GCL-Sampler: Discovering Kernel Similarity for Sampled GPU Simulation via Graph Contrastive Learning

Jiaqi Wang; Jingwei Sun; Jiyu Luo; Han Li; Guangzhong Sun

cs.PFFeb 28, 2026

GCL-Sampler: Discovering Kernel Similarity for Sampled GPU Simulation via Graph Contrastive Learning

Authors: Jiaqi Wang, Jingwei Sun, Jiyu Luo, Han Li, Guangzhong Sun

Organizations: University of Science and Technology of China, Hefei, China

Abstract

GPU architectural simulation is orders of magnitude slower than native execution, necessitating workload sampling for practical speedups. Existing methods rely on hand-crafted features with limited expressiveness, yielding either aggressive sampling with high errors or conservative sampling with constrained speedups. To address these issues, we propose GCL-Sampler, a sampling framework that leverages Relational Graph Convolutional Networks with contrastive learning to automatically discover high-dimensional kernel similarities from trace graphs. By encoding instruction sequences and data dependencies into graph embeddings, GCL-Sampler captures rich structural and semantic properties of program execution, enabling both high fidelity and substantial speedup. Evaluations on extensive benchmarks show that GCL-Sampler achieves 258.94x average speedup against full workload with 0.37% error, outperforming state-of-the-art methods, PKA (129.23x, 20.90%), Sieve (94.90x, 4.10%) and STEM+ROOT (56.57x, 0.38%).

Figures & tables

Figure 1. GCL-Sampler achieves near-ideal performance (red star) with minimal error and maximum speedup compared to existing methods.

Figure 2. Overview of GCL-Sampler. GCL-Sampler transforms program traces into HRGs and leverages RGCN-based contrastive learning to generate kernel embeddings, followed by K-Means clustering.

Component	Description
$tb_{x}$ , $tb_{y}$ , $tb_{z}$	CTA coordinates in the grid
warp_id	Warp identifier within the CTA
PC	Program counter
mask	32-bit active lane mask
#dests	Number of destination registers
[dest_regs]	Destination register identifiers

Table 1. Trace components for each instruction.

Figure 3. An example of graph construction from traces.

	CPU (Intel)	GPU	Architecture
P1	E5-2680 v4	RTX 2080Ti	Turing
P2	i9-12900KF	RTX 3080Ti	Ampere
P3	W5-3435X	RTX 4090	Ada Lovelace

Table 2. Hardware configurations for experiments.

Figure 4. Sampling error of GCL-Sampler, PKA, Sieve and STEM+ROOT. A woman and a girl in white dresses sit in an open car.

Figure 5. Speedup of GCL-Sampler, PKA, Sieve and STEM+ROOT. A woman and a girl in white dresses sit in an open car.

Program	P1		P2		P3
Program	E (%)	SU	E (%)	SU	E (%)	SU
bfs	1.07	3.04 $\times$	<0.01	3.17 $\times$	0.05	3.51 $\times$
lu	0.92	2224.67 $\times$	0.11	1644.30 $\times$	<0.01	1658.56 $\times$
phi-2	0.63	18.41 $\times$	10.44	13.42 $\times$	9.74	14.01 $\times$
qwen1.5	0.38	5.19 $\times$	2.20	3.76 $\times$	1.93	3.81 $\times$
3mm	0.71	3.05 $\times$	1.61	3.00 $\times$	0.67	3.03 $\times$

Table 3. Sampling error and speedup across GPU microarchitecture. P1 denotes Turing, P2 denotes Ampere, and P3 denotes Ada Lovelace. E(%)=Error(in %), SU=Speedup.

Figure 6. Microarchitectural metrics comparison between sampled and full simulation for cfd and pythia workloads.

KernelFoundry: Hardware-aware evolutionary GPU kernel optimization
Mar 12, 2026Nina Wiedemann, Quentin Leboutet, Michael Paulitsch +2GPU Kernel Optimization GPU Acceleration
Intel Corporation
HIERA: Workload-Aware Planning Across Implementation Spaces for GPU Kernel Optimization
Aug 21, 2026Jinghao Wang, Qiqi Gu, Chenpeng Wu +3GPU Kernel Optimization GPU Kernel Generation
Shanghai Key Laboratory of Scalable Computing and Systems · School of Computer Science, Shanghai Jiao Tong University
Record-Remix-Replay: Hierarchical GPU Kernel Optimization using Evolutionary Search
Apr 13, 2026Daniel Nichols, Konstantinos Parasyris, Caetano Melone +3GPU Kernel Optimization Evolutionary Optimization
Lawrence Livermore National Laboratory, Livermore, USA

Mar 12, 2026J/K move · Enter open · S save

KernelFoundry: Hardware-aware evolutionary GPU kernel optimization

Nina Wiedemann, Quentin Leboutet, Michael Paulitsch, Diana Wofk+1

Intel Corporation

GPU Kernel Optimization GPU Acceleration Evolutionary Optimization Large Language Model-Guided Optimization LLM-Based Program Synthesis GPU Kernel Generation+1

Open paper AskPDF arXiv

GPU kernel optimization challenges LLMs beyond standard coding tasks, as it requires an understanding of hardware architecture, parallel computing optimization strategies, and profiling outputs. However, most existing approaches leveraging LLMs for kernel generation apply standard prompting and feedback loops, considering hardware only through profiling feedback. We introduce KernelFoundry, an evolutionary framework that efficiently explores the space of GPU kernels through (1) MAP-Elites quality diversity search with kernel-specific behavioral dimensions to sustain exploration; (2) meta-prompt evolution that co-evolves prompts with kernels to uncover task-specific optimization strategies, and (3) a template-based parameter optimization approach to tune kernels to inputs and hardware. We evaluate this framework on Kernel-Bench, robust-kbench and custom tasks, generating SYCL kernels as a cross-platform GPU programming paradigm, and CUDA kernels for comparison to prior work. Our approach consistently outperforms the baseline methods and achieves an average speedup of 2.3 on KernelBench for SYCL. Moreover, KernelFoundry is implemented as a distributed framework with remote access to diverse hardware, allowing quick benchmarking and featuring a flexible user input layer to support kernel generation for a wide range of real use cases beyond benchmarking.

Abstract

Explore similar work

KernelFoundry: Hardware-aware evolutionary GPU kernel optimization

HIERA: Workload-Aware Planning Across Implementation Spaces for GPU Kernel Optimization

Record-Remix-Replay: Hierarchical GPU Kernel Optimization using Evolutionary Search