7 ms·

It happens again and again, never believe benchmarks by original author/owner. They are misleading in most cases. Deno vs bun was same case, each showed their

It happens again and again, never believe benchmarks by original author/owner. They are misleading in most cases. Deno vs bun was same case, each showed their runtime as faster

franciscop 4y ago
Nah, read a bit of the Deno vs Bun thing and it's pretty clear. Bun said that Bun was way faster (on 1 thread), and it is. But Deno were butt-hurt for these claims and said they were way faster (in 2-4 threads, vs Bun's 1 thread), and they are. But that's a ridiculous comparison for multiple reason: - Bun does recognize they don't have multi-processor yet, so yeah cannot compare that (remember it's still alpha software!). So they compared what they had so far with a similar setup in Deno, pretty fair for that stage IMHO! - In real-life you'd have either 1 thread for smaller apps, or setup a proxy with multiple instances if the program was single-threaded, so the comparison Deno made was in no way realistic.
- pier25 4y ago
  > or setup a proxy with multiple instances if the program was single-threaded Not having to do that is quite an advantage for Deno though regardless of the perf claims.
- pierrebai 4y ago
  You are claiming that if your product A does not have feature X that could make it faster, it is okay to claim that your product A is faster than product B by hampering B under the impediment of not using X? "My database A is is faster than B! (When B is not using indexes, because my database A does not have indexes, so it would be unfair to let B uses indexes..." is the equivalent here.
  - franciscop 4y ago
    Oh cmon no I'm not at all. I'm saying: "I'm building a new product. It's half-way there, but so far I've finished single-threaded processing! Look, at single-thread it's faster than anything else by a large margin". That's it. Indexes in a DB are like the main thing to do, while many/most servers that are run online are still single-threaded, so not a good comparison.
- gr__or 4y ago
  I think you're underestimating the effort needed for parallelization here. Deno is written in a language that was built for "fearless concurrency" (Rust), whereas Bun is written in a language that barely has a concurrency story (Zig)

It happens again and again, never believe benchmarks by original author/owner. They are misleading in most cases. Deno vs bun was same case, each showed their runtime as faster

franciscop 4y ago

Nah, read a bit of the Deno vs Bun thing and it's pretty clear. Bun said that Bun was way faster (on 1 thread), and it is. But Deno were butt-hurt for these claims and said they were way faster (in 2-4 threads, vs Bun's 1 thread), and they are. But that's a ridiculous comparison for multiple reason: - Bun does recognize they don't have multi-processor yet, so yeah cannot compare that (remember it's still alpha software!). So they compared what they had so far with a similar setup in Deno, pretty fair for that stage IMHO! - In real-life you'd have either 1 thread for smaller apps, or setup a proxy with multiple instances if the program was single-threaded, so the comparison Deno made was in no way realistic.

pier25 4y ago
> or setup a proxy with multiple instances if the program was single-threaded Not having to do that is quite an advantage for Deno though regardless of the perf claims.
pierrebai 4y ago
You are claiming that if your product A does not have feature X that could make it faster, it is okay to claim that your product A is faster than product B by hampering B under the impediment of not using X? "My database A is is faster than B! (When B is not using indexes, because my database A does not have indexes, so it would be unfair to let B uses indexes..." is the equivalent here.
- franciscop 4y ago
  Oh cmon no I'm not at all. I'm saying: "I'm building a new product. It's half-way there, but so far I've finished single-threaded processing! Look, at single-thread it's faster than anything else by a large margin". That's it. Indexes in a DB are like the main thing to do, while many/most servers that are run online are still single-threaded, so not a good comparison.
gr__or 4y ago
I think you're underestimating the effort needed for parallelization here. Deno is written in a language that was built for "fearless concurrency" (Rust), whereas Bun is written in a language that barely has a concurrency story (Zig)