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: (2048, 512) x (512, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 512, dtype=torch.float32, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchron... | 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} | 0.355793 | 2,147,483,648 | 14,680,064 | 146.285714 | 4 | {"M": 2048, "N": 1024, "K": 512, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 512) x (512, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 512, dtype=torch.float32, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchron... | 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} | 0.255535 | 2,147,483,648 | 14,680,064 | 146.285714 | 4 | {"M": 2048, "N": 1024, "K": 512, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 512) x (512, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 512, dtype=torch.float32, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchron... | 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} | 0.159802 | 2,147,483,648 | 14,680,064 | 146.285714 | 4 | {"M": 2048, "N": 1024, "K": 512, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 512) x (512, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 512, dtype=torch.float32, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchron... | 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} | 0.201706 | 2,147,483,648 | 14,680,064 | 146.285714 | 4 | {"M": 2048, "N": 1024, "K": 512, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 512) x (512, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 512, dtype=torch.float32, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchron... | 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} | 0.188347 | 2,147,483,648 | 14,680,064 | 146.285714 | 4 | {"M": 2048, "N": 1024, "K": 512, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 512) x (512, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 512, dtype=torch.float32, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchron... | 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} | 0.155098 | 2,147,483,648 | 14,680,064 | 146.285714 | 4 | {"M": 2048, "N": 1024, "K": 512, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 512) x (512, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 512, dtype=torch.float32, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchron... | 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} | 0.056151 | 2,147,483,648 | 14,680,064 | 146.285714 | 4 | {"M": 2048, "N": 1024, "K": 512, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 512) x (512, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 512, dtype=torch.float32, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchron... | 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} | 0.123431 | 2,147,483,648 | 14,680,064 | 146.285714 | 4 | {"M": 2048, "N": 1024, "K": 512, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 512) x (512, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 512, dtype=torch.float32, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchron... | 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} | 0.061808 | 2,147,483,648 | 14,680,064 | 146.285714 | 4 | {"M": 2048, "N": 1024, "K": 512, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 512) x (512, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 512, dtype=torch.float32, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchron... | 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} | 0.082201 | 2,147,483,648 | 14,680,064 | 146.285714 | 4 | {"M": 2048, "N": 1024, "K": 512, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 512) x (512, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 512, dtype=torch.float32, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchron... | 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} | 0.090833 | 2,147,483,648 | 14,680,064 | 146.285714 | 4 | {"M": 2048, "N": 1024, "K": 512, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 512) x (512, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 512, dtype=torch.float32, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchron... | 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} | 0.090532 | 2,147,483,648 | 14,680,064 | 146.285714 | 4 | {"M": 2048, "N": 1024, "K": 512, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 512) x (512, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 512, dtype=torch.float16, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchron... | 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} | 0.104693 | 2,147,483,648 | 7,340,032 | 292.571429 | 2 | {"M": 2048, "N": 1024, "K": 512, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 512) x (512, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 512, dtype=torch.float16, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchron... | 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} | 0.064971 | 2,147,483,648 | 7,340,032 | 292.571429 | 2 | {"M": 2048, "N": 1024, "K": 512, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 512) x (512, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 512, dtype=torch.float16, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchron... | 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} | 0.090953 | 2,147,483,648 | 7,340,032 | 292.571429 | 2 | {"M": 2048, "N": 1024, "K": 512, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 512) x (512, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 512, dtype=torch.float16, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchron... | 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} | 0.070333 | 2,147,483,648 | 7,340,032 | 292.571429 | 2 | {"M": 2048, "N": 1024, "K": 512, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 512) x (512, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 512, dtype=torch.float16, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchron... | 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} | 0.06316 | 2,147,483,648 | 7,340,032 | 292.571429 | 2 | {"M": 2048, "N": 1024, "K": 512, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 512) x (512, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 512, dtype=torch.float16, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchron... | 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} | 0.067199 | 2,147,483,648 | 7,340,032 | 292.571429 | 2 | {"M": 2048, "N": 1024, "K": 512, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 512) x (512, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 512, dtype=torch.float16, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchron... | 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} | 0.041334 | 2,147,483,648 | 7,340,032 | 292.571429 | 2 | {"M": 2048, "N": 1024, "K": 512, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 512) x (512, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 512, dtype=torch.float16, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchron... | 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} | 0.066485 | 2,147,483,648 | 7,340,032 | 292.571429 | 2 | {"M": 2048, "N": 1024, "K": 512, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 512) x (512, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 512, dtype=torch.float16, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchron... | 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} | 0.057631 | 2,147,483,648 | 7,340,032 | 292.571429 | 2 | {"M": 2048, "N": 1024, "K": 512, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 512) x (512, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 512, dtype=torch.float16, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchron... | 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} | 0.05663 | 2,147,483,648 | 7,340,032 | 292.571429 | 2 | {"M": 2048, "N": 1024, "K": 512, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 512) x (512, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 512, dtype=torch.float16, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchron... | 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} | 0.033173 | 2,147,483,648 | 7,340,032 | 292.571429 | 2 | {"M": 2048, "N": 1024, "K": 512, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 512) x (512, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 512, dtype=torch.float16, device='cuda')
B = torch.randn(512, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.synchron... | 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} | 0.063271 | 2,147,483,648 | 7,340,032 | 292.571429 | 2 | {"M": 2048, "N": 1024, "K": 512, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 1024) x (1024, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 1024, dtype=torch.float32, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.988059 | 4,294,967,296 | 20,971,520 | 204.8 | 4 | {"M": 2048, "N": 1024, "K": 1024, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 1024) x (1024, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 1024, dtype=torch.float32, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.452706 | 4,294,967,296 | 20,971,520 | 204.8 | 4 | {"M": 2048, "N": 1024, "K": 1024, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 1024) x (1024, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 1024, dtype=torch.float32, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.186777 | 4,294,967,296 | 20,971,520 | 204.8 | 4 | {"M": 2048, "N": 1024, "K": 1024, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 1024) x (1024, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 1024, dtype=torch.float32, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.30725 | 4,294,967,296 | 20,971,520 | 204.8 | 4 | {"M": 2048, "N": 1024, "K": 1024, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 1024) x (1024, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 1024, dtype=torch.float32, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.356122 | 4,294,967,296 | 20,971,520 | 204.8 | 4 | {"M": 2048, "N": 1024, "K": 1024, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 1024) x (1024, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 1024, dtype=torch.float32, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.205188 | 4,294,967,296 | 20,971,520 | 204.8 | 4 | {"M": 2048, "N": 1024, "K": 1024, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 1024) x (1024, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 1024, dtype=torch.float32, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.124461 | 4,294,967,296 | 20,971,520 | 204.8 | 4 | {"M": 2048, "N": 1024, "K": 1024, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 1024) x (1024, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 1024, dtype=torch.float32, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.211688 | 4,294,967,296 | 20,971,520 | 204.8 | 4 | {"M": 2048, "N": 1024, "K": 1024, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 1024) x (1024, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 1024, dtype=torch.float32, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.112026 | 4,294,967,296 | 20,971,520 | 204.8 | 4 | {"M": 2048, "N": 1024, "K": 1024, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 1024) x (1024, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 1024, dtype=torch.float32, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.149623 | 4,294,967,296 | 20,971,520 | 204.8 | 4 | {"M": 2048, "N": 1024, "K": 1024, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 1024) x (1024, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 1024, dtype=torch.float32, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.156392 | 4,294,967,296 | 20,971,520 | 204.8 | 4 | {"M": 2048, "N": 1024, "K": 1024, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 1024) x (1024, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 1024, dtype=torch.float32, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.203608 | 4,294,967,296 | 20,971,520 | 204.8 | 4 | {"M": 2048, "N": 1024, "K": 1024, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 1024) x (1024, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 1024, dtype=torch.float16, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.161047 | 4,294,967,296 | 10,485,760 | 409.6 | 2 | {"M": 2048, "N": 1024, "K": 1024, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 1024) x (1024, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 1024, dtype=torch.float16, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.092472 | 4,294,967,296 | 10,485,760 | 409.6 | 2 | {"M": 2048, "N": 1024, "K": 1024, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 1024) x (1024, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 1024, dtype=torch.float16, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.103046 | 4,294,967,296 | 10,485,760 | 409.6 | 2 | {"M": 2048, "N": 1024, "K": 1024, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 1024) x (1024, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 1024, dtype=torch.float16, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.0843 | 4,294,967,296 | 10,485,760 | 409.6 | 2 | {"M": 2048, "N": 1024, "K": 1024, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 1024) x (1024, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 1024, dtype=torch.float16, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.030807 | 4,294,967,296 | 10,485,760 | 409.6 | 2 | {"M": 2048, "N": 1024, "K": 1024, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 1024) x (1024, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 1024, dtype=torch.float16, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.123807 | 4,294,967,296 | 10,485,760 | 409.6 | 2 | {"M": 2048, "N": 1024, "K": 1024, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 1024) x (1024, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 1024, dtype=torch.float16, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.074093 | 4,294,967,296 | 10,485,760 | 409.6 | 2 | {"M": 2048, "N": 1024, "K": 1024, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 1024) x (1024, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 1024, dtype=torch.float16, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.158313 | 4,294,967,296 | 10,485,760 | 409.6 | 2 | {"M": 2048, "N": 1024, "K": 1024, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 1024) x (1024, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 1024, dtype=torch.float16, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.057054 | 4,294,967,296 | 10,485,760 | 409.6 | 2 | {"M": 2048, "N": 1024, "K": 1024, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 1024) x (1024, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 1024, dtype=torch.float16, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.055129 | 4,294,967,296 | 10,485,760 | 409.6 | 2 | {"M": 2048, "N": 1024, "K": 1024, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 1024) x (1024, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 1024, dtype=torch.float16, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.04606 | 4,294,967,296 | 10,485,760 | 409.6 | 2 | {"M": 2048, "N": 1024, "K": 1024, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 1024) x (1024, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 1024, dtype=torch.float16, device='cuda')
B = torch.randn(1024, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.085145 | 4,294,967,296 | 10,485,760 | 409.6 | 2 | {"M": 2048, "N": 1024, "K": 1024, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 2048) x (2048, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 2048, dtype=torch.float32, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 1.559171 | 8,589,934,592 | 33,554,432 | 256 | 4 | {"M": 2048, "N": 1024, "K": 2048, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 2048) x (2048, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 2048, dtype=torch.float32, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.886449 | 8,589,934,592 | 33,554,432 | 256 | 4 | {"M": 2048, "N": 1024, "K": 2048, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 2048) x (2048, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 2048, dtype=torch.float32, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.519752 | 8,589,934,592 | 33,554,432 | 256 | 4 | {"M": 2048, "N": 1024, "K": 2048, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 2048) x (2048, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 2048, dtype=torch.float32, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.753181 | 8,589,934,592 | 33,554,432 | 256 | 4 | {"M": 2048, "N": 1024, "K": 2048, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 2048) x (2048, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 2048, dtype=torch.float32, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.686981 | 8,589,934,592 | 33,554,432 | 256 | 4 | {"M": 2048, "N": 1024, "K": 2048, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 2048) x (2048, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 2048, dtype=torch.float32, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.453385 | 8,589,934,592 | 33,554,432 | 256 | 4 | {"M": 2048, "N": 1024, "K": 2048, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 2048) x (2048, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 2048, dtype=torch.float32, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.171888 | 8,589,934,592 | 33,554,432 | 256 | 4 | {"M": 2048, "N": 1024, "K": 2048, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 2048) x (2048, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 2048, dtype=torch.float32, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.437884 | 8,589,934,592 | 33,554,432 | 256 | 4 | {"M": 2048, "N": 1024, "K": 2048, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 2048) x (2048, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 2048, dtype=torch.float32, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.138021 | 8,589,934,592 | 33,554,432 | 256 | 4 | {"M": 2048, "N": 1024, "K": 2048, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 2048) x (2048, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 2048, dtype=torch.float32, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.217042 | 8,589,934,592 | 33,554,432 | 256 | 4 | {"M": 2048, "N": 1024, "K": 2048, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 2048) x (2048, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 2048, dtype=torch.float32, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.229313 | 8,589,934,592 | 33,554,432 | 256 | 4 | {"M": 2048, "N": 1024, "K": 2048, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 2048) x (2048, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 2048, dtype=torch.float32, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.321843 | 8,589,934,592 | 33,554,432 | 256 | 4 | {"M": 2048, "N": 1024, "K": 2048, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 2048) x (2048, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 2048, dtype=torch.float16, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.279987 | 8,589,934,592 | 16,777,216 | 512 | 2 | {"M": 2048, "N": 1024, "K": 2048, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 2048) x (2048, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 2048, dtype=torch.float16, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.176137 | 8,589,934,592 | 16,777,216 | 512 | 2 | {"M": 2048, "N": 1024, "K": 2048, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 2048) x (2048, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 2048, dtype=torch.float16, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.206413 | 8,589,934,592 | 16,777,216 | 512 | 2 | {"M": 2048, "N": 1024, "K": 2048, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 2048) x (2048, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 2048, dtype=torch.float16, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.078206 | 8,589,934,592 | 16,777,216 | 512 | 2 | {"M": 2048, "N": 1024, "K": 2048, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 2048) x (2048, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 2048, dtype=torch.float16, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.083575 | 8,589,934,592 | 16,777,216 | 512 | 2 | {"M": 2048, "N": 1024, "K": 2048, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 2048) x (2048, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 2048, dtype=torch.float16, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.169812 | 8,589,934,592 | 16,777,216 | 512 | 2 | {"M": 2048, "N": 1024, "K": 2048, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 2048) x (2048, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 2048, dtype=torch.float16, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.090096 | 8,589,934,592 | 16,777,216 | 512 | 2 | {"M": 2048, "N": 1024, "K": 2048, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 2048) x (2048, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 2048, dtype=torch.float16, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.172466 | 8,589,934,592 | 16,777,216 | 512 | 2 | {"M": 2048, "N": 1024, "K": 2048, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 2048) x (2048, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 2048, dtype=torch.float16, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.12987 | 8,589,934,592 | 16,777,216 | 512 | 2 | {"M": 2048, "N": 1024, "K": 2048, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 2048) x (2048, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 2048, dtype=torch.float16, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.071374 | 8,589,934,592 | 16,777,216 | 512 | 2 | {"M": 2048, "N": 1024, "K": 2048, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 2048) x (2048, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 2048, dtype=torch.float16, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.046686 | 8,589,934,592 | 16,777,216 | 512 | 2 | {"M": 2048, "N": 1024, "K": 2048, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 2048) x (2048, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 2048, dtype=torch.float16, device='cuda')
B = torch.randn(2048, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.170342 | 8,589,934,592 | 16,777,216 | 512 | 2 | {"M": 2048, "N": 1024, "K": 2048, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 4096) x (4096, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 4096, dtype=torch.float32, device='cuda')
B = torch.randn(4096, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 2.850719 | 17,179,869,184 | 58,720,256 | 292.571429 | 4 | {"M": 2048, "N": 1024, "K": 4096, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 4096) x (4096, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 4096, dtype=torch.float32, device='cuda')
B = torch.randn(4096, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 1.431605 | 17,179,869,184 | 58,720,256 | 292.571429 | 4 | {"M": 2048, "N": 1024, "K": 4096, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 4096) x (4096, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 4096, dtype=torch.float32, device='cuda')
B = torch.randn(4096, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.633057 | 17,179,869,184 | 58,720,256 | 292.571429 | 4 | {"M": 2048, "N": 1024, "K": 4096, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 4096) x (4096, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 4096, dtype=torch.float32, device='cuda')
B = torch.randn(4096, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 1.240949 | 17,179,869,184 | 58,720,256 | 292.571429 | 4 | {"M": 2048, "N": 1024, "K": 4096, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 4096) x (4096, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 4096, dtype=torch.float32, device='cuda')
B = torch.randn(4096, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 1.286604 | 17,179,869,184 | 58,720,256 | 292.571429 | 4 | {"M": 2048, "N": 1024, "K": 4096, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 4096) x (4096, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 4096, dtype=torch.float32, device='cuda')
B = torch.randn(4096, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.815755 | 17,179,869,184 | 58,720,256 | 292.571429 | 4 | {"M": 2048, "N": 1024, "K": 4096, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 4096) x (4096, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 4096, dtype=torch.float32, device='cuda')
B = torch.randn(4096, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.309937 | 17,179,869,184 | 58,720,256 | 292.571429 | 4 | {"M": 2048, "N": 1024, "K": 4096, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 4096) x (4096, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 4096, dtype=torch.float32, device='cuda')
B = torch.randn(4096, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.495012 | 17,179,869,184 | 58,720,256 | 292.571429 | 4 | {"M": 2048, "N": 1024, "K": 4096, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 4096) x (4096, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 4096, dtype=torch.float32, device='cuda')
B = torch.randn(4096, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.318517 | 17,179,869,184 | 58,720,256 | 292.571429 | 4 | {"M": 2048, "N": 1024, "K": 4096, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 4096) x (4096, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 4096, dtype=torch.float32, device='cuda')
B = torch.randn(4096, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.394341 | 17,179,869,184 | 58,720,256 | 292.571429 | 4 | {"M": 2048, "N": 1024, "K": 4096, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 4096) x (4096, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 4096, dtype=torch.float32, device='cuda')
B = torch.randn(4096, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.566066 | 17,179,869,184 | 58,720,256 | 292.571429 | 4 | {"M": 2048, "N": 1024, "K": 4096, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 4096) x (4096, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 4096, dtype=torch.float32, device='cuda')
B = torch.randn(4096, 1024, dtype=torch.float32, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.715249 | 17,179,869,184 | 58,720,256 | 292.571429 | 4 | {"M": 2048, "N": 1024, "K": 4096, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 4096) x (4096, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 4096, dtype=torch.float16, device='cuda')
B = torch.randn(4096, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.357059 | 17,179,869,184 | 29,360,128 | 585.142857 | 2 | {"M": 2048, "N": 1024, "K": 4096, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 4096) x (4096, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 4096, dtype=torch.float16, device='cuda')
B = torch.randn(4096, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.303924 | 17,179,869,184 | 29,360,128 | 585.142857 | 2 | {"M": 2048, "N": 1024, "K": 4096, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 4096) x (4096, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 4096, dtype=torch.float16, device='cuda')
B = torch.randn(4096, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.549422 | 17,179,869,184 | 29,360,128 | 585.142857 | 2 | {"M": 2048, "N": 1024, "K": 4096, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 4096) x (4096, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 4096, dtype=torch.float16, device='cuda')
B = torch.randn(4096, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.129761 | 17,179,869,184 | 29,360,128 | 585.142857 | 2 | {"M": 2048, "N": 1024, "K": 4096, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 4096) x (4096, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 4096, dtype=torch.float16, device='cuda')
B = torch.randn(4096, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.153306 | 17,179,869,184 | 29,360,128 | 585.142857 | 2 | {"M": 2048, "N": 1024, "K": 4096, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 4096) x (4096, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 4096, dtype=torch.float16, device='cuda')
B = torch.randn(4096, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.256455 | 17,179,869,184 | 29,360,128 | 585.142857 | 2 | {"M": 2048, "N": 1024, "K": 4096, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 4096) x (4096, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 4096, dtype=torch.float16, device='cuda')
B = torch.randn(4096, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.12924 | 17,179,869,184 | 29,360,128 | 585.142857 | 2 | {"M": 2048, "N": 1024, "K": 4096, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 4096) x (4096, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 4096, dtype=torch.float16, device='cuda')
B = torch.randn(4096, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.355262 | 17,179,869,184 | 29,360,128 | 585.142857 | 2 | {"M": 2048, "N": 1024, "K": 4096, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 4096) x (4096, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 4096, dtype=torch.float16, device='cuda')
B = torch.randn(4096, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.180247 | 17,179,869,184 | 29,360,128 | 585.142857 | 2 | {"M": 2048, "N": 1024, "K": 4096, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 4096) x (4096, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 4096, dtype=torch.float16, device='cuda')
B = torch.randn(4096, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.05274 | 17,179,869,184 | 29,360,128 | 585.142857 | 2 | {"M": 2048, "N": 1024, "K": 4096, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 4096) x (4096, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 4096, dtype=torch.float16, device='cuda')
B = torch.randn(4096, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.05188 | 17,179,869,184 | 29,360,128 | 585.142857 | 2 | {"M": 2048, "N": 1024, "K": 4096, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 4096) x (4096, 1024) -> (2048, 1024)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 4096, dtype=torch.float16, device='cuda')
B = torch.randn(4096, 1024, dtype=torch.float16, device='cuda')
C = matmul_kernel(A, B)
torch.cuda.sync... | 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} | 0.328783 | 17,179,869,184 | 29,360,128 | 585.142857 | 2 | {"M": 2048, "N": 1024, "K": 4096, "dtype": "float16"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 64) x (64, 2048) -> (2048, 2048)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 64, dtype=torch.float32, device='cuda')
B = torch.randn(64, 2048, 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} | 0.16672 | 536,870,912 | 17,825,792 | 30.117647 | 4 | {"M": 2048, "N": 2048, "K": 64, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 64) x (64, 2048) -> (2048, 2048)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 64, dtype=torch.float32, device='cuda')
B = torch.randn(64, 2048, 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} | 0.118313 | 536,870,912 | 17,825,792 | 30.117647 | 4 | {"M": 2048, "N": 2048, "K": 64, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 64) x (64, 2048) -> (2048, 2048)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 64, dtype=torch.float32, device='cuda')
B = torch.randn(64, 2048, 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} | 0.067272 | 536,870,912 | 17,825,792 | 30.117647 | 4 | {"M": 2048, "N": 2048, "K": 64, "dtype": "float32"} |
import torch
def matmul_kernel(A, B):
# Matrix multiplication: (2048, 64) x (64, 2048) -> (2048, 2048)
C = torch.matmul(A, B)
return C
A = torch.randn(2048, 64, dtype=torch.float32, device='cuda')
B = torch.randn(64, 2048, 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} | 0.063965 | 536,870,912 | 17,825,792 | 30.117647 | 4 | {"M": 2048, "N": 2048, "K": 64, "dtype": "float32"} |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.