All times are UTC+02:00




Post new topic  Reply to topic  [ 7 posts ] 
Author Message
 Post subject: CPU
PostPosted: Sun Apr 21, 2013 3:23 am 

Joined: Mon Feb 04, 2013 11:25 pm
Posts: 153
Will ST with composite clipper run on a Core 2 quad Q 6700? I notice several post where users are running core 2 duo (probably clocked).I have a eng friend wanting to set up a box on the cheap.thanks


Top
   
 Post subject: Re: CPU
PostPosted: Sun Apr 21, 2013 3:43 am 
User avatar

Joined: Tue Mar 17, 2009 2:56 pm
Posts: 4229
Right now i am running Full FM preset on E5400 slighltly clocked from 2,7 to 3.0 with new clipper strictness at 18 with Declipper ON.
Q6700 - http://ark.intel.com/products/30790/
E5400 - http://ark.intel.com/products/40478/
That CPU is 2.66GHz but also with 8MB more cache and higher FSB, also 65nm so more heath.


Top
   
 Post subject: Re: CPU
PostPosted: Sun Apr 21, 2013 3:44 am 
Site Admin
User avatar

Joined: Mon Mar 17, 2008 1:40 am
Posts: 11425
http://www.cpubenchmark.net/
Look at this, and calculate the performance per core. ST uses only 2, if each core is fast enough it should work.


Top
   
 Post subject: Re: CPU
PostPosted: Sun Apr 21, 2013 3:52 am 

Joined: Mon Feb 04, 2013 11:25 pm
Posts: 153
http://www.cpubenchmark.net/cpu.php?cpu ... 40+2.66GHz

Looks like it would work fine.


Top
   
 Post subject: Re: CPU
PostPosted: Sun Apr 21, 2013 12:24 pm 
Site Admin
User avatar

Joined: Mon Mar 17, 2008 1:40 am
Posts: 11425
Q6700: 3360, divide by 4 cores gives 840
E5400: 1606, divide by 2 cores gives 804, but it's overclocked to 3 GHz so *3/2.7 = 893.

Of course, the CPU is not the only thing that affects the total execution speed - memory speed is also important, and CPU's might behave differently due to different cache sizes etc. This benchmark is only an indication.

I actually would never have expected it to run on that E5400... My laptop is an i7 2630QM (2 GHz):
i7-2630QM: 5643, divide by 4 cores gives 1411.
Which is quite a lot more - I should be able to run it even in single core mode based on this. But I can't. On the other hand, in multicore mode I have enough headroom to push the declipper quality settings close to the maximum, which has quite a big effect on the CPU load.


If latency is not an issue and you're running with Latency set to 4096 samples, you can always reduce the Quality slider quite a lot - at high latency settings this has very little effect on the audio. If you reduce Quality to - say - 50%, the CPU load is 75% of the normal CPU load, which helps a lot.


Just for comparison: My old Pentium 4:
Intel Pentium 4 2.40GHz: 241


Top
   
 Post subject: Re: CPU
PostPosted: Sun Apr 21, 2013 3:31 pm 

Joined: Sun Feb 10, 2013 6:58 am
Posts: 152
Location: Sydney, Australia
Using your guide to the PassMark figures:

I am running the software on an Intel Q9650 which rates 4,314 or 1078.5 per core. I want to build a rackmount PC for the software and was looking at putting in an i3 3220 which rates 4,239 or 2,164.5 per core (only 2 cores). An Intel i7 3770K rates at 9,630 or 2,407.5 per core.

As you're saying ST SA only uses 2 cores, is the i3 3220 the better choice here?

Thanks,
Clipper


Top
   
 Post subject: Re: CPU
PostPosted: Sun Apr 21, 2013 9:57 pm 
Site Admin
User avatar

Joined: Mon Mar 17, 2008 1:40 am
Posts: 11425
The i7 is still slightly faster per core! But probably a lot more expensive too. By the way, one thing that makes these comparisons very difficult and unreliable is Hyperthreading - if a PC has hyperthreading it will score higher in these benchmarks but it will have nearly no effect on Stereo Tool.

Now, if the price difference is small I would probably go for the i7 since there's more room for future improvements. No idea if ST will ever use it though.

I heard from someone who is using the composite clipper on a Q9650, he says it's running properly. But he does have a lot more latency (8 ms) due to the longer time needed to perform all the calculations. I might - maybe - add a 4 core version some day to reduce this effect (not planning to do this right now though). On a faster CPU the effect will be much less than these 8 ms. And even if I would do this, it won't help for most other filters (most filters use 1 core per channel, and there are only 2 channels in a stereo signal).

There's another thing that I want to do which reduces artifacts when using lower latency settings, which will cause the CPU load (and number of cores used) to double. But I don't know if I'll ever add that to the software version.


Top
   
Display posts from previous:  Sort by  
Post new topic  Reply to topic  [ 7 posts ] 

All times are UTC+02:00


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
Powered by phpBB® Forum Software © phpBB Limited