GD vs AGG performance...

[Andreas Albarello]

If I'm not mistaken, you actually don't use linework antialiasing with 
GD, while AGG uses antialiasing all the time and there's no way to turn 
it off (and antialiasing is the main reason why the AGG output looks so 
great).

Now, antialiasing ain't cheap. My guess is, if you'd enable antialiasing 
with GD, you'd get the same rendering times you are getting with AGG, 
but AGG would still be looking much better.

Bottom line is, I'd say, if you're fine with the output you are getting 
with GD and have never thought about enabling antialiasing, than there's 
no reason to switch to AGG, IMHO, as that would mean getting 
antialiasing all the time, whether you like it or not.

Best regards,
Andreas Albarello

John Cole wrote:
> I'm just going by the tuning info reported by mapserver, but there is a
> noticeable difference between them.
> 
> Try:
> MS4 (MS4W 2.2.2) total time: 0.250s
> http://map.uai.com:28080/cgi-bin/mapserv.exe?map=/mapserver/tiger/tiger.map&mode=browse&layers=bound+hydro+landmark+roads+roads_anno+city+hydro_anno+landmark_anno+landmark_point
> 
> MS5 AGG (MS4W 2.2.6) total time: 1.266s
> http://map.uai.com:28080/cgi-bin2/mapserv.exe?map=/mapserver/tiger/tiger5.map&mode=browse&layers=bound+hydro+landmark+roads+roads_anno+city+hydro_anno+landmark_anno+landmark_point
> 
> MS5 GD (MS4W 2.2.6) total time: 0.250s
> http://map.uai.com:28080/cgi-bin2/mapserv.exe?map=/mapserver/tiger/tiger5gd.map&mode=browse&layers=bound+hydro+landmark+roads+roads_anno+city+hydro_anno+landmark_anno+landmark_point
> 
> Except for GD->AGG (and the debug difference between 4 and 5), the map files
> are identical.
> 
> (BTW, these are using the same html template, so if you navigate, you won't
> be using the same MS/renderer combo).
> 
> Is there something else that needs to be done for AGG to perform similar to
> GD?
> 
> John
> 
> 
> 
> Rob McCulley-2 wrote:
>> That hasn't been my experience at all.  I wrote a quick little python
>> script that used mapscript to produce the same five maps twice, once
>> with agg and once with gd.  The mapfiles were identical, except for the
>> outputformat section, and the times were within 2% of each other.
>>
>> Rob McCulley
>>
>>
>> -----Original Message-----
>> From: UMN MapServer Users List [mailto:MAPSERVER-USERS at LISTS.UMN.EDU] On
>> Behalf Of John Cole
>> Sent: Thursday, August 23, 2007 12:56 PM
>> To: MAPSERVER-USERS at LISTS.UMN.EDU
>> Subject: [UMN_MAPSERVER-USERS] GD vs AGG performance...
>>
>> Perhaps it's too early to discuss performance differences, but I was
>> curious
>> if the current AGG rendering speed is what to expect.
>>
>> Looking at a few of my mapfiles, AGG is aprox 5 times slower than GD on
>> MS5
>> (with MS4 GD just a hair faster than MS5 GD).
>>
>> The maps look great, but after spending a lot of time getting maps below
>> 100ms, it's hurts to see them over a second again :-(  Just want my cake
>> and
>> get to eat it too :-)
>>
>> Thanks,
>>
>> John
>> -- 
>> View this message in context:
>> http://www.nabble.com/GD-vs-AGG-performance...-tf4319265.html#a12299688
>> Sent from the Mapserver - User mailing list archive at Nabble.com.
>>
>>
> 

-- 
--------------------------------------------------------------------
Andreas Albarello
Analysis & SW Development

Territorium Online srl/GmbH
Via Buozzi/Buozzistraße 12 - I 39100 Bolzano/Bozen
email: andreas.albarello at territoriumonline.com
web:   www.territoriumonline.com
--------------------------------------------------------------------