ComfyUI Node: HSWQ Loader for INT8 Quantized ControlNets
Zestyclose_Bake3680 · reddit · 2026-08-27
Addressing ComfyUI's inability to load INT8 quantized ControlNet models (e.g., Qwen Image Fun ControlNet), a new loader node ComfyUI-HSWQ-Loader has been released.
Core Features:
- Native INT8 Support: Keeps weights in VRAM using TensorWiseINT8Layout, significantly reducing memory usage.
- Fast Execution: Utilizes comfykitchen's int8linear kernel with online activation rotation.
- Compatibility: Fully backward compatible with FP16/BF16/FP8 models, serving as a drop-in replacement for the stock loader.
The node fixes PyTorch gradient errors and correctly injects MixedPrecisionOps configurations.
More from Infra
- Two vLLM recipes for Blackwell: NVFP4 KV cache buys 262K context and more streams — SeanHighness · 2026-08-27
- Dev forks Nvidia drivers to enable PCIe P2P on GeForce for SlimServe — QuixiAI · 2026-08-27
- View: Single dev with 200 B300s could beat Alibaba's post-training team — kalomaze · 2026-08-27
- GLM 5.3 Flash Benchmark: Hits 881 tok/s on Dual DGX — teortaxesTex · 2026-08-27
- Rakyll advises: If you have any CPU nodes, hold on to them — rakyll · 2026-08-27
- OpenRouter Serves 200B Total Tokens, Adds Qwen 3.5 35B — gajesh · 2026-08-27