b10534

ggml-org/llama.cppb10534Aug 21, 2026by github-actions[bot]

AI Summary

Optimizes CUDA decoding performance by tuning the crossover point between mul_mat_vec_q and MMQ batch kernels, adding hardware-specific switch points to handle different GPUs and quantization types.

Key Highlights

CUDA runtime override (GGML_CUDA_MMVQ_MAX) to tune the mvq->MMQ decode crossover
Hardware-specific switch points for Blackwell, DGX Spark, and Ada GPUs
Performance improvement of +23-41% at B=8 for Q4_K dense models
Clamping and validation for runtime override values

New Features

CUDA performance optimization for decode paths

Full Release Notes

<details open>

CUDA: adding switch points per HW and quant type to tune the mvq->MMQ decode crossover (#26079)

* CUDA: runtime GGML_CUDA_MMVQ_MAX to tune the mvq->MMQ decode crossover

Add a runtime override of the mul_mat_vec_q -> MMQ batch crossover
(default MMVQ_MAX_BATCH_SIZE). Lowering it routes batches above the
threshold from the CUDA-core vector kernel to the int8 MMQ tensor-core
path, which is faster once quantized decode becomes compute-bound at
B>1 (measured +23-41% at B=8 on RTX 5090 for Q4_K dense, no low-batch loss).

The value is parsed once and clamped to [1, MMVQ_MAX_BATCH_SIZE], since
mul_mat_vec_q asserts ncols_dst <= that; invalid input warns and falls
back to the default. The override is applied consistently in both the
mul_mat_vec_q and MUL_MAT_ID dispatch paths. Default behavior unchanged.

* Added Blackwell specific switch point, to reduce dependence on runtime env var.

* Add per-HW switch point values for DGX Spark and removing runtime env var

* Adding switch points for Ada, tested on RTX 4090

* Modifying DGX Spark numbers based on latest run and adding some comments and small functional changes relating to MoE

* Reverting an unnecessary conditional

* Update ggml/src/ggml-cuda/mmvq.cu

---------

Co-authored-by: praneshgo <227579474+praneshgo@users.noreply.github.com>
Co-authored-by: Oliver Simons <osimons@nvidia.com>

</details>

**Website:**
- <https://llama.app>

**Attestations:**
- <https://github.com/ggml-org/llama.cpp/attestations/42022971>

**macOS/iOS:**
- [macOS Apple Silicon (arm64)](https://github.com/ggml-org/llama.cpp/releases/download/b10534/llama-b10534-bin-macos-arm64.tar.gz)
- macOS Apple Silicon (arm64, KleidiAI enabled) [DISABLED](https://github.com/ggml-org/llama.cpp/pull/23780)
- [macOS Intel (x64)](https://github.com/ggml-org/llama.cpp/releases/download/b10534/llama-b10534-bin-macos-x64.tar.gz)
- [iOS XCFramework](https://github.com/ggml-org/llama.cpp/releases/download/b10534/llama-b10534-xcframework.zip)

**Linux:**
- [Ubuntu x64 (CPU)](https://github.com/ggml-org/llama.cpp/releases/download/b10534/llama-b10534-bin-ubuntu-x64.tar.gz)
- [Ubuntu arm64 (CPU)](https://github.com/ggml-org/llama.cpp/releases/download/b10534/llama-b10534-bin-ubuntu-arm64.tar.gz)
- [Ubuntu s390x (CPU)](https://github.com/ggml-org/llama.cpp/releases/download/b10534/llama-b10534-bin-ubuntu-s390x.tar.gz)
- [Ubuntu x64 (Vulkan)](https://github.com/ggml-org/llama.cpp/releases/download/b10534/llama-b10534-bin-ubuntu-vulkan-x64.tar.gz)
- [Ubuntu arm64 (Vulkan)](https://github.com/ggml-org/llama.cpp/releases/download/b10534/llama-b10534-bin-ubuntu-vulkan-arm64.tar.gz)
- Ubuntu x64 (ROCm 7.14)[DISABLED](https://github.com/ggml-org/llama.cpp/pull/26969)
- [Ubuntu x64 (OpenVINO)](https://github.com/ggml-org/llama.cpp/releases/download/b10534/llama-b10534-bin-ubuntu-openvino-2026.3-x64.tar.gz)
- [Ubuntu x64 (SYCL FP32)](https://github.com/ggml-org/llama.cpp/releases/download/b10534/llama-b10534-bin-ubuntu-sycl-fp32-x64.tar.gz)
- [Ubuntu x64 (SYCL FP16)](https://github.com/ggml-org/llama.cpp/releases/download/b10534/llama-b10534-bin-ubuntu-sycl-fp16-x64.tar.gz)

**Android:**
- [Android arm64 (CPU)](https://github.com/ggml-org/llama.cpp/releases/download/b10534/llama-b10534-bin-android-arm64.tar.gz)

**Windows:**
- [Windows x64 (CPU)](https://github.com/ggml-org/llama.cpp/releases/download/b10534/llama-b10534-bin-win-cpu-x64.zip)
- [Windows arm64 (CPU)](https://github.com/ggml-org/llama.cpp/releases/download/b10534/llama-b10534-bin-win-cpu-arm64.zip)
- [Windows arm64 (OpenCL Adreno)](https://github.com/ggml-org/llama.cpp/releases/download/b10534/llama-b10534-bin-win-opencl-adreno-arm64.zip)
- [Windows x64 (CUDA 12)](https://github.com/ggml-org/llama.cpp/releases/download/b10534/llama-b10534-bin-win-cuda-12.4-x64.zip) - [CUDA 12.4 DLLs](https://github.com/ggml-org/llama.cpp/releases/download/b10534/cudart-llama-bin-win-cuda-12.4-x64.zip)
- [Windows x64 (CUDA 13)](https://github.com/ggml-org/llama.cpp/releases/download/b10534/llama-b10534-bin-win-cuda-13.3-x64.zip) - [CUDA 13.3 DLLs](https://github.com/ggml-org/llama.cpp/releases/download/b10534/cudart-llama-bin-win-cuda-13.3-x64.zip)
- [Windows arm64 (CUDA 13) (preview)](https://github.com/ggml-org/llama.cpp/releases/download/b10534/llama-b10534-bin-win-cuda-13.4-arm64.zip) - [CUDA 13.4 DLLs](https://github.com/ggml-org/llama.cpp/releases/download/b10534/cudart-llama-bin-win-cuda-13.4-arm64.zip)
- [Windows x64 (Vulkan)](https://github.com/ggml-org/llama.cpp/releases/download/b10534/llama-b10534-bin-win-vulkan-x64.zip)
- [Windows x64 (OpenVINO)](https://github.com/ggml-org/llama.cpp/releases/download/b10534/llama-b10534-bin-win-openvino-2026.3-x64.zip)
- [Windows x64 (SYCL)](https://github.com/ggml-org/llama.cpp/releases/download/b10534/llama-b10534-bin-win-sycl-x64.zip)
- [Windows x64 (ROCm 7.14)](https://github.com/ggml-org/llama.cpp/releases/download/b10534/llama-b10534-bin-win-rocm-7.14-x64.zip)

**openEuler:**
- [DISABLED](https://github.com/ggml-org/llama.cpp/pull/23705)
- openEuler x86 (310p)
- openEuler x86 (910b, ACL Graph)
- openEuler aarch64 (310p)
- openEuler aarch64 (910b, ACL Graph)

**UI:**
- [UI](https://github.com/ggml-org/llama.cpp/releases/download/b10534/llama-b10534-ui.tar.gz)