Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.lang.php > #10316 > unrolled thread

OPcache in php5.5.1 on Windows

Back to article view | Back to comp.lang.php

Contents

  OPcache in php5.5.1 on Windows Erwin Moller <erwinmollerusenet@xs4all.nl> - 2013-07-31 14:50 +0200
    Re: OPcache in php5.5.1 on Windows Jerry Stuckle <jstucklex@attglobal.net> - 2013-07-31 09:29 -0400
    Re: OPcache in php5.5.1 on Windows Goran <goran@nospam.com> - 2013-08-01 16:00 +0200
    Re: OPcache in php5.5.1 on Windows Adam Harvey <usenet@adamharvey.name> - 2013-08-01 21:21 +0000
      Re: OPcache in php5.5.1 on Windows Erwin Moller <erwinmollerusenet@xs4all.nl> - 2013-08-06 23:13 +0200

#10316 — OPcache in php5.5.1 on Windows

Hello cpl,

Does anybody in here have experience with OPCache under windows?
http://nl3.php.net/opcache
It is bundled in PHP5.5 (and via PECL in earlier versions).

I installed PHP5.5.1 on a SBS2008 server under IIS7 (fast cgi).
After activating opcache, I used the suggested directives in php.ini, as 
described here:
http://nl3.php.net/manual/en/opcache.configuration.php

It does show up now in phpinfo() as:

Opcode Caching 	Up and Running
Optimization 	Enabled
Startup 	OK
Shared memory model 	win32
Cache hits 	12
Cache misses 	15
Used memory 	5752832
Free memory 	61356032
Wasted memory 	0
Cached scripts 	15
Cached keys 	21
Max keys 	3907
OOM restarts 	0
Hash keys restarts 	0
Manual restarts 	0

So far so good (I think).


Now I wonder: Is there any way to find out WHAT kind of performance gain 
it gives?

I have a few ideas but I am not very enthusiastic about any of them:

1) Testing the time spend in the script: Not good, that is exclusive the 
time needed to interpret the script (I expect, but I couldn't find a 
decisive answer yet. If anybody knows, let me know please)

2) From end-user experience: Not good: If I time the responsetimes with 
a clockwatch on my browser, I merely measure noise I expect: time spend 
on the network, is the server busy doing other stuff, is the database 
busy, etc.

3) Making a huge complicated script, and test it. Hoping the time needed 
to interpret it is so long I can measure the difference as described in 
2. (I haven't done this yet)

What would be a good way to test the speedgains by using OPCache?
Any ideas?

Thanks for your time.

Regards,
Erwin Moller



-- 
"That which can be asserted without evidence, can be dismissed without 
evidence."
-- Christopher Hitchens

[toc] | [next] | [standalone]

#10318

On 7/31/2013 8:50 AM, Erwin Moller wrote:
> Hello cpl,
>
> Does anybody in here have experience with OPCache under windows?
> http://nl3.php.net/opcache
> It is bundled in PHP5.5 (and via PECL in earlier versions).
>
> I installed PHP5.5.1 on a SBS2008 server under IIS7 (fast cgi).
> After activating opcache, I used the suggested directives in php.ini, as
> described here:
> http://nl3.php.net/manual/en/opcache.configuration.php
>
> It does show up now in phpinfo() as:
>
> Opcode Caching     Up and Running
> Optimization     Enabled
> Startup     OK
> Shared memory model     win32
> Cache hits     12
> Cache misses     15
> Used memory     5752832
> Free memory     61356032
> Wasted memory     0
> Cached scripts     15
> Cached keys     21
> Max keys     3907
> OOM restarts     0
> Hash keys restarts     0
> Manual restarts     0
>
> So far so good (I think).
>
>
> Now I wonder: Is there any way to find out WHAT kind of performance gain
> it gives?
>
> I have a few ideas but I am not very enthusiastic about any of them:
>
> 1) Testing the time spend in the script: Not good, that is exclusive the
> time needed to interpret the script (I expect, but I couldn't find a
> decisive answer yet. If anybody knows, let me know please)
>
> 2) From end-user experience: Not good: If I time the responsetimes with
> a clockwatch on my browser, I merely measure noise I expect: time spend
> on the network, is the server busy doing other stuff, is the database
> busy, etc.
>
> 3) Making a huge complicated script, and test it. Hoping the time needed
> to interpret it is so long I can measure the difference as described in
> 2. (I haven't done this yet)
>
> What would be a good way to test the speedgains by using OPCache?
> Any ideas?
>
> Thanks for your time.
>
> Regards,
> Erwin Moller
>
>
>

Think about this: No matter how you do it, the only important 
measurement is response time to the end user (system load is related, 
and can also affect response time).  Building a huge, complicated script 
which shows a 70% performance game may make you feel better - but will 
not be representative of the real world.


-- 
==================
Remove the "x" from my email address
Jerry Stuckle
JDS Computer Training Corp.
jstucklex@attglobal.net
==================

[toc] | [prev] | [next] | [standalone]

#10324

Best way to benchmark it is to hit your testing server with your 
production access log.

[toc] | [prev] | [next] | [standalone]

#10330

On Wed, 31 Jul 2013 14:50:19 +0200, Erwin Moller wrote:
> What would be a good way to test the speedgains by using OPCache?
> Any ideas?

You may want to consider using a tool like ab (which comes with Apache, at 
least on most Linux distros) and hitting a representative sample of pages 
that you're interested in. ab provides a range of options for how many 
simultaneous requests it will make, the total number of options, and so 
on, and will then give you a report at the end showing best case, average, 
and worst case response times.

If you do multiple runs with OPcache enabled and disabled, you should be 
able to get a pretty good idea of what sort of performance improvement 
you're getting.

Adam

[toc] | [prev] | [next] | [standalone]

#10425

On 8/1/2013 11:21 PM, Adam Harvey wrote:
> On Wed, 31 Jul 2013 14:50:19 +0200, Erwin Moller wrote:
>> What would be a good way to test the speedgains by using OPCache?
>> Any ideas?
>
> You may want to consider using a tool like ab (which comes with Apache, at
> least on most Linux distros) and hitting a representative sample of pages
> that you're interested in.

Thank you, Adam.

Apache Bench exists for windows (named unsurprisingly ab.exe).
I'll look into it.

Regards,
Erwin Moller



-- 
"That which can be asserted without evidence, can be dismissed without 
evidence."
-- Christopher Hitchens

[toc] | [prev] | [standalone]

Back to top | Article view | comp.lang.php

csiph-web