Re: [TYPO3-core] RFC: #0011505: TYPO3 4.3 is slower than 4.2.8 -> re-implement native db caching

[ries van Twisk-2]

On Jul 8, 2009, at 12:13 PM, Steffen Kamper wrote:

> Hi,
>
> Rupert Germann schrieb:
>> hi steffen,
>>
>> Steffen Kamper wrote:
>>> PID   uncached        cached (Cacheframework)         cached (patch)
>>> =
>>> =
>>> ====================================================================
>>> 1     1909ms          1365ms                          1342ms
>>
>> 1365ms for a cached page?!
>> doesn't look very cached to me if compared to uncached ;-)
>>
>
> yes, it's slooooow
> It's my development having tons of extensions in it (not all  
> installed)
> But for comparison it's good ;)
>
> vg Steffen

Steffen,

even with tons of extensions, should the cached version still not
be in the order of ms??

However, you did mention you posted parse times, would
you not want to post retrieval times?

Ries



_______________________________________________
TYPO3-dev mailing list
TYPO3-dev@...
http://lists.netfielders.de/cgi-bin/mailman/listinfo/typo3-dev

[Steffen Kamper-3]

Hi Ries,

Ries van Twisk schrieb:
>
> Steffen,
>
> even with tons of extensions, should the cached version still not
> be in the order of ms??
>
> However, you did mention you posted parse times, would
> you not want to post retrieval times?
>

thx for outsourcing this to dev.
My impression is that most of the time the exts are parsed for the
localconfs. The page delivery may be fast, but this file operations take
a long time. I have to examine this.

And yes - i quickly outputted the parse time in index_ts.php which shows
  complete parse time. For cache related i should use own timers in code
to isolate cache only. But as the patched version saved some ms it's
obvious that it speeds it up a bit.

vg Steffen
_______________________________________________
TYPO3-dev mailing list
TYPO3-dev@...
http://lists.netfielders.de/cgi-bin/mailman/listinfo/typo3-dev

[Mathias Schreiber [wmdb >]]

Steffen Kamper schrieb:
> thx for outsourcing this to dev.
> My impression is that most of the time the exts are parsed for the
> localconfs. The page delivery may be fast, but this file operations take
> a long time. I have to examine this.

Take a look at #11474 and #11745.
I add a setting in EM_CONF to disable loading unneeded ext_tables and
ext_localconf from BE extensions for example.
I generate temp_cached_HASH.php files like before but also generate
temp_cached_FE_HASH.php for loading in the FE.
These files are 1/4 in a blank install thus speeding up frontend rendring.

> And yes - i quickly outputted the parse time in index_ts.php which shows
>  complete parse time. For cache related i should use own timers in code
> to isolate cache only. But as the patched version saved some ms it's
> obvious that it speeds it up a bit.

We will start working on a tarball delivering a testing installation so
all benchmarks are made upon common ground.

cheers
Mathias


--
TYPO3 certified interogator
T3DD09 Entertainer
_______________________________________________
TYPO3-dev mailing list
TYPO3-dev@...
http://lists.netfielders.de/cgi-bin/mailman/listinfo/typo3-dev

[Steffen Kamper-3]

Hi Mattes,

Mathias Schreiber [wmdb >] schrieb:

> Steffen Kamper schrieb:
>> thx for outsourcing this to dev.
>> My impression is that most of the time the exts are parsed for the
>> localconfs. The page delivery may be fast, but this file operations
>> take a long time. I have to examine this.
>
> Take a look at #11474 and #11745.
> I add a setting in EM_CONF to disable loading unneeded ext_tables and
> ext_localconf from BE extensions for example.
> I generate temp_cached_HASH.php files like before but also generate
> temp_cached_FE_HASH.php for loading in the FE.
> These files are 1/4 in a blank install thus speeding up frontend rendring.
>

will do when BT is back

>> And yes - i quickly outputted the parse time in index_ts.php which
>> shows  complete parse time. For cache related i should use own timers
>> in code to isolate cache only. But as the patched version saved some
>> ms it's obvious that it speeds it up a bit.
>
> We will start working on a tarball delivering a testing installation so
> all benchmarks are made upon common ground.
>

that would be great. But how to apply patches there? Best would be to
put it in a forge project in repository!

vg Steffen
_______________________________________________
TYPO3-dev mailing list
TYPO3-dev@...
http://lists.netfielders.de/cgi-bin/mailman/listinfo/typo3-dev