TOGSim C++ trace-generation pipeline (P0-P3): explicit dataflow producer + barriers#267

Open

YWHyuk wants to merge 64 commits into

feature/togsim-cpp-trace

Commits on Jul 7, 2026

[Docs] C++ trace pipeline design (runtime-tag pairing, ABI)
YWHyuk
committed
[TOGSim] C++ trace pipeline: front end, runtime, loader, bridge, Core feed
YWHyuk
committed
[TOGSim] Per-iteration tag pairing for multi-tile-K and conv
YWHyuk
committed
[TOGSim] Work-item outlining and ABI v12 dispatch
YWHyuk
committed
[TOGSim] SRAM-capacity and SA weight-buffer throttle for the trace path
YWHyuk
committed
[Tooling] TOGSim trace timeline (Perfetto) and the trace emits it needs
YWHyuk
committed
[TOGSim] Make the C++ trace path the default and stabilize it
YWHyuk
committed
[TOGSim] Make the trace runtime test self-contained
YWHyuk
committed
[Frontend] Trace cache-safe replay and compile-race fixes
YWHyuk
committed
[TOGSim] Redesign trace-bridge dependency, barrier, SRAM-version, and runtime model
YWHyuk
committed
[Frontend] Run the spad-overflow check in timing-only mode, budget at spad/2
YWHyuk
committed
[Frontend] Generate the trace.cpp ABI/API banner from togsim_runtime.h
YWHyuk
committed
[TOGSim] Pick 1- vs 2-dispatch concurrency by per-dispatch spad footprint
YWHyuk
committed
[Frontend] Budget fused-epilogue spad buffers honestly in GEMM tile selection
YWHyuk
committed
[Frontend] Stop charging the kernel stack frame against the spad budget
YWHyuk
committed
[CI] Bump spike pin to v1.0.3
YWHyuk
committed
[Frontend] Add per-kernel CPU functional verify sub-option
YWHyuk
committed
[Frontend] Budget fused-prologue spad buffers in BMM tile selection
YWHyuk
committed
[Docs] Document per-kernel functional verify
YWHyuk
committed
[Frontend] Condense BMM prologue spad-budget comments to one line
YWHyuk
committed
[TOGSim] Drop the ABI version changelog comment
YWHyuk
committed
Merge pull request #278 from PSAL-POSTECH/fix/bmm-prologue-spad-budget
YWHyuk
committed
[TOGSim] Skip per-instruction trace string build when trace logging is off
YWHyuk
committed
[TOGSim] Skip the per-cycle issue scan when nothing can newly issue
YWHyuk
committed
[TOGSim] Fix use-after-erase in the zero-cycle COMP skip path
YWHyuk
committed
[Frontend] Make the C++ trace the sole main TOG path; drop legacy ONNX TOG
YWHyuk
committed
[Frontend] Represent kernel body as ordered steps; multi-step indirect
YWHyuk
committed
[Frontend] Carry gather/scatter offset as an explicit transfer descriptor
YWHyuk
committed
[Frontend] Clean up indirect access: symbol-set detection, CONFIG4 stride, compute-loop multi-dim offset
YWHyuk
committed
[Frontend] Drop memref.dma_start: lower togsim.transfer directly to Gemmini
YWHyuk
committed
[Frontend] Port the trace path (build_tog) to togsim.transfer/togsim.wait
YWHyuk
committed
[Frontend] Detect the indirect offset operand by the `indirect` attr, not operand count
YWHyuk
committed
[Frontend] Fix stale build_skeleton docstrings (memref.dma_wait -> togsim.wait)
YWHyuk
committed
[Frontend] Lower togsim.transfer to a DMA descriptor + CONFIG_DESC (no CONFIG/2/3/4)
YWHyuk
committed
[Build] Bump spike to v1.0.4 (DMA descriptor / CONFIG_DESC)
YWHyuk
committed
[Build] Bump spike to v1.0.5 (masked-DMA clamp + MVOUT accumulate)
YWHyuk
committed
[Frontend] Masked-DMA dynamic clamp: drop tile divisibility, fill non-dividing tails
YWHyuk
committed
[Frontend] Extend the masked-DMA clamp to conv (explicit extents + fine-grained)
YWHyuk
committed
[Frontend] Clamp the fused-epilogue output store on non-dividing shapes
YWHyuk
committed
[Test] Golden unit test for the lower_transfer_to_gemmini pass
YWHyuk
committed
[Frontend] Masked-DMA review fixes: exp target-match, explicit local_dims, cleanup
YWHyuk
committed
[Test] Non-dividing masked-DMA regression suite + CI
YWHyuk
committed
[Frontend] Drop the dead self.masks splice in the reduction-fusion epilogue
YWHyuk
committed
[Frontend] Reduce >4D togsim.transfer to <=4D before build_tog (fix pixel_shuffle trace)
YWHyuk
committed
[Frontend] Per-kernel verify: run the CPU golden fully on CPU (device-safe)
YWHyuk
committed

Commits on Jul 8, 2026

Commits on Jul 9, 2026

Commits on Jul 10, 2026