fix nix build: Add spirv-headers to vulkanBuildInputs - #85
Merged
TheTom merged 1 commit intoApr 20, 2026
Merged
Conversation
ref: TheTom#81 spirv-headers were introduced by TheTom@1f30ac0 but not added to the nix build environment
cguentherTUChemnitz
force-pushed
the
patch-1
branch
from
April 17, 2026 14:29
1900b0d to
f9ee92b
Compare
Author
|
in my eyes this can already be merged, since it fixes the problem discovered at: #81 Nevertheless is additional broken at a later stage. It seems that there is still some rte.glsl import problem of not finding the file. It seems this branch expectes a certain shader that is just not present / generated. |
Owner
|
Verified build locally — no issues. Thanks @cguentherTUChemnitz! |
fukuro-kun
pushed a commit
to fukuro-kun/fukuro-llama-cpp-turboquant
that referenced
this pull request
Jul 5, 2026
fix nix build: Add spirv-headers to vulkanBuildInputs
fukuro-kun
pushed a commit
to fukuro-kun/fukuro-llama-cpp-turboquant
that referenced
this pull request
Jul 19, 2026
Deskriptoren werden direkt in den Command Buffer geschrieben statt Deskriptor-Sets zu allokieren und zu binden. Eliminiert: - vkAllocateDescriptorSets (Pool-Management) - vkUpdateDescriptorSets (pro Dispatch) - vkCmdBindDescriptorSets (pro Dispatch) Implementierung: - Extension-Check in Device-Initialisierung (GGML_VK_DISABLE_PUSH_DESCRIPTOR zum Deaktivieren) - Push-Descriptor-Layout (dsl_push) mit PUSH_DESCRIPTOR_BIT - Pipeline-Layouts verwenden dsl_push wenn supported - Dispatch: pushDescriptorSetKHR statt updateDescriptorSets + bindDescriptorSets - Descriptor-Set-Allokation übersprungen wenn push_descriptor_supported - Device-Log erweitert: 'push_desc: 1/0' MAX_PARAMETER_COUNT=12 < 32 (maxPushDescriptors Minimum) — innerhalb des Limits. RADV unterstützt VK_KHR_push_descriptor seit Mesa 17.1 (2017).
fukuro-kun
pushed a commit
to fukuro-kun/fukuro-llama-cpp-turboquant
that referenced
this pull request
Jul 19, 2026
fukuro-kun
pushed a commit
to fukuro-kun/fukuro-llama-cpp-turboquant
that referenced
this pull request
Jul 19, 2026
…xPushDescriptors
fukuro-kun
pushed a commit
to fukuro-kun/fukuro-llama-cpp-turboquant
that referenced
this pull request
Jul 19, 2026
…auf RADV Benchmark Mars (1B Q4_K_M): pp512 ±0.1%, tg128 ±0.2%, pp4096 ±0.3%. RADV descriptor set allocation ist bereits sehr effizient (pre-allocated pools). Implementierung korrekt (push_desc: 1 bestätigt) aber kein messbarer Speedup.
fukuro-kun
pushed a commit
to fukuro-kun/fukuro-llama-cpp-turboquant
that referenced
this pull request
Jul 19, 2026
fukuro-kun
pushed a commit
to fukuro-kun/fukuro-llama-cpp-turboquant
that referenced
this pull request
Jul 19, 2026
…ors check P1: VK_KHR_push_descriptor wurde in device_extensions eingefügt bevor maxPushDescriptors geprüft wurde — bei zu geringem Limit wurde die Extension aktiviert aber nicht genutzt. Jetzt: push_back erst nach erfolgreichem Check. P2: max_push_descriptors Feld entfernt (dead code — geschrieben aber nie gelesen).
fukuro-kun
pushed a commit
to fukuro-kun/fukuro-llama-cpp-turboquant
that referenced
this pull request
Jul 20, 2026
…hed ⏭️ + TheTom#76 CPU Fusion ⏭️ 3 parallele Tiefen-Evals für Tier-3 Items: AtomicBot-ai#74 Vulkan Descriptor Indexing (Bindless) ❌ VERWORFEN: - Redundant mit TheTom#85 Push Descriptors (✅ implementiert 2026-07-14) - Push Descriptors eliminieren dieselben CPU-Aufrufe - TheTom#85-Benchmark auf Mars RADV: ±0.1-0.3% (Rauschen) - Bindless würde über Push-Descriptors hinaus <0.5% bringen - Workload-Mismatch: Bindless für draw-heavy Rendering, nicht Compute - Mars/Venus bandwidth-bound, nicht descriptor-bound - Aufwand revidiert: 2-4 → 3-5 Wochen (Shader-Rewrite aller .comp-Files) AtomicBot-ai#75 Non-blocking Pipeline Scheduling ⏭️ SPÄTER: - PR ggml-org#19922 closed (2026-03-03, unmerged, 4+ Mo stale) - Fork hat bereits Upstream-Pipeline-Parallelismus - Konflikt mit TheTom#79 TP (✅+23-32% tg, split-mode-exklusiv) - NVIDIA ungetestet, PP-Regression auf 4x MI50 gemeldet - 2-GPU-Setup → geringer Bubble-Hebel - Aufwand revidiert: 3-4 → 4-6 Wochen TheTom#76 CPU Backend Operator Fusion ⏭️ SPÄTER: - RMS_NORM+MUL Fusion bereits im Fork (PR ggml-org#22423 upstream-merged) - MoE Gated FFN riskant: PR ggml-org#20596 zeigt Regressionen auf Consumer-CPUs (M2: 0.98-1.00x, qwen3moe 30B: 0.75-0.96x bei t=2-4) - Nur auf 96-Core-EPYC konsistente Gains (1.04-1.08x) - Styx/Uranus haben Consumer-CPUs → wahrscheinlich Regression - Re-Eval wenn PR ggml-org#20596 gemerged mit Regression-Freiheit
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.
ref: #81
spirv-headers were introduced by 1f30ac0
but not added to the nix build environment
closes: #81