Skip to content
View ichen522's full-sized avatar

Highlights

  • Pro

Block or report ichen522

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse

Pinned Loading

  1. ARM_CPU ARM_CPU Public

    Quad-threaded 5-stage ARMv4T-subset CPU in Verilog. Runs GCC-compiled C on 4 hardware threads with zero-overhead context switching and a micro-op engine for LDM/STM.

    Verilog 1

  2. SRAM_Compiler SRAM_Compiler Public

    SRAM memory compiler in 45 nm (GPDK045): parameterized SKILL layout generators (0.71 µm² 6T bitcell, sense amp, decoder), Innovus-routed control with per-bank self-timing. 1/2/4/8-bank macros DRC/L…

    Verilog

  3. llama3.2-int4-moe llama3.2-int4-moe Public

    INT4 weight quantization and LoRA-based Mixture-of-Experts for Llama 3.2-1B, implemented from scratch in PyTorch.

    Python

  4. AXI_DMA AXI_DMA Public

    4-channel AXI4 DMA in SystemVerilog. Verified with Verilator, a DPI-C reference model, SVA, 100% functional coverage, mutation testing and a formal fairness proof. Signed off at 500 MHz in FreePDK4…

    SystemVerilog

  5. sm-dft-flow sm-dft-flow Public

    End-to-end ASIC DFT flow for a GPU Tensor Core SM: RTL synthesis, scan insertion, P&R, ATPG (99.92% fault coverage), STA, SDF simulation, and MBIST using FreePDK45.

    Python

  6. asic_design_flow asic_design_flow Public

    RTL-to-GDSII ASIC design of a GPU Tensor Core using Synopsys Design Compiler and Cadence Innovus, achieving 400 MHz post-route timing closure in 45nm.

    Tcl