[ruby-core:85666] Re: [Ruby trunk Feature#12589] VM performance improvement proposal

From: Vladimir Makarov <vmakarov@...>
Date: 2018-02-20 04:40:50 UTC
List: ruby-core #85666

On 02/19/2018 05:17 PM, sam.saffron@gmail.com wrote:
> Issue #12589 has been updated by sam.saffron (Sam Saffron).
>
>
> I just measured your branch using Discourse bench at: https://github.com/discourse/discourse/blob/master/script/bench.rb
>
> Looks like it is a bit slower than master:
Trace insn are still generated on the branch.  The current trunk does 
not generate them.  Removing trace insns by Koichi improved performance 
by about 10%.  I believe the branch will be faster when the trace insns 
are also removed there.  But it is hard to predict what the actual 
improvement will be after that.
> RTL:
>
> ```
> timings:
>    load_rails: 3749
> ruby-version: 2.5.0-p-1
> rss_kb: 247952
> pss_kb: 236618
> memorysize: 5.88 GB
> virtual: vmware
> architecture: amd64
> operatingsystem: Ubuntu
> processor0: Intel(R) Core(TM) i7-4770K CPU @ 3.50GHz
> physicalprocessorcount: 1
> kernelversion: 4.4.0
>
> ```
>
> Master
>
> ```
> timings:
>    load_rails: 3789
> ruby-version: 2.6.0-p-1
> rss_kb: 276588
> pss_kb: 265237
> memorysize: 5.88 GB
> virtual: vmware
> architecture: amd64
> operatingsystem: Ubuntu
> processor0: Intel(R) Core(TM) i7-4770K CPU @ 3.50GHz
> physicalprocessorcount: 1
> kernelversion: 4.4.0
>
> ```
>
> Very interesting to see the significant memory improvement, is that expected?
No. Actually I expected more memory usage for RTL.  It is hard for me to 
say a reason for memory improvement for now.  When the current trunk 
will be merged into the branch, I could speculate more.  Now the branch 
code is far behind (about 13 months) the current trunk.

>   only env var I am running is: `RUBY_GLOBAL_METHOD_CACHE_SIZE: 131072`
>

Unsubscribe: <mailto:ruby-core-request@ruby-lang.org?subject=unsubscribe>
<http://lists.ruby-lang.org/cgi-bin/mailman/options/ruby-core>

In This Thread

Prev Next