feat: add calibrate() to OutlierTurboQuant for data-driven channel split#64
Open
brosequist wants to merge 1 commit intoTheTom:mainfrom
Open
feat: add calibrate() to OutlierTurboQuant for data-driven channel split#64brosequist wants to merge 1 commit intoTheTom:mainfrom
brosequist wants to merge 1 commit intoTheTom:mainfrom
Conversation
Previously the outlier/inlier channel split was set at construction time and never adjusted. calibrate(calibration_vectors) now computes per-channel RMS, flags channels whose RMS exceeds 3× the median as outliers, and updates the split on the compressor — matching the dynamic-threshold approach described in the LLM.int8() and SmoothQuant literature. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
clonekang
pushed a commit
to clonekang/turboquant_llama
that referenced
this pull request
May 3, 2026
Origin's April upstream-sync rebase interleaved two changes that left the Vulkan turbo3 KV path broken: * ggml-org/llama.cpp upstream PR #21572 (1f30ac0) moved fp16 RTE rounding to a runtime SPIR-V patch and dropped the _rte shader variants plus rte.glsl itself. * TheTom/llama-cpp-turboquant PR TheTom#62 (ff8bb73) added turbo3 KV support against a base that still had those variants. After the rebase, the tree had dangling cpy_f32_*_rte_len / _data references, a two-arg SET_ROWS macro called with one arg, a #include "rte.glsl" in a shader whose header no longer exists, and MMQ shader variants generated for turbo3_0 even though the flash_attn MMQ path has no turbo3 code. The result was that ggml-vulkan.cpp failed to compile on a clean checkout (spirv-headers + all of the above) and the shader-gen emitted garbage variants. Separately, turbo3 flash-attn pipelines were only wired up for FA_SCALAR. On a coopmat-capable device (e.g. RADV on a 7900 XTX) the tuning heuristic picks FA_COOPMAT1 for most shapes, which landed in ggml_vk_flash_attn with an uninitialized pipeline (wg_denoms={0,0,0}) and tripped the Br == wg_denoms[0] assertion as soon as a prefill ubatch was dispatched. End-to-end llama-cli on Vulkan + -ctk turbo3 aborted on the first real forward pass. Changes: * Drop the if (float_controls_rte_fp16) / else branches around cpy_f32_quant pipeline creation and collapse SET_ROWS to a single variant, matching upstream post-1f30ac0ce. * Remove the #include "rte.glsl" from copy_to_quant.comp. * Skip the MMQ flash_attn shader variant for turbo3_0 in the shader generator (no MMQ code path for it). * Register CREATE_FA(GGML_TYPE_TURBO3_0, turbo3_0, FA_COOPMAT1, _cm1) and the _cm2 counterpart alongside the other quant types. Verified on AMD 7900 XTX (gfx1100 / RADV NAVI31, ROCm 7.2.1 + Vulkan 1.4.341, spirv-headers 1.4.341.0): * Full HIP+Vulkan build is clean with no shader compile errors. * test-backend-ops -o SET_ROWS -b Vulkan0 : 147/147 * test-backend-ops -o FLASH_ATTN_EXT -b Vulkan0 -p type_KV=turbo3 : 530 cases pass (previously aborted on case 3). * test-backend-ops -o FLASH_ATTN_EXT -b ROCm0 -p type_KV=turbo3 : still green (no HIP regression). * llama-cli on Qwen3-8B Q4_K_M with -ngl 99 -fa on -ctk turbo3 -ctv turbo3 on Vulkan0 no longer aborts. The remaining head_dim=128 correctness issue on the Vulkan turbo3 decode path is pre-existing and orthogonal to this change. llama-bench on Qwen3.5-27B Q4_K_M, 7900 XTX OC, HIP backend: F16 tg128=20.98 turbo3 tg128=20.13 turbo4 tg128=20.17 Refs: TheTom/llama-cpp-turboquant issues TheTom#50, TheTom#64, TheTom#81
clonekang
pushed a commit
to clonekang/turboquant_llama
that referenced
this pull request
May 3, 2026
Origin's April upstream-sync rebase interleaved two changes that left the Vulkan turbo3 KV path broken: * ggml-org/llama.cpp upstream PR #21572 (1f30ac0) moved fp16 RTE rounding to a runtime SPIR-V patch and dropped the _rte shader variants plus rte.glsl itself. * TheTom/llama-cpp-turboquant PR TheTom#62 (ff8bb73) added turbo3 KV support against a base that still had those variants. After the rebase, the tree had dangling cpy_f32_*_rte_len / _data references, a two-arg SET_ROWS macro called with one arg, a #include "rte.glsl" in a shader whose header no longer exists, and MMQ shader variants generated for turbo3_0 even though the flash_attn MMQ path has no turbo3 code. The result was that ggml-vulkan.cpp failed to compile on a clean checkout (spirv-headers + all of the above) and the shader-gen emitted garbage variants. Separately, turbo3 flash-attn pipelines were only wired up for FA_SCALAR. On a coopmat-capable device (e.g. RADV on a 7900 XTX) the tuning heuristic picks FA_COOPMAT1 for most shapes, which landed in ggml_vk_flash_attn with an uninitialized pipeline (wg_denoms={0,0,0}) and tripped the Br == wg_denoms[0] assertion as soon as a prefill ubatch was dispatched. End-to-end llama-cli on Vulkan + -ctk turbo3 aborted on the first real forward pass. Changes: * Drop the if (float_controls_rte_fp16) / else branches around cpy_f32_quant pipeline creation and collapse SET_ROWS to a single variant, matching upstream post-1f30ac0ce. * Remove the #include "rte.glsl" from copy_to_quant.comp. * Skip the MMQ flash_attn shader variant for turbo3_0 in the shader generator (no MMQ code path for it). * Register CREATE_FA(GGML_TYPE_TURBO3_0, turbo3_0, FA_COOPMAT1, _cm1) and the _cm2 counterpart alongside the other quant types. Verified on AMD 7900 XTX (gfx1100 / RADV NAVI31, ROCm 7.2.1 + Vulkan 1.4.341, spirv-headers 1.4.341.0): * Full HIP+Vulkan build is clean with no shader compile errors. * test-backend-ops -o SET_ROWS -b Vulkan0 : 147/147 * test-backend-ops -o FLASH_ATTN_EXT -b Vulkan0 -p type_KV=turbo3 : 530 cases pass (previously aborted on case 3). * test-backend-ops -o FLASH_ATTN_EXT -b ROCm0 -p type_KV=turbo3 : still green (no HIP regression). * llama-cli on Qwen3-8B Q4_K_M with -ngl 99 -fa on -ctk turbo3 -ctv turbo3 on Vulkan0 no longer aborts. The remaining head_dim=128 correctness issue on the Vulkan turbo3 decode path is pre-existing and orthogonal to this change. llama-bench on Qwen3.5-27B Q4_K_M, 7900 XTX OC, HIP backend: F16 tg128=20.98 turbo3 tg128=20.13 turbo4 tg128=20.17 Refs: TheTom/llama-cpp-turboquant issues TheTom#50, TheTom#64, TheTom#81
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
1 participant
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.
Summary
OutlierTurboQuant.calibrate(calibration_vectors)for data-driven outlier channel detection.Test plan
pytest tests/test_outlier.py— includes tests for calibrate() updating the channel mask and handling all-inlier / all-outlier edge cases.🤖 Generated with Claude Code