | 536,870,912 | 3,145,728 | 170.666667 | 2 | {"M": 256, "N": 1024, "K": 1024, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 1024) x (1024, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 1024, dtype=torch.float16, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA H100 SXM | {"gpu_name": "NVIDIA H100 SXM", "cuda_cores": 16896, "tensor_cores": 528, "memory_gb": 80, "memory_bandwidth_gbps": 3350, "base_clock_mhz": 1095, "boost_clock_mhz": 1830, "sm_count": 132, "fp32_tflops": 67, "fp16_tflops": 989, "tdp_watts": 700, "compute_capability": 9.0, "l2_cache_mb": 50} | 0.0658 | 536,870,912 | 3,145,728 | 170.666667 | 2 | {"M": 256, "N": 1024, "K": 1024, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 1024) x (1024, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 1024, dtype=torch.float16, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA H100 PCIe | {"gpu_name": "NVIDIA H100 PCIe", "cuda_cores": 14592, "tensor_cores": 456, "memory_gb": 80, "memory_bandwidth_gbps": 2039, "base_clock_mhz": 1095, "boost_clock_mhz": 1620, "sm_count": 114, "fp32_tflops": 48, "fp16_tflops": 756, "tdp_watts": 350, "compute_capability": 9.0, "l2_cache_mb": 50} | 0.044899 | 536,870,912 | 3,145,728 | 170.666667 | 2 | {"M": 256, "N": 1024, "K": 1024, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 1024) x (1024, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 1024, dtype=torch.float16, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA RTX A6000 | {"gpu_name": "NVIDIA RTX A6000", "cuda_cores": 10752, "tensor_cores": 336, "memory_gb": 48, "memory_bandwidth_gbps": 768, "base_clock_mhz": 1410, "boost_clock_mhz": 1860, "sm_count": 84, "fp32_tflops": 38.7, "fp16_tflops": 77.4, "tdp_watts": 300, "compute_capability": 8.6, "l2_cache_mb": 6} | 0.045417 | 536,870,912 | 3,145,728 | 170.666667 | 2 | {"M": 256, "N": 1024, "K": 1024, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 2048) x (2048, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 2048, dtype=torch.float32, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA T4 | {"gpu_name": "NVIDIA T4", "cuda_cores": 2560, "tensor_cores": 320, "memory_gb": 16, "memory_bandwidth_gbps": 320, "base_clock_mhz": 585, "boost_clock_mhz": 1590, "sm_count": 40, "fp32_tflops": 8.1, "fp16_tflops": 65, "tdp_watts": 70, "compute_capability": 7.5, "l2_cache_mb": 4} | 0.223073 | 1,073,741,824 | 11,534,336 | 93.090909 | 4 | {"M": 256, "N": 1024, "K": 2048, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 2048) x (2048, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 2048, dtype=torch.float32, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA V100 | {"gpu_name": "NVIDIA V100", "cuda_cores": 5120, "tensor_cores": 640, "memory_gb": 32, "memory_bandwidth_gbps": 900, "base_clock_mhz": 1230, "boost_clock_mhz": 1530, "sm_count": 80, "fp32_tflops": 15.7, "fp16_tflops": 125, "tdp_watts": 300, "compute_capability": 7.0, "l2_cache_mb": 6} | 0.149148 | 1,073,741,824 | 11,534,336 | 93.090909 | 4 | {"M": 256, "N": 1024, "K": 2048, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 2048) x (2048, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 2048, dtype=torch.float32, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA A10G | {"gpu_name": "NVIDIA A10G", "cuda_cores": 9216, "tensor_cores": 288, "memory_gb": 24, "memory_bandwidth_gbps": 600, "base_clock_mhz": 885, "boost_clock_mhz": 1695, "sm_count": 80, "fp32_tflops": 31.2, "fp16_tflops": 62.5, "tdp_watts": 150, "compute_capability": 8.6, "l2_cache_mb": 6} | 0.098288 | 1,073,741,824 | 11,534,336 | 93.090909 | 4 | {"M": 256, "N": 1024, "K": 2048, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 2048) x (2048, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 2048, dtype=torch.float32, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA A100 40GB | {"gpu_name": "NVIDIA A100 40GB", "cuda_cores": 6912, "tensor_cores": 432, "memory_gb": 40, "memory_bandwidth_gbps": 1555, "base_clock_mhz": 765, "boost_clock_mhz": 1410, "sm_count": 108, "fp32_tflops": 19.5, "fp16_tflops": 312, "tdp_watts": 400, "compute_capability": 8.0, "l2_cache_mb": 40} | 0.139782 | 1,073,741,824 | 11,534,336 | 93.090909 | 4 | {"M": 256, "N": 1024, "K": 2048, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 2048) x (2048, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 2048, dtype=torch.float32, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA A100 80GB | {"gpu_name": "NVIDIA A100 80GB", "cuda_cores": 6912, "tensor_cores": 432, "memory_gb": 80, "memory_bandwidth_gbps": 2039, "base_clock_mhz": 765, "boost_clock_mhz": 1410, "sm_count": 108, "fp32_tflops": 19.5, "fp16_tflops": 312, "tdp_watts": 400, "compute_capability": 8.0, "l2_cache_mb": 40} | 0.131503 | 1,073,741,824 | 11,534,336 | 93.090909 | 4 | {"M": 256, "N": 1024, "K": 2048, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 2048) x (2048, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 2048, dtype=torch.float32, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA L4 | {"gpu_name": "NVIDIA L4", "cuda_cores": 7424, "tensor_cores": 232, "memory_gb": 24, "memory_bandwidth_gbps": 300, "base_clock_mhz": 795, "boost_clock_mhz": 2040, "sm_count": 58, "fp32_tflops": 30.3, "fp16_tflops": 121, "tdp_watts": 72, "compute_capability": 8.9, "l2_cache_mb": 48} | 0.074402 | 1,073,741,824 | 11,534,336 | 93.090909 | 4 | {"M": 256, "N": 1024, "K": 2048, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 2048) x (2048, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 2048, dtype=torch.float32, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA L40S | {"gpu_name": "NVIDIA L40S", "cuda_cores": 18176, "tensor_cores": 568, "memory_gb": 48, "memory_bandwidth_gbps": 864, "base_clock_mhz": 1110, "boost_clock_mhz": 2520, "sm_count": 142, "fp32_tflops": 91.6, "fp16_tflops": 183.2, "tdp_watts": 350, "compute_capability": 8.9, "l2_cache_mb": 96} | 0.03041 | 1,073,741,824 | 11,534,336 | 93.090909 | 4 | {"M": 256, "N": 1024, "K": 2048, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 2048) x (2048, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 2048, dtype=torch.float32, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA RTX 3090 | {"gpu_name": "NVIDIA RTX 3090", "cuda_cores": 10496, "tensor_cores": 328, "memory_gb": 24, "memory_bandwidth_gbps": 936, "base_clock_mhz": 1395, "boost_clock_mhz": 1695, "sm_count": 82, "fp32_tflops": 35.6, "fp16_tflops": 71, "tdp_watts": 350, "compute_capability": 8.6, "l2_cache_mb": 6} | 0.122236 | 1,073,741,824 | 11,534,336 | 93.090909 | 4 | {"M": 256, "N": 1024, "K": 2048, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 2048) x (2048, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 2048, dtype=torch.float32, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA RTX 4090 | {"gpu_name": "NVIDIA RTX 4090", "cuda_cores": 16384, "tensor_cores": 512, "memory_gb": 24, "memory_bandwidth_gbps": 1008, "base_clock_mhz": 2235, "boost_clock_mhz": 2520, "sm_count": 128, "fp32_tflops": 82.6, "fp16_tflops": 165.2, "tdp_watts": 450, "compute_capability": 8.9, "l2_cache_mb": 72} | 0.063899 | 1,073,741,824 | 11,534,336 | 93.090909 | 4 | {"M": 256, "N": 1024, "K": 2048, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 2048) x (2048, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 2048, dtype=torch.float32, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA H100 SXM | {"gpu_name": "NVIDIA H100 SXM", "cuda_cores": 16896, "tensor_cores": 528, "memory_gb": 80, "memory_bandwidth_gbps": 3350, "base_clock_mhz": 1095, "boost_clock_mhz": 1830, "sm_count": 132, "fp32_tflops": 67, "fp16_tflops": 989, "tdp_watts": 700, "compute_capability": 9.0, "l2_cache_mb": 50} | 0.071266 | 1,073,741,824 | 11,534,336 | 93.090909 | 4 | {"M": 256, "N": 1024, "K": 2048, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 2048) x (2048, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 2048, dtype=torch.float32, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA H100 PCIe | {"gpu_name": "NVIDIA H100 PCIe", "cuda_cores": 14592, "tensor_cores": 456, "memory_gb": 80, "memory_bandwidth_gbps": 2039, "base_clock_mhz": 1095, "boost_clock_mhz": 1620, "sm_count": 114, "fp32_tflops": 48, "fp16_tflops": 756, "tdp_watts": 350, "compute_capability": 9.0, "l2_cache_mb": 50} | 0.089568 | 1,073,741,824 | 11,534,336 | 93.090909 | 4 | {"M": 256, "N": 1024, "K": 2048, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 2048) x (2048, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 2048, dtype=torch.float32, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA RTX A6000 | {"gpu_name": "NVIDIA RTX A6000", "cuda_cores": 10752, "tensor_cores": 336, "memory_gb": 48, "memory_bandwidth_gbps": 768, "base_clock_mhz": 1410, "boost_clock_mhz": 1860, "sm_count": 84, "fp32_tflops": 38.7, "fp16_tflops": 77.4, "tdp_watts": 300, "compute_capability": 8.6, "l2_cache_mb": 6} | 0.097675 | 1,073,741,824 | 11,534,336 | 93.090909 | 4 | {"M": 256, "N": 1024, "K": 2048, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 2048) x (2048, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 2048, dtype=torch.float16, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA T4 | {"gpu_name": "NVIDIA T4", "cuda_cores": 2560, "tensor_cores": 320, "memory_gb": 16, "memory_bandwidth_gbps": 320, "base_clock_mhz": 585, "boost_clock_mhz": 1590, "sm_count": 40, "fp32_tflops": 8.1, "fp16_tflops": 65, "tdp_watts": 70, "compute_capability": 7.5, "l2_cache_mb": 4} | 0.049446 | 1,073,741,824 | 5,767,168 | 186.181818 | 2 | {"M": 256, "N": 1024, "K": 2048, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 2048) x (2048, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 2048, dtype=torch.float16, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA V100 | {"gpu_name": "NVIDIA V100", "cuda_cores": 5120, "tensor_cores": 640, "memory_gb": 32, "memory_bandwidth_gbps": 900, "base_clock_mhz": 1230, "boost_clock_mhz": 1530, "sm_count": 80, "fp32_tflops": 15.7, "fp16_tflops": 125, "tdp_watts": 300, "compute_capability": 7.0, "l2_cache_mb": 6} | 0.048666 | 1,073,741,824 | 5,767,168 | 186.181818 | 2 | {"M": 256, "N": 1024, "K": 2048, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 2048) x (2048, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 2048, dtype=torch.float16, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA A10G | {"gpu_name": "NVIDIA A10G", "cuda_cores": 9216, "tensor_cores": 288, "memory_gb": 24, "memory_bandwidth_gbps": 600, "base_clock_mhz": 885, "boost_clock_mhz": 1695, "sm_count": 80, "fp32_tflops": 31.2, "fp16_tflops": 62.5, "tdp_watts": 150, "compute_capability": 8.6, "l2_cache_mb": 6} | 0.055164 | 1,073,741,824 | 5,767,168 | 186.181818 | 2 | {"M": 256, "N": 1024, "K": 2048, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 2048) x (2048, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 2048, dtype=torch.float16, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA A100 40GB | {"gpu_name": "NVIDIA A100 40GB", "cuda_cores": 6912, "tensor_cores": 432, "memory_gb": 40, "memory_bandwidth_gbps": 1555, "base_clock_mhz": 765, "boost_clock_mhz": 1410, "sm_count": 108, "fp32_tflops": 19.5, "fp16_tflops": 312, "tdp_watts": 400, "compute_capability": 8.0, "l2_cache_mb": 40} | 0.048786 | 1,073,741,824 | 5,767,168 | 186.181818 | 2 | {"M": 256, "N": 1024, "K": 2048, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 2048) x (2048, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 2048, dtype=torch.float16, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA A100 80GB | {"gpu_name": "NVIDIA A100 80GB", "cuda_cores": 6912, "tensor_cores": 432, "memory_gb": 80, "memory_bandwidth_gbps": 2039, "base_clock_mhz": 765, "boost_clock_mhz": 1410, "sm_count": 108, "fp32_tflops": 19.5, "fp16_tflops": 312, "tdp_watts": 400, "compute_capability": 8.0, "l2_cache_mb": 40} | 0.061502 | 1,073,741,824 | 5,767,168 | 186.181818 | 2 | {"M": 256, "N": 1024, "K": 2048, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 2048) x (2048, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 2048, dtype=torch.float16, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA L4 | {"gpu_name": "NVIDIA L4", "cuda_cores": 7424, "tensor_cores": 232, "memory_gb": 24, "memory_bandwidth_gbps": 300, "base_clock_mhz": 795, "boost_clock_mhz": 2040, "sm_count": 58, "fp32_tflops": 30.3, "fp16_tflops": 121, "tdp_watts": 72, "compute_capability": 8.9, "l2_cache_mb": 48} | 0.036876 | 1,073,741,824 | 5,767,168 | 186.181818 | 2 | {"M": 256, "N": 1024, "K": 2048, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 2048) x (2048, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 2048, dtype=torch.float16, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA L40S | {"gpu_name": "NVIDIA L40S", "cuda_cores": 18176, "tensor_cores": 568, "memory_gb": 48, "memory_bandwidth_gbps": 864, "base_clock_mhz": 1110, "boost_clock_mhz": 2520, "sm_count": 142, "fp32_tflops": 91.6, "fp16_tflops": 183.2, "tdp_watts": 350, "compute_capability": 8.9, "l2_cache_mb": 96} | 0.035561 | 1,073,741,824 | 5,767,168 | 186.181818 | 2 | {"M": 256, "N": 1024, "K": 2048, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 2048) x (2048, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 2048, dtype=torch.float16, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA RTX 3090 | {"gpu_name": "NVIDIA RTX 3090", "cuda_cores": 10496, "tensor_cores": 328, "memory_gb": 24, "memory_bandwidth_gbps": 936, "base_clock_mhz": 1395, "boost_clock_mhz": 1695, "sm_count": 82, "fp32_tflops": 35.6, "fp16_tflops": 71, "tdp_watts": 350, "compute_capability": 8.6, "l2_cache_mb": 6} | 0.054311 | 1,073,741,824 | 5,767,168 | 186.181818 | 2 | {"M": 256, "N": 1024, "K": 2048, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 2048) x (2048, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 2048, dtype=torch.float16, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA RTX 4090 | {"gpu_name": "NVIDIA RTX 4090", "cuda_cores": 16384, "tensor_cores": 512, "memory_gb": 24, "memory_bandwidth_gbps": 1008, "base_clock_mhz": 2235, "boost_clock_mhz": 2520, "sm_count": 128, "fp32_tflops": 82.6, "fp16_tflops": 165.2, "tdp_watts": 450, "compute_capability": 8.9, "l2_cache_mb": 72} | 0.053266 | 1,073,741,824 | 5,767,168 | 186.181818 | 2 | {"M": 256, "N": 1024, "K": 2048, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 2048) x (2048, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 2048, dtype=torch.float16, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA H100 SXM | {"gpu_name": "NVIDIA H100 SXM", "cuda_cores": 16896, "tensor_cores": 528, "memory_gb": 80, "memory_bandwidth_gbps": 3350, "base_clock_mhz": 1095, "boost_clock_mhz": 1830, "sm_count": 132, "fp32_tflops": 67, "fp16_tflops": 989, "tdp_watts": 700, "compute_capability": 9.0, "l2_cache_mb": 50} | 0.036583 | 1,073,741,824 | 5,767,168 | 186.181818 | 2 | {"M": 256, "N": 1024, "K": 2048, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 2048) x (2048, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 2048, dtype=torch.float16, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA H100 PCIe | {"gpu_name": "NVIDIA H100 PCIe", "cuda_cores": 14592, "tensor_cores": 456, "memory_gb": 80, "memory_bandwidth_gbps": 2039, "base_clock_mhz": 1095, "boost_clock_mhz": 1620, "sm_count": 114, "fp32_tflops": 48, "fp16_tflops": 756, "tdp_watts": 350, "compute_capability": 9.0, "l2_cache_mb": 50} | 0.034841 | 1,073,741,824 | 5,767,168 | 186.181818 | 2 | {"M": 256, "N": 1024, "K": 2048, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 2048) x (2048, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 2048, dtype=torch.float16, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA RTX A6000 | {"gpu_name": "NVIDIA RTX A6000", "cuda_cores": 10752, "tensor_cores": 336, "memory_gb": 48, "memory_bandwidth_gbps": 768, "base_clock_mhz": 1410, "boost_clock_mhz": 1860, "sm_count": 84, "fp32_tflops": 38.7, "fp16_tflops": 77.4, "tdp_watts": 300, "compute_capability": 8.6, "l2_cache_mb": 6} | 0.058204 | 1,073,741,824 | 5,767,168 | 186.181818 | 2 | {"M": 256, "N": 1024, "K": 2048, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 4096) x (4096, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 4096, dtype=torch.float32, device='cuda')
B = torch.randn(4096, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA T4 | {"gpu_name": "NVIDIA T4", "cuda_cores": 2560, "tensor_cores": 320, "memory_gb": 16, "memory_bandwidth_gbps": 320, "base_clock_mhz": 585, "boost_clock_mhz": 1590, "sm_count": 40, "fp32_tflops": 8.1, "fp16_tflops": 65, "tdp_watts": 70, "compute_capability": 7.5, "l2_cache_mb": 4} | 0.369775 | 2,147,483,648 | 22,020,096 | 97.52381 | 4 | {"M": 256, "N": 1024, "K": 4096, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 4096) x (4096, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 4096, dtype=torch.float32, device='cuda')
B = torch.randn(4096, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA V100 | {"gpu_name": "NVIDIA V100", "cuda_cores": 5120, "tensor_cores": 640, "memory_gb": 32, "memory_bandwidth_gbps": 900, "base_clock_mhz": 1230, "boost_clock_mhz": 1530, "sm_count": 80, "fp32_tflops": 15.7, "fp16_tflops": 125, "tdp_watts": 300, "compute_capability": 7.0, "l2_cache_mb": 6} | 0.272455 | 2,147,483,648 | 22,020,096 | 97.52381 | 4 | {"M": 256, "N": 1024, "K": 4096, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 4096) x (4096, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 4096, dtype=torch.float32, device='cuda')
B = torch.randn(4096, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA A10G | {"gpu_name": "NVIDIA A10G", "cuda_cores": 9216, "tensor_cores": 288, "memory_gb": 24, "memory_bandwidth_gbps": 600, "base_clock_mhz": 885, "boost_clock_mhz": 1695, "sm_count": 80, "fp32_tflops": 31.2, "fp16_tflops": 62.5, "tdp_watts": 150, "compute_capability": 8.6, "l2_cache_mb": 6} | 0.135179 | 2,147,483,648 | 22,020,096 | 97.52381 | 4 | {"M": 256, "N": 1024, "K": 4096, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 4096) x (4096, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 4096, dtype=torch.float32, device='cuda')
B = torch.randn(4096, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA A100 40GB | {"gpu_name": "NVIDIA A100 40GB", "cuda_cores": 6912, "tensor_cores": 432, "memory_gb": 40, "memory_bandwidth_gbps": 1555, "base_clock_mhz": 765, "boost_clock_mhz": 1410, "sm_count": 108, "fp32_tflops": 19.5, "fp16_tflops": 312, "tdp_watts": 400, "compute_capability": 8.0, "l2_cache_mb": 40} | 0.202454 | 2,147,483,648 | 22,020,096 | 97.52381 | 4 | {"M": 256, "N": 1024, "K": 4096, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 4096) x (4096, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 4096, dtype=torch.float32, device='cuda')
B = torch.randn(4096, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA A100 80GB | {"gpu_name": "NVIDIA A100 80GB", "cuda_cores": 6912, "tensor_cores": 432, "memory_gb": 80, "memory_bandwidth_gbps": 2039, "base_clock_mhz": 765, "boost_clock_mhz": 1410, "sm_count": 108, "fp32_tflops": 19.5, "fp16_tflops": 312, "tdp_watts": 400, "compute_capability": 8.0, "l2_cache_mb": 40} | 0.173996 | 2,147,483,648 | 22,020,096 | 97.52381 | 4 | {"M": 256, "N": 1024, "K": 4096, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 4096) x (4096, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 4096, dtype=torch.float32, device='cuda')
B = torch.randn(4096, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA L4 | {"gpu_name": "NVIDIA L4", "cuda_cores": 7424, "tensor_cores": 232, "memory_gb": 24, "memory_bandwidth_gbps": 300, "base_clock_mhz": 795, "boost_clock_mhz": 2040, "sm_count": 58, "fp32_tflops": 30.3, "fp16_tflops": 121, "tdp_watts": 72, "compute_capability": 8.9, "l2_cache_mb": 48} | 0.128389 | 2,147,483,648 | 22,020,096 | 97.52381 | 4 | {"M": 256, "N": 1024, "K": 4096, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 4096) x (4096, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 4096, dtype=torch.float32, device='cuda')
B = torch.randn(4096, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA L40S | {"gpu_name": "NVIDIA L40S", "cuda_cores": 18176, "tensor_cores": 568, "memory_gb": 48, "memory_bandwidth_gbps": 864, "base_clock_mhz": 1110, "boost_clock_mhz": 2520, "sm_count": 142, "fp32_tflops": 91.6, "fp16_tflops": 183.2, "tdp_watts": 350, "compute_capability": 8.9, "l2_cache_mb": 96} | 0.059709 | 2,147,483,648 | 22,020,096 | 97.52381 | 4 | {"M": 256, "N": 1024, "K": 4096, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 4096) x (4096, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 4096, dtype=torch.float32, device='cuda')
B = torch.randn(4096, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA RTX 3090 | {"gpu_name": "NVIDIA RTX 3090", "cuda_cores": 10496, "tensor_cores": 328, "memory_gb": 24, "memory_bandwidth_gbps": 936, "base_clock_mhz": 1395, "boost_clock_mhz": 1695, "sm_count": 82, "fp32_tflops": 35.6, "fp16_tflops": 71, "tdp_watts": 350, "compute_capability": 8.6, "l2_cache_mb": 6} | 0.12375 | 2,147,483,648 | 22,020,096 | 97.52381 | 4 | {"M": 256, "N": 1024, "K": 4096, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 4096) x (4096, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 4096, dtype=torch.float32, device='cuda')
B = torch.randn(4096, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA RTX 4090 | {"gpu_name": "NVIDIA RTX 4090", "cuda_cores": 16384, "tensor_cores": 512, "memory_gb": 24, "memory_bandwidth_gbps": 1008, "base_clock_mhz": 2235, "boost_clock_mhz": 2520, "sm_count": 128, "fp32_tflops": 82.6, "fp16_tflops": 165.2, "tdp_watts": 450, "compute_capability": 8.9, "l2_cache_mb": 72} | 0.044904 | 2,147,483,648 | 22,020,096 | 97.52381 | 4 | {"M": 256, "N": 1024, "K": 4096, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 4096) x (4096, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 4096, dtype=torch.float32, device='cuda')
B = torch.randn(4096, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA H100 SXM | {"gpu_name": "NVIDIA H100 SXM", "cuda_cores": 16896, "tensor_cores": 528, "memory_gb": 80, "memory_bandwidth_gbps": 3350, "base_clock_mhz": 1095, "boost_clock_mhz": 1830, "sm_count": 132, "fp32_tflops": 67, "fp16_tflops": 989, "tdp_watts": 700, "compute_capability": 9.0, "l2_cache_mb": 50} | 0.111096 | 2,147,483,648 | 22,020,096 | 97.52381 | 4 | {"M": 256, "N": 1024, "K": 4096, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 4096) x (4096, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 4096, dtype=torch.float32, device='cuda')
B = torch.randn(4096, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA H100 PCIe | {"gpu_name": "NVIDIA H100 PCIe", "cuda_cores": 14592, "tensor_cores": 456, "memory_gb": 80, "memory_bandwidth_gbps": 2039, "base_clock_mhz": 1095, "boost_clock_mhz": 1620, "sm_count": 114, "fp32_tflops": 48, "fp16_tflops": 756, "tdp_watts": 350, "compute_capability": 9.0, "l2_cache_mb": 50} | 0.099127 | 2,147,483,648 | 22,020,096 | 97.52381 | 4 | {"M": 256, "N": 1024, "K": 4096, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 4096) x (4096, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 4096, dtype=torch.float32, device='cuda')
B = torch.randn(4096, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA RTX A6000 | {"gpu_name": "NVIDIA RTX A6000", "cuda_cores": 10752, "tensor_cores": 336, "memory_gb": 48, "memory_bandwidth_gbps": 768, "base_clock_mhz": 1410, "boost_clock_mhz": 1860, "sm_count": 84, "fp32_tflops": 38.7, "fp16_tflops": 77.4, "tdp_watts": 300, "compute_capability": 8.6, "l2_cache_mb": 6} | 0.102002 | 2,147,483,648 | 22,020,096 | 97.52381 | 4 | {"M": 256, "N": 1024, "K": 4096, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 4096) x (4096, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 4096, dtype=torch.float16, device='cuda')
B = torch.randn(4096, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA T4 | {"gpu_name": "NVIDIA T4", "cuda_cores": 2560, "tensor_cores": 320, "memory_gb": 16, "memory_bandwidth_gbps": 320, "base_clock_mhz": 585, "boost_clock_mhz": 1590, "sm_count": 40, "fp32_tflops": 8.1, "fp16_tflops": 65, "tdp_watts": 70, "compute_capability": 7.5, "l2_cache_mb": 4} | 0.076579 | 2,147,483,648 | 11,010,048 | 195.047619 | 2 | {"M": 256, "N": 1024, "K": 4096, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 4096) x (4096, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 4096, dtype=torch.float16, device='cuda')
B = torch.randn(4096, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA V100 | {"gpu_name": "NVIDIA V100", "cuda_cores": 5120, "tensor_cores": 640, "memory_gb": 32, "memory_bandwidth_gbps": 900, "base_clock_mhz": 1230, "boost_clock_mhz": 1530, "sm_count": 80, "fp32_tflops": 15.7, "fp16_tflops": 125, "tdp_watts": 300, "compute_capability": 7.0, "l2_cache_mb": 6} | 0.075705 | 2,147,483,648 | 11,010,048 | 195.047619 | 2 | {"M": 256, "N": 1024, "K": 4096, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 4096) x (4096, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 4096, dtype=torch.float16, device='cuda')
B = torch.randn(4096, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA A10G | {"gpu_name": "NVIDIA A10G", "cuda_cores": 9216, "tensor_cores": 288, "memory_gb": 24, "memory_bandwidth_gbps": 600, "base_clock_mhz": 885, "boost_clock_mhz": 1695, "sm_count": 80, "fp32_tflops": 31.2, "fp16_tflops": 62.5, "tdp_watts": 150, "compute_capability": 8.6, "l2_cache_mb": 6} | 0.083112 | 2,147,483,648 | 11,010,048 | 195.047619 | 2 | {"M": 256, "N": 1024, "K": 4096, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 4096) x (4096, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 4096, dtype=torch.float16, device='cuda')
B = torch.randn(4096, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA A100 40GB | {"gpu_name": "NVIDIA A100 40GB", "cuda_cores": 6912, "tensor_cores": 432, "memory_gb": 40, "memory_bandwidth_gbps": 1555, "base_clock_mhz": 765, "boost_clock_mhz": 1410, "sm_count": 108, "fp32_tflops": 19.5, "fp16_tflops": 312, "tdp_watts": 400, "compute_capability": 8.0, "l2_cache_mb": 40} | 0.043039 | 2,147,483,648 | 11,010,048 | 195.047619 | 2 | {"M": 256, "N": 1024, "K": 4096, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 4096) x (4096, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 4096, dtype=torch.float16, device='cuda')
B = torch.randn(4096, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA A100 80GB | {"gpu_name": "NVIDIA A100 80GB", "cuda_cores": 6912, "tensor_cores": 432, "memory_gb": 80, "memory_bandwidth_gbps": 2039, "base_clock_mhz": 765, "boost_clock_mhz": 1410, "sm_count": 108, "fp32_tflops": 19.5, "fp16_tflops": 312, "tdp_watts": 400, "compute_capability": 8.0, "l2_cache_mb": 40} | 0.040183 | 2,147,483,648 | 11,010,048 | 195.047619 | 2 | {"M": 256, "N": 1024, "K": 4096, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 4096) x (4096, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 4096, dtype=torch.float16, device='cuda')
B = torch.randn(4096, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA L4 | {"gpu_name": "NVIDIA L4", "cuda_cores": 7424, "tensor_cores": 232, "memory_gb": 24, "memory_bandwidth_gbps": 300, "base_clock_mhz": 795, "boost_clock_mhz": 2040, "sm_count": 58, "fp32_tflops": 30.3, "fp16_tflops": 121, "tdp_watts": 72, "compute_capability": 8.9, "l2_cache_mb": 48} | 0.094673 | 2,147,483,648 | 11,010,048 | 195.047619 | 2 | {"M": 256, "N": 1024, "K": 4096, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 4096) x (4096, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 4096, dtype=torch.float16, device='cuda')
B = torch.randn(4096, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA L40S | {"gpu_name": "NVIDIA L40S", "cuda_cores": 18176, "tensor_cores": 568, "memory_gb": 48, "memory_bandwidth_gbps": 864, "base_clock_mhz": 1110, "boost_clock_mhz": 2520, "sm_count": 142, "fp32_tflops": 91.6, "fp16_tflops": 183.2, "tdp_watts": 350, "compute_capability": 8.9, "l2_cache_mb": 96} | 0.069504 | 2,147,483,648 | 11,010,048 | 195.047619 | 2 | {"M": 256, "N": 1024, "K": 4096, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (256, 4096) x (4096, 1024) -> (256, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(256, 4096, dtype=torch.float16, device='cuda')
B = torch.randn(4096, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchro... | matmul | NVIDIA RTX 3090 | {"gpu_name": "NVIDIA RTX 3090", "cuda_cores": 10496, "tensor_cores": 328, "memory_gb": 24, "memory_bandwidth_gbps": 936, "base_clock_mhz": 1395, "boost_clock_mhz": 1695, "sm_count": 82, "fp32_tflops": 35.6, "fp16_tflops": 71, "tdp_watts": 350, "compute_capability": 8.6, "l2_cache_mb": 6} | 0.080834 | 2,147,483,648 | 11,010,048 | 195.047619 | 2 | {"M": 256, "N": 1024, "K": 4096, "dtype": "float16"} |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.