vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

make github CI happy

Browse source
This commit is contained in:
zhou.weiguo 2024-04-25 15:28:45 +08:00
parent 5bc7b76117
commit 11a11b0b54
No known key found for this signature in database
GPG key ID: 952EA81D18BB2FA8
1 changed files with 2 additions and 2 deletions
Download patch file Download diff file Expand all files Collapse all files

4
README-qnn.md
View file

@ -11,7 +11,7 @@
## Background ## Background
Android maintained its position as the leading mobile operating system worldwide in the fourth quarter of 2023 with <b><a href="https://www.statista.com/statistics/272698/global-market-share-held-by-mobile-operating-systems-since-2009/">a market share of 70.1 percent </a></b> . Qualcomm is No.1 mobile SoC semiconductor company in our planet currently. Android maintained its position as the leading mobile operating system worldwide in the fourth quarter of 2023 with <b><a href="https://www.statista.com/statistics/272698/global-market-share-held-by-mobile-operating-systems-since-2009/">a market share of 70.1 percent </a></b> . Qualcomm is No.1 mobile SoC semiconductor company in our planet currently.
**QNN**(Qualcomm Neural Network, aka Qualcomm AI Engine Direct) SDK is verified to work with the following versions of the ML frameworks: **QNN**(Qualcomm Neural Network, aka Qualcomm AI Engine Direct) SDK is verified to work with the following versions of the ML frameworks:
@ -115,7 +115,7 @@ Please add the **[QNN]** prefix/tag in issues/PRs titles to help the community c
- only support FP32 / FP16 and the input and output tensors must be of the <b>same data type</b> - only support FP32 / FP16 and the input and output tensors must be of the <b>same data type</b>
- lack of [implementation of other GGML-OPs using QNN API](https://github.com/zhouwg/llama.cpp/blob/qualcomm_qnn_backend_for_ggml/ggml-qnn.cpp#L3452). this work is very similar to <a href="https://github.com/zhouwg/llama.cpp/blob/qualcomm_qnn_backend_for_ggml/ggml-qnn.cpp#L2983">GGML_OP_ADD / GGML_OP_MUL / GGML_OP_MULMAT</a> in ggml-qnn.cpp - lack of [implementation of other GGML-OPs using QNN API](https://github.com/zhouwg/llama.cpp/blob/qualcomm_qnn_backend_for_ggml/ggml-qnn.cpp#L3452). this work is very similar to <a href="https://github.com/zhouwg/llama.cpp/blob/qualcomm_qnn_backend_for_ggml/ggml-qnn.cpp#L2983">GGML_OP_ADD / GGML_OP_MUL / GGML_OP_MULMAT</a> in ggml-qnn.cpp
- multithreading not working with QNN GPU&HTP (aka DSP) backend - multithreading not working with QNN GPU&HTP (aka DSP) backend