Proposed Benchmark testing for T2

[bk]

Is it the 64 bit mixing engine?

Nope. I got nearly identical results with 64 bit on or off.

[bk]

I'm not sure if the comments about t1 being more efficient then t2 are completely fair.

I tried it on my setup and I did get apr 10 more tracks on t1 than t2, but only when using native plugins.

I tried using the free classic plugins (so anyone can try the same) and there was no difference at all between t2 and t1.

After further messing around it seems pretty clear that the native plugins take up quite a bit more cpu in t2 than T1.

To me this doesn't matter because I never use the native effects, but this could be why some people are seeing a performance difference between the two versions!

This is a very good point. It had occurred to me, but I never got around to testing it. Thanks for doing that. I'll do some tests of my own to confirm, but I already feel better about T2!
Frankly, I was wondering why I never seemed to have cpu isssues when switching to T2, but that makes sense now. (I don't typically use the native plugs either)

[zOap]

Heres my results:

39 tracks, 24/44.1 with the above plugins.

My system:
Win XP Home SP2
Pentium HT 3.0
1,5 GB RAM
DEll Mobo(8300)
2 200 8MB Cache drives
I have done some tweaks, getting rid of unwanted graphics and some hd tweaks. I've also affiniated most of my system to run on cpu 0, while T is running both cpu 0 and 1.

[polaris20]

Heres my results:

39 tracks, 24/44.1 with the above plugins.

My system:
Win XP Home SP2
Pentium HT 3.0
1,5 GB RAM
DEll Mobo(8300)
2 200 8MB Cache drives
I have done some tweaks, getting rid of unwanted graphics and some hd tweaks. I've also affiniated most of my system to run on cpu 0, while T is running both cpu 0 and 1.

Which plugs did you use, the Kjaerhus plugs? I'll have to try that again with those.

[zOap]

Heres my results:

39 tracks, 24/44.1 with the above plugins.

My system:
Win XP Home SP2
Pentium HT 3.0
1,5 GB RAM
DEll Mobo(8300)
2 200 8MB Cache drives
I have done some tweaks, getting rid of unwanted graphics and some hd tweaks. I've also affiniated most of my system to run on cpu 0, while T is running both cpu 0 and 1.

Which plugs did you use, the Kjaerhus plugs? I'll have to try that again with those.

No, the first ones, eq, reverb, delay, Chorus, Compressor, Volume/Pan

[polaris20]

Redoing the test, but instead I wanted to remove the reverb, and go to an additional delay, to more closely replicate the PTLE/DaveC test. So I removed the T2 reverb, left the others as is, and then added an MDA delay.

Athlon 2500+, 1gig of RAM: 27 tracks, still in the blue, no freezing.

Reverb, at least the stock T2 effects of it, really saps the hell out of the cpu!

EDITstill exactly 4 more tracks with 1.6, FWIW. But if I can get 27 tracks out of T2 without reverb, and maybe using the freeze function when I do use reverb, it's not that big a deal. I imagine I can really get a boatload of tracks using the freeze function on completed tracks.

[polaris20]

Another thing; running the original test, but with freezing, I got 40 tracks, and counting. I gave up at that point, because it didn't seem there was an end in sight.

So I am uber happy with that. I don't mind freezing when using tons of tracks of reverb, because I am usually not heavy on reverb anyway.

I am going to download and try the demo of FX Teleport though, because that might be really cool.

[bk]

Here's an update from me:
I switched out the native Tracktion plugs with similar vsts that I typically use on a day to day basis, and got identical results in both T1 and T2.

Engaging 64 bit mixing gave me one less track in T2.

This has been pretty enlightening for me. I've been able to pinpoint some cpu issues with vsts that I was using, and substitute ones that sound just as good, but use considerably less cpu.

I own GlissEQ, but was often grabbing and using Posifopit EQ because it was more conveniently placed in my plugin folder. Wow. Huge difference in cpu usage of those plugs. 20 or so tracks of Posifopit really chews the cpu, while Gliss barely dents it.

I got similar results between Crunchessor and a lttle known freeware comp called "Vanilla". (I still like the sound of Vanilla, just probably won't reach for it as often).

I'm not trying to turn this into a voxengo ad, just relating my personal experience. This has been quite eye-opening for me. Thanks for getting me thinking, Polaris.

[valley]

a lot of smaller developers can't afford the heavily optimized commercial compilers. Even identical code may well run slower when compiled with some freeware, or lite compiler versions.

Factor in the fact the commercial developers have a little more incentive for spending the time to convert often repeated sections of code into native assembly language, and it is little surprise that at least some freeware plugs aren't quite as efficient as those of more established developers.

Be glad that people make freeware plugs, and if you use them a lot, send the dev some money. Only that way will they be motivated to purchase the (horribly expensive) tools, and invest the time needed to optimize their work.

[polaris20]

a lot of smaller developers can't afford the heavily optimized commercial compilers. Even identical code may well run slower when compiled with some freeware, or lite compiler versions.

Factor in the fact the commercial developers have a little more incentive for spending the time to convert often repeated sections of code into native assembly language, and it is little surprise that at least some freeware plugs aren't quite as efficient as those of more established developers.

Be glad that people make freeware plugs, and if you use them a lot, send the dev some money. Only that way will they be motivated to purchase the (horribly expensive) tools, and invest the time needed to optimize their work.

I am going to school for my 2nd BS in Computer Science, and I'll be taking a crack at VST's. Hopefully I can do something cool and useful.

[valley]

Hey good luck with that!

I'm starting a part-time BS in spring. Dunno quite how I'm going to schedule it around my work hours but we'll see...

[polaris20]

I'd really love a Mac user to give this a shot (the test archive I did), because I'd like to see how it compares to WinXP/2K.

[rockstar_not]

Oh woe is me.

I just finished setting up my brand spanking new PC

Asus A8V Deluxe MOBO
Athlon 64 3000+ processor
1GB DDR400 RAM
XP Professional
120 GB SATA Hard Drive
Gigabyte FX5200 8x AGP video card.

Tweaked XP for audio as suggested here:
http://www.musicxp.net/tuning_tips.php

Installed the same sound card as was in my last PC, which was an AMD Thunderbird 1.6 GHz Athlon, 256 MB Ram Win98SE

Thought I had the same latency settings etc,

and pulled up T2 'Tesserract' and I couldn't believe it, but it was popping and crackling and I was seeing regular CPU spikes over 50%.

This used to run clean on my old box with T1.6

I was so depressed, I just turned off the machine and went to bed.

-Scott

[zOap]

There seems to be alot of versions out there of the test as for now. It would be very nice if we all could agree of a united test, because at the moment I don\t know what to read out from everybody elses results.

I think the original version is good, we all use reverb now and then so a good test should have one, yes it sucks the life out of any cpu, but that doesn\t say that we don\t need one.

To polaris who had 40 tracks with freezes: Was the cpu metre still in blue when you got the freezes? Cause I can fill up my comp too with lots and lots of tracks while it freezes, but then it will be useless? Won\t it?

[polaris20]

To polaris who had 40 tracks with freezes: Was the cpu metre still in blue when you got the freezes? Cause I can fill up my comp too with lots and lots of tracks while it freezes, but then it will be useless? Won\t it?

Yes, it was still in the blue with the freezes. Why would it be useless? If you need to further edit that track, unfreeze it, do what you gotta do, then re-freeze it. That's the whole point of freezing tracks.