-
The hax CI job takes ~15 minutes right now because setting up hax is so slow.
Use cached binaries rather than rebuilding every time.
https://github.com/cryspen/bertie/blob/8ef1493caf8d7f0fdaae2103…
-
## Feature Request
### Summary
In a block sync of 5000 blocks on Osmosis mainnet, we see 23 seconds out of 1050 in EventBus.PublishEventTx (`2.1%` of the ApplyBlock time), this should likely…
-
Tests to write for a test suite:
- [ ] Test everything in Ben’s presentation
- [ ] Reach out to Ben
What features are expected? Everything should appear
Possible variation in accuracy
How do we s…
-
**Goals**
Jeff is suggesting to make a quick edit on Spacely to concatenate the send commands into one line and to parse the receive afterwards:
- [x] Benchmark read/write before the change
- [x]…
-
Let's speedup the compilation of the C++ source code of LFortran. We identified at least the following potential speedups:
* [ ] split `asr.h` into smaller parts, and only use what we need; the sto…
-
# Feature or enhancement
### Proposal:
The code for packing and unpacking floating point numbers is typically a single memory operation on modern CPUs. However, the current implementation is much mo…
ruema updated
3 months ago
-
Thanks for your interesting work! I have tried to infer a 75 frames video on A100 in 512*768, which will take about 3mins. I also tried to use more cards, however, it only generates more videos :( . D…
-
So, running this with a CPU only is very very slow. It took 60 minutes for it to make a 16-minute video. There are GPUs that exist. However, that is a platform specific problem - specific hardware use…
-
Recently come across this and its great for skipping sponsor stuff, but doesn't look like it catch most ads? if its not possible to get around them could a function to speed them up be possible? was s…
-
Is it possible to enable CUDA for diarization? It would mean a sizable speedup for that step.
Trying to set --onnx-provider to cuda gives the following message:
`D:\a\sherpa-onnx\sherpa-onnx\sherp…