Improve WTX#599
Conversation
|
/benchmark -f wtx-http2 --save |
|
👋 |
|
🚀 Benchmark run triggered for |
Benchmark ResultsFramework:
Full log |
|
I'll add some more frameworks on this test for a broader comparison, could be a load generator issue or not optimal config on connections/streams |
|
Hum, seems unlikely given the results of the other HTTP/2 frameworks. It is probably a bad default HTTP/2 configuration like the maximum number of concurrent streams or a recent change I made in the |
|
Hum, let me tweak some HTTP/2 parameters to see their runtime impact. I will be back after a couple of hours. |
|
Added a few more h2c entries and more conn variety for better comparison https://www.http-arena.com/leaderboard/#v=h2&t=baseline-h2c&type=infrastructure,production,tuned |
Wow... The performance of h2o is really impressive
Not so much in the JSON scenario, strange. |
|
/benchmark -f wtx-http2 --save |
|
👋 |
|
🚀 Benchmark run triggered for |
Benchmark ResultsFramework:
Full log |
|
Nah, I give up. That was the best I could do given the constraints. At least it is clear that HTTP/2 parameters weren't the problem. |
|
Perhaps I will get back after a couple of months. The code changes are still worth merging though. |
Yea, the json implementation is really bad for it, jsut threw it in |


Description
Looks like the performance of
WTXis terrible, which is strange given other similar benchmarks and the time I spent optimizing it.Well, this is the last attempt without diving into profiling applications . Let's see if things will improve.