Hardware●●●●Gem
We built Talos – a full CNN inference engine running on silicon
Strips away PyTorch flexibility entirely; full CNN inference as deterministic hardware logic in SystemVerilog.
WizardryZero to OneBig Brain
luthiraabeykoon
103mo ago
Back to browse
We built Talos – a full CNN inference engine running on silicon
by luthiraabeykoon·Feb 23, 2026·1 point·0 comments
AI Analysis
●●●●GemWizardryZero to OneBold Bet
CNN inference fully hardcoded as silicon logic, not software optimized for hardware.
Strengths
- •Strips runtime overhead entirely—every multiply, buffer, and data path is deterministic digital logic on FPGA, not scheduler-bound.
- •Built in two weeks under extreme constraint; demonstrates genuine hardware debugging craft (nanosecond timing closure, waveform hunting).
- •Flips the conventional wisdom: hardware accelerators usually adapt software logic, but Talos rethinks inference from the circuit level up.
Weaknesses
- •Unclear production viability—two-week timeline and Show HN framing suggest proof-of-concept, not shipping product with real benchmark comparisons.
- •No public performance claims against GPU baselines; latency and throughput numbers needed to evaluate practical advantage.
Category
Target Audience
ML engineers, hardware designers, inference optimization specialists
Similar To
NVIDIA TensorRT · Google TPU · Xilinx Vitis AI
Similar Projects
Hardware●●Solid
ML accelerator on a RISC-V FPGA SoC – zero-cycle matmul, boots Linux
Zero-cycle matrix multiplication in combinatorial logic on Lattice ECP5 is genuinely wild.
WizardryBig BrainNiche Gem
dstrbad
402mo ago
Hardware●●Solid
Low-rank approximation for 3x3 FPGA convolutions (33% less DSP usage)
Clever ML+hardware co-design, but a blog post without open-source code, benchmarks, or deployment examples.
WizardryNiche Gem
el_dockerr
113mo ago
Infrastructure●●●Banger
Open-source logic synthesis – formal logic to FPGA
Open-source logic synthesis running on FPGAs when Yosys dominates the space.
Zero to OneWizardryBold Bet
major4x
102mo ago
Data●●Solid
Benchmarking Apple Silicon unified mem for GPU-accelerated SQL analysis
The repo does one practical thing well: quantify the real-world impact of Apple Silicon's unified memory on analytics by running six TPC-H queries plus a GPU-favorable QX and shipping the raw charts and code. It's specific and empirical — you get MLX vs NumPy vs DuckDB numbers and PNGs, not just hand-wavy claims — but it's narrowly scoped to M4 hardware and small-ish scales, so its conclusions are useful for experimentation rather than sweeping generalization.
WizardryNiche Gem
sadopc
313mo ago
AI/ML●●●Banger
RunAnwhere – Faster AI Inference on Apple Silicon
Custom Metal shaders beat llama.cpp and MLX—1.67x faster on M4 Max.
WizardrySlickZero to One
sanchitmonga22
2401533mo ago