[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"$f2g3safznunmw1":3},{"slug":4,"category":5,"publishedAt":6,"titleZh":7,"titleEn":8,"summaryZh":9,"summaryEn":10,"models":11,"hardwares":15,"bodyZh":16,"bodyEn":17,"sourceName":18,"sourceUrl":19,"updatedAt":20},"four-5080-qengine-p2p-host-staging-report-2026-11-06","RUNTIME","2026-11-05T23:04:09.828Z","四张 RTX 5080 的新自报：关闭 P2P，回退的未必是 copy engine","New four-RTX-5080 report: disabling P2P may fall back to host staging, not the copy engine","club-3090 #1952 在窗口内发布同机 P2P 开关对照。作者将差异归于主机暂存传输，并明确未测试强制 copy engine。报告是单机社区自报，不是通用显卡排名。","club-3090 #1952 reports an on\u002Foff P2P comparison on one machine. The author identifies host-staged transfers and explicitly did not test forced copy-engine transport. This is one community report, not a universal GPU ranking.",[12],{"id":13,"name":14},11,"Qwen3.8-Flash-Next",[],"11 月 5 日 14:29 UTC 发布的 club-3090 #1952，给出 QEngine 1.3.4.3 在四张 RTX 5080 上的 P2P 开关对照。它的新价值不是又多一个吞吐数字，而是把“关掉 P2P 后到底走哪条路”写进日志证据。这是单份社区自报，AraoAI 未独立复现；原文未给测量的精确起止时间，因此本次只报道新公开报告，不把表格冒充窗口内新测量入库。\n\n1. 条件必须和结果一起看。报告使用 Ubuntu 26.04.1、四张 16GB 卡，经同一 PCIe Gen5 x8 交换芯片连接；模型为 Qwen3.8-Flash-Next INT4，固定 TP、六个席位、主机内存 KV 池及关闭思考。作者交错执行两次开、两次关，共四次启动。这个配置不是引擎四卡默认的流水线并行，席位设置也不能直接等同于测试时六请求同时运行。\n\n2. 下降与提升不是同一分母。按原文两次启动均值，关闭 P2P 相对开启时，10K 预填充下降 32.4%，叙事生成下降 14.7%。这些只描述该机、该提示词与测试阶段；不能反写成开启后就提升同样百分比，更不能把 10K 压力测试的预填充与叙事生成拼成一条同场景记录。\n\n3. 关键证据是传输路径。原文日志称，关闭后集合通信经 pinned host memory 暂存，不是自动切到 copy engine。作者明确没有测试 FLASH_COMM=ce，因此不能得出 P2P 优于所有 copy-engine 实现。90K 预填充每次启动仅一次测量，也不应当成稳定长期均值。\n\n排错先看实际协商链路、拓扑及传输日志，再改开关；不要只凭一个速度反推通信方式。QEngine 当前许可证为 PolyForm Noncommercial 1.0.0，不能默认用于商业部署。该报告能帮助设计对照实验，不能替代另一台机器的验证，更不能据此直接下单四卡主机。\n\n来源：[社区报告 #1952](https:\u002F\u002Fgithub.com\u002Fnoonghunna\u002Fclub-3090\u002Fissues\u002F1952)；[引擎许可证](https:\u002F\u002Fgithub.com\u002Fbpx03\u002FQEngine\u002Fblob\u002Fmain\u002FLICENSE)。","club-3090 #1952, published at 14:29 UTC on November 5, presents a P2P on\u002Foff comparison using QEngine 1.3.4.3 and four RTX 5080 GPUs. Its useful contribution is the logged fallback transport, not merely another throughput result. This is one community self-report, not independently reproduced by AraoAI. Exact measurement start and end times are absent, so we report the newly published evidence without importing it as measurements made inside this daily window.\n\n1. Conditions matter. The machine runs Ubuntu 26.04.1 with four 16GB cards behind a PCIe Gen5 x8 switch. Qwen3.8-Flash-Next INT4 uses fixed tensor parallelism, six configured seats, host-RAM KV and thinking disabled. The author interleaves two enabled and two disabled boots. This differs from the engine default pipeline mode for four cards. Six configured seats do not establish six simultaneous benchmark requests.\n\n2. A decline and an increase use different denominators. The reported two-boot means show that disabling P2P reduced 10K-prompt prefill by 32.4% and narrative decode by 14.7% relative to enabled P2P. These figures apply to this rig and these phases. They are not an equal-percentage gain in the reverse direction, and narrative decode must not be paired with a separate 10K stress-test prefill to manufacture one workload record.\n\n3. Transport logs are decisive. The disabled arm stages collectives through pinned host memory, rather than automatically switching to the copy engine. FLASH_COMM=ce was not tested, so the report does not establish that P2P beats every copy-engine implementation. The 90K prefill leg has only one measurement per boot.\n\nCheck negotiated links, topology and actual transport logs before changing switches. QEngine currently uses PolyForm Noncommercial 1.0.0; commercial deployment is not automatically permitted. This report helps design an experiment, but cannot validate another machine or justify a four-GPU purchase by itself.\n\nSources: https:\u002F\u002Fgithub.com\u002Fnoonghunna\u002Fclub-3090\u002Fissues\u002F1952 ; https:\u002F\u002Fgithub.com\u002Fbpx03\u002FQEngine\u002Fblob\u002Fmain\u002FLICENSE","club-3090 社区自报 \u002F QEngine LICENSE","https:\u002F\u002Fgithub.com\u002Fnoonghunna\u002Fclub-3090\u002Fissues\u002F1952","2026-11-05T23:04:09.829Z"]