Repository navigation
hip: accept gfx1010 and gfx1011 (RDNA1) in the arch gate - #1142
Closed
AcckiyGerman wants to merge 1 commit into
Closed
AcckiyGerman wants to merge 1 commit into
AcckiyGerman wants to merge 1 commit into
Conversation
gfx1010 (Navi 10) is measured on an RX 5700 XT: 8 GB card, Coder IQ1_M, 32K context, prompt 115 / 144 / 149 tok/s and decode 18.0 / 20.3 / 23.3 tok/s at 4K / 16K / 30K prompt tokens, 6 of 6 needle checks at 8K and 30K, 61 of 61 ctest on the card. Nothing beyond the arch list is needed: the portable scalar dp4a that main falls to measured 162.7 / 24.3 against 158.3 / 24.0 with the __gfx1012__ SDWA control (Niko1221#442) - inside the run-to-run spread - and 16 of 16 HIP tests pass on it. gfx1011 (Navi 12) is untested on hardware: same RDNA1 ISA, builds with 0 errors on ROCm 7.14's gfx101X-dgpu wheels. Two box facts, for whoever picks this up: hipBLASLt 1.4.0 refuses all 26 of the engine's GEMM shapes on gfx1010 (heuristic_count = 0), so there is no tuning table and prompts run on plain hipBLAS; and in a PC that also has an AMD iGPU, ROCm's HSA runtime segfaults in GpuAgent::InitDma while enumerating the pair - below HIP, so HIP_VISIBLE_DEVICES cannot reach it - and the engine must start with ROCR_VISIBLE_DEVICES naming the one KFD node (rocminfo for the index). Co-Authored-By: Qwen3.8-Flash-Next <noreply@qwen.ai>
Owner
|
Thank you. The change in this PR is already in 0.1.40.2 as a port of the original PR with authorship kept, so this re-file is closed as a duplicate of that port. |
2 of 3 tasks
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Reopens #996 on the new
main(it was closed by the history cleanup, as your comment described). One commit, cherry-picked ontoupstream/main; the only conflict was the arch list line, which now keepsgfx1034andgfx1151alongside the new entries.Adds
gfx1010andgfx1011to_strata_hip_unvalidated. No source changes are needed.gfx1010 (Navi 10) — tested. RX 5700 XT 8 GB, Ryzen 5 7600X, 30 GB RAM, Ubuntu 26.04, ROCm 7.14.0a20260612 (
gfx101X-dgpuwheels), Coder IQ1_M, 32,768-token context,--mmap-experts --vram-reserve-mib 1024 --spec 4. Official community benchmark script, greedy, 256 output tokens, 3 runs, medians:61 of 61 ctest pass on the card, 6 of 6
tools/needle_bench.pychecks at 8k and 30k. The card holds 607 of the model's 12,288 experts (profile-prefilled), decode cache hit 31%.gfx1011 (Navi 12, Radeon PRO V520 / Pro 5600M) — untested on hardware. Same RDNA1 ISA, builds with 0 errors on the same wheels.
Two facts the arch list can't carry: hipBLASLt 1.4.0 refuses all 26 prefill GEMM shapes on gfx1010 (
heuristic_count = 0), so there is no tuning table; and with an AMD iGPU in the boxhsa_initsegfaults enumerating the pair - below HIP, soHIP_VISIBLE_DEVICEScannot reach it - and the engine needsROCR_VISIBLE_DEVICESnaming one KFD node. The #442 SDWAdp4aturned out not to be needed: portable 162.7/24.3, SDWA 158.3/24.0, 16/16 HIP tests on the portable build.The full report directory is ready if you want it under
bench/results/.