eGPU: A 750 MHz Class Soft GPGPU for FPGA

被引:1
|
作者
Langhammer, Martin [1 ,2 ]
Constantinides, George A. [2 ]
机构
[1] Intel Corp, Programmable Solut Grp, Santa Clara, CA 95051 USA
[2] Imperial Coll London, Elect & Elect Engn, London, England
关键词
GPGPU; FPGA;
D O I
10.1109/FPL60245.2023.00047
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
This paper introduces the eGPU, a SIMT soft processor designed for FPGAs. Soft processors typically achieve modest operating frequencies, a fraction of the headline performance claimed by modern FPGA families, and obtain correspondingly modest performance results. We propose a GPGPU architecture structured specifically to take advantage of both the soft logic and embedded features of the FPGA. We also consider the physical location of the embedded memories and DSP Blocks relative to the location and number of soft logic elements in order to have a design with balanced resources. Our goal is to create a high performance soft processor able to implement complex portions of FPGA system designs, such as the linear solvers commonly used in wireless systems, through push-button compilation from software. The eGPU architecture is a streaming multiprocessor (SM) machine with 512 threads. Each SM contains 16 scalar processors (SP). Both IEEE754 FP32 and INT32 integer arithmetic are supported. We demonstrate a single SM eGPU in an Intel Agilex device, requiring 5600 ALMs and 24 DSP Blocks, which closes timing at over 770 MHz from a completely unconstrained compile. Multiple eGPUs can also be tightly packed together into a single Agilex FPGA logic region, with minimal speed penalty.
引用
收藏
页码:277 / 282
页数:6
相关论文
共 50 条
  • [1] Scalable Analytic Placement for FPGA on GPGPU
    Pattison, Ryan
    Fobel, Christian
    Grewal, Gary
    Areibi, Shawki
    2015 INTERNATIONAL CONFERENCE ON RECONFIGURABLE COMPUTING AND FPGAS (RECONFIG), 2015,
  • [2] FlexGrip: A Soft GPGPU for FPGAs
    Andryc, Kevin
    Merchant, Murtaza
    Tessier, Russell
    PROCEEDINGS OF THE 2013 INTERNATIONAL CONFERENCE ON FIELD-PROGRAMMABLE TECHNOLOGY (FPT), 2013, : 230 - 237
  • [3] Bridging the GPGPU-FPGA Efficiency Gap
    Fletcher, Christopher W.
    Lebedev, Ilia
    Asadi, Narges B.
    Burke, Daniel R.
    Wawrzynek, John
    FPGA 11: PROCEEDINGS OF THE 2011 ACM/SIGDA INTERNATIONAL SYMPOSIUM ON FIELD PROGRAMMABLE GATE ARRAYS, 2011, : 119 - 122
  • [4] 奔Ⅲ Coppermine 800MHz & 750MHz
    小玩子
    现代电子技术, 2000, (04) : 18 - 19
  • [5] A 750-MHZ MAP OF THE CRAB
    AGAFONOV, MI
    ASLANYAN, AM
    GULYAN, AG
    IVANOV, VP
    MARTIROSYAN, RM
    PODVOISKAYA, OA
    STANKEVICH, KS
    SOVIET ASTRONOMY LETTERS, 1986, 12 (02): : 112 - 114
  • [6] Integer Computations with Soft GPGPU on FPGAs
    Al Kadi, Muhammed
    Huebner, Michael
    2016 INTERNATIONAL CONFERENCE ON FIELD-PROGRAMMABLE TECHNOLOGY (FPT), 2016, : 28 - 35
  • [7] A Comparison of FPGA and GPGPU Designs for Bayesian Occupancy Filters
    Medina, Luis
    Diez-Ochoa, Miguel
    Correal, Raul
    Cuenca-Asensi, Sergio
    Serrano, Alejandro
    Godoy, Jorge
    Martinez-Alvarez, Antonio
    Villagra, Jorge
    SENSORS, 2017, 17 (11):
  • [8] 750-MHZ NMR MAGNET DEVELOPMENT
    ZHUKOVSKY, A
    IWASA, Y
    BOBROV, ES
    LUDLAM, J
    WILLIAMS, JEC
    HIROSE, R
    ZHAO, ZP
    IEEE TRANSACTIONS ON MAGNETICS, 1992, 28 (01) : 644 - 647
  • [9] Predicting the Soft Error Vulnerability of GPGPU Applications
    Topcu, Burak
    Oz, Isil
    30TH EUROMICRO INTERNATIONAL CONFERENCE ON PARALLEL, DISTRIBUTED AND NETWORK-BASED PROCESSING (PDP 2022), 2022, : 108 - 115
  • [10] Soft error vulnerability prediction of GPGPU applications
    Burak Topçu
    Işıl Öz
    The Journal of Supercomputing, 2023, 79 : 6965 - 6990