Expired50% confidenceReleaseExact time
This update adds int4 quantization support for convrot and includes targeted int8/int4 optimizations for NVIDIA Turing architecture GPUs.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
7/16/2026
First Seen
Valid until: 10/14/2026(expired)
Sources
Related Claims
UnverifiedRTX 30系列（Ampere架构）的Tensor Core对INT4的原生支持有限，INT4运算通常需要拆包模拟67% similarUnverifiedA developer has converted several mainstream image models to INT4 ConvRot W4A4 format and adapted them for the latest version of ComfyUI.67% similarUnverifiedThe Apple M4 Pro's Neural Engine reaches 38 TOPS, the Qualcomm Snapdragon X Elite integrates a 45 TOPS NPU, and Intel Core Ultra has also integrated an NPU to support local AI inference.66% similarUnverifiedComfyUI-INT4-Fast supports native mixed-precision checkpoints, determining the compute path at per-layer granularity: backbone modules are routed to INT4 Tensor Cores, while sensitive layers (patch projection) are directed to Triton's INT8 execution path in INT8 format.66% similarUnverifiedComfyUI-INT4-Fast was independently improved upon BobJohnson24's ComfyUI-INT8-Fast, and relies on comfy-kitchen under the hood to invoke the GPU's Tensor Cores.66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/535366API
curl https://kongchang.com/api/v1/knowledge/claims/535366MCP
get_claim(id=535366)