Saved in:
Bibliographic Details
Main Authors: Ahmad, Adeel, Kamal, Ahmad Tameem, Amir, Nouman, Zafar, Bilal, Nasir, Saad Bin
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2508.14899
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918128194158592
author Ahmad, Adeel
Kamal, Ahmad Tameem
Amir, Nouman
Zafar, Bilal
Nasir, Saad Bin
author_facet Ahmad, Adeel
Kamal, Ahmad Tameem
Amir, Nouman
Zafar, Bilal
Nasir, Saad Bin
contents This project enables RISC-V microkernel support in IREE, an MLIR-based machine learning compiler and runtime. The approach begins by enabling the lowering of MLIR linalg dialect contraction ops to linalg.mmt4d op for the RISC-V64 target within the IREE pass pipeline, followed by the development of optimized microkernels for RISC-V. The performance gains are compared with upstream IREE and Llama.cpp for the Llama-3.2-1B-Instruct model.
format Preprint
id arxiv_https___arxiv_org_abs_2508_14899
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Accelerating GenAI Workloads by Enabling RISC-V Microkernel Support in IREE
Ahmad, Adeel
Kamal, Ahmad Tameem
Amir, Nouman
Zafar, Bilal
Nasir, Saad Bin
Hardware Architecture
Artificial Intelligence
This project enables RISC-V microkernel support in IREE, an MLIR-based machine learning compiler and runtime. The approach begins by enabling the lowering of MLIR linalg dialect contraction ops to linalg.mmt4d op for the RISC-V64 target within the IREE pass pipeline, followed by the development of optimized microkernels for RISC-V. The performance gains are compared with upstream IREE and Llama.cpp for the Llama-3.2-1B-Instruct model.
title Accelerating GenAI Workloads by Enabling RISC-V Microkernel Support in IREE
topic Hardware Architecture
Artificial Intelligence
url https://arxiv.org/abs/2508.14899