Skip to content

Conversation

@jjjxp03
Copy link

@jjjxp03 jjjxp03 commented Dec 5, 2025

For CANN backend:Currently, mul_mat_id performs a device-to-host copy, which does not support full-image inference.Optimize mul_mat_id quantization by referencing the non-quantized version and removing the device_to_host copy operation.

Verify the correctness of the modification using the existing test tool:
./bin/test-backend-ops test -b CANN0 -o MUL_MAT_ID
Passed the q_8 test. Currently, only q_8 has been implemented and tested. Tensor self is not implemented for DT_INT4; to support q_4, the corresponding type implementation needs to be added.

@github-actions github-actions bot added ggml changes relating to the ggml tensor library for machine learning Ascend NPU issues specific to Ascend NPUs labels Dec 5, 2025
@TianHao324
Copy link
Contributor

TianHao324 commented Dec 8, 2025

The results of running the above-mentioned modifications on the qwen2.5:0.5b f16 model are as follows.

user
Building a website can be done in 10 simple steps:
assistant
Here’s a concise breakdown of the 10 steps you can follow to build a website:

1. **Define Your Goals and Audience**
   - **Define**: Understand your target audience, what you want to achieve, and who your website will serve.

2. **Choose a Domain Name and Hosting**
   - **Domain Name**: Secure a memorable and relevant domain name.
   - **Hosting**: Select a reliable hosting provider that suits your budget and needs.

3. **Select a Content Creation Platform**
   - **CMS (Content Management System)**: Choose a platform like WordPress or Wix for easy content management.

4. **Design Your Website**
   - **Design**: Plan and create a visually appealing and user-friendly website design.

5. **Choose a Theme or Template**
   - **Themes**: Select a theme or template that fits your brand and needs.

6. **Create Content**
   - **Write**: Write the content that will be on your website.
   - **Create**: Craft the content that meets your goals.

7. **Optimize Your Website**
   - **SEO**: Use keyword research, meta tags, and on-page SEO strategies to improve your website’s search engine ranking.

8. **Create and Publish Pages**
   - **Pages**: Organize your content into different pages.
   - **Publish**: Share your website on various platforms like Google Sites, WordPress, or other platforms.

9. **Implement Social Media Integration**
   - **Social Media**: Add social media features like Facebook, Twitter, and Instagram to your site.
   - **Social Media Management**: Set up social media management tools to manage your social media accounts.

10. **Test Your Website**
    - **Testing**: Test your website thoroughly to ensure it meets your requirements and is user-friendly.
    - **Iterate**: Based on feedback, make necessary adjustments to improve user experience.

By following these steps, you can create a functional and user-friendly website.

llama_perf_sampler_print:    sampling time =     104.30 ms /   421 runs   (    0.25 ms per token,  4036.47 tokens per second)
llama_perf_context_print:        load time =    9360.35 ms
llama_perf_context_print: prompt eval time =      15.28 ms /    21 tokens (    0.73 ms per token,  1374.08 tokens per second)
llama_perf_context_print:        eval time =    4707.41 ms /   399 runs   (   11.80 ms per token,    84.76 tokens per second)
llama_perf_context_print:       total time =    8466.42 ms /   420 tokens
llama_perf_context_print:    graphs reused =        386

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

Ascend NPU issues specific to Ascend NPUs ggml changes relating to the ggml tensor library for machine learning

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants