Dual 5060Ti Inference: Is PCIe Bandwidth a Bottleneck?
Ok-Conflict391 · reddit · 2026-08-21
A user running a 27B parameter model (Q6 quantized) on dual RTX 5060Ti GPUs achieved 40-50 tps. HWinfo monitoring revealed that both the PCIe 5x16 and PCIe 4x4 slots were fully saturated during inference. The user suspects the second x4 slot is a bottleneck and is considering moving the GPU to a spare PCIe 5x4 NVMe slot via a riser.
More from Infra
- Muse Spark Open runs locally on a Mac M3 Ultra via llama.cpp: frontier model on desktop — pcuenq · 2026-10-16
- Muse-Spark open-weights drop: 816B-A42B MoE model gets 359GB MLX 3-bit build — ns123abc · 2026-10-16
- vLLM ships Day-0 LoRA RL for Muse Spark Open, lifting AIME from 0.56 to 0.85 — vllm_project · 2026-10-16
- DSpark draft model for Muse Spark Open proposes 7 tokens per pass with 23-42% acceptance — vllm_project · 2026-10-16
- Cloudflare Stream's WebRTC ultra-low-latency live streaming is now Generally Available — ritakozlov · 2026-10-16
- YC-backed Overshoot launches inference optimization, claims $5M saved for AI teams — ycombinator · 2026-10-16