Skip to content
STIMSMITH

DiffTest-H

Paper
First seen 6/6/2026
Last seen 6/6/2026
Evidence 15 chunks

NEIGHBORHOOD

No graph connections found for this entity yet. It may appear in future ingestion runs.

explore full graph →

RELATIONSHIPS

43 connections
XiangShan evaluates → 100% 3e
DiffTest-H is deployed and evaluated on XiangShan, a 6-wide out-of-order RISC-V processor.
register update uses → 100% 2e
DiffTest-H uses register update events as verification events.
memory access (MMIO) uses → 100% 2e
DiffTest-H handles MMIO access as a non-deterministic event.
communication overhead uses → 100% 2e
DiffTest-H addresses communication overhead as its primary optimization target.
hardware-software communication uses → 100% 2e
DiffTest-H optimizes hardware-software communication for co-simulation.
Design Under Test (DUT) uses → 100% 2e
DiffTest-H verifies the design under test by comparing its states against REF.
golden reference model (REF) uses → 100% 2e
DiffTest-H uses a golden reference model for comparison against the DUT.
TLB operations uses → 90% 2e
DiffTest-H covers TLB operations as part of its verification event types.
NutShell evaluates → 90% 2e
DiffTest-H is evaluated on NutShell, a scalar in-order processor.
Cadence Palladium uses → 100% 2e
DiffTest-H is deployed on Cadence Palladium emulator for hardware-accelerated verification.
Xilinx VU19P FPGA uses → 100% 2e
DiffTest-H is deployed on Xilinx VU19P FPGA for hardware-accelerated verification.
Verilator compares with → 100% 2e
DiffTest-H is compared against 16-thread Verilator for simulation speed.
Fromajo compares with → 100% 2e
DiffTest-H is compared against the Fromajo co-simulation framework on FPGA.
DiffTest compares with → 100% 2e
DiffTest-H is compared to the baseline DiffTest framework for speedup.
FPGA prototyping uses → 100% 2e
DiffTest-H is deployed on FPGA for hardware-accelerated co-simulation.
hardware emulation uses → 100% 2e
DiffTest-H is deployed on hardware emulator for hardware-accelerated co-simulation.
DPI-C interface uses → 90% 2e
DiffTest-H uses DPI-C as a hardware-software communication interface.
DiffTest-H uses non-blocking hardware-software parallelism to hide software processing latency.
verification events uses → 100% 2e
DiffTest-H processes verification events for co-simulation.
non-deterministic events (NDEs) uses → 100% 2e
DiffTest-H handles non-deterministic events that must be synchronized to REF.
ISA checker uses → 100% 2e
DiffTest-H uses an ISA checker on the software side to verify DUT correctness.
instruction commit uses → 100% 2e
DiffTest-H uses instruction commit events as verification events from the DUT.
Batch (tight packing of verification events) introduces → 100% 2e
DiffTest-H introduces the Batch technique for tightly packing verification events.
Squash (order-decoupled fusion) introduces → 100% 2e
DiffTest-H introduces the Squash technique for order-decoupled event fusion.
DiffTest-H introduces the Replay technique for lightweight debugging.
Co-simulation uses → 100% 2e
DiffTest-H is a hardware-accelerated co-simulation framework.
structural semantics uses → 100% 2e
DiffTest-H leverages structural semantics to optimize communication.
order semantics uses → 100% 2e
DiffTest-H leverages order semantics to decouple communication from checking order.
LogGP model uses → 100% 2e
DiffTest-H uses the LogGP model to analyze hardware-software communication overhead.
Yinan Xu authored by → 100% 1e
Yinan Xu is listed as an author of the DiffTest-H paper.
Kehan Feng authored by → 100% 1e
Kehan Feng is listed as an author of the DiffTest-H paper.
Luoshan Cai authored by → 100% 1e
Luoshan Cai is listed as an author of the DiffTest-H paper.
Yaoyang Zhou authored by → 100% 1e
Yaoyang Zhou is listed as an author of the DiffTest-H paper.
Yungang Bao authored by → 100% 1e
Yungang Bao is listed as an author of the DiffTest-H paper.
DUT trace dumping and reloading uses → 100% 1e
DiffTest-H uses DUT trace dumping and reloading to support iterative debugging.
cache coherence verification uses → 90% 1e
DiffTest-H covers cache coherence as one of its verification event types.
Spike (ISS) uses → 85% 1e
DiffTest-H uses Spike as a reference ISS.
NEMU uses → 85% 1e
DiffTest-H uses NEMU as a reference ISS.
XDMA protocol uses → 90% 1e
DiffTest-H FPGA platform uses XDMA protocol for hardware-software communication.
GFIFO (emulator non-blocking primitive) uses → 90% 1e
DiffTest-H uses GFIFO primitive on Palladium for non-blocking transmission.
behavioral semantics uses → 100% 1e
DiffTest-H leverages behavioral semantics to enable lightweight debugging.
Chisel HDL uses → 100% 1e
DiffTest-H is implemented using Chisel HDL.
Kunlin You authored by → 100% 1e
Kunlin You is listed as an author of the DiffTest-H paper.