Distributed Intel Arc Pro B70s matched single GPU!
This is a dev post classified by Jev as AI dev tools (a tool drop), kept by the Dev Radar because it carries real work, not commentary.
Distributed Intel Arc Pro B70s matched single GPU! Someone tested Cascadia 0.2.4 with Qwen3.8-27B INT4 across 🎮 2× Intel Arc Pro B70 32GB. Is this for real? You have to see the results 👇 On 1 B70 → ⚡ 23.78 tps Split 32/32 layers across 2 x B70s → ⚡ 24.02 tps That’s 101% of the single-GPU decode rate. The interesting part is that the usual multi-GPU penalty disappeared. Cascadia measured only 🔗 0.249 ms/frame relay overhead For comparison, the author previously measured ordinary llama.cpp RPC on the same two cards at ... 🎮 1× B70: 76.1 tps 🎮 2× B70 split: 40.8 tps 🎯 Capacity withou
Posted by David Hendrickson (11.2k followers) 1 h ago · 0 likes · 248 views · view the original post on X. Kept by the Dev Radar as AI dev tools.
More dev work like this
- I thought they said they are on a break (slow down） — @u1tra_instinct
- BREAKING: SpaceXAI just released a major Grok Build update, with Grok 4.7 now available… — @cb_doge
- roughly similar on AI R&D but a beast on agentic coding task, very impressive — @eliebakouch
- Did you quietly remove the Fable from the subscription along with the release, or why… — @tarasshyn
- New tutorial | Package segmentation with Ultralytics YOLO26 📦 — @ultralytics
- anthropic just mogged everyone with opus 5.5 — @notjazii
- Introducing https://theopenfrontier.com! — @nutlope
- @rafalwilinski — @rafalwilinski
Every post is read and classified by Jev (TypeSafe): what it is, which market it belongs to, and whether the link is a real tool. 21.2k posts from 5k X accounts over the last 21 days, 2.4k tools, 12 markets. Collected every 5 minutes, fully re-ranked every hour — last update 2026-09-22 17:37 UTC. Full method.