ROIpad ← Back to Search

Insight for: Dual RTX 3090 (TP=2) vs single: a controlled A/B on your harness — +16-35%, DFlash2 residency, and the KV_MEM pin

Performance scaling of Qwen3.8-27B with vLLM using dual RTX 3090s (Tensor Parallelism TP=2) compared to a single card, specifically addressing the impact of KV_MEM pinning and overall throughput gains.

Analyzed: Aug 29, 2026

This detailed A/B test confirms significant performance gains (16-35% throughput increase) by scaling Qwen3.8-27B inference from single to dual RTX 3090s using Tensor Parallelism. The identification and resolution of the KV_MEM pinning issue, which previously stranded 8 GiB per card, directly improves resource utilization and validates the multi-GPU scaling strategy. This demonstrates a clear path for B2B customers to enhance inference capacity and throughput on readily available hardware. The collaborative, data-driven approach to identifying and fixing performance bottlenecks, even in complex multi-GPU setups, reinforces the value of open-source contributions and responsive development for critical infrastructure components like vLLM.

Dual RTX 3090 (TP=2) Tensor Parallelism controlled A/B MTP DFlash2 (greedy, decode tok/s) KV_MEM pin vLLM applies `--kv-cache-memory` per worker strands ~8 GiB per card launcher detects `--tensor-parallel-size`/`-tp` disable-custom-all-reduce p2p (but no nvlink) thermal throttling

GitHub Issue

Parent Entity

Dual RTX 3090 (TP=2) vs single: a controlled A/B on your harness — +16-35%, DFlash2 residency, and the KV_MEM pin

State: Open