ImperasFPM Fast Processor Models

on YouTube ↗10m

" summary

Imperas FPM uses JIT binary translation of RISC‑V to x86 and claims to boot Linux in seconds where RTL runs at kHz and takes days, and emulators/FPGAs take hours. The simulator architecture keeps model and engine separate, so performance scales linearly to 64–128 cores for large multicore arrays. Verification needs are huge: one add instruction yields 10–30 cover points, ~20–40 SystemVerilog lines per point, leading to ~300,000 lines to cover an RV64GC‑class ISA. Synopsys Imperas DV generates that coverage.

AI-written summary · about

" chapters (7)

1. 0:00Introduction to ImperasFPM
2. 1:08Industry trends in RISC-V
3. 3:15Why fast CPU models matter
4. 5:22CPU verification challenges
5. 6:06What makes models fast
6. 7:12CPU verification complexity
7. 9:18Conclusion and Q&A

" related talks

• 03mImperasRick O'Connor, Larry Lapides · Embedded World 2019
• 09mDemo: Imperas/OVP RISC-V Models: Not Your Grandparents’ ISSJon Taylor · RISC-V Summit North America 2023
• 15mA Common Software Development Environment for Many-core RISC-V based Hardware and Virtual PlatformsSimon Davidmann, Gajinder Panesar · 8th RISC-V Workshop Barcelona 2018
• 06mImperas Interviewembedded world 2023
• 11mGetting Started with RISC-V Custom InstructionsJon Taylor · embedded world 2023
• 21mThe New Verification Ecosystem that Supports RISC-V Verification fo...Lee Moore, John Sotiropoulos · RISC-V Summit 2022
• 13mEarly RISC-V Software Verification with SIM-VLukas Jünger · embedded world 2023
• 10mRISC-V Virtual Platforms for Early RISC-V Embedded Software DevelopmentKevin McDermott, Lee Moore, Hugh O’Keeffe · 8th RISC-V Workshop Barcelona 2018