arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2607.07643cs.AR
收藏

ATLAS:用于DL优化FPGA的自动化高级综合

ATLAS: Automated HLS for DL-Optimized FPGAs

Arizona State University(亚利桑那州立大学)

Ruthwik Reddy Sunketa, Aman Arora

AI总结:

研究针对FPGA架构中加速DL推理的问题,提出ATLAS全自动流程,以高级DL模型描述为起点,利用GEMM作为通用抽象层,由hls4ml - GEMM和GEMM IP生成器组成,实现无需手动RTL设计的硬件实现,评估了11个DL设计。

摘要:

FPGA架构越来越多地集成特定领域的片上硬模块以加速DL推理,尤其是在DL计算中占主导地位的GEMM。为实现这些硬模块的性能提升,需要手动RTL设计。虽然使用C/C++编程和HLS工具提高了FPGA工程师的抽象级别和生产力,但HLS工具本身不支持为自定义硬模块生成代码。之前的工作表明HLS工具中的黑盒机制可用于针对自定义硬模块,但仍需在用户编写的HLS C中进行显式函数调用并手动创建RTL IP库,且DL模型的每一层都需重复此大量工作。此外,对于DL,更高级别的编程接口(如Pytorch/Keras而非C/C++)更有利于提高可编程性和用户接受度。我们提出了ATLAS,这是一种从高级DL模型描述到在具有自定义片上DL优化硬模块的FPGA上进行硬件实现的全自动流程,无需最终用户进行手动RTL设计或显式硬模块实例化。我们的方法使用GEMM作为通用抽象层,包括两个组件:(1)hls4ml - GEMM,一个编译器前端,将DL层转换为具有与架构无关的GEMM函数调用的HLS C代码;(2)GEMM IP生成器,一个架构感知后端,生成具有平铺逻辑、控制有限状态机和调度元数据的基于硬模块的RTL包装器。我们针对11个DL设计评估了该流程,包括单个全连接、卷积和注意力层,以及完整的CNN、MLP和Transformer模型,目标是使用Catapult进行HLS和VTR进行实现的具有张量切片的FPGA架构。

英文摘要:

FPGA architectures increasingly incorporate domain-specific in-fabric hardblocks to accelerate DL inference, particularly GEMM, which dominates DL computation. To realize the performance gains of these hardblocks, manual RTL design is required: the programmer must understand the hardblock microarchitecture, instantiate them in RTL, and manage tiling and control logic. While programming in C/C++ and using HLS tools has increased the abstraction level and productivity of FPGA engineers, HLS tools do not support code generation for custom hardblocks natively. Prior work has demonstrated that blackbox mechanisms in HLS tools can be used to target custom hardblocks, but this still requires explicit function calls in user-written HLS C and manual creation of RTL IP libraries, significant effort that must be repeated for every layer in a DL model. Furthermore, for DL, an even high-level programming interface, e.g., Pytorch/Keras instead of C/C++, is desirable for improved programmability and user adoption. We present ATLAS, a fully automated flow from a high-level DL model description to a hardware implementation on an FPGA with custom in-fabric DL-optimized hardblocks, requiring no manual RTL design or explicit hardblock instantiation from the end user. Our approach uses GEMM as a universal abstraction layer and comprises two components: (1) hls4ml-GEMM, a compiler frontend that transforms DL layers into HLS C code with architecture-agnostic GEMM function calls, and (2) a GEMM IP Generator, an architecture-aware backend that produces hardblock-based RTL wrappers with tiling logic, control FSMs, and scheduling metadata. We evaluate the flow across 11 DL designs, including individual fully connected, convolution, and attention layers, as well as full CNN, MLP, and Transformer models targeting an FPGA architecture with Tensor Slices using Catapult for HLS and VTR for implementation.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑