4 ms·
250 vs. 50000 req/s seems like a too big of a difference to me. Sure Go is faster than Node but Node is no slough either, you might want to dig in some deeper w
by RedShift1 5y ago
250 vs. 50000 req/s seems like a too big of a difference to me. Sure Go is faster than Node but Node is no slough either, you might want to dig in some deeper why you only got 250 req/s with Node.
- akvadrako 5y agoThat could mostly be due to multithreading. That comes free with go but requires a different model in node.
- lultimouomo 5y ago50000/250=200, that is a lot of cores!
- fileeditview 5y agoThat calculation assumes that Go and Node have the same performance and the only gain is by parallelism. But I agree.. the performance difference seems strange. On a 1 core basis I'd expect something like a factor 2-5 gain.
- adonovan 5y agoIdentical algorithms (e.g. quicksort) run ~40x faster in Go than Ruby in a single thread due to more efficient CPU usage, even when you avoid the kinds of steps that would allocate objects in Ruby. Multiply by 64 cores and it’s easy to observe a 1000x improvement with very little difference in the code.
- KwisaksHaderach 5y agoDoesn't sound unrealistic if you have a mix load of IO and raw processing.