Unverified50% confidenceFactExact time
WASM后端1.6 tokens/s的低性能部分是因为页面未启用跨源隔离，导致ONNX Runtime Web使用单线程执行。
1
Sources
50%
Confidence
Long-term
Relevance
10/17/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe regular build of onnxruntime-web (ort-wasm-simd-threaded.{mjs,wasm}) includes the CPU kernel for GatherBlockQuantized and can support running 4-bit Gemma on WebAssembly.60% similarUnverifiedThe Node.js runtime has performance shortcomings in CPU-intensive scenarios.56% similarUnverifiedWhen the embedding service is unavailable, session-indexer automatically falls back to FTS5's BM25 full-text search.54% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1051515API
curl https://kongchang.com/api/v1/knowledge/claims/1051515MCP
get_claim(id=1051515)