Framework Benchmarks

[Iain Barnett]

Anyone else had a look at these?

http://www.techempower.com/blog/2013/03/28/framework-benchmarks/

Interesting.

Regards,
Iain

[kashyap k]

Things I observed:

1. They haven't used "json_ext" which improves serialization performance.

2. No numbers with much faster json implementations like Oj

3. Ruby compiled with "-O3" config parameter is faster than the normal Ruby via rvm/rbenv and there is no information about that.

4. The issues in their github repo paint a similar picture with regards to other frameworks as well.

Nice perspective though since the servlet performance is almost unthinkable with Ruby.

--
You received this message because you are subscribed to the Google Groups "sinatrarb" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sinatrarb+...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

--
Kashyap KMBC

[Iain Barnett]

On 30 Mar 2013, at 08:31, kashyap k wrote:

Things I observed:

1. They haven't used "json_ext" which improves serialization performance.

2. No numbers with much faster json implementations like Oj

3. Ruby compiled with "-O3" config parameter is faster than the normal Ruby via rvm/rbenv and there is no information about that.

4. The issues in their github repo paint a similar picture with regards to other frameworks as well.

Nice perspective though since the servlet performance is almost unthinkable with Ruby.

They'd all be interesting things to try.

I'd also like to see the difference between different ORM's, especially Sequel, but the main difference I'd make is to the line that makes the call to `rand`

    World.find(Random.rand(10000) + 1)

Personally, I'd like all the specs to preload that list of numbers so they're working from the same list and `rand` isn't a factor.

Nice to see that Sinatra generally beats Rails though :)

Regards,

Iain