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: (64, 512) x (512, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 512, dtype=torch.float32, device='cuda')
B = torch.randn(512, 256, 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} | 0.049165 | 16,777,216 | 720,896 | 23.272727 | 4 | {"M": 64, "N": 256, "K": 512, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 512) x (512, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 512, dtype=torch.float32, device='cuda')
B = torch.randn(512, 256, 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} | 0.056979 | 16,777,216 | 720,896 | 23.272727 | 4 | {"M": 64, "N": 256, "K": 512, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 512) x (512, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 512, dtype=torch.float32, device='cuda')
B = torch.randn(512, 256, 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} | 0.038767 | 16,777,216 | 720,896 | 23.272727 | 4 | {"M": 64, "N": 256, "K": 512, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 512) x (512, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 512, dtype=torch.float32, device='cuda')
B = torch.randn(512, 256, 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} | 0.04615 | 16,777,216 | 720,896 | 23.272727 | 4 | {"M": 64, "N": 256, "K": 512, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 512) x (512, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 512, dtype=torch.float16, device='cuda')
B = torch.randn(512, 256, 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} | 0.021489 | 16,777,216 | 360,448 | 46.545455 | 2 | {"M": 64, "N": 256, "K": 512, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 512) x (512, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 512, dtype=torch.float16, device='cuda')
B = torch.randn(512, 256, 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} | 0.048222 | 16,777,216 | 360,448 | 46.545455 | 2 | {"M": 64, "N": 256, "K": 512, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 512) x (512, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 512, dtype=torch.float16, device='cuda')
B = torch.randn(512, 256, 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} | 0.062246 | 16,777,216 | 360,448 | 46.545455 | 2 | {"M": 64, "N": 256, "K": 512, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 512) x (512, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 512, dtype=torch.float16, device='cuda')
B = torch.randn(512, 256, 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} | 0.053986 | 16,777,216 | 360,448 | 46.545455 | 2 | {"M": 64, "N": 256, "K": 512, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 512) x (512, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 512, dtype=torch.float16, device='cuda')
B = torch.randn(512, 256, 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} | 0.030852 | 16,777,216 | 360,448 | 46.545455 | 2 | {"M": 64, "N": 256, "K": 512, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 512) x (512, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 512, dtype=torch.float16, device='cuda')
B = torch.randn(512, 256, 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} | 0.02668 | 16,777,216 | 360,448 | 46.545455 | 2 | {"M": 64, "N": 256, "K": 512, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 512) x (512, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 512, dtype=torch.float16, device='cuda')
B = torch.randn(512, 256, 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} | 0.05201 | 16,777,216 | 360,448 | 46.545455 | 2 | {"M": 64, "N": 256, "K": 512, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 512) x (512, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 512, dtype=torch.float16, device='cuda')
B = torch.randn(512, 256, 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} | 0.038126 | 16,777,216 | 360,448 | 46.545455 | 2 | {"M": 64, "N": 256, "K": 512, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 512) x (512, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 512, dtype=torch.float16, device='cuda')
B = torch.randn(512, 256, 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} | 0.035524 | 16,777,216 | 360,448 | 46.545455 | 2 | {"M": 64, "N": 256, "K": 512, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 512) x (512, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 512, dtype=torch.float16, device='cuda')
B = torch.randn(512, 256, 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} | 0.022703 | 16,777,216 | 360,448 | 46.545455 | 2 | {"M": 64, "N": 256, "K": 512, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 512) x (512, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 512, dtype=torch.float16, device='cuda')
B = torch.randn(512, 256, 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} | 0.06638 | 16,777,216 | 360,448 | 46.545455 | 2 | {"M": 64, "N": 256, "K": 512, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 512) x (512, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 512, dtype=torch.float16, device='cuda')
B = torch.randn(512, 256, 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} | 0.054826 | 16,777,216 | 360,448 | 46.545455 | 2 | {"M": 64, "N": 256, "K": 512, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 1024) x (1024, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 1024, dtype=torch.float32, device='cuda')
B = torch.randn(1024, 256, 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} | 0.055149 | 33,554,432 | 1,376,256 | 24.380952 | 4 | {"M": 64, "N": 256, "K": 1024, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 1024) x (1024, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 1024, dtype=torch.float32, device='cuda')
B = torch.randn(1024, 256, 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} | 0.039236 | 33,554,432 | 1,376,256 | 24.380952 | 4 | {"M": 64, "N": 256, "K": 1024, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 1024) x (1024, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 1024, dtype=torch.float32, device='cuda')
B = torch.randn(1024, 256, 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} | 0.046769 | 33,554,432 | 1,376,256 | 24.380952 | 4 | {"M": 64, "N": 256, "K": 1024, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 1024) x (1024, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 1024, dtype=torch.float32, device='cuda')
B = torch.randn(1024, 256, 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} | 0.061069 | 33,554,432 | 1,376,256 | 24.380952 | 4 | {"M": 64, "N": 256, "K": 1024, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 1024) x (1024, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 1024, dtype=torch.float32, device='cuda')
B = torch.randn(1024, 256, 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} | 0.066096 | 33,554,432 | 1,376,256 | 24.380952 | 4 | {"M": 64, "N": 256, "K": 1024, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 1024) x (1024, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 1024, dtype=torch.float32, device='cuda')
B = torch.randn(1024, 256, 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} | 0.052327 | 33,554,432 | 1,376,256 | 24.380952 | 4 | {"M": 64, "N": 256, "K": 1024, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 1024) x (1024, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 1024, dtype=torch.float32, device='cuda')
B = torch.randn(1024, 256, 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} | 0.040008 | 33,554,432 | 1,376,256 | 24.380952 | 4 | {"M": 64, "N": 256, "K": 1024, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 1024) x (1024, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 1024, dtype=torch.float32, device='cuda')
B = torch.randn(1024, 256, 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} | 0.057005 | 33,554,432 | 1,376,256 | 24.380952 | 4 | {"M": 64, "N": 256, "K": 1024, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 1024) x (1024, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 1024, dtype=torch.float32, device='cuda')
B = torch.randn(1024, 256, 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} | 0.036749 | 33,554,432 | 1,376,256 | 24.380952 | 4 | {"M": 64, "N": 256, "K": 1024, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 1024) x (1024, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 1024, dtype=torch.float32, device='cuda')
B = torch.randn(1024, 256, 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} | 0.056228 | 33,554,432 | 1,376,256 | 24.380952 | 4 | {"M": 64, "N": 256, "K": 1024, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 1024) x (1024, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 1024, dtype=torch.float32, device='cuda')
B = torch.randn(1024, 256, 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} | 0.05398 | 33,554,432 | 1,376,256 | 24.380952 | 4 | {"M": 64, "N": 256, "K": 1024, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 1024) x (1024, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 1024, dtype=torch.float32, device='cuda')
B = torch.randn(1024, 256, 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} | 0.058631 | 33,554,432 | 1,376,256 | 24.380952 | 4 | {"M": 64, "N": 256, "K": 1024, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 1024) x (1024, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 1024, dtype=torch.float16, device='cuda')
B = torch.randn(1024, 256, 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} | 0.060572 | 33,554,432 | 688,128 | 48.761905 | 2 | {"M": 64, "N": 256, "K": 1024, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 1024) x (1024, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 1024, dtype=torch.float16, device='cuda')
B = torch.randn(1024, 256, 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} | 0.052072 | 33,554,432 | 688,128 | 48.761905 | 2 | {"M": 64, "N": 256, "K": 1024, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 1024) x (1024, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 1024, dtype=torch.float16, device='cuda')
B = torch.randn(1024, 256, 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} | 0.028381 | 33,554,432 | 688,128 | 48.761905 | 2 | {"M": 64, "N": 256, "K": 1024, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 1024) x (1024, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 1024, dtype=torch.float16, device='cuda')
B = torch.randn(1024, 256, 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} | 0.036174 | 33,554,432 | 688,128 | 48.761905 | 2 | {"M": 64, "N": 256, "K": 1024, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 1024) x (1024, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 1024, dtype=torch.float16, device='cuda')
B = torch.randn(1024, 256, 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} | 0.039502 | 33,554,432 | 688,128 | 48.761905 | 2 | {"M": 64, "N": 256, "K": 1024, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 1024) x (1024, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 1024, dtype=torch.float16, device='cuda')
B = torch.randn(1024, 256, 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} | 0.031324 | 33,554,432 | 688,128 | 48.761905 | 2 | {"M": 64, "N": 256, "K": 1024, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 1024) x (1024, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 1024, dtype=torch.float16, device='cuda')
B = torch.randn(1024, 256, 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} | 0.055736 | 33,554,432 | 688,128 | 48.761905 | 2 | {"M": 64, "N": 256, "K": 1024, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 1024) x (1024, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 1024, dtype=torch.float16, device='cuda')
B = torch.randn(1024, 256, 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} | 0.035204 | 33,554,432 | 688,128 | 48.761905 | 2 | {"M": 64, "N": 256, "K": 1024, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 1024) x (1024, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 1024, dtype=torch.float16, device='cuda')
B = torch.randn(1024, 256, 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} | 0.041777 | 33,554,432 | 688,128 | 48.761905 | 2 | {"M": 64, "N": 256, "K": 1024, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 1024) x (1024, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 1024, dtype=torch.float16, device='cuda')
B = torch.randn(1024, 256, 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} | 0.05386 | 33,554,432 | 688,128 | 48.761905 | 2 | {"M": 64, "N": 256, "K": 1024, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 1024) x (1024, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 1024, dtype=torch.float16, device='cuda')
B = torch.randn(1024, 256, 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} | 0.059521 | 33,554,432 | 688,128 | 48.761905 | 2 | {"M": 64, "N": 256, "K": 1024, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 1024) x (1024, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 1024, dtype=torch.float16, device='cuda')
B = torch.randn(1024, 256, 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} | 0.023428 | 33,554,432 | 688,128 | 48.761905 | 2 | {"M": 64, "N": 256, "K": 1024, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 2048) x (2048, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 2048, dtype=torch.float32, device='cuda')
B = torch.randn(2048, 256, 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} | 0.043281 | 67,108,864 | 2,686,976 | 24.97561 | 4 | {"M": 64, "N": 256, "K": 2048, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 2048) x (2048, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 2048, dtype=torch.float32, device='cuda')
B = torch.randn(2048, 256, 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} | 0.062969 | 67,108,864 | 2,686,976 | 24.97561 | 4 | {"M": 64, "N": 256, "K": 2048, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 2048) x (2048, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 2048, dtype=torch.float32, device='cuda')
B = torch.randn(2048, 256, 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} | 0.027027 | 67,108,864 | 2,686,976 | 24.97561 | 4 | {"M": 64, "N": 256, "K": 2048, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 2048) x (2048, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 2048, dtype=torch.float32, device='cuda')
B = torch.randn(2048, 256, 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} | 0.059784 | 67,108,864 | 2,686,976 | 24.97561 | 4 | {"M": 64, "N": 256, "K": 2048, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 2048) x (2048, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 2048, dtype=torch.float32, device='cuda')
B = torch.randn(2048, 256, 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} | 0.029657 | 67,108,864 | 2,686,976 | 24.97561 | 4 | {"M": 64, "N": 256, "K": 2048, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 2048) x (2048, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 2048, dtype=torch.float32, device='cuda')
B = torch.randn(2048, 256, 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} | 0.043819 | 67,108,864 | 2,686,976 | 24.97561 | 4 | {"M": 64, "N": 256, "K": 2048, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 2048) x (2048, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 2048, dtype=torch.float32, device='cuda')
B = torch.randn(2048, 256, 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} | 0.029901 | 67,108,864 | 2,686,976 | 24.97561 | 4 | {"M": 64, "N": 256, "K": 2048, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 2048) x (2048, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 2048, dtype=torch.float32, device='cuda')
B = torch.randn(2048, 256, 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} | 0.057408 | 67,108,864 | 2,686,976 | 24.97561 | 4 | {"M": 64, "N": 256, "K": 2048, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 2048) x (2048, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 2048, dtype=torch.float32, device='cuda')
B = torch.randn(2048, 256, 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} | 0.0407 | 67,108,864 | 2,686,976 | 24.97561 | 4 | {"M": 64, "N": 256, "K": 2048, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 2048) x (2048, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 2048, dtype=torch.float32, device='cuda')
B = torch.randn(2048, 256, 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} | 0.028206 | 67,108,864 | 2,686,976 | 24.97561 | 4 | {"M": 64, "N": 256, "K": 2048, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 2048) x (2048, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 2048, dtype=torch.float32, device='cuda')
B = torch.randn(2048, 256, 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} | 0.04728 | 67,108,864 | 2,686,976 | 24.97561 | 4 | {"M": 64, "N": 256, "K": 2048, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 2048) x (2048, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 2048, dtype=torch.float32, device='cuda')
B = torch.randn(2048, 256, 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} | 0.033361 | 67,108,864 | 2,686,976 | 24.97561 | 4 | {"M": 64, "N": 256, "K": 2048, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 2048) x (2048, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 2048, dtype=torch.float16, device='cuda')
B = torch.randn(2048, 256, 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} | 0.046028 | 67,108,864 | 1,343,488 | 49.95122 | 2 | {"M": 64, "N": 256, "K": 2048, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 2048) x (2048, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 2048, dtype=torch.float16, device='cuda')
B = torch.randn(2048, 256, 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} | 0.054945 | 67,108,864 | 1,343,488 | 49.95122 | 2 | {"M": 64, "N": 256, "K": 2048, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 2048) x (2048, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 2048, dtype=torch.float16, device='cuda')
B = torch.randn(2048, 256, 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} | 0.064176 | 67,108,864 | 1,343,488 | 49.95122 | 2 | {"M": 64, "N": 256, "K": 2048, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 2048) x (2048, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 2048, dtype=torch.float16, device='cuda')
B = torch.randn(2048, 256, 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} | 0.072419 | 67,108,864 | 1,343,488 | 49.95122 | 2 | {"M": 64, "N": 256, "K": 2048, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 2048) x (2048, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 2048, dtype=torch.float16, device='cuda')
B = torch.randn(2048, 256, 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} | 0.057027 | 67,108,864 | 1,343,488 | 49.95122 | 2 | {"M": 64, "N": 256, "K": 2048, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 2048) x (2048, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 2048, dtype=torch.float16, device='cuda')
B = torch.randn(2048, 256, 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} | 0.040797 | 67,108,864 | 1,343,488 | 49.95122 | 2 | {"M": 64, "N": 256, "K": 2048, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 2048) x (2048, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 2048, dtype=torch.float16, device='cuda')
B = torch.randn(2048, 256, 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} | 0.049649 | 67,108,864 | 1,343,488 | 49.95122 | 2 | {"M": 64, "N": 256, "K": 2048, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 2048) x (2048, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 2048, dtype=torch.float16, device='cuda')
B = torch.randn(2048, 256, 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} | 0.035863 | 67,108,864 | 1,343,488 | 49.95122 | 2 | {"M": 64, "N": 256, "K": 2048, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 2048) x (2048, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 2048, dtype=torch.float16, device='cuda')
B = torch.randn(2048, 256, 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} | 0.031909 | 67,108,864 | 1,343,488 | 49.95122 | 2 | {"M": 64, "N": 256, "K": 2048, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 2048) x (2048, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 2048, dtype=torch.float16, device='cuda')
B = torch.randn(2048, 256, 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} | 0.026621 | 67,108,864 | 1,343,488 | 49.95122 | 2 | {"M": 64, "N": 256, "K": 2048, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 2048) x (2048, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 2048, dtype=torch.float16, device='cuda')
B = torch.randn(2048, 256, 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} | 0.056302 | 67,108,864 | 1,343,488 | 49.95122 | 2 | {"M": 64, "N": 256, "K": 2048, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 2048) x (2048, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 2048, dtype=torch.float16, device='cuda')
B = torch.randn(2048, 256, 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} | 0.046516 | 67,108,864 | 1,343,488 | 49.95122 | 2 | {"M": 64, "N": 256, "K": 2048, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 4096) x (4096, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 4096, dtype=torch.float32, device='cuda')
B = torch.randn(4096, 256, 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} | 0.069168 | 134,217,728 | 5,308,416 | 25.283951 | 4 | {"M": 64, "N": 256, "K": 4096, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 4096) x (4096, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 4096, dtype=torch.float32, device='cuda')
B = torch.randn(4096, 256, 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} | 0.043128 | 134,217,728 | 5,308,416 | 25.283951 | 4 | {"M": 64, "N": 256, "K": 4096, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 4096) x (4096, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 4096, dtype=torch.float32, device='cuda')
B = torch.randn(4096, 256, 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} | 0.069554 | 134,217,728 | 5,308,416 | 25.283951 | 4 | {"M": 64, "N": 256, "K": 4096, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 4096) x (4096, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 4096, dtype=torch.float32, device='cuda')
B = torch.randn(4096, 256, 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} | 0.034861 | 134,217,728 | 5,308,416 | 25.283951 | 4 | {"M": 64, "N": 256, "K": 4096, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 4096) x (4096, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 4096, dtype=torch.float32, device='cuda')
B = torch.randn(4096, 256, 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} | 0.048369 | 134,217,728 | 5,308,416 | 25.283951 | 4 | {"M": 64, "N": 256, "K": 4096, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 4096) x (4096, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 4096, dtype=torch.float32, device='cuda')
B = torch.randn(4096, 256, 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} | 0.072946 | 134,217,728 | 5,308,416 | 25.283951 | 4 | {"M": 64, "N": 256, "K": 4096, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 4096) x (4096, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 4096, dtype=torch.float32, device='cuda')
B = torch.randn(4096, 256, 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} | 0.057719 | 134,217,728 | 5,308,416 | 25.283951 | 4 | {"M": 64, "N": 256, "K": 4096, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 4096) x (4096, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 4096, dtype=torch.float32, device='cuda')
B = torch.randn(4096, 256, 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} | 0.057821 | 134,217,728 | 5,308,416 | 25.283951 | 4 | {"M": 64, "N": 256, "K": 4096, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 4096) x (4096, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 4096, dtype=torch.float32, device='cuda')
B = torch.randn(4096, 256, 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} | 0.034409 | 134,217,728 | 5,308,416 | 25.283951 | 4 | {"M": 64, "N": 256, "K": 4096, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 4096) x (4096, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 4096, dtype=torch.float32, device='cuda')
B = torch.randn(4096, 256, 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} | 0.031324 | 134,217,728 | 5,308,416 | 25.283951 | 4 | {"M": 64, "N": 256, "K": 4096, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 4096) x (4096, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 4096, dtype=torch.float32, device='cuda')
B = torch.randn(4096, 256, 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} | 0.024861 | 134,217,728 | 5,308,416 | 25.283951 | 4 | {"M": 64, "N": 256, "K": 4096, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 4096) x (4096, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 4096, dtype=torch.float32, device='cuda')
B = torch.randn(4096, 256, 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} | 0.059956 | 134,217,728 | 5,308,416 | 25.283951 | 4 | {"M": 64, "N": 256, "K": 4096, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 4096) x (4096, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 4096, dtype=torch.float16, device='cuda')
B = torch.randn(4096, 256, 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} | 0.042935 | 134,217,728 | 2,654,208 | 50.567901 | 2 | {"M": 64, "N": 256, "K": 4096, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 4096) x (4096, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 4096, dtype=torch.float16, device='cuda')
B = torch.randn(4096, 256, 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} | 0.04326 | 134,217,728 | 2,654,208 | 50.567901 | 2 | {"M": 64, "N": 256, "K": 4096, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 4096) x (4096, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 4096, dtype=torch.float16, device='cuda')
B = torch.randn(4096, 256, 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} | 0.050983 | 134,217,728 | 2,654,208 | 50.567901 | 2 | {"M": 64, "N": 256, "K": 4096, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 4096) x (4096, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 4096, dtype=torch.float16, device='cuda')
B = torch.randn(4096, 256, 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} | 0.035342 | 134,217,728 | 2,654,208 | 50.567901 | 2 | {"M": 64, "N": 256, "K": 4096, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 4096) x (4096, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 4096, dtype=torch.float16, device='cuda')
B = torch.randn(4096, 256, 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} | 0.039259 | 134,217,728 | 2,654,208 | 50.567901 | 2 | {"M": 64, "N": 256, "K": 4096, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 4096) x (4096, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 4096, dtype=torch.float16, device='cuda')
B = torch.randn(4096, 256, 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} | 0.030835 | 134,217,728 | 2,654,208 | 50.567901 | 2 | {"M": 64, "N": 256, "K": 4096, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 4096) x (4096, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 4096, dtype=torch.float16, device='cuda')
B = torch.randn(4096, 256, 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} | 0.057169 | 134,217,728 | 2,654,208 | 50.567901 | 2 | {"M": 64, "N": 256, "K": 4096, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 4096) x (4096, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 4096, dtype=torch.float16, device='cuda')
B = torch.randn(4096, 256, 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} | 0.034104 | 134,217,728 | 2,654,208 | 50.567901 | 2 | {"M": 64, "N": 256, "K": 4096, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 4096) x (4096, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 4096, dtype=torch.float16, device='cuda')
B = torch.randn(4096, 256, 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} | 0.055529 | 134,217,728 | 2,654,208 | 50.567901 | 2 | {"M": 64, "N": 256, "K": 4096, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 4096) x (4096, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 4096, dtype=torch.float16, device='cuda')
B = torch.randn(4096, 256, 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} | 0.05032 | 134,217,728 | 2,654,208 | 50.567901 | 2 | {"M": 64, "N": 256, "K": 4096, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 4096) x (4096, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 4096, dtype=torch.float16, device='cuda')
B = torch.randn(4096, 256, 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} | 0.065532 | 134,217,728 | 2,654,208 | 50.567901 | 2 | {"M": 64, "N": 256, "K": 4096, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 4096) x (4096, 256) -> (64, 256)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 4096, dtype=torch.float16, device='cuda')
B = torch.randn(4096, 256, 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} | 0.03552 | 134,217,728 | 2,654,208 | 50.567901 | 2 | {"M": 64, "N": 256, "K": 4096, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 64) x (64, 512) -> (64, 512)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 64, dtype=torch.float32, device='cuda')
B = torch.randn(64, 512, 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} | 0.055172 | 4,194,304 | 278,528 | 15.058824 | 4 | {"M": 64, "N": 512, "K": 64, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 64) x (64, 512) -> (64, 512)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 64, dtype=torch.float32, device='cuda')
B = torch.randn(64, 512, 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} | 0.018334 | 4,194,304 | 278,528 | 15.058824 | 4 | {"M": 64, "N": 512, "K": 64, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 64) x (64, 512) -> (64, 512)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 64, dtype=torch.float32, device='cuda')
B = torch.randn(64, 512, 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} | 0.043196 | 4,194,304 | 278,528 | 15.058824 | 4 | {"M": 64, "N": 512, "K": 64, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 64) x (64, 512) -> (64, 512)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 64, dtype=torch.float32, device='cuda')
B = torch.randn(64, 512, 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} | 0.046326 | 4,194,304 | 278,528 | 15.058824 | 4 | {"M": 64, "N": 512, "K": 64, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 64) x (64, 512) -> (64, 512)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 64, dtype=torch.float32, device='cuda')
B = torch.randn(64, 512, 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} | 0.052682 | 4,194,304 | 278,528 | 15.058824 | 4 | {"M": 64, "N": 512, "K": 64, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 64) x (64, 512) -> (64, 512)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 64, dtype=torch.float32, device='cuda')
B = torch.randn(64, 512, 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} | 0.036135 | 4,194,304 | 278,528 | 15.058824 | 4 | {"M": 64, "N": 512, "K": 64, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 64) x (64, 512) -> (64, 512)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 64, dtype=torch.float32, device='cuda')
B = torch.randn(64, 512, 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} | 0.028839 | 4,194,304 | 278,528 | 15.058824 | 4 | {"M": 64, "N": 512, "K": 64, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 64) x (64, 512) -> (64, 512)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 64, dtype=torch.float32, device='cuda')
B = torch.randn(64, 512, 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} | 0.045766 | 4,194,304 | 278,528 | 15.058824 | 4 | {"M": 64, "N": 512, "K": 64, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 64) x (64, 512) -> (64, 512)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 64, dtype=torch.float32, device='cuda')
B = torch.randn(64, 512, 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} | 0.017405 | 4,194,304 | 278,528 | 15.058824 | 4 | {"M": 64, "N": 512, "K": 64, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 64) x (64, 512) -> (64, 512)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 64, dtype=torch.float32, device='cuda')
B = torch.randn(64, 512, 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} | 0.054943 | 4,194,304 | 278,528 | 15.058824 | 4 | {"M": 64, "N": 512, "K": 64, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 64) x (64, 512) -> (64, 512)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 64, dtype=torch.float32, device='cuda')
B = torch.randn(64, 512, 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} | 0.056278 | 4,194,304 | 278,528 | 15.058824 | 4 | {"M": 64, "N": 512, "K": 64, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (64, 64) x (64, 512) -> (64, 512)
C = torch.matmul(A, B)
return C
A = torch.randn(64, 64, dtype=torch.float32, device='cuda')
B = torch.randn(64, 512, 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} | 0.054808 | 4,194,304 | 278,528 | 15.058824 | 4 | {"M": 64, "N": 512, "K": 64, "dtype": "float32"} |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.