high performance computing on graphics processing units: hgpu.org

hgpu.org » LLM

Omniwise: Predicting GPU Kernels Performance with LLMs

Zixian Wang, Cole Ramos, Muhammad A. Awad, Keith Lowery

View

Tags: AMD, AMD Radeon Instinct MI250, AMD Radeon Instinct MI300X, Artificial intelligence, Benchmarking, Computer science, LLM, Neural networks, Performance, ROCm

June 29, 2025 by hgpu

A First Look at Bugs in LLM Inference Engines

Mugeng Liu, Siqi Zhong, Weichen Bi, Yixuan Zhang, Zhiyang Chen, Zhenpeng Chen, Xuanzhe Liu, Yun Ma

View

Download (PDF)

Tags: AI, Computer science, LLM, Software Engineering

June 22, 2025 by hgpu

CUDA-LLM: LLMs Can Write Efficient CUDA Kernels

Wentao Chen, Jiace Zhu, Qi Fan, Yehan Ma, An Zou

View

Download (PDF)

Tags: Artificial intelligence, Code generation, Computer sceince, CUDA, LLM, nVidia, nVidia GeForce GTX 1660, nVidia GeForce RTX 3090 Ti

June 15, 2025 by hgpu

HPCTransCompile: An AI Compiler Generated Dataset for High-Performance CUDA Transpilation and LLM Preliminary Exploration

Jiaqi Lv, Xufeng He, Yanchen Liu, Xu Dai, Yang Hu, Shouyi Yin

View

Download (PDF)

Source codes

Tags: AI, Benchmarking, Compilers, Computer science, CUDA, Deep learning, LLM, nVidia, nVidia A100, Package, performance portability

June 15, 2025 by hgpu

Acceleration as a Service (XaaS) Source Containers

Eiman Alnuaimi

View

Download (PDF)

Source codes

Tags: Computer science, Heterogeneous systems, HPC, Intel, Intel Data Center GPU Max 1550, LLM, MPI, nVidia, nVidia GH200, nVidia V100, Optimization, Package, performance portability, Thesis

June 8, 2025 by hgpu

MemAscend: System Memory Optimization for SSD-Offloaded LLM Fine-Tuning

Yong-Cheng Liaw, Shuo-Han Chen

View

Download (PDF)

Tags: Artificial intelligence, Benchmarking, Computer science, LLM, Memory, nVidia, nVidia H100, nVidia RTX A5000

June 8, 2025 by hgpu

Performance of Confidential Computing GPUs

Antonio Martínez Ibarra, Julian James Stephen, Aurora González Vidal, K. R. Jayaram, Antonio Fernando Skarmeta Gómez

View

Download (PDF)

Tags: Computer science, CUDA, LLM, nVidia, nVidia H100, Performance, Security

May 25, 2025 by hgpu

Can Large Language Models Predict Parallel Code Performance?

Gregory Bolet, Giorgis Georgakoudis, Harshitha Menon, Konstantinos Parasyris, Niranjan Hasabnis, Hayden Estes, Kirk W. Cameron, Gal Oren

View

Download (PDF)

Source codes

Tags: Benchmarking, Computer science, CUDA, LLM, nVidia, nVidia GeForce RTX 3080, OpenMP, Package, Performance, performance portability

May 18, 2025 by hgpu

GPU Performance Portability needs Autotuning

Burkhard Ringlein, Thomas Parnell, Radu Stoica

View

Download (PDF)

Tags: AMD Radeon Instinct MI250, ATI, Auto-Tuning, Computer science, CUDA, DSL, HIP, LLM, nVidia, nVidia A100, Performance, performance portability

May 18, 2025 by hgpu

LIFT: LLM-Based Pragma Insertion for HLS via GNN Supervised Fine-Tuning

Neha Prakriya, Zijian Ding, Yizhou Sun, Jason Cong

View

Download (PDF)

Tags: AMD Radeon Instinct MI250, ATI, Computer science, FPGA, HLS, LLM, Machine learning, Neural networks, PyTorch

May 4, 2025 by hgpu

Scaling On-Device GPU Inference for Large Generative Models

Jiuqiang Tang, Raman Sarokin, Ekaterina Ignasheva, Grant Jensen, Lin Chen, Juhyun Lee, Andrei Kulik, Matthias Grundmann

View

Download (PDF)

Tags: AI, Computer science, CUDA, LLM, Machine learning, nVidia, nVidia GeForce RTX 4090, OpenCL

May 4, 2025 by hgpu

Data-efficient LLM Fine-tuning for Code Generation

Weijie Lv, Xuan Xia, Sheng-Jun Huang

View

Download (PDF)

Source codes

Tags: Code generation, Computer science, CUDA, LLM, nVidia, nVidia A100, Package, Python, PyTorch

April 27, 2025 by hgpu

torchPDLP: Primal-Dual Linear Programming in PyTorch. In collaboration with AMD and IPAM

Accelerating a Linear Programming Algorithm on AMD GPUs

Benchmarks for Dissecting CPU-GPU Unified Physical Memory on AMD MI300A APUs

Dissecting CPU-GPU Unified Physical Memory on AMD MI300A APUs

kvcached: Elastic KV cache for dynamic GPU sharing and efficient multi-LLM inference

Towards Efficient and Practical GPU Multitasking in the Era of LLM

Bandicoot: C++ library for GPU accelerated linear algebra

Bandicoot: A Templated C++ Library for GPU Linear Algebra

Luthier: Bridging Auto-Tuning and Vendor Libraries for Efficient Deep Learning Inference

Fused Kernel Library (FKL)

The Fused Kernel Library: A C++ API to Develop Highly-Efficient GPU Libraries

GPUHammer: Rowhammer Attacks on GPU Memories are Practical

Block: Balance Loader of LLM Serving with Context, Knowledge and Predictive Scheduling

Block: Balancing Load in LLM Serving with Context, Knowledge and Predictive Scheduling

SIGMo: Scalable Isomorphism Graph Matching on GPUs

SIGMo: High-Throughput Batched Subgraph Isomorphism on GPUs for Molecular Matching

DGEMM without FP64 Arithmetic - using FP64 Emulation and FP8 Tensor Cores with Ozaki Scheme

DGEMM without FP64 Arithmetic – using FP64 Emulation and FP8 Tensor Cores with Ozaki Scheme

See all packages

* * *

high performance computing on graphics processing units: hgpu.org

Omniwise: Predicting GPU Kernels Performance with LLMs

A First Look at Bugs in LLM Inference Engines

CUDA-LLM: LLMs Can Write Efficient CUDA Kernels

HPCTransCompile: An AI Compiler Generated Dataset for High-Performance CUDA Transpilation and LLM Preliminary Exploration

Acceleration as a Service (XaaS) Source Containers

MemAscend: System Memory Optimization for SSD-Offloaded LLM Fine-Tuning

Performance of Confidential Computing GPUs

Can Large Language Models Predict Parallel Code Performance?

GPU Performance Portability needs Autotuning

LIFT: LLM-Based Pragma Insertion for HLS via GNN Supervised Fine-Tuning

Scaling On-Device GPU Inference for Large Generative Models

Data-efficient LLM Fine-tuning for Code Generation

Recent source codes

torchPDLP: Primal-Dual Linear Programming in PyTorch. In collaboration with AMD and IPAM

Benchmarks for Dissecting CPU-GPU Unified Physical Memory on AMD MI300A APUs

kvcached: Elastic KV cache for dynamic GPU sharing and efficient multi-LLM inference

Bandicoot: C++ library for GPU accelerated linear algebra

Luthier: Bridging Auto-Tuning and Vendor Libraries for Efficient Deep Learning Inference

Fused Kernel Library (FKL)

GPUHammer: Rowhammer Attacks on GPU Memories are Practical

Block: Balance Loader of LLM Serving with Context, Knowledge and Predictive Scheduling

SIGMo: Scalable Isomorphism Graph Matching on GPUs

DGEMM without FP64 Arithmetic - using FP64 Emulation and FP8 Tensor Cores with Ozaki Scheme

Most viewed papers (last 30 days)