Skip to content

Navigation Menu

Appearance settings

GitHub Copilot
Write better code with AI

GitHub Spark New
Build and deploy intelligent apps

GitHub Models New
Manage and compare prompts

GitHub Advanced Security
Find and fix vulnerabilities

Actions
Automate any workflow
Codespaces
Instant dev environments

Issues
Plan and track work

Code Review
Manage code changes

Discussions
Collaborate outside of code

Code Search
Find more, search less
Explore

Why GitHub

Documentation

GitHub Skills

Blog
Integrations

GitHub Marketplace

MCP Registry
View all features
By company size

Enterprises

Small and medium teams

Startups

Nonprofits
By use case

App Modernization

DevSecOps

DevOps

CI/CD

View all use cases
By industry

Healthcare

Financial services

Manufacturing

Government

View all industries
View all solutions
Topics

AI

DevOps

Security

Software Development

View all
Explore

Learning Pathways

Events & Webinars

Ebooks & Whitepapers

Customer Stories

Partners

Executive Insights
GitHub Sponsors
Fund open source developers
The ReadME Project
GitHub community articles
Repositories

Topics

Trending

Collections
Enterprise platform
AI-powered developer platform
Available add-ons

GitHub Advanced Security
Enterprise-grade security features

Copilot for business
Enterprise-grade AI features

Premium Support
Enterprise-grade 24/7 support
Pricing

Search code, repositories, users, issues, pull requests...

Search

Clear

Search syntax tips

Provide feedback

We read every piece of feedback, and take your input very seriously.

Include my email address so I can be contacted

Saved searches

Use saved searches to filter your results more quickly

Name

Query

To see all available qualifiers, see our documentation.

Appearance settings

You signed in with another tab or window. Reload to refresh your session. You signed out in another tab or window. Reload to refresh your session. You switched accounts on another tab or window. Reload to refresh your session.

Dismiss alert

l3utterfly / llama.cpp Public

forked from ggml-org/llama.cpp

Fork 0
Star 0

Code
Pull requests
Actions
Projects
Security
Insights

Additional navigation options

Code
Pull requests
Actions
Projects
Security
Insights

merge from upstream#72

Merged

l3utterfly merged 180 commits intol3utterfly:ggml-hexagonl3utterfly/llama.cpp:ggml-hexagonfrom

jeffzhou2000:self-buildjeffzhou2000/ggml-hexagon:self-buildCopy head branch name to clipboard

Jun 27, 2025

Conversation Commits180 (180)Checks Files changed

Merged

merge from upstream#72
l3utterfly merged 180 commits intol3utterfly:ggml-hexagonfrom
jeffzhou2000:self-build

Commits

Commits on Jun 27, 2025

ggml-qnn: add Qualcomm QNN backend for GGML
jeffzhou2000
committed
ggml-qnn: santiy check
jeffzhou2000
committed
ggml-qnn: update script build-run-android.sh to compare peformance of ggml-qnn
jeffzhou2000
committed
ggml-qnn: fix minor issue in test-backend-ops.cpp
jeffzhou2000
committed
ggml-qnn: merge QNN RPC feature from https://github.com/zhouwg/kantv/blob/ggml-qnn-quantize/core/ggml/llamacpp/ggml-qnn.cpp
jeffzhou2000
committed
ggml-qnn: sync from branch kantvai-ggmlqnn-npurpc
jeffzhou2000
committed
ggml-qnn: a concise approach to offload mulmat to QNN backend(sync from branch kantvai-ggmlqnn-npurpc, https://github.com/kantv-ai/llama.cpp/wiki/offloading-mulmat-to-QNN-backend)
jeffzhou2000
committed
ggml-qnn: remove redundant codes
jeffzhou2000
committed
ggml-qnn: sync from branch kantvai-ggmlqnn-npurpc
jeffzhou2000
committed
ggml-qnn: sync from branch kantvai-ggmlqnn-npurpc
jeffzhou2000
committed
ggml-qnn: sync from branch kantvai-ggmlqnn-npurpc
jeffzhou2000
committed
ggml-qnn: add Qualcomm QNN backend for GGML
jeffzhou2000
committed
ggml-qnn: merge QNN RPC feature from https://github.com/zhouwg/kantv/blob/ggml-qnn-quantize/core/ggml/llamacpp/ggml-qnn.cpp
jeffzhou2000
committed
ggml-qnn: sync from branch kantvai-ggmlqnn-npurpc
jeffzhou2000
committed
ggml-qnn: a concise approach to offload mulmat to QNN backend(sync from branch kantvai-ggmlqnn-npurpc, https://github.com/kantv-ai/llama.cpp/wiki/offloading-mulmat-to-QNN-backend)
jeffzhou2000
committed
ggml-qnn: remove redundant codes
jeffzhou2000
committed
ggml-qnn: sync from branch kantvai-ggmlqnn-npurpc
jeffzhou2000
committed
ggml-qnn: sync from branch kantvai-ggmlqnn-npurpc
jeffzhou2000
committed
ggml-qnn: sync from branch kantvai-ggmlqnn-npurpc
jeffzhou2000
committed
ggml-qnn: fix a minior typo in internal doc
jeffzhou2000
committed
ggml-qnn: refine function ggml_qnn_create_general_tensor() to avoid complex/redundant pointer operation
jeffzhou2000
committed
ggml-qnn: fix a minor typo in source code
jeffzhou2000
committed
build: avoid ggml-qnn backend breaking other backend's builds
jeffzhou2000
committed
ggml-qnn: remove redundant codes to make PR reviewers happy
jeffzhou2000
committed
ggml-qnn: refine code format
jeffzhou2000
committed
ggml-qnn: offload quantized type mulmat to QNN backend
jeffzhou2000
committed
ggml-qnn: refine source code structure to make code more clearly
jeffzhou2000
committed
ggml-qnn: enable release build with necessary logs to make reviewers happy
jeffzhou2000
committed
ggml-qnn: enable all quantize type with 2d mulmat
jeffzhou2000
committed
ggml-qnn: enable log output of GGMLQNN_LOG_INFO in command line mode for benchmark more conveniently
jeffzhou2000
committed
ggml-qnn: Windows port --- step2
jeffzhou2000
committed
ggml-qnn: merge UT code and corresponding script from local dev branch to make workflow easily
jeffzhou2000
committed
ggml-qnn: merge ggml_qnn_mul_mat_4d from local dev branch to make workflow easily
jeffzhou2000
committed
ggml-qnn: submit AI-assisted ggml_qnn_mul_mat_4d(not worked currently) which generated by Grok 3
jeffzhou2000
committed
ggml-qnn: AI-assisted ggml_qnn_mul_mat_4d by Grok 3 --- step2
jeffzhou2000
committed
ggml-qnn: AI-assisted ggml_qnn_mul_mat_4d by Grok 3 --- step3
jeffzhou2000
committed
ggml-qnn: AI-assisted ggml_qnn_mul_mat_4d by Grok 3 --- step4
jeffzhou2000
committed
ggml-qnn: AI-assisted ggml_qnn_mul_mat_4d by Grok 3 --- step5
jeffzhou2000
committed
ggml-qnn: AI-assisted ggml_qnn_mul_mat_4d by Grok 3 --- step6
jeffzhou2000
committed
ggml-qnn: AI-assisted ggml_qnn_mul_mat_4d by Grok 3 --- step7
jeffzhou2000
committed
ggml-qnn: AI-assisted ggml_qnn_mul_mat_4d by Grok 3 --- step8
jeffzhou2000
committed
ggml-qnn: AI-assisted ggml_qnn_mul_mat_4d by Grok 3 --- good in step9
jeffzhou2000
committed
ggml-qnn: AI-assisted ggml_qnn_mul_mat_4d by Grok 3 --- narrow down to make AI happy
jeffzhou2000
committed
ggml-qnn: AI-assisted ggml_qnn_mul_mat_4d by Grok 3 --- step10
jeffzhou2000
committed
ggml-qnn: AI-assisted ggml_qnn_mul_mat_4d by Grok 3 --- narrow down to make AI happy
jeffzhou2000
committed
ggml-qnn: AI-assisted ggml_qnn_mul_mat_4d by Grok 3 --- step11
jeffzhou2000
committed
ggml-qnn: AI-assisted ggml_qnn_mul_mat_4d by Grok 3 --- both ok in step12
jeffzhou2000
committed
ggml-qnn: AI-assisted ggml_qnn_mul_mat_4d by Grok 3 ---finalizing version also both ok in step13
jeffzhou2000
committed
ggml-qnn: refine ggml_qnn_mul_mat and ggml_qnn_general_node according to Grok 3's style
jeffzhou2000
committed
ggml-qnn: remove no-needed comments
jeffzhou2000
committed
ggml-qnn: Windows port --- step3
jeffzhou2000
committed
ggml-qnn: remove un-needed function
jeffzhou2000
committed
ggml-qnn:rebase to upstream
jeffzhou2000
committed
ggml-qnn: fix a minior issue during rebase to upstream
jeffzhou2000
committed
ggml-qnn: update script according to https://github.com/ggml-org/llama.cpp/pull/12155
jeffzhou2000
committed
ggml-qnn: fix a minior issue in ggmlqnn_create_general_tensor()
jeffzhou2000
committed
ggml-qnn: active member variable _device_id in class qnn_instance
jeffzhou2000
committed
ggml-qnn: refine ggml_qnn_general_node and ggml_qnn_mul_mat to make code more clearly
jeffzhou2000
committed
ggml-qnn: Windows port --- step4
jeffzhou2000
committed
ggml-qnn: Windows port -- step5
jeffzhou2000
committed
ggml-qnn: WoA(Windows on ARM) -- step6
jeffzhou2000
committed
ggml-qnn: rebase to upstream
jeffzhou2000
committed
ggml-qnn: pr to upstream
jeffzhou2000
committed
ggml-qnn: rebase to upstream
jeffzhou2000
committed
ggml-qnn: self code-review
jeffzhou2000
committed
ggml-qnn: rebase upstream
jeffzhou2000
committed
ggml-qnn: add approach through Hexagon cDSP
jeffzhou2000
committed
ggml-qnn: refine general approach through Hexagon cDSP
jeffzhou2000
committed
ggml-qnn: refine the entire ggml-qnn.cpp to make code more clear
jeffzhou2000
committed
ggml-qnn: refine the entire ggml-qnn.cpp to make code more clear
jeffzhou2000
committed
ggml-qnn: add build script for libggmlop_skel.so
jeffzhou2000
committed
ggml-qnn: remove redundant functions in this PR and make codes more clear
jeffzhou2000
committed
ggml-qnn: original ggml_compute_forward_add and ggml_compute_forward_mul_mat works fine on Hexagon cDSP at the first time
jeffzhou2000
committed
ggml-qnn: modify build-run-android.sh to verify mulmat and validate mulmat performance on cDSP easily
jeffzhou2000
committed
ggml-qnn: make host code(ggml-qnn.cpp) more clear and more stable
jeffzhou2000
committed
ggml-qnn: refine code according to self code-review and make code more clear
jeffzhou2000
committed
ggml-qnn: offload more ggml op to Hexagon cDSP
jeffzhou2000
committed
ggml-hexagon: code on AP(arm-cpu) side is stable now
jeffzhou2000
committed
ggml-hexagon: optimize GGML_OP_ADD on cDSP side
jeffzhou2000
committed
ggml-hexagon: simplify hexagon-kernel build logic in CMakeLists.txt
jeffzhou2000
committed
ggml-hexagon: release ggml-hexagon v0.98
jeffzhou2000
committed
ggml-hexagon: release ggml-hexagon v0.99
jeffzhou2000
committed
ggml-hexagon: try to offload q6_k mulmat to cDSP
jeffzhou2000
committed
ggml-hexagon: fix minior issue in ggml-hexagon.cpp after self code-review
jeffzhou2000
committed
ggml-hexagon: check validation of ggml-hexagon.cfg before create appropriate backend
jeffzhou2000
committed
ggml-hexagon: fix all compiler warnings in ggml-hexagon.cpp
jeffzhou2000
committed
ggml-hexagon: enable only one backend device for HWACCEL_CDSP and enable rpc ion memory pool for HWACCEL_CDSP
jeffzhou2000
committed
ggml-hexagon: rpc ion memory pool and test-backend-ops works fine in HWACCEL_CDSP approach at the first time
jeffzhou2000
committed
ggml-hexagon: make comprision of mulmat performance between HWACCEL_QNN and HWACCEL_CDSP easily
jeffzhou2000
committed
ggml-hexagon: release ggml-hexagon v1.00
jeffzhou2000
committed
ggml-hexagon: rebase to upstream
jeffzhou2000
committed
ggml-hexagon: check configuration of enable_rpc_dma_mempool in function ggmlhexagon_check_valid_appcfg
jeffzhou2000
committed
ggml-hexagon: uniform rpc_ion_memsize and rpc_ion_usage between HWACCEL_CDSP and HWACCEL_QNN
jeffzhou2000
committed
ggml-hexagon: make buffer mechanism more clear in HWACCEL_CDSP approach
jeffzhou2000
committed
ggml-hexagon: add perf function in hexagon kernerls on cDSP side
jeffzhou2000
committed
ggml-hexagon: fix a stupid issue of why set rpc latency failure and improve NPU performance significantly
jeffzhou2000
committed
ggml-hexagon: make helper function ggmlhexagon_get_timestring() thread-safe
jeffzhou2000
committed
ggml-hexagon: fix a typo in ggml-hexagon.cpp
jeffzhou2000
committed
ggml-hexagon: list all known todo and fixme tasks in ggml-hexagon.cpp
jeffzhou2000
committed
ggml-hexagon: fix units MB -> MiB
jeffzhou2000
committed
ggml-hexagon: try to make ggml-hexagon backend works fine in a standard Android APP
jeffzhou2000
committed
ggml-hexagon: remove reduament code and make debug log more clear
jeffzhou2000
committed
ggml-hexagon: add gemma-3-4b-it-Q8_0.gguf to verify q8_0 mulmat on cDSP
jeffzhou2000
committed
ggml-hexagon:add skeleton code of offload GGML_OP_SOFT_MAX/GGML_OP_RMS_NORM/GGML_OP_POOL_2D to Hexagon cDSP
jeffzhou2000
committed
ggml-hexagon: release ggml-dsp v0.60 on cDSP side
jeffzhou2000
committed
ggml-hexagon: merge build logic in kernels/Makefile to ggml-hexagon/CMakeLists.txt and remove Makefile
jeffzhou2000
committed
ggml-hexagon: fix a typo in ggml-hexagon.cpp
jeffzhou2000
committed
ggml-hexagon: uniform NDEBUG usage in ggml-hexagon.cpp and ggml-dsp.c
jeffzhou2000
committed
ggml-hexagon: add profiler feature for purpose of visualize NPU performance and release ggml-hexagon.cpp v1.80
jeffzhou2000
committed
ggml-hexagon: remove so-called dma memory pool to avoid confusion and ambiguity
jeffzhou2000
committed
ggml-hexagon: make function ggmlhexagon_init_rpcmempool in ggml-hexagon.cpp more robust
jeffzhou2000
committed
ggml-hexagon: fix potential resource leak in class hexagon_profiler
jeffzhou2000
committed
ggml-hexagon: enable multi-threading feature on cDSP side
jeffzhou2000
committed
ggml-hexagon: upgrade QNN SDK to v2.33.0.250327
jeffzhou2000
committed
ggml-hexagon: fix typo in ggml-hexagon.cpp
jeffzhou2000
committed
ggml-dsp: probe QuRT RTOS information in function ggmlop_dsp_open
jeffzhou2000
committed
ggml-hexagon: setting enable_rpc_ion_mempool to 1 and make test-backend-ops happy
jeffzhou2000
committed
ggml-hexagon: check whether user's specified htp arch is valid in CMakeLists.txt
jeffzhou2000
committed
ggml-hexagon: sync with upstream
jeffzhou2000
committed
ggml-hexagon: refine pinned-memory feature
jeffzhou2000
committed
ggml-hexagon: refine build system in ggml-hexagon
jeffzhou2000
committed
ggml-hexagon: remove redundant code in struct ggml_backend_hexagon_buffer_context
jeffzhou2000
committed
ggml-hexagon: upgrade Android NDK to android-ndk-r28
jeffzhou2000
committed
ggml-dsp: split ggml-dsp.c into multiple files and cleanup
jeffzhou2000
committed
ggml-dsp: refine ggml-dsp and make ggml-dsp more clear
jeffzhou2000
committed
ggml-hexagon: fix a minior issue in dev ops
jeffzhou2000
committed
ggml-hexagon: fix a build issue in CI
jeffzhou2000
committed
ggml-dsp: cleanup code
jeffzhou2000
committed
ggml-hexagon: sync with upstream
jeffzhou2000
committed
ggml-dsp: cleanup code
jeffzhou2000
committed
ggml-dsp:refine ggmlhexagon_dsp_add_f32
jeffzhou2000
committed
ggml-dsp: refine logic of thread_counts
jeffzhou2000
committed
ggml-hexagon: release v1.06 and ready for code review
jeffzhou2000
committed
ggml-dsp: make GGML_OP_ADD more faster on cDSP side
jeffzhou2000
committed
ggml-hexagon: sync from project kantv(make ggml-hexagon backend can works in a standard Android APP)
jeffzhou2000
committed
sync with upstream llama.cpp and sync ggml-hexagon.cpp from project kantv
jeffzhou2000
committed
sync with upstream
jeffzhou2000
committed
sync with upstream
jeffzhou2000
committed
ggml-hexagon: upgrade QNN SDK to v2.34.0.250424
jeffzhou2000
committed
sync with upstream
jeffzhou2000
committed
ggml-hexagon: sync from project kantv(fix a long-term issue which introduced in https://github.com/kantv-ai/kantv/pull/281)
jeffzhou2000
committed
ggml-hexagon: sync with upstream llama.cpp
jeffzhou2000
committed
build: enable self-contained-build to simplify workflow
jeffzhou2000
committed
sync with upstream
jeffzhou2000
committed
add prebuilt binary libggmlop-skel.so
jeffzhou2000
committed
refine ggml-hexagon.cfg for the prebuilt binary libggmlop-skel.so
jeffzhou2000
committed
refine scripts to avoid confusion
jeffzhou2000
committed
ggml-hexagon: add set_hexagon_cfg(int new_hexagon_backend, int new_hwaccel_approach) in ggml-hexagon.h for further usage
jeffzhou2000
committed
project: rename libggmlop-skel.so to libggmldsp-skel.so and add ggmlhexagon-benchmark.cpp
jeffzhou2000
committed
project: release libggmldsp-skel.so v0.97
jeffzhou2000
committed
ggml-hexagon: upgrade QNN SDK to v2.35.0.250530
jeffzhou2000
committed
project: fix typo and build issue
jeffzhou2000
committed
ggmlhexagon-benchmark: add running timestamp and enable ggmlhexagon-benchmark works fine on Linux
jeffzhou2000
committed
ggml-hexagon: update ggml-hexagon.cpp to v1.11 and refine related codes and scripts
jeffzhou2000
committed
llama-bench: add running timestamp to analysis regression issue in llama.cpp:https://github.com/zhouwg/ggml-hexagon/discussions/34#discussioncomment-13449650
jeffzhou2000
committed
project: add prebuilt LLM models for compare inference peformance between QNN-CPU,QNN-GPU,QNN-NPU,cDSP,ggml
jeffzhou2000
committed
script: refine scripts/build-run-android.sh
jeffzhou2000
committed
project: sync with upstream
jeffzhou2000
committed
troubleshooting: add ggml-20250531 to troubleshooting performance regression issue in upstream
jeffzhou2000
committed
script: simplify workflow
jeffzhou2000
committed
project: sync with upstream
jeffzhou2000
committed
project: sync with upstream
jeffzhou2000
committed
project: sync with upstream
jeffzhou2000
committed
project: add prebuilt LLM model t5-277M-F32.gguf for compare inference peformance between QNN-CPU,QNN-GPU,QNN-NPU,cDSP,ggml
jeffzhou2000
committed
script: refine scripts/build-run-android.sh
jeffzhou2000
committed
project: adapt to thread safety test in upstream
jeffzhou2000
committed
project: remove unused ggml-20250531 which added for troubleshooting regression issue in upstream
jeffzhou2000
committed
ggml-hexagon: fix issue which introduced by test-thread-safety in the upstream
jeffzhou2000
committed
project: add codes for developers/experts's effort on cDSP side
jeffzhou2000
committed
build: refine script for developers/experts's effort on cDSP side
jeffzhou2000
committed
script: fix a minor issue in scripts/build-run-android.sh
jeffzhou2000
committed
script: refine script according to https://github.com/quic/ai-hub-apps/tree/main/tutorials/llm_on_genie
jeffzhou2000
committed
ggml-hexagon: add mulmat_algotype for further usage
jeffzhou2000
committed
project: sync with upstream
jeffzhou2000
committed
ggml-dsp: fix typo
jeffzhou2000
committed
project: release libggmldsp-skel.so v0.98
jeffzhou2000
committed
project: sync with upstream
jeffzhou2000
committed
project: sync with upstream
jeffzhou2000
committed
project: sync with upstream
jeffzhou2000
committed
test: verify Google gemma-3n on Android phone
jeffzhou2000
committed

Footer

© 2025 GitHub, Inc.

Footer navigation

Terms
Privacy
Security
Status
Community
Docs
Contact

You can’t perform that action at this time.