Will features such as distributed shared memory, TMA, and WGMMA be retained in the upcoming B100 series and future architectures? I couldn’t find detailed information about this in the whitepaper. Thank you!
Related topics
| Topic | Replies | Views | Activity | |
|---|---|---|---|---|
| Question about the following release features of the next release | 0 | 1442 | May 31, 2007 | |
| Distributed shared memory will lose after the kernel in H100? | 1 | 225 | September 12, 2023 | |
| SHFL for Ampere and Volta | 0 | 438 | December 22, 2022 | |
| Does 4080 support Distributed Shared Memory？ | 6 | 574 | December 26, 2022 | |
| What is the best way to re-use a tensor core C fragment now as A or B input when their types differ? | 5 | 706 | November 24, 2023 | |
| Configurable Shared Memory Size / SM | 1 | 805 | November 8, 2021 | |
| Cuda Unified memory with Pascal | 3 | 820 | August 13, 2016 | |
| Unified memory for CC 6.1 | 4 | 1395 | December 6, 2016 | |
| Is Shared Memory persistent throughout a session? | 1 | 7393 | August 24, 2011 | |
| Mobile Processor support? | 1 | 3488 | April 9, 2007 |