Qwen 3.6 35b slower on 7900XTX than 3060Ti? User investigates Vulkan config
Jebbyk1 · reddit · 2026-08-25
A user found that the inference speed of the Qwen 3.6 35b (a3b q4-k-m) model on an AMD 7900XTX (20t/s, 100% load) was lower than on an NVIDIA 3060Ti (30t/s, 50% load). Neither Vulkan nor ROCm in a Linux llama.cpp environment resolved the performance regression, despite the theoretical expectation that the 7900XTX should be 2-3 times faster.
More from Infra
- Next-gen GraalVM benchmarks: 4x faster startup, 2x less memory, no build step needed — RealGeneKim · 2026-10-15
- How Microsoft optimized Windows for Nvidia's Arm-based RTX Spark AI laptops — tomwarren · 2026-10-15
- How Microsoft optimized Windows for Nvidia's Arm-based RTX Spark laptops — The Verge AI · 2026-10-15
- Global smartphone shipments plunge 22% in worst quarterly decline on record as AI demand fuels memory costs — Polymarket · 2026-10-15
- SK Hynix drops 11% again, now below 3x FY2027 earnings as AI memory selloff deepens — JOBhakdi · 2026-10-15
- ASML Sells Zero Lithography Machines into Europe as US Hosts 75%+ of AI Compute — NinaDSchick · 2026-10-15