Blog
Notes from the bench
Longer pieces live here. Succinct ROCm/vLLM version updates are also listed so you can scan dates without opening the full changelog.
ROCm version notes
Oct 12, 2026 · snapshot
Consumer stacks — Qwen3.6 daily driver, Flash agent, calculatorNorth star is now the two stacks from our test rig. llama.cpp Vulkan daily driver; Strata for Flash; /calculator and /stacks are first-class pages.
Oct 12, 2026 · snapshot
ROCm 10.1.0 notes, CLI v0.2.0, error paste, version matrix10.1.0 lists R9700 as gfx1201 and Radeon Ubuntu 26.04.1 / 24.04.5 / 22.04.5. Published ROCm CLI is v0.2.0, not a 1.0.0 tag. New pages for search, errors, backends, quants, RSS.
Sep 22, 2026
vLLM ROCm Docker try-tag v0.30.0Dated snapshot. AMD-validated vLLM 0.27 / ROCm 10.0.0 unchanged. Community image moved to vllm/vllm-openai-rocm:v0.30.0. Named Qwen defaults from that week were later withdrawn.
Sep 21, 2026
Weekly refresh — vLLM 0.27 reliability (dated)Dated snapshot. ROCm 10.0.0 unchanged for gfx1201. Reliability serve is AMD-aligned vLLM 0.27. vLLM 0.29 ROCm Docker remains the community cutting-edge pin (no newer stable that week). Named Qwen defaults from that week were later withdrawn. R9700 training KIs documented.
Sep 9, 2026
vLLM 0.29.0 ROCm extrasUpstream vLLM 0.29 ships ROCm wheels and Docker. Extra index like wheels.vllm.ai/rocm/0.29.0/rocm723; image vllm/vllm-openai-rocm:v0.29.0. Plain pip install vllm is CUDA. AMD 10.0 docs still documented 0.27.
Sep 2, 2026
ROCm 7.14.1 patch line2 Sep 2026 patch on the 7.14 TheRock line. Still a valid pin if your wheel/image was built against it. Not a reason to install Ubuntu archive 7.1.
Aug 26, 2026
ROCm Core SDK 10.0.0Production TheRock line. R9700 / gfx1201 first-class. amdrocm-* packages, one gfx family, scoped /opt/rocm prefix. AMD’s notes pair PyTorch 2.13 and vLLM 0.27 — that is the supported combo, not a forever pin.
Longer posts
Sep 20, 2026 · process · rocm
The weekly card is not scripture
Why this site tracks ROCm and vLLM as dated snapshots, and how to refresh the YAML without redesigning pages.
Sep 18, 2026 · llama.cpp · vulkan · rocm
Vulkan is an on-ramp, not a consolation prize
Why llama.cpp on RADV still belongs in a ROCm-oriented home lab, especially on RDNA 4.
Sep 15, 2026 · rocm · changelog
ROCm versions are a matching problem
A changelog-style note on why “install the latest of everything” is the least reliable way to get gfx1201 working.
Sep 12, 2026 · naming · hardware
The name is Radeon AI PRO R9700
Why forums say “R9700 Pro,” why that is not an RX 9700, and which string to put in a bug report.