-
Recent profiling [profvis_arduous.dingo_20240815.html.zip](https://github.com/user-attachments/files/16641071/profvis_arduous.dingo_20240815.html.zip) from exploration-tooling uncovered a couple slow …
-
# JavaScript benchmark 性能对比 · 一个工具 - VIP
在线 jsperf 工具,可以对比多个 JS 代码片段的 perf benchmark 性能数据,选择性能最高的写法!
[https://atool.vip/perf/](https://atool.vip/perf/)
-
Discuss TG Llama e2e perf in this issue
![image](https://github.com/user-attachments/assets/ff77d09e-3aba-4170-adb8-bb08fed09a75)
First tracy shows that concatting output on host is a huge 250ms…
-
### Is your proposal related to a problem?
(Write your answer here.)
We can enable prisma to enhance an existing application or explore its capabilities through a demo.
### Describe the sol…
-
```
u64 do_cuda_uvm_init(int, u64): Assertion `page_size == 64*KB || page_size == 128*KB || page_size == 256*KB' failed.
Aborted (core dumped)
```
-
Parsing perf stat JSON output throws an error when using the perf stat command Wrapper (PerfStatWrap)
Error thrown:
```
Traceback (most recent call last):
File "/home/taoufik/Documents/PAAE/be…
-
Command:
`perf record -o perf.data -e cycles -e shed:sched_switch --switch-events --sample-cpu -m 8M --aio -z --call-graph dwarf $(realpath ./extract)`
Output:
```
event syntax error: 'shed…
-
Running perf for the following flatbuffer [out.ttnn](https://drive.google.com/file/d/1xpW59AFICEhlsmpkriD1ssJy2GpK-1mR/view?usp=sharing) create a csv [ops_perf_results_2024_08_29_10_47_08.csv](https:/…
-
We're currently specifying in the CLI a fixed txn batch size for an incoming block payload. This has some limitations:
- we are not taking advantage of groups of "small" txns, especially in terms o…
-
- Explore ways to improve the performance of the pipeline, such as:
- Avoid parsing the entire BAM file.
- Introduce multithreading in Kestrel genotyping to handle larger datasets more efficient…