When enough is enough (CPU power)

[zeronyne]

So I'm building a new PC to play Doom 3 (and eventually, Gigasampler), and it got me to thinking about my studio rig.

 

I'm using a dual 1.8 gig G5 for Final Cut, After Effects and 3d animation, a dual 1 gig Quicksilver G4 for Logic, Reason, Live, Dp, etc., and I have a Sawtooth G4 upgraded to 1 gig in my proper office for music also.

 

So I pick up the phone to order the new dual G5, and I stopped myself. For the first time in my life as a musician, I did not feel compelled to upgrade my main music box. I can get over 100 audio tracks playing in Logic at once (I use UW SCSI). At a more reasonable 48 tracks, I can easily get all the plug in effects going that I want. And with the new track freeze function, softsynth limits are a thing of the past.

 

I never feel hampered by rendering times or jittery graphics, and even the CPU hogs like the CS-80V or any of the convolution reverbs just make me think a bit before using them.

 

So I've finally reached the point where having the additional power is not worth the heartache of configuring yet another system. I'm sure this is even more true for Wintel folks, who could, if they were insane enough, upgrade every other month.

 

I've always gone by the first rule of small business: "Do not buy any overhead until it is absolutely clear that you are losing money without it". I have hit that limit, and I am a benchmarking hound, so it's quite significant. It's a good time to be a computer-based musician, folks.

 

I told my wife this today, and she looked like she was staring at a replicant of her husband. she said she was proud of me that I wasn't just buying a new computer for he hell of it.

 

Of course, I didn't tell her about the stuff from NewEgg that's coming tomorrow...mmm....Doooooommm.

[Alndln]

Originally posted by zeronyne:

Wintel folks, who could, if they were insane enough, upgrade every other month.

Well,it's it's cheaper for us for one thing,and putting a new box together is easier than assembling a child's toy,for me anyway,but I feel the same way as far as not rushing to upgrade to the latest and greatest just because it exsists.Maybe if I ran a commercial studio and actually needed all the power i could get,but that's not the case for me.In fact I still use my AMD T-Bred B 2800 and 2400 and have no real lack of power(for my projects which avereage 20-30 stereo tracks/a few VSTi's and plugs),or any stability issues whatsoever.My machines still feel new to me.This wasn't always the case though,for either platform.I think technology has caught up enough to give us somewhat of a breather,but who knows what's around the bend.

[gtrmac]

Originally posted by zeronyne:

I've always gone by the first rule of small business: "Do not buy any overhead until it is absolutely clear that you are losing money without it". I have hit that limit, and I am a benchmarking hound, so it's quite significant. It's a good time to be a computer-based musician, folks.

I'm embarressed to tell you what I'm using! My computer is over three years old and I don't have any compelling reason to upgrade. My projects are usually limited to 24 to 32 tracks and I use VSTi like Halion and LM-4 but I always have plenty of CPU headroom. And I'm still using Windows 2000 to boot (pun intended). I'll probably upgrade at the end of this year just for the hell of it though. I plan to update my software and add some VSTi and plugins and I guess it's just about time for it. I am also expecting a project which will finance the upgrade with money left over but who knows I may still hold out and follow zeronyne's First Rule.

[Geoff Grace]

Originally posted by Alndlnbot:

I think technology has caught up enough to give us somewhat of a breather,but who knows what's around the bend.

How about this? As someone who currently has G.A.S. for the East West/Quantum Leap Symphonic Orchestra library , I'm currently considering the huge host system recommendations from the June 2004 Sound On Sound Review:

 

Posted at SoundOnSound.com (eSub required for access until 2005):

In the US, East West (in conjunction with music PC company Vision DAW) are offering full computer systems designed to run EWQLSO, and these machines feature a 2.8GHz Pentium 4 processor, 2GB of DDR433 memory, and a 36GB Raptor drive for the sample data. These type of specs are certainly more commonplace these days for dedicated digital-audio or sampling workstations, and give a better idea of the specification East West feel is appropriate for their library.

 

I'd also recommend at least two computers dedicated to the library to get the most out of the sounds. You could run Strings and Percussion on one, and Brass and Woodwinds on the other a good idea given that the strings and brass are typically the most used sections in orchestral writing, especially in Hollywood! In an ideal world, though, the best performance would easily be obtained by having four computers, so you could dedicate each orchestral section to its own computer. This doesn't mean that single-computer users won't get much out of the library, it just means that you might be doing rather a lot of bouncing!

Later in the review it becomes apparent that authors Dave Stewart & Mark Wherry are recommending that users consider 2-4 computers in addition to the computer that hosts their DAWs:

 

At 5.41GB, EWQLSO's percussion is the smallest of its four volumes. According to Doug Rogers, power-users who run the library on multiple computers generally find they can install the percussion on the same system as their audio sequencer.

Why is so much power required? Here's why - this library contains 24 bit, six channel surround samples of each orchestral instrument!

 

Well, I guess it's still much cheaper than hiring a real orchestra!

 

Best,

 

Geoff

[Mark LaCoste]

Zerodyne, it's great to hear someone say this. I haven't upgraded in a while myself. Still running an AMD 2100+ (abut 1.7 gHz), I am tempted to upgrade, but am working around any limitations I find. Still, I have a question on this subject, and since both a Mac and a PC guy have posted, this may be educational.

 

For the past year or more I tried to record direct to disk, using DX and VST effects in real time. I tried to imagine that latency was not an issue. My track count before dropouts was not stellar. Two weeks ago I gave up on real time 'in the box' effects, plugged my preamps into the digital mixer, and started using it's effects during tracking. Not only did this free up processing for DXi's, but it sounds better in the headphones without that flang-ey feeling between my voice and the monitor signal in the cans. My system can run at the hammerfall's minimum latencey, but the delay is perceptable. This would not get any better even if I upgraded to systems like yours, Zerodyne and Alndlnbot... Right?

 

I guess I'm wondering if you would feel like upgrading if you had a lowly, single-CPU system running at 1.7 G. Do you guys monitor through your soundcards while recording, or do you use zero latency monitoring through your mixer or audio interface?

[Alndln]

Originally posted by Geoff Grace:

Why is so much power required? Here's why - this library contains 24 bit, six channel surround samples of each orchestral instrument!

With a large orchestreral sample library your dealing with a lot of polyphony and multiple streams at once which not only eats CPU,but puts a lot of strain on your disk which in turn ets even more CPU.If I myself were get build a machine for a large Giga library,I'd go with an Athlon FX or 64(not dual Opteron) hands down,because those chips eat a lot less CPU at low latencies than anything availible.2 FX or 64 machines linked by FX Teleport with 4 scsi drives(or Raptors when their big enough)should make a nice hot rod for large Giga library's.Thankfully for me i have no such desires.

[Geoff Grace]

Thanks Alndln!!

 

Originally posted by Mark LaCoste:

Zerodyne, it's great to hear someone say this.

Mark, I'm sorry about your cold!

 

I'm sure once it clears up, you'll be able to write "zeroNyne" (that's "0 - 9") again!

 

Best,

 

Geoff

[realtrance]

I'd hold off on upgrading just a little bit. The new Intel 915P chipset, supporting PCI-X (AGP 16X, as opposed to a maximum of AGP8X, for example), the new DDR2 RAM (can be pushed currently to 600MHZ), the cooler and more efficient Intel P4 LG packaging, the 6800 and X800 Pro video cards (supporting PCI-X from Nvidia and ATI, respectively), USB 2.0... 802.11g... there's a host of stuff that's going to be standard that is currently available at a premium and will come down dramatically in the next six months, that's worth waiting for.

 

rt

[zeronyne]

zerodyne sounds cool! Sort of like the company that finds the Terminator's hand.

 

To answer your question, Mark, I use zero latency monitoring. I personally cannot stand even a little bit of perceptible latency, so I route through my MOTU 424 card's near-zero latency feature. I don't think we're at the stage where monitoring with plug ins while recording is a viable thing, but someone here may have some uber-computer based DAW that can do that.

 

And Geoff, now you have given me the perfect impetus to buy 3 more Athlon 64 boxes...although the last time I recorded any orchestral simulations for money, the Proteus 2 was the new big thing!

[Geoff Grace]

Originally posted by zeronyne:

And Geoff, now you have given me the perfect impetus to buy 3 more Athlon 64 boxes...although the last time I recorded any orchestral simulations for money, the Proteus 2 was the new big thing!

Oh, your wife's gonna be real happy you read my post!

 

Best,

 

Geoff

[Stephen Fortner]

I know whatcha mean about not feeling the need to upgrade. The want, certainly, but not the need. My main machine is a dual 1GHz Quicksilver G4 with 2GB of memory and a IDE Western Digital "caviar" drive with the 8MB cache for audio (in addition to the system drive). I'm running DP4 and Cubase SX 2, and get everything done I need to, smoothly and quickly. I don't work at 88.2 or 96K a lot; perhaps I'd see the need if I did.

 

I have a secondary machine for soft synths that I got as a gift: a Toshiba P25 series 17" laptop with a 2.8 hyperthreading P4, so that takes some load off the Mac. In fact, I feel that in 2-1/2 years, I have yet to max out my Mac and feel a "business need" to upgrade. I'm sure it'll happen, just not yet.

 

Now, the dual-Opteron neKo I had in here for review (go buy the current issue of Keyboard and keep me working, everyone!) was just freakin' absurd: huge gobs of power on demand. BTW, if you ever are watching that sitcom with Method Man and Redman on Fox, and see a neKo in a shot, that's the one I had!

[Geoff Grace]

Originally posted by Stephen Fortner:

Now, the dual-Opteron neKo I had in here for review (go buy the current issue of Keyboard and keep me working, everyone!) was just freakin' absurd. BTW, if you ever are watching that sitcom with Method Man and Redman on Fox, and see a neKo in a shot, that's the one I had!

I read that review, you lucky man! Nice job!

 

BTW, that very same issue of Keyboard also has reviews for the also unique but expensive Haken Continuum, the very same East West/Quantum Leap Symphonic Orchestra I mentioned above, the Korg Legacy Collection, FXpansion BFD, and Arturia Minimoog V, to mention but a few. Now that's a lineup of great gear! Interestingly, only the East West/Quantum Leap Symphonic Orchestra and the Korg Legacy Collection received the coveted Key Buy awards! Man, whoever says that Keyboard hands out those things too easily ought to take a look at the current issue!

 

Best,

 

Geoff

[TheFunkman]

How about things on the video side? Final Cut is moving into the Hi Def realm (I don't know whether that's hi-def compressed or uncompressed), are there any hard-drive throughput or storage issues because of your video demands?

 

P.S.: Speaking of "computer power," my day-job business sent a large delivery of computer automation equipment to a TV station. Things got pretty tense when some of the employees figured out that the gear they were getting would mean the end of their jobs. I watched the same thing happen in the 90's with radio stations: now Moore's Law has caught up to the television industry.

 

http://www.computermuseum.li/Testpage/UNIVAC-1-Census.jpg

[not coaster MODERATOR]

Originally posted by Mark LaCoste:

Zerodyne, it's great to hear someone say this. I haven't upgraded in a while myself. Still running an AMD 2100+ (abut 1.7 gHz), I am tempted to upgrade, but am working around any limitations I find. Still, I have a question on this subject, and since both a Mac and a PC guy have posted, this may be educational.

 

For the past year or more I tried to record direct to disk, using DX and VST effects in real time. I tried to imagine that latency was not an issue. My track count before dropouts was not stellar. Two weeks ago I gave up on real time 'in the box' effects, plugged my preamps into the digital mixer, and started using it's effects during tracking. Not only did this free up processing for DXi's, but it sounds better in the headphones without that flang-ey feeling between my voice and the monitor signal in the cans. My system can run at the hammerfall's minimum latencey, but the delay is perceptable. This would not get any better even if I upgraded to systems like yours, Zerodyne and Alndlnbot... Right?

 

I guess I'm wondering if you would feel like upgrading if you had a lowly, single-CPU system running at 1.7 G. Do you guys monitor through your soundcards while recording, or do you use zero latency monitoring through your mixer or audio interface?

the key for me is that my preamp has 2 outs. one sends a dry signal to logic, and the other goes to my outboard mixer with outboard effects. so i can hear real-time no latency effects while tracking ( or even rehearsing) and print the dry.

[Mark LaCoste]

I'm feeling much better now and can type Zeronyne with the best of them. Also, I feel a strange non-desire to upgrade my CPU and system board.

[Geoff Grace]

[Alndln]

Originally posted by Mark LaCoste:

Still running an AMD 2100+ (abut 1.7 gHz),

For the past year or more I tried to record direct to disk, using DX and VST effects in real time. I tried to imagine that latency was not an issue. My track count before dropouts was not stellar.the delay is perceptable. Do you guys monitor through your soundcards while recording, or do you use zero latency monitoring through your mixer or audio interface?

I depend on live input of plugs all the time,but I monitor through my software(Sonar)and not my cards mixer at 2.9 ms,and since it's through Sonar which compensates for any delay,there is no delay.As I said before,I have 2 machines,a 266FSB 2400 T-Bred B and a 333FSB 2800 T-Bred B so were not talking that much of a leap over your 2000 TBred(or Palomino).I mostly run a reverb plug through the auxilary at all times,and only run select and needed plugs on live input such as a delay for the vocalist ot Amplitube for guitar,or a GRM plug that's needed for performance ect.,I also use Sonars version of freeze for most of my VSTi's(render/archive)so I'm only running 2 or 3 plugs on live input at a time(EQ's/Comps ect. aren't really neccessary at that stage.My projects run anywhere from 10 or 32 or so stereo tracks.By the time i'm ready to mix my VSTi's are done and I can raise my latencey to about 30-40 ms for mixing so I can load up on the needed plugs for that(EQ's comps ect.).For convolution reverb I clone my track,and apply it full wet on the clone and use the clone as a wet send,not just to save CPU in general but because a lot of IR's cause CPU spikes.When all is said and done even my 2400(2.0gz)is more than enough for the way I work.Although I was all ready to jump into an FX or Opteron setup,I held back because everythings just fine for the way I work,but if I needed a ton of live input plugs or run an orchesteral library I consider upgrading,and to the 64/FX/Opteron family only,mainly because that family of chips are uneffected by low latencey v/s CPU hit.I'm sure a G5 would be very similar in that regard on a 64 bit OS running a 64 bit app(new Logic?).If you don't quite need mega power an Athlon 64 upgrade is quite cheap these days.

[the stranger]

Good thread.

[kylen]

I'm gonna need to upgrade everything was fine at 24bit 44.1KHz but...I just began working at 32bit float 96KHz and it's beginning to sound like I can almost reach out and touch the performers. Does that mean I'll have to go to 192KHz next...when does it get real enough - when I can feel the hot breath of the performers ?

 

It takes computer power to get there but I'm still thinking about that one...maybe everything doesn't have to be so pristine. How about replicating the old Shure Brothers P.A. with the overdriven horn in the high school gymnasium - that sure sounded good to me at the time