← Back to Leaderboard

The AI CUDA Engineer 👷

99_TripletMarginLoss • warp_shfl_triplet_loss_optimized_edit_2_edit_1

Level 1 • Task 99
import torch
import torch.nn as nn
import torch.nn.functional as F


def module_fn(
    anchor: torch.Tensor, positive: torch.Tensor, negative: torch.Tensor, margin: float
) -> torch.Tensor:
    """
    Computes the Triplet Margin Loss for metric learning tasks.

    Args:
        anchor (torch.Tensor): Anchor values.
        positive (torch.Tensor): Positive values.
        negative (torch.Tensor): Negative values.
        margin (float): Margin value.

    Returns:
        torch.Tensor: Triplet Margin Loss.
    """
    return F.triplet_margin_loss(anchor, positive, negative, margin=margin)


class Model(nn.Module):
    """
    A model that computes Triplet Margin Loss for metric learning tasks.
    """

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

    def forward(self, anchor, positive, negative, fn=module_fn):
        return fn(anchor, positive, negative, self.margin)


batch_size = 128
input_shape = (4096,)
dim = 1
margin = 1.0


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


def get_init_inputs():
    return [margin]

import torch
import torch.nn as nn

class Model(nn.Module):
    """
    A model that computes Triplet Margin Loss for metric learning tasks.

    Parameters:
        margin (float): The margin between the positive and negative samples.
    """
    def __init__(self, margin=1.0):
        super(Model, self).__init__()
        self.loss_fn = torch.nn.TripletMarginLoss(margin=margin)

    def forward(self, anchor, positive, negative):
        return self.loss_fn(anchor, positive, negative)

batch_size = 128
input_shape = (4096, )
dim = 1

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

def get_init_inputs():
    return [1.0]  # Default margin
Download Evaluation Download PyTorch Download CUDA Download Profiles

Kernel Information

Operation Name 99_TripletMarginLoss
Level ID 1
Task ID 99
Kernel Name warp_shfl_triplet_loss_optimized_edit_2_edit_1
CUDA Speedup (Native) 3.147x
CUDA Speedup (Compile) 2.128x
CUDA Runtime 0.016 ms
PyTorch Runtime (Native) 0.050 ms
PyTorch Runtime (Compile) 0.034 ms
Correct True
Max Diff (vs. Reference) 0.073000
Model bedrock/anthropic.claude-3-5-sonnet-20241022-v2:0
Temperature 0.00
View Experiment Progress Details

Related Kernels (Level 1, Task 99 • 99_TripletMarginLoss)

Rank Kernel Name Runtime (ms) Speedup Native Speedup Compile
🥇 warp_shfl_tripletloss_base 0.01 3.87 2.62
🥇 vectorized_warp_shfl_reduction_edit_1 0.01 3.87 2.62
🥇 warp_shfl_tripletloss_edit_1 0.01 3.87 2.62
🥇 warp_optimized_tripletloss_edit_1 0.01 3.87 2.62
🥇 vectorized_warp_shfl_reduction_base 0.01 3.87 2.62
6 combined_warp_shfl_shared_memory_edit_1 0.01 3.60 2.43
6 combined_warp_shfl_shared_memory_base 0.01 3.60 2.43
6 warp_optimized_tripletloss_base 0.01 3.60 2.43
6 optimized_tripletloss_base 0.01 3.60 2.43
10 aligned_coalesced_triplet_base_base 0.01 3.36 2.27
10 ldg_aligned_triplet_base 0.01 3.36 2.27
10 block_size_experiment_triplet_loss_base 0.01 3.36 2.27
10 warp_shfl_triplet_loss_base 0.01 3.36 2.27
14 warp_shfl_triplet_loss_optimized_edit_2_edit_1 0.02 3.15 2.13
14 modular_aligned_triplet_base_base 0.02 3.15 2.13
14 optimized_triplet_margin_loss_base 0.02 3.15 2.13
14 uniform_control_flow_triplet_base 0.02 3.15 2.13
14 shared_memory_optimized_triplet_loss_base 0.02 3.15 2.13
14 shared_memory_optimized_triplet_loss_edit_1 0.02 3.15 2.13
20 hybrid_reduction_triplet_loss_base 0.02 2.96 2.00 
#include <torch/extension.h>
#include <cuda.h>
#include <cuda_runtime.h>

// Adjusted kernel parameters for improved performance based on hardware requirements

template <typename scalar_t>
__global__ void triplet_margin_loss_kernel(
    const scalar_t* anchor,
    const scalar_t* positive,
    const scalar_t* negative,
    scalar_t* output,
    const float margin,
    const int batch_size,
    const int feat_size) {
    
    const int tid = blockIdx.x * blockDim.x + threadIdx.x;
    const int batch_idx = tid / feat_size;
    const int feat_idx = tid % feat_size;
    const int warp_size = 32;
    const unsigned mask = 0xffffffff;
    
    if (batch_idx < batch_size && feat_idx < feat_size) {
        const int idx = batch_idx * feat_size + feat_idx;
        const scalar_t a = anchor[idx];
        const scalar_t p = positive[idx];
        const scalar_t n = negative[idx];
        
        // Compute distance components
        const scalar_t d_pos = a - p;
        const scalar_t d_neg = a - n;
        
        // Squared distances
        scalar_t dist_pos = d_pos * d_pos;
        scalar_t dist_neg = d_neg * d_neg;
        
        // Use shared memory for reduction
        __shared__ scalar_t shared_pos[512];
        __shared__ scalar_t shared_neg[512];
        
        // Initialize shared memory
        shared_pos[threadIdx.x] = dist_pos;
        shared_neg[threadIdx.x] = dist_neg;
        __syncthreads();
        
        // Parallel reduction in shared memory
        for (int stride = blockDim.x/2; stride > 0; stride >>= 1) {
            if (threadIdx.x < stride) {
                shared_pos[threadIdx.x] += shared_pos[threadIdx.x + stride];
                shared_neg[threadIdx.x] += shared_neg[threadIdx.x + stride];
            }
            __syncthreads();
        }
        
        if (feat_idx == 0) {
            // Final loss computation for this batch element
            const scalar_t loss = max(scalar_t(0.0), sqrt(dist_pos) - sqrt(dist_neg) + margin);
            output[batch_idx] = loss;
        }
    }
}

torch::Tensor triplet_margin_loss_cuda(
    torch::Tensor anchor,
    torch::Tensor positive,
    torch::Tensor negative,
    float margin) {
    
    TORCH_CHECK(anchor.device().is_cuda(), "anchor must be a CUDA tensor");
    TORCH_CHECK(positive.device().is_cuda(), "positive must be a CUDA tensor");
    TORCH_CHECK(negative.device().is_cuda(), "negative must be a CUDA tensor");
    
    const int batch_size = anchor.size(0);
    const int feat_size = anchor.size(1);
    auto output = torch::zeros({batch_size}, anchor.options());
    
    const int threads = 512;  // optimized block size
    const int blocks = (batch_size * feat_size + threads - 1) / threads;
    
    AT_DISPATCH_FLOATING_TYPES(anchor.scalar_type(), "triplet_margin_loss_kernel", ([&] {
        triplet_margin_loss_kernel<scalar_t><<<blocks, threads>>>(
            anchor.data_ptr<scalar_t>(),
            positive.data_ptr<scalar_t>(),
            negative.data_ptr<scalar_t>(),
            output.data_ptr<scalar_t>(),
            margin,
            batch_size,
            feat_size);
    }));
    
    return output.mean();
}

PYBIND11_MODULE(TORCH_EXTENSION_NAME, m) {
    m.def("forward", &triplet_margin_loss_cuda, "Triplet margin loss forward (CUDA)");
}
Performance Metrics
Metric Value Unit Variance Samples
Executed Ipc Active 1.926 inst/cycle 0.000 5
Executed Ipc Elapsed 1.374 inst/cycle 0.000 5
Issue Slots Busy 48.598 % 0.190 5
Issued Ipc Active 1.942 inst/cycle 0.000 5
SM Busy 48.598 % 0.190 5
Memory Throughput 771496646710.068 byte/second 11071283447668240384.000 5
Mem Busy 14.152 % 0.003 5
Max Bandwidth 23.132 % 0.012 5
L1/TEX Hit Rate 0.000 % 0.000 5
L2 Hit Rate 12.996 % 0.007 5
Mem Pipes Busy 33.476 % 0.020 5
Warp Cycles Per Issued Instruction 28.686 cycle 0.012 5
Warp Cycles Per Executed Instruction 28.950 cycle 0.012 5
Avg. Active Threads Per Warp 31.280 0.000 5
Avg. Not Predicated Off Threads Per Warp 29.260 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 5.000 block 0.000 5
Block Limit Shared Mem 12.000 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 87.570 % 0.028 5
Achieved Active Warps Per SM 56.044 warp 0.011 5
Analysis Rules
Rule Description
INF HighPipeUtilization ALU is the highest-utilized pipeline (32.3%) 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 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.
WRN Occupancy This kernel's theoretical occupancy is not impacted by any block limit. The difference between calculated theoretical (100.0%) and measured achieved occupancy (87.7%) can be the result of warp scheduling overheads or workload imbalances during the kernel execution. Load imbalances can occur between warps within a block as well as across blocks of the same kernel. See the CUDA Best Practices Guide (https://docs.nvidia.com/cuda/cuda-c-best-practices-guide/index.html#occupancy) for more details on optimizing occupancy.
Operation / Metric Value Unit
aten::zeros
CPU Time 4941889.80 μs
Device Time 199466.77 μs
Self CPU Time 144393.66 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 5272294.55 μs
Device Time 7174376.98 μs
Self CPU Time 290326.48 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 4981972.64 μs
Device Time 7174376.98 μs
Self CPU Time 365543.14 μs
Self Device Time 7174376.98 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 5345794.35 μs
Device Time 457613.34 μs
Self CPU Time 5345794.35 μs
Self Device Time 457613.34 μs
CPU Memory Usage 0 B
Device Memory Usage 0 B
Self CPU Memory Usage 0 B
Self Device Memory Usage 0 B
void triplet_margin_loss_kernel<float>(float const*, float const*, float const*, float*, float, int, int)
CPU Time 0.00 μs
Device Time 666194.34 μs
Self CPU Time 0.00 μs
Self Device Time 666194.34 μs
CPU Memory Usage 0 B
Device Memory Usage 0 B
Self CPU Memory Usage 0 B
Self Device Memory Usage 0 B
aten::mean
CPU Time 1117276.81 μs
Device Time 364164.85 μs
Self CPU Time 693944.93 μs
Self Device Time 364164.85 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 6974910.21 μs
Self CPU Time 0.00 μs
Self Device Time 6974910.21 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 
45286 warnings generated when compiling for host.
Suppressed 45324 warnings (45277 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_1/task_99/b4_s3_warp_shfl_triplet_loss_optimized_edit_2/edit_1/edit_1.cu:9:5 bugprone-easily-swappable-parameters
9 | const scalar_t* anchor,
| ^~~~~~~~~~~~~~~~~~~~~~~
10 | const scalar_t* positive,
| ~~~~~~~~~~~~~~~~~~~~~~~~~
11 | const scalar_t* negative,
| ~~~~~~~~~~~~~~~~~~~~~~~~
/home/robert_sakana_ai/llm_cuda/experiments/20250212_optimize_b5_s4_e1_v2/level_1/task_99/b4_s3_warp_shfl_triplet_loss_optimized_edit_2/edit_1/edit_1.cu:9:21: note: the first parameter in the range is 'anchor'
9 | const scalar_t* anchor,
| ^~~~~~
/home/robert_sakana_ai/llm_cuda/experiments/20250212_optimize_b5_s4_e1_v2/level_1/task_99/b4_s3_warp_shfl_triplet_loss_optimized_edit_2/edit_1/edit_1.cu:11:21: note: the last parameter in the range is 'negative'
11 | const scalar_t* negative,
| ^~~~~~~~
/home/robert_sakana_ai/llm_cuda/experiments/20250212_optimize_b5_s4_e1_v2/level_1/task_99/b4_s3_warp_shfl_triplet_loss_optimized_edit_2/edit_1/edit_1.cu:13:5: warning: 2 adjacent parameters of 'triplet_margin_loss_kernel' of convertible types are easily swapped by mistake [bugprone-easily-swappable-parameters]
13 | const float margin,
| ^~~~~~~~~~~~~~~~~~~
14 | const int batch_size,
| ~~~~~~~~~~~~~~~~~~~~
/home/robert_sakana_ai/llm_cuda/experiments/20250212_optimize_b5_s4_e1_v2/level_1/task_99/b4_s3_warp_shfl_triplet_loss_optimized_edit_2/edit_1/edit_1.cu:13:17: note: the first parameter in the range is 'margin'
13 | const float margin,
| ^~~~~~
/home/robert_sakana_ai/llm_cuda/experiments/20250212_optimize_b5_s4_e1_v2/level_1/task_99/b4_s3_warp_shfl_triplet_loss_optimized_edit_2/edit_1/edit_1.cu:14:15: note: the last parameter in the range is 'batch_size'
14 | const int batch_size,
| ^~~~~~~~~~
/home/robert_sakana_ai/llm_cuda/experiments/20250212_optimize_b5_s4_e1_v2/level_1/task_99/b4_s3_warp_shfl_triplet_loss_optimized_edit_2/edit_1/edit_1.cu:14:5: note: 'const float' and 'const int' may be implicitly converted: 'const float' (as 'float') -> 'const int' (as 'int'), 'const int' (as 'int') -> 'const float' (as 'float')
14 | const int batch_size,
| ^
/home/robert_sakana_ai/llm_cuda/experiments/20250212_optimize_b5_s4_e1_v2/level_1/task_99/b4_s3_warp_shfl_triplet_loss_optimized_edit_2/edit_1/edit_1.cu:17:21: warning: narrowing conversion from 'unsigned int' to signed type 'int' is implementation-defined [bugprone-narrowing-conversions]
17 | const int tid = blockIdx.x * blockDim.x + threadIdx.x;
| ^
/home/robert_sakana_ai/llm_cuda/experiments/20250212_optimize_b5_s4_e1_v2/level_1/task_99/b4_s3_warp_shfl_triplet_loss_optimized_edit_2/edit_1/edit_1.cu:47:27: warning: narrowing conversion from 'unsigned int' to signed type 'int' is implementation-defined [bugprone-narrowing-conversions]
47 | for (int stride = blockDim.x/2; stride > 0; stride >>= 1) {
| ^
/home/robert_sakana_ai/llm_cuda/experiments/20250212_optimize_b5_s4_e1_v2/level_1/task_99/b4_s3_warp_shfl_triplet_loss_optimized_edit_2/edit_1/edit_1.cu:73:28: warning: narrowing conversion from 'int64_t' (aka 'long') to signed type 'int' is implementation-defined [bugprone-narrowing-conversions]
73 | const int batch_size = anchor.size(0);
| ^
/home/robert_sakana_ai/llm_cuda/experiments/20250212_optimize_b5_s4_e1_v2/level_1/task_99/b4_s3_warp_shfl_triplet_loss_optimized_edit_2/edit_1/edit_1.cu:74:27: warning: narrowing conversion from 'int64_t' (aka 'long') to signed type 'int' is implementation-defined [bugprone-narrowing-conversions]
74 | const int feat_size = anchor.size(1);
| ^
/home/robert_sakana_ai/llm_cuda/experiments/20250212_optimize_b5_s4_e1_v2/level_1/task_99/b4_s3_warp_shfl_triplet_loss_optimized_edit_2/edit_1/edit_1.cu:80:5: 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]
80 | AT_DISPATCH_FLOATING_TYPES(anchor.scalar_type(), "triplet_margin_loss_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__, \
| ^