← Back to Leaderboard

The AI CUDA Engineer 👷

18_Matmul_with_transposed_both • matmul_transpose_ldg_optimization_base

Level 1 • Task 18
import torch
import torch.nn as nn
import torch.nn.functional as F


def module_fn(A: torch.Tensor, B: torch.Tensor) -> torch.Tensor:
    """
    Performs a single matrix multiplication with transposed A and B (C = A.T * B.T).

    Args:
        A: Input tensor of shape (K, M).
        B: Input tensor of shape (N, K).

    Returns:
        Output tensor of shape (M, N).
    """
    return torch.matmul(A.T, B.T)


class Model(nn.Module):
    """
    Simple model that performs a single matrix multiplication (C = A * B)
    """

    def __init__(self):
        super(Model, self).__init__()

    def forward(self, A: torch.Tensor, B: torch.Tensor, fn=module_fn) -> torch.Tensor:
        return fn(A, B)


M = 1024
K = 4096
N = 2048


def get_inputs():
    A = torch.randn(K, M)
    B = torch.randn(N, K)
    return [A, B]


def get_init_inputs():
    return []  # No special initialization inputs needed

import torch
import torch.nn as nn

class Model(nn.Module):
    """
    Simple model that performs a single matrix multiplication (C = A * B)
    """
    def __init__(self):
        super(Model, self).__init__()
    
    def forward(self, A: torch.Tensor, B: torch.Tensor) -> torch.Tensor:
        """
        Performs matrix multiplication.

        Args:
            A: Input tensor of shape (M, K).
            B: Input tensor of shape (K, N).

        Returns:
            Output tensor of shape (M, N).
        """
        return torch.matmul(A.T, B.T)

M = 1024
K = 4096
N = 2048

def get_inputs():
    A = torch.randn(K, M)
    B = torch.randn(N, K)
    return [A, B]

def get_init_inputs():
    return []  # No special initialization inputs needed
Download Evaluation Download PyTorch Download CUDA Download Profiles

Kernel Information

Operation Name 18_Matmul_with_transposed_both
Level ID 1
Task ID 18
Kernel Name matmul_transpose_ldg_optimization_base
CUDA Speedup (Native) 0.191x
CUDA Speedup (Compile) 0.217x
CUDA Runtime 1.905 ms
PyTorch Runtime (Native) 0.365 ms
PyTorch Runtime (Compile) 0.413 ms
Correct True
Max Diff (vs. Reference) 0.001000
Model azure-gpt-4o-2024-08-06
Temperature 1.00
View Experiment Progress Details

Related Kernels (Level 1, Task 18 • 18_Matmul_with_transposed_both)

Rank Kernel Name Runtime (ms) Speedup Native Speedup Compile
🥇 optimized_matmul_transpose_base 1.87 0.19 0.22
🥈 matmul_transpose_ldg_128bit_align_base 1.90 0.19 0.22
🥉 matmul_transpose_ldg_optimization_base 1.91 0.19 0.22
🥉 optimized_matmul_transpose_base 1.91 0.19 0.22
5 optimized_matmul_transpose_base 1.91 0.19 0.22
6 stride_loop_matmul_transpose_base 1.91 0.19 0.22
6 warp_divergence_optimized_matmul_base 1.91 0.19 0.22
8 optimized_matmul_transpose_base 1.91 0.19 0.22
9 18_matmul_transposed_block32_base 1.96 0.19 0.21
10 combined_matmul_transpose_base 1.97 0.19 0.21
11 optimized_matmul_transpose_base 1.99 0.18 0.21
11 reduced_sync_matmul_transposed_edit_1 1.99 0.18 0.21
13 reduced_sync_matmul_transposed_base 1.99 0.18 0.21
13 optimized_matmul_transpose_edit_1 1.99 0.18 0.21
15 combined_matmul_transpose_edit_1 2.04 0.18 0.20
16 18_Matmul_transposed_coalesced_base 2.15 0.17 0.19
17 modular_matmul_transposed_base 2.15 0.17 0.19
18 18_Matmul_transposed_coalesced_edit_1 2.16 0.17 0.19
19 matmul_transpose_modular_base 2.23 0.16 0.18
20 modular_vectorized_matmul_transposed_base_base 2.24 0.16 0.18 
#include <torch/extension.h>
#include <cuda.h>
#include <cuda_runtime.h>

// Define tile sizes
#define BLOCK_SIZE_M 32  // Output tile height (each block computes 32 rows)
#define BLOCK_SIZE_N 16  // Output tile width (each block computes 16 columns)
#define BLOCK_SIZE_K 16  // Reduction tile depth

// Kernel: each thread computes a 2x1 sub-tile (2 rows, 1 column) of C
// A is (K x M): element A[k, m] = A[k * M + m]
// B is (N x K): element B[n, k] = B[n * K + k]
// C is (M x N): element C[m, n] = C[m * N + n]

template <typename scalar_t>
__global__ void matmul_transpose_ldg_optimization_kernel(
    const scalar_t* __restrict__ A,
    const scalar_t* __restrict__ B,
    scalar_t* __restrict__ C,
    int M,
    int N,
    int K) {

    // Determine the starting indices for this block's tile in C
    int m_start = blockIdx.y * BLOCK_SIZE_M;  // row start in C
    int n_start = blockIdx.x * BLOCK_SIZE_N;  // col start in C

    // Thread indices within the block
    int tx = threadIdx.x; // Expected range: [0, 15]
    int ty = threadIdx.y; // Expected range: [0, 15]

    // Each thread computes two rows: row0 and row1
    int row0 = m_start + tx;             // first row computed by this thread
    int row1 = row0 + (BLOCK_SIZE_M / 2);  // second row computed (offset by 16)
    int col = n_start + ty;              // column index in C

    // Accumulators for the two output elements
    scalar_t acc0 = 0;
    scalar_t acc1 = 0;

    // Declare shared memory tiles
    __shared__ scalar_t A_tile[BLOCK_SIZE_K][BLOCK_SIZE_M]; // Size: 16 x 32
    __shared__ scalar_t B_tile[BLOCK_SIZE_N][BLOCK_SIZE_K];   // Size: 16 x 16

    // Total threads in a block
    int tId = threadIdx.y * blockDim.x + threadIdx.x; // Range: 0 to 255
    int blockSize = blockDim.x * blockDim.y;            // = 256

    int numTiles = (K + BLOCK_SIZE_K - 1) / BLOCK_SIZE_K;
    for (int tile = 0; tile < numTiles; tile++) {
        // Load A tile into shared memory
        // A tile dimensions: BLOCK_SIZE_K x BLOCK_SIZE_M (16 x 32 = 512 elements)
        int totalAElements = BLOCK_SIZE_K * BLOCK_SIZE_M; // 512
        for (int idx = tId; idx < totalAElements; idx += blockSize) {
            int kd = idx / BLOCK_SIZE_M;  // k-index within the tile
            int md = idx % BLOCK_SIZE_M;  // m-index within the tile
            int global_m = m_start + md;  // global m index
            int global_k = tile * BLOCK_SIZE_K + kd;  // global k index
            if (global_m < M && global_k < K)
                A_tile[kd][md] = __ldg(&A[global_k * M + global_m]);
            else
                A_tile[kd][md] = 0;
        }

        // Load B tile into shared memory
        // B tile dimensions: BLOCK_SIZE_N x BLOCK_SIZE_K (16 x 16 = 256 elements)
        int totalBElements = BLOCK_SIZE_N * BLOCK_SIZE_K; // 256
        for (int idx = tId; idx < totalBElements; idx += blockSize) {
            int nd = idx / BLOCK_SIZE_K;  // n-index within the tile
            int kd = idx % BLOCK_SIZE_K;  // k-index within the tile
            int global_n = n_start + nd;  // global n index
            int global_k = tile * BLOCK_SIZE_K + kd;  // global k index
            if (global_n < N && global_k < K)
                B_tile[nd][kd] = __ldg(&B[global_n * K + global_k]);
            else
                B_tile[nd][kd] = 0;
        }

        __syncthreads();

        // Compute the partial results for this tile
        for (int k = 0; k < BLOCK_SIZE_K; k++) {
            scalar_t a_val0 = A_tile[k][tx];                     // for row0
            scalar_t a_val1 = A_tile[k][tx + (BLOCK_SIZE_M / 2)];  // for row1
            scalar_t b_val = B_tile[ty][k];
            acc0 += a_val0 * b_val;
            acc1 += a_val1 * b_val;
        }
        __syncthreads();
    }

    // Write the results to global memory
    if (row0 < M && col < N) {
        C[row0 * N + col] = acc0;
    }
    if (row1 < M && col < N) {
        C[row1 * N + col] = acc1;
    }
}


// PyTorch binding

torch::Tensor matmul_transpose_cuda(torch::Tensor A, torch::Tensor B) {
    // Dimensions:
    // A: (K x M), B: (N x K), therefore C: (M x N)
    int K = A.size(0);
    int M = A.size(1);
    int N = B.size(0);

    auto C = torch::empty({M, N}, A.options());

    // Define block dimensions: use 16x16 threads per block
    dim3 threads(16, 16);
    // Grid dimensions based on tile sizes
    dim3 blocks((N + BLOCK_SIZE_N - 1) / BLOCK_SIZE_N, (M + BLOCK_SIZE_M - 1) / BLOCK_SIZE_M);

    AT_DISPATCH_FLOATING_TYPES(A.scalar_type(), "matmul_transpose_ldg_optimization_kernel", ([&] {
        matmul_transpose_ldg_optimization_kernel<scalar_t><<<blocks, threads>>>(
            A.data_ptr<scalar_t>(),
            B.data_ptr<scalar_t>(),
            C.data_ptr<scalar_t>(),
            M, N, K);
    }));

    return C;
}

PYBIND11_MODULE(TORCH_EXTENSION_NAME, m) {
    m.def("forward", &matmul_transpose_cuda, "Matrix multiplication with transposed inputs using multi-output kernel and ldg optimization (CUDA)");
}
Performance Metrics
Metric Value Unit Variance Samples
Executed Ipc Active 2.582 inst/cycle 0.000 5
Executed Ipc Elapsed 2.554 inst/cycle 0.000 5
Issue Slots Busy 64.582 % 0.008 5
Issued Ipc Active 2.582 inst/cycle 0.000 5
SM Busy 64.582 % 0.008 5
Memory Throughput 71068218101.104 byte/second 687096220118865664.000 5
Mem Busy 83.582 % 0.014 5
Max Bandwidth 75.126 % 0.011 5
L1/TEX Hit Rate 1.692 % 0.003 5
L2 Hit Rate 86.284 % 0.077 5
Mem Pipes Busy 73.392 % 0.010 5
Warp Cycles Per Issued Instruction 22.552 cycle 0.000 5
Warp Cycles Per Executed Instruction 22.554 cycle 0.000 5
Avg. Active Threads Per Warp 32.000 0.000 5
Avg. Not Predicated Off Threads Per Warp 31.120 0.000 5
Max Active Clusters 0.000 cluster 0.000 5
Max Cluster Size 8.000 block 0.000 5
Overall GPU Occupancy 0.000 % 0.000 5
Cluster Occupancy 0.000 % 0.000 5
Block Limit SM 32.000 block 0.000 5
Block Limit Registers 8.000 block 0.000 5
Block Limit Shared Mem 16.000 block 0.000 5
Block Limit Warps 8.000 block 0.000 5
Theoretical Active Warps per SM 64.000 warp 0.000 5
Theoretical Occupancy 100.000 % 0.000 5
Achieved Occupancy 91.072 % 0.004 5
Achieved Active Warps Per SM 58.288 warp 0.002 5
Analysis Rules
Rule Description
INF HighPipeUtilization ALU is the highest-utilized pipeline (37.1%) based on active cycles, taking into account the rates of its different instructions. It executes integer and logic operations. It is well-utilized, but should not be a bottleneck.
INF Occupancy This kernel's theoretical occupancy is not impacted by any block limit.
INF CPIStall Check the Warp Stall Sampling (All Cycles) table for the top stall locations in your source based on sampling data. The Kernel Profiling Guide (https://docs.nvidia.com/nsight-compute/ProfilingGuide/index.html#metrics-reference) provides more details on each stall reason.
Operation / Metric Value Unit
aten::to
CPU Time 265075.92 μs
Device Time 6060.28 μs
Self CPU Time 50.09 μs
Self Device Time 0.00 μs
CPU Memory Usage 0 B
Device Memory Usage 0 B
Self CPU Memory Usage 0 B
Self Device Memory Usage 0 B
cudaLaunchKernel
CPU Time 5825758.27 μs
Device Time 8842.08 μs
Self CPU Time 5825758.27 μs
Self Device Time 8842.08 μs
CPU Memory Usage 0 B
Device Memory Usage 0 B
Self CPU Memory Usage 0 B
Self Device Memory Usage 0 B
void matmul_transpose_ldg_optimization_kernel<float>(float const*, float const*, float*, int, int, int)
CPU Time 0.00 μs
Device Time 6179881.65 μs
Self CPU Time 0.00 μs
Self Device Time 6179881.65 μs
CPU Memory Usage 0 B
Device Memory Usage 0 B
Self CPU Memory Usage 0 B
Self Device Memory Usage 0 B
cudaDeviceSynchronize
CPU Time 520382.65 μs
Device Time 0.00 μs
Self CPU Time 520382.65 μs
Self Device Time 0.00 μs
CPU Memory Usage 0 B
Device Memory Usage 0 B
Self CPU Memory Usage 0 B
Self Device Memory Usage 0 B
cudaEventRecord
CPU Time 11244.42 μs
Device Time 16206.47 μs
Self CPU Time 11244.42 μs
Self Device Time 16206.47 μs
CPU Memory Usage 0 B
Device Memory Usage 0 B
Self CPU Memory Usage 0 B
Self Device Memory Usage 0 B
aten::zero_
CPU Time 5658662.74 μs
Device Time 251169.60 μs
Self CPU Time 7484.48 μs
Self Device Time 0.00 μs
CPU Memory Usage 0 B
Device Memory Usage 0 B
Self CPU Memory Usage 0 B
Self Device Memory Usage 0 B
aten::fill_
CPU Time 5651180.15 μs
Device Time 251169.60 μs
Self CPU Time 11331.09 μs
Self Device Time 251169.60 μs
CPU Memory Usage 0 B
Device Memory Usage 0 B
Self CPU Memory Usage 0 B
Self Device Memory Usage 0 B
void at::native::vectorized_elementwise_kernel<4, at::native::FillFunctor<int>, at::detail::Array<char*, 1> >(int, at::native::FillFunctor<int>, at::detail::Array<char*, 1>)
CPU Time 0.00 μs
Device Time 251169.60 μs
Self CPU Time 0.00 μs
Self Device Time 251169.60 μs
CPU Memory Usage 0 B
Device Memory Usage 0 B
Self CPU Memory Usage 0 B
Self Device Memory Usage 0 B
Status: Completed 
45288 warnings generated when compiling for host.
Suppressed 45322 warnings (45275 in non-user code, 47 NOLINT).
Use -header-filter=.* to display errors from all non-system headers. Use -system-headers to display errors from system headers as well.
/home/robert_sakana_ai/llm_cuda/experiments/20250202_optimize_b10_s4_e0_sweep/level_1/task_18/b6_s3_matmul_transpose_ldg_optimization/base/base.cu:17:5 bugprone-easily-swappable-parameters
17 | const scalar_t* __restrict__ A,
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
18 | const scalar_t* __restrict__ B,
| ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
/home/robert_sakana_ai/llm_cuda/experiments/20250202_optimize_b10_s4_e0_sweep/level_1/task_18/b6_s3_matmul_transpose_ldg_optimization/base/base.cu:17:34: note: the first parameter in the range is 'A'
17 | const scalar_t* __restrict__ A,
| ^
/home/robert_sakana_ai/llm_cuda/experiments/20250202_optimize_b10_s4_e0_sweep/level_1/task_18/b6_s3_matmul_transpose_ldg_optimization/base/base.cu:18:34: note: the last parameter in the range is 'B'
18 | const scalar_t* __restrict__ B,
| ^
/home/robert_sakana_ai/llm_cuda/experiments/20250202_optimize_b10_s4_e0_sweep/level_1/task_18/b6_s3_matmul_transpose_ldg_optimization/base/base.cu:25:19: warning: narrowing conversion from 'unsigned int' to signed type 'int' is implementation-defined [bugprone-narrowing-conversions]
25 | int m_start = blockIdx.y * BLOCK_SIZE_M; // row start in C
| ^
/home/robert_sakana_ai/llm_cuda/experiments/20250202_optimize_b10_s4_e0_sweep/level_1/task_18/b6_s3_matmul_transpose_ldg_optimization/base/base.cu:26:19: warning: narrowing conversion from 'unsigned int' to signed type 'int' is implementation-defined [bugprone-narrowing-conversions]
26 | int n_start = blockIdx.x * BLOCK_SIZE_N; // col start in C
| ^
/home/robert_sakana_ai/llm_cuda/experiments/20250202_optimize_b10_s4_e0_sweep/level_1/task_18/b6_s3_matmul_transpose_ldg_optimization/base/base.cu:29:14: warning: narrowing conversion from 'unsigned int' to signed type 'int' is implementation-defined [bugprone-narrowing-conversions]
29 | int tx = threadIdx.x; // Expected range: [0, 15]
| ^
/home/robert_sakana_ai/llm_cuda/experiments/20250202_optimize_b10_s4_e0_sweep/level_1/task_18/b6_s3_matmul_transpose_ldg_optimization/base/base.cu:30:14: warning: narrowing conversion from 'unsigned int' to signed type 'int' is implementation-defined [bugprone-narrowing-conversions]
30 | int ty = threadIdx.y; // Expected range: [0, 15]
| ^
/home/robert_sakana_ai/llm_cuda/experiments/20250202_optimize_b10_s4_e0_sweep/level_1/task_18/b6_s3_matmul_transpose_ldg_optimization/base/base.cu:46:15: warning: narrowing conversion from 'unsigned int' to signed type 'int' is implementation-defined [bugprone-narrowing-conversions]
46 | int tId = threadIdx.y * blockDim.x + threadIdx.x; // Range: 0 to 255
| ^
/home/robert_sakana_ai/llm_cuda/experiments/20250202_optimize_b10_s4_e0_sweep/level_1/task_18/b6_s3_matmul_transpose_ldg_optimization/base/base.cu:47:21: warning: narrowing conversion from 'unsigned int' to signed type 'int' is implementation-defined [bugprone-narrowing-conversions]
47 | int blockSize = blockDim.x * blockDim.y; // = 256
| ^
/home/robert_sakana_ai/llm_cuda/experiments/20250202_optimize_b10_s4_e0_sweep/level_1/task_18/b6_s3_matmul_transpose_ldg_optimization/base/base.cu:107:13: warning: narrowing conversion from 'int64_t' (aka 'long') to signed type 'int' is implementation-defined [bugprone-narrowing-conversions]
107 | int K = A.size(0);
| ^
/home/robert_sakana_ai/llm_cuda/experiments/20250202_optimize_b10_s4_e0_sweep/level_1/task_18/b6_s3_matmul_transpose_ldg_optimization/base/base.cu:108:13: warning: narrowing conversion from 'int64_t' (aka 'long') to signed type 'int' is implementation-defined [bugprone-narrowing-conversions]
108 | int M = A.size(1);
| ^
/home/robert_sakana_ai/llm_cuda/experiments/20250202_optimize_b10_s4_e0_sweep/level_1/task_18/b6_s3_matmul_transpose_ldg_optimization/base/base.cu:109:13: warning: narrowing conversion from 'int64_t' (aka 'long') to signed type 'int' is implementation-defined [bugprone-narrowing-conversions]
109 | int N = B.size(0);
| ^
/home/robert_sakana_ai/llm_cuda/experiments/20250202_optimize_b10_s4_e0_sweep/level_1/task_18/b6_s3_matmul_transpose_ldg_optimization/base/base.cu:118:5: warning: inside a lambda, '__func__' expands to the name of the function call operator; consider capturing the name of the enclosing function explicitly [bugprone-lambda-function-name]
118 | AT_DISPATCH_FLOATING_TYPES(A.scalar_type(), "matmul_transpose_ldg_optimization_kernel", ([&] {
| ^
/home/robert_sakana_ai/miniconda3/envs/llm2cuda/lib/python3.11/site-packages/torch/include/ATen/Dispatch.h:237:34: note: expanded from macro 'AT_DISPATCH_FLOATING_TYPES'
237 | AT_DISPATCH_SWITCH(TYPE, NAME, AT_DISPATCH_CASE_FLOATING_TYPES(__VA_ARGS__))
| ^
/home/robert_sakana_ai/miniconda3/envs/llm2cuda/lib/python3.11/site-packages/torch/include/ATen/Dispatch.h:233:3: note: expanded from macro 'AT_DISPATCH_CASE_FLOATING_TYPES'
233 | AT_DISPATCH_CASE(at::ScalarType::Double, __VA_ARGS__) \
| ^
/home/robert_sakana_ai/miniconda3/envs/llm2cuda/lib/python3.11/site-packages/torch/include/ATen/Dispatch.h:74:3: note: expanded from macro 'AT_DISPATCH_CASE'
74 | AT_PRIVATE_CASE_TYPE_USING_HINT(enum_type, scalar_t, __VA_ARGS__)
| ^
note: (skipping 1 expansions in backtrace; use -fmacro-backtrace-limit=0 to see all)
/home/robert_sakana_ai/miniconda3/envs/llm2cuda/lib/python3.11/site-packages/torch/include/ATen/Dispatch.h:58:7: note: expanded from macro 'AT_PRIVATE_CHECK_SELECTIVE_BUILD'
58 | AT_ERROR( \
| ^
/home/robert_sakana_ai/miniconda3/envs/llm2cuda/lib/python3.11/site-packages/torch/include/c10/util/Exception.h:711:32: note: expanded from macro 'AT_ERROR'
711 | C10_EXPAND_MSVC_WORKAROUND(TORCH_CHECK(false, ::c10::str(__VA_ARGS__))); \
| ^
/home/robert_sakana_ai/miniconda3/envs/llm2cuda/lib/python3.11/site-packages/torch/include/c10/util/Exception.h:536:9: note: expanded from macro 'TORCH_CHECK'
536 | __func__, \
| ^