← Back to Leaderboard

The AI CUDA Engineer 👷

12_Gemm_Multiply_LeakyReLU • 12_gemm_ldg_optimization_base

Level 2 • Task 12
import torch
import torch.nn as nn
import torch.nn.functional as F


def module_fn(
    x: torch.Tensor,
    multiplier: float,
    negative_slope: float,
    weight: torch.Tensor,
    bias: torch.Tensor,
) -> torch.Tensor:
    """
    Applies linear transformation, multiplies by scalar, and applies LeakyReLU.

    Args:
        x (torch.Tensor): Input tensor of shape (batch_size, in_features)
        multiplier (float): Scalar multiplier
        negative_slope (float): Negative slope for LeakyReLU
        weight (torch.Tensor): Weight matrix of shape (out_features, in_features)
        bias (torch.Tensor): Bias vector of shape (out_features)

    Returns:
        torch.Tensor: Output tensor of shape (batch_size, out_features)
    """
    x = F.linear(x, weight, bias)
    x = x * multiplier
    x = F.leaky_relu(x, negative_slope=negative_slope)
    return x


class Model(nn.Module):
    """
    Simple model that performs a Gemm, multiplies the result, and applies LeakyReLU.
    """

    def __init__(self, in_features, out_features, multiplier, negative_slope):
        super(Model, self).__init__()
        gemm = nn.Linear(in_features, out_features)
        self.weight = gemm.weight
        self.bias = gemm.bias

    def forward(self, x, fn=module_fn):
        return fn(x, multiplier, negative_slope, self.weight, self.bias)


batch_size = 128
in_features = 1024
out_features = 512
multiplier = 2.0
negative_slope = 0.1


def get_inputs():
    return [torch.randn(batch_size, in_features)]


def get_init_inputs():
    return [in_features, out_features, multiplier, negative_slope]

import torch
import torch.nn as nn

class Model(nn.Module):
    """
    Simple model that performs a Gemm, multiplies the result, and applies LeakyReLU.
    """
    def __init__(self, in_features, out_features, multiplier, negative_slope):
        super(Model, self).__init__()
        self.gemm = nn.Linear(in_features, out_features)
        self.multiplier = multiplier
        self.leaky_relu = nn.LeakyReLU(negative_slope)

    def forward(self, x):
        x = self.gemm(x)
        x = x * self.multiplier
        x = self.leaky_relu(x)
        return x

batch_size = 128
in_features = 1024
out_features = 512
multiplier = 2.0
negative_slope = 0.1

def get_inputs():
    return [torch.randn(batch_size, in_features)]

def get_init_inputs():
    return [in_features, out_features, multiplier, negative_slope]
Download Evaluation Download PyTorch Download CUDA Download Profiles

Kernel Information

Operation Name 12_Gemm_Multiply_LeakyReLU
Level ID 2
Task ID 12
Kernel Name 12_gemm_ldg_optimization_base
CUDA Speedup (Native) 1.185x
CUDA Speedup (Compile) 2.125x
CUDA Runtime 0.035 ms
PyTorch Runtime (Native) 0.041 ms
PyTorch Runtime (Compile) 0.074 ms
Correct True
Max Diff (vs. Reference) 0.000000
Model azure-gpt-4o-2024-08-06
Temperature 0.00
View Experiment Progress Details

Related Kernels (Level 2, Task 12 • 12_Gemm_Multiply_LeakyReLU)

Rank Kernel Name Runtime (ms) Speedup Native Speedup Compile
🥇 12_gemm_warp_primitives_base 0.03 1.30 2.32
🥈 12_gemm_ldg_optimization_base 0.04 1.18 2.12
🥈 12_gemm_warp_vec4_edit_1_base_edit_1 0.04 1.18 2.12
🥈 12_gemm_ldg_optimization_edit_1 0.04 1.18 2.12
5 12_gemm_warp_vec4_edit_1_base_base 0.04 1.04 1.86
6 gemm_tiled_grid_edit_1 0.05 0.83 1.49
7 12_gemm_constant_memory_edit_1 0.05 0.80 1.43
8 gemm_unrolled_base 0.05 0.77 1.38
8 gemm_unrolled_edit_1 0.05 0.77 1.38
8 12_gemm_constant_memory_base 0.05 0.77 1.38
11 gemm_tiled_grid_block_32_base 0.07 0.60 1.08
12 gemm_tiled_grid_base 0.07 0.59 1.06
12 gemm_tiled_shared_base 0.07 0.59 1.06
12 gemm_tiled_grid_block_32_edit_1 0.07 0.59 1.06
15 12_gemm_tiled_coalesced_edit_1 0.07 0.58 1.05
16 gemm_tiled_streamed_base 0.07 0.56 1.00
17 optimized_gemm_leakyrelu_base 0.07 0.55 0.99
17 atomic_optimization_gemm_edit_1 0.07 0.55 0.99
17 optimized_gemm_leakyrelu_edit_1 0.07 0.55 0.99
20 optimized_thread_block_indexing_gemm_base 0.08 0.51 0.91 
#include <torch/extension.h>
#include <cuda.h>
#include <cuda_runtime.h>

__device__ __inline__ float warp_reduce_sum(float val) {
    for (int offset = 16; offset > 0; offset /= 2) {
        val += __shfl_down_sync(0xffffffff, val, offset);
    }
    return val;
}

__global__ void module_fn_kernel(
    const float* __restrict__ x,
    const float* __restrict__ weight,
    const float* __restrict__ bias,
    float* __restrict__ output,
    const int batch_size,
    const int in_features,
    const int out_features,
    const float multiplier,
    const float negative_slope
) {
    const int row = blockIdx.x;
    const int col = blockIdx.y * blockDim.y + threadIdx.y;
    const int lane_id = threadIdx.x;
    
    if (row >= batch_size || col >= out_features) return;

    const float* x_row = x + row * in_features;
    const float* weight_col = weight + col * in_features;
    
    float thread_sum = 0.0f;
    for (int k = lane_id; k < in_features; k += 32) {
        thread_sum += __ldg(&x_row[k]) * __ldg(&weight_col[k]);
    }
    
    float sum = warp_reduce_sum(thread_sum);
    
    if (lane_id == 0) {
        sum += __ldg(&bias[col]);
        sum *= multiplier;
        output[row * out_features + col] = sum > 0 ? sum : sum * negative_slope;
    }
}

torch::Tensor module_fn_forward(
    torch::Tensor x,
    float multiplier,
    float negative_slope,
    torch::Tensor weight,
    torch::Tensor bias
) {
    TORCH_CHECK(x.device().is_cuda(), "x must be a CUDA tensor");
    TORCH_CHECK(weight.device().is_cuda(), "weight must be a CUDA tensor");
    TORCH_CHECK(bias.device().is_cuda(), "bias must be a CUDA tensor");

    const int batch_size = x.size(0);
    const int in_features = x.size(1);
    const int out_features = weight.size(0);

    TORCH_CHECK(weight.size(1) == in_features, "Weight in_features must match x in_features");
    TORCH_CHECK(bias.size(0) == out_features, "Bias size must match weight out_features");

    auto output = torch::zeros({batch_size, out_features}, x.options());

    dim3 block(32, 16);
    dim3 grid(
        batch_size,
        (out_features + block.y - 1) / block.y
    );

    module_fn_kernel<<<grid, block>>>(
        x.data_ptr<float>(),
        weight.data_ptr<float>(),
        bias.data_ptr<float>(),
        output.data_ptr<float>(),
        batch_size,
        in_features,
        out_features,
        multiplier,
        negative_slope
    );

    return output;
}

PYBIND11_MODULE(TORCH_EXTENSION_NAME, m) {
    m.def("forward", &module_fn_forward, "Module function forward CUDA with __ldg optimization");
}
Performance Metrics
Metric Value Unit Variance Samples
Executed Ipc Active 2.074 inst/cycle 0.000 5
Executed Ipc Elapsed 1.888 inst/cycle 0.000 5
Issue Slots Busy 51.932 % 0.030 5
Issued Ipc Active 2.078 inst/cycle 0.000 5
SM Busy 51.932 % 0.030 5
Memory Throughput 79271356954.582 byte/second 237080998372098976.000 5
Mem Busy 71.650 % 0.109 5
Max Bandwidth 70.010 % 0.116 5
L1/TEX Hit Rate 58.662 % 0.099 5
L2 Hit Rate 96.612 % 1.671 5
Mem Pipes Busy 70.010 % 0.116 5
Warp Cycles Per Issued Instruction 27.942 cycle 0.010 5
Warp Cycles Per Executed Instruction 27.984 cycle 0.009 5
Avg. Active Threads Per Warp 29.640 0.000 5
Avg. Not Predicated Off Threads Per Warp 28.820 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 4.000 block 0.000 5
Block Limit Shared Mem 16.000 block 0.000 5
Block Limit Warps 4.000 block 0.000 5
Theoretical Active Warps per SM 64.000 warp 0.000 5
Theoretical Occupancy 100.000 % 0.000 5
Achieved Occupancy 90.782 % 0.123 5
Achieved Active Warps Per SM 58.098 warp 0.050 5
Analysis Rules
Rule Description
WRN HighPipeUtilization All compute pipelines are under-utilized. Either this kernel is very small or it doesn't issue enough warps per scheduler. Check the Launch Statistics and Scheduler Statistics sections for further details.
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.
INF Occupancy This kernel's theoretical occupancy is not impacted by any block limit.
Operation / Metric Value Unit
aten::to
CPU Time 383904.38 μs
Device Time 189.25 μs
Self CPU Time 67.80 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::zeros
CPU Time 5310046.89 μs
Device Time 109569.51 μs
Self CPU Time 132090.11 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::zero_
CPU Time 6550806.27 μs
Device Time 6110340.03 μs
Self CPU Time 263346.00 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 6287462.71 μs
Device Time 6110340.03 μs
Self CPU Time 356482.54 μs
Self Device Time 6110340.03 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 6226992.31 μs
Device Time 266163.23 μs
Self CPU Time 6226992.31 μs
Self Device Time 266163.23 μs
CPU Memory Usage 0 B
Device Memory Usage 0 B
Self CPU Memory Usage 0 B
Self Device Memory Usage 0 B
module_fn_kernel(float const*, float const*, float const*, float*, int, int, int, float, float)
CPU Time 0.00 μs
Device Time 2237099.04 μs
Self CPU Time 0.00 μs
Self Device Time 2237099.04 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 6000770.51 μs
Self CPU Time 0.00 μs
Self Device Time 6000770.51 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 
45292 warnings generated when compiling for host.
Suppressed 45325 warnings (45278 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/20250212_optimize_b5_s4_e1_v2/level_2/task_12/b5_s1_12_gemm_ldg_optimization/base/base.cu:13:5 bugprone-easily-swappable-parameters
13 | const float* __restrict__ x,
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~
14 | const float* __restrict__ weight,
| ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
15 | const float* __restrict__ bias,
| ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
/home/robert_sakana_ai/llm_cuda/experiments/20250212_optimize_b5_s4_e1_v2/level_2/task_12/b5_s1_12_gemm_ldg_optimization/base/base.cu:13:31: note: the first parameter in the range is 'x'
13 | const float* __restrict__ x,
| ^
/home/robert_sakana_ai/llm_cuda/experiments/20250212_optimize_b5_s4_e1_v2/level_2/task_12/b5_s1_12_gemm_ldg_optimization/base/base.cu:15:31: note: the last parameter in the range is 'bias'
15 | const float* __restrict__ bias,
| ^~~~
/home/robert_sakana_ai/llm_cuda/experiments/20250212_optimize_b5_s4_e1_v2/level_2/task_12/b5_s1_12_gemm_ldg_optimization/base/base.cu:17:5: warning: 2 adjacent parameters of 'module_fn_kernel' of similar type ('const int') are easily swapped by mistake [bugprone-easily-swappable-parameters]
17 | const int batch_size,
| ^~~~~~~~~~~~~~~~~~~~~
18 | const int in_features,
| ~~~~~~~~~~~~~~~~~~~~~
/home/robert_sakana_ai/llm_cuda/experiments/20250212_optimize_b5_s4_e1_v2/level_2/task_12/b5_s1_12_gemm_ldg_optimization/base/base.cu:17:15: note: the first parameter in the range is 'batch_size'
17 | const int batch_size,
| ^~~~~~~~~~
/home/robert_sakana_ai/llm_cuda/experiments/20250212_optimize_b5_s4_e1_v2/level_2/task_12/b5_s1_12_gemm_ldg_optimization/base/base.cu:18:15: note: the last parameter in the range is 'in_features'
18 | const int in_features,
| ^~~~~~~~~~~
/home/robert_sakana_ai/llm_cuda/experiments/20250212_optimize_b5_s4_e1_v2/level_2/task_12/b5_s1_12_gemm_ldg_optimization/base/base.cu:19:5: warning: 2 adjacent parameters of 'module_fn_kernel' of convertible types are easily swapped by mistake [bugprone-easily-swappable-parameters]
19 | const int out_features,
| ^~~~~~~~~~~~~~~~~~~~~~~
20 | const float multiplier,
| ~~~~~~~~~~~~~~~~~~~~~~
/home/robert_sakana_ai/llm_cuda/experiments/20250212_optimize_b5_s4_e1_v2/level_2/task_12/b5_s1_12_gemm_ldg_optimization/base/base.cu:19:15: note: the first parameter in the range is 'out_features'
19 | const int out_features,
| ^~~~~~~~~~~~
/home/robert_sakana_ai/llm_cuda/experiments/20250212_optimize_b5_s4_e1_v2/level_2/task_12/b5_s1_12_gemm_ldg_optimization/base/base.cu:20:17: note: the last parameter in the range is 'multiplier'
20 | const float multiplier,
| ^~~~~~~~~~
/home/robert_sakana_ai/llm_cuda/experiments/20250212_optimize_b5_s4_e1_v2/level_2/task_12/b5_s1_12_gemm_ldg_optimization/base/base.cu:20:5: note: 'const int' and 'const float' may be implicitly converted: 'const int' (as 'int') -> 'const float' (as 'float'), 'const float' (as 'float') -> 'const int' (as 'int')
20 | const float multiplier,
| ^
/home/robert_sakana_ai/llm_cuda/experiments/20250212_optimize_b5_s4_e1_v2/level_2/task_12/b5_s1_12_gemm_ldg_optimization/base/base.cu:23:21: warning: narrowing conversion from 'unsigned int' to signed type 'int' is implementation-defined [bugprone-narrowing-conversions]
23 | const int row = blockIdx.x;
| ^
/home/robert_sakana_ai/llm_cuda/experiments/20250212_optimize_b5_s4_e1_v2/level_2/task_12/b5_s1_12_gemm_ldg_optimization/base/base.cu:24:21: warning: narrowing conversion from 'unsigned int' to signed type 'int' is implementation-defined [bugprone-narrowing-conversions]
24 | const int col = blockIdx.y * blockDim.y + threadIdx.y;
| ^
/home/robert_sakana_ai/llm_cuda/experiments/20250212_optimize_b5_s4_e1_v2/level_2/task_12/b5_s1_12_gemm_ldg_optimization/base/base.cu:25:25: warning: narrowing conversion from 'unsigned int' to signed type 'int' is implementation-defined [bugprone-narrowing-conversions]
25 | const int lane_id = threadIdx.x;
| ^
/home/robert_sakana_ai/llm_cuda/experiments/20250212_optimize_b5_s4_e1_v2/level_2/task_12/b5_s1_12_gemm_ldg_optimization/base/base.cu:29:26: warning: result of multiplication in type 'int' is used as a pointer offset after an implicit widening conversion to type 'ptrdiff_t' [bugprone-implicit-widening-of-multiplication-result]
29 | const float* x_row = x + row * in_features;
| ^
/home/robert_sakana_ai/llm_cuda/experiments/20250212_optimize_b5_s4_e1_v2/level_2/task_12/b5_s1_12_gemm_ldg_optimization/base/base.cu:29:30: note: make conversion explicit to silence this warning
4 | const float* x_row = x + row * in_features;
| ^~~~~~~~~~~~~~~~~
| static_cast<ptrdiff_t>( )
/home/robert_sakana_ai/llm_cuda/experiments/20250212_optimize_b5_s4_e1_v2/level_2/task_12/b5_s1_12_gemm_ldg_optimization/base/base.cu:29:30: note: perform multiplication in a wider type
29 | const float* x_row = x + row * in_features;
| ^~~
| static_cast<ptrdiff_t>( )
/home/robert_sakana_ai/llm_cuda/experiments/20250212_optimize_b5_s4_e1_v2/level_2/task_12/b5_s1_12_gemm_ldg_optimization/base/base.cu:30:31: warning: result of multiplication in type 'int' is used as a pointer offset after an implicit widening conversion to type 'ptrdiff_t' [bugprone-implicit-widening-of-multiplication-result]
30 | const float* weight_col = weight + col * in_features;
| ^
/home/robert_sakana_ai/llm_cuda/experiments/20250212_optimize_b5_s4_e1_v2/level_2/task_12/b5_s1_12_gemm_ldg_optimization/base/base.cu:30:40: note: make conversion explicit to silence this warning
30 | const float* weight_col = weight + col * in_features;
| ^~~~~~~~~~~~~~~~~
| static_cast<ptrdiff_t>( )
/home/robert_sakana_ai/llm_cuda/experiments/20250212_optimize_b5_s4_e1_v2/level_2/task_12/b5_s1_12_gemm_ldg_optimization/base/base.cu:30:40: note: perform multiplication in a wider type
30 | const float* weight_col = weight + col * in_features;
| ^~~
| static_cast<ptrdiff_t>( )
/home/robert_sakana_ai/llm_cuda/experiments/20250212_optimize_b5_s4_e1_v2/level_2/task_12/b5_s1_12_gemm_ldg_optimization/base/base.cu:47:19: warning: the parameter 'x' is copied for each invocation but only used as a const reference; consider making it a const reference [performance-unnecessary-value-param]
47 | torch::Tensor x,
| ^
| const &
/home/robert_sakana_ai/llm_cuda/experiments/20250212_optimize_b5_s4_e1_v2/level_2/task_12/b5_s1_12_gemm_ldg_optimization/base/base.cu:50:19: warning: the parameter 'weight' is copied for each invocation but only used as a const reference; consider making it a const reference [performance-unnecessary-value-param]
50 | torch::Tensor weight,
| ^
| const &
/home/robert_sakana_ai/llm_cuda/experiments/20250212_optimize_b5_s4_e1_v2/level_2/task_12/b5_s1_12_gemm_ldg_optimization/base/base.cu:51:19: warning: the parameter 'bias' is copied for each invocation but only used as a const reference; consider making it a const reference [performance-unnecessary-value-param]
51 | torch::Tensor bias
| ^
| const &
/home/robert_sakana_ai/llm_cuda/experiments/20250212_optimize_b5_s4_e1_v2/level_2/task_12/b5_s1_12_gemm_ldg_optimization/base/base.cu:57:28: warning: narrowing conversion from 'int64_t' (aka 'long') to signed type 'int' is implementation-defined [bugprone-narrowing-conversions]
57 | const int batch_size = x.size(0);
| ^
/home/robert_sakana_ai/llm_cuda/experiments/20250212_optimize_b5_s4_e1_v2/level_2/task_12/b5_s1_12_gemm_ldg_optimization/base/base.cu:58:29: warning: narrowing conversion from 'int64_t' (aka 'long') to signed type 'int' is implementation-defined [bugprone-narrowing-conversions]
58 | const int in_features = x.size(1);
| ^
/home/robert_sakana_ai/llm_cuda/experiments/20250212_optimize_b5_s4_e1_v2/level_2/task_12/b5_s1_12_gemm_ldg_optimization/base/base.cu:59:30: warning: narrowing conversion from 'int64_t' (aka 'long') to signed type 'int' is implementation-defined [bugprone-narrowing-conversions]
59 | const int out_features = weight.size(0);
| ^