ICACT20210136 Question.1
Questioner: mrobaei.dev@gmail.com 2021-02-09 오후 10:11:57 |
ICACT20210136 Answer.1
Answer by Auhor zhanyueying@csu.ac.cn 2021-02-09 오후 10:11:57
Chrome  Click!! |
| Dear author, for some reason my question has been deleted? Considering the hardware requirement (you have used GPU with 24GB memory), what are the applications for the proposed methods? |
Thank you for your question! I don’t know why your first question disappears. But I remember that there was a question about the hardware requirements of mobile phones. My motivation is to simulate and evaluate the performance of LDPC codes. The proposed method can be used for simulation applications. If there is a demand for more than tens of million mobile devices, I think an encoder based on ASICs would be a lower cost choice. |
ICACT20210136 Question.2
Questioner: yfujii@kke.co.jp 2021-02-09 오후 10:43:43 |
ICACT20210136 Answer.2
Answer by Auhor zhanyueying@csu.ac.cn 2021-02-09 오후 10:43:43
Chrome Click!! |
| Thank you for your presentation.
Do you have any plan to embed your LDPC Encoder into 5G software platform such as OpenAirInterface?(OAI) I believe, OAI already has its own LDPC Encoder/Decoder in CUDA. If you know, could you tell me what's your implementation's advantage on the OAI's one?
|
Thank you for your question and telling me about OAI! I have not used OAI on my device before. But I read the codes of OAI today and I found that it used hard coding. It stores the parity check matrices data in the source code rather than obtaining it from a configuration file. So, I believe that my implementation is more extensible. For example, my encoder can read new matrices different from the existing matrices. In addition, my lab is also developing a USRP system and I plan to embed my encoder into it. |
ICACT20210136 Question.3
Questioner: Yiile@outlook.com 2021-02-09 오후 10:56:52 |
ICACT20210136 Answer.3
Answer by Auhor zhanyueying@csu.ac.cn 2021-02-09 오후 10:56:52
Chrome Click!! |
| Dear author. What are the features of high throughput and flexibility in your works? Could you explain them in detail? |
Thank you for your question. My LDPC encoder achieves a 38–62Gbps throughput on a single GPU, and I believe that it’s a high throughput. The feature of flexibility means that it can encode LDPC codes with different circulant permutation matrix sizes and different code rates. I hope I have answered your question. |
