The aliasing thread

[WilliamK]

Is this CPU % for a single voice @ 44.1kHz? Can't be, that's really slow! Also, what sort of CPU are you testing on?

Please, read my original post. Those tests are for 128 voices.

Wk

[Scr1pt3r]

WilliamK, I thought about what you suggested as to comparing CPU resources and I think its a good idea.
The reason is that for example this thread discusses about aliasing and therefore developers learn and change their imlementations, a CPU comparing thread could also make some developers to change their implementations of some VSTi.
Why dont you run on Wusik.com a page describing the CPU precentage and put its aliasing picture attached to it ?
That would make good comparison if on the same amount of aliasing different samplers use different amount of CPU usage.

I suggest you run it on Wusik.Com as I doubt george is gonna run it on his page.

[george]

I find this CPU comparison pointless since (if I'm not mistaken) Jeskola XS-1 beats all of them.

[pough]

I find this CPU comparison pointless since (if I'm not mistaken) Jeskola XS-1 beats all of them.

Does that mean your sampler comparison page is pointless because sfz beats all of them?

[Scr1pt3r]

I find this CPU comparison pointless since (if I'm not mistaken) Jeskola XS-1 beats all of them.

What an illogical remark George
And does the aliasing comparison is pointless because highlife beats almost all synths when its rendered offline ?

Ok developers , pack your bags and go home, High life beats all of you.

[george]

Does that mean your sampler comparison page is pointless because sfz beats all of them?

Aliasing has been already discussed here very well and proved how necessary is by developers and users, so I'd agree with you.

CPU usage is quite easy to understand, user side speaking. Aliasing is not.

[stefancrs]

Does that mean your sampler comparison page is pointless because sfz beats all of them?

Aliasing has been already discussed here very well and proved how necessary is by developers and users, so I'd agree with you.

CPU usage is quite easy to understand, user side speaking. Aliasing is not.

They would probably understand a cpu usage / noise ratio though

I'd like to bump what I wrote before, would like to get some feedback on this idea from for instance George, René, Dave and other (imo) highly regarded developers.

So here it goes again:
What about "converting" the sample to contain the integrated value (per sample) instead of the current value (per sample)? Then one could calculate those integrated values with high precision and then downsample from "infinite" oversampling using a box filter (since you'd just take the average of the output since the last output sample). Ofcourse reading from the sample would use interpolation of some sort, and with some oversampling as well one could probably reduce the aliasing that occurs due to using box filtering...

[george]

I find this CPU comparison pointless since (if I'm not mistaken) Jeskola XS-1 beats all of them.

What an illogical remark George
And does the aliasing comparison is pointless because highlife beats almost all synths when its rendered offline ?

You seem to miss how Jeskola XS-1 has been overlooked in William's "CPU comparison" (I'd better call that WusikStation propaganda). However it seems you prefer to focus into my statement instead that omission.

Ok developers , pack your bags and go home, High life beats all of you.

Talking about illogical remarks...

[René]

When you've downloaded them let me know so I can take them down. I don't want the rest of the muon website to suffer bandwidth-wise by serving up a 16mb file to thousands of curious aliasing-analysts

I've got the files, you can safely delete them.

I couldn't honestly perceive any difference between the two files. However, I find the test does only represent a very particular case: there's never an exposed piano note, the piano sound doesn't have any frequency contents above 5kHz nor below 100Hz, and the SoundFont isn't transposing it more than a few semitones. What's the SoundFont? I'd like to check it out.

I'm trying to reproduce WilliamK measurements. Please anyone enlighten me, how one does load a wav file into Wusikstation?

Regarding the CPU usage table, I think it would be cool to have one. However, it's not too easy to perform: the test for ALL contenders should be performed in the same computer, and we would need several computer configurations: SDRAM, DDRAM, AMD XP, P3, P4, etc. We should also set all configurations with the same OS and Host, and make every background service in the OS match, disable network cards, turn off internet connections, having identical audio and video boards and drivers, etc. Really a megatask, dign of ZD labs.

-René

[mauseoleum]

< brave tag > you have my best wishes, may the Gods of Sonic Excellence guide you well on this quest < / brave tag >

[Scr1pt3r]

You seem to miss how Jeskola XS-1 has been overlooked in William's "CPU comparison" (I'd better call that WusikStation propaganda).

You're right , I didn't notice that.
William , where is the Jeskola XS-1 CPU comparison ?
I hope it doesn't beat Wusikstation because that would make me conclude you left it out on purpose.

[george]

Regarding the CPU usage table, I think it would be cool to have one. However, it's not too easy to perform: the test for ALL contenders should be performed in the same computer, and we would need several computer configurations: SDRAM, DDRAM, AMD XP, P3, P4, etc. We should also set all configurations with the same OS and Host, and make every background service in the OS match, disable network cards, turn off internet connections, having identical audio and video boards and drivers, etc. Really a megatask, dign of ZD labs.

I couldn't agree more. That measurent is also very easy to do for anyone having a host including a CPU meter tho. Alias analisys is a bit more... laborious

[Scr1pt3r]

When you've downloaded them let me know so I can take them down. I don't want the rest of the muon website to suffer bandwidth-wise by serving up a 16mb file to thousands of curious aliasing-analysts

I've got the files, you can safely delete them.

I couldn't honestly perceive any difference between the two files. However, I find the test does only represent a very particular case: there's never an exposed piano note, the piano sound doesn't have any frequency contents above 5kHz nor below 100Hz, and the SoundFont isn't transposing it more than a few semitones. What's the SoundFont? I'd like to check it out.

I'm trying to reproduce WilliamK measurements. Please anyone enlighten me, how one does load a wav file into Wusikstation?

Regarding the CPU usage table, I think it would be cool to have one. However, it's not too easy to perform: the test for ALL contenders should be performed in the same computer, and we would need several computer configurations: SDRAM, DDRAM, AMD XP, P3, P4, etc. We should also set all configurations with the same OS and Host, and make every background service in the OS match, disable network cards, turn off internet connections, having identical audio and video boards and drivers, etc. Really a megatask, dign of ZD labs.

-René

Does it realy need to be that accurate ?
I mean, if I'll run those tests on my machine it should be relatively the same on all the machines ( And I'm not talking of comparing SSE supported machines with those who dont... )

I'm running on a 2.6Ghz HT machine, so could run those tests if needed.

[WilliamK]

You're right , I didn't notice that. William , where is the Jeskola XS-1 CPU comparison ? I hope it doesn't beat Wusikstation because that would make me conclude you left it out on purpose.

Sorry guys, but I never used Jeskola, didn't even know about it. So it was a mistake only, not by purpose. I have to check it.

George, this is not Wusikstation propaganda. If it is, your page is propaganda to SFZ, as it shows a non real-time graphic right on the top. But don't take it the wrong way ok? I'm just pointing something up.

One thing I agree, the RAM speed could pose a problem on making a CPU ratio. But since I use Cache Prefetch on my code, any RAM speed would get the same ratio with the same processor. I'm impressed that few VSTs does that, as it only introduces one sample delay...

Wk

[stefancrs]

Come on. Please notice my post