-
Relates to https://github.com/webmachinelearning/webnn/issues/252
Some accelerators use non-standard float point types (e.g. bfloat16 and TF32). They are important to achieve high performance (e.g.…
-
My job failed due to FAILED_CONTROLLER after a preemption.
sky spot logs --controller 5 doesn't show any error:
> (small-yt, pid=45563) I 03-12 15:33:03 spot_utils.py:92] =======================…
-
## Issue description
There are a slew of new AI hardware accelerators available that can beat the GPUs handily either in terms of raw performance or performance per Watt. There are also many new ef…
-
The [Heterogeneous parallel programming with open standards using oneAPI and Data Parallel C++](https://www.w3.org/2020/06/machine-learning-workshop/talks/heterogeneous_parallel_programming_with_open_…
-
Hi,
I'm trying to replace private spad in accelerators with private cache, according to Sam's suggestion in [this thread](https://github.com/harvard-acc/smaug/issues/102). I also got from this thread…
-
(nanodet) simon@Simon:~/nanodet$ python tools/train.py config/nanodet-m-416.yml
[root][07-16 11:17:37]INFO:Using Tensorboard, logs will be saved in workspace/nanodet_m_416/logs
[root][07-16 11:17:37…
-
Congratulations on the name change! That reminded me, perhaps we should support OpenCL 2.0 on Android. I should also definitely make Metal runnable on iOS, perhaps we introduce both at the same time (…
-
- PyTorch-Forecasting version: 0.8.4
- PyTorch version: 1.8.0
- Python version: 3.8.8
- Operating System: CentOS
### Expected behavior
I'm working through the _Demand forecasting with the T…
-
`psa_crypto_init()` is allowed to, and even encouraged to, initialize the RNG. This allows many implementations to provide a RNG that will never fail. (Implementations that comply to some security sta…
-
**Is your feature request related to a problem? Please describe.**
Allow the provider to tell the user which types of FI_HMEM they support, and the operations they support on each type of HMEM.
**…