TR Forums

Before I start I just want you to know I am familiar with CreativeCow and the Adobe forums. I've had some help there. But I find you guys to be the most up-to-date on your info and your recommendations. I know most of the builds are for games but I've read excellent info on workstation builds too.

Here's what I know so far:
- I'm slightly nervous about using expensive parts for my first build, but I can't afford to build a "practice" machine. I have an Asus P4SP-MX (478) that my parents gave me when it died. I never tried to get it working; I just studied it, familiarized myself with the landscape inside a case, learned what I could from downloading the Asus manual...
- I will be buying parts this week. There's no real rush so I can take my time but I'm ready. I have the money and my budget is $2500.
- I am 99% sure I'm buying Intel - the 1090T, 1075T are awfully tempting but I like Intel. I think the higher-clocked i7s beat the 6-core AMDs in most of the apps I care about. Plus I may very well buy that 980X monster...if I can convince myself it's okay to use a thousand dollar part on my first real build.
- I will be using Win 7 Pro with at least 12G RAM (After Effects loves RAM and CS5 is 64-bit so it can finally use it all).
- I will have at least one SSD for the OS and programs. (one thing I don't quite understand: to take advantage of those read speeds does that mean all my assets have to be on that drive too - while I'm working on them?
-I already own a good graphics monitor and about a TB of external storage.

I have to decide if I'm using server parts or not. My thinking is "8 cores are better than 4" but are they better than 6 when the 6 are the 980X?

USING DAILY: Adobe Production Premium CS4/5 (After Effects, Premier Pro, Photoshop, Illustrator, Flash, Encore, Soundbooth), Cinema 4D, Audacity (sound recording).
DOING: 3D animation in Cinema 4D, video (SD) conversion and rendering, Flash Video, HDR rendering in Photoshop, compositing in After Effects, moving to Premier Pro and out to Encore for various destinations: DVD authoring, iPhone, web, etc.

I do very little HD, if any. I'm not buying a dedicated capture card. Most of my work is in Cinema 4D, After Effects and Flash, while Bridge, Photoshop and Illustrator are always open for quick sketches and/or image tweaks.

I also do quite a bit of audio recording. It's not a primary concern but I don't want integrated audio - I need a good sound card.

So it seems like for what I do, "the more cores the better." That's why I'm thinking server parts - 2 quads as opposed to 1 Core i7 9xx. But besides having the capacity for multiple processors, do I have to use ECC memory then? Are the power requirements different as well on server boards? I haven't researched the server parts (obviously) very much.

I thought about the i7 930 and bumping it up but I don't think I should actually plan on overclocking for my first build.


There are 2 things I really hope you'll give me your opinion on:

ONE - is it stupid to use super-expensive parts on my first build?
I'm really only worried about applying the thermal grease correctly and getting the cooler attached.

TWO - should I go for 2 Xeon or that 980X? If I do a server board, what else does that commit me to? Special RAM? Special power requirements? Restrictions on audio?

THANK YOU VERY MUCH FOR ANY THOUGHTS YOU CAN SHARE! Seriously, I know that was a lot to read. But I'm finally serious about doing this and I finally have the money to do it. So as much hand-holding as you can do for me...I will be very grateful.

Thanks guys! Have a great day.

Or should I lower my sites and just buy this?

http://www.bhphotovideo.com/c/product/693860-REG/Rain_Recording_FR_SL2_P01_Solstice_Pro_Multimedia_and.html#features

It's not an i7, which I really want, and I'm not so sure a 9800GT - while it's prob good enough for what I'm doing - it seems like a bit of a ripoff, to pay $1800 and get a so-so GPU like that. Also, of course no SSD...

While I love the idea of spending $700 less and just walking down to B&H and carrying this thing home...I'm sick of being locked into a manufacturer's idea of what a workstation is. Probably for the same $1800 I could put together something nicer...esp since I already have storage space.

I don't know....help!

The Utility Player would smoke that $1,800 build you linked to. And that costs only $800 + OS + keyboard.

Even an similar Dell/HP should be better priced and include a warranty that's longer than 1 year.

Edit: are you sure you need such a beefy processor and 12GB of RAM?

I hear ya Wib - it's crazy to spend $1800 and get a $190 processor - I'm not buying that thing. I'm not sure what you're paying for on that machine...$1800 for what exactly? 1.5TB of 7200rpm storage? Win 7 Pro? A 9800GT?

Am I sure I need that much? well... I mean do I *need* it? Kind of, yes, because eventually (soon I hope) time will be money, and Cinema 4D and After Effects really love the mulit-core processing. And After Effects soaks up the RAM - when you're previewing your work, the more RAM the better. You really can't have too much.

As far as the 980X goes...I could certainly limit my ambition on that. But I do want to go with an i7 with its triple-channel memory. The throughput has to be better than the Lynnfields - at least I think so - when it comes to rendering 3D and processing filters on animated models... And since I'm not confident about overclocking I may have to spend the big bucks to get a super-fast processor...

If time's not money yet, I recommend doing a little more research before spending two grand on hardware. First, see how long these tasks take on your current hardware. Second, try to find some representative benchmarks on a review site that includes (something similar to) your current hardware and the high-end hardware you're considering. Finally, decide if the total time saved is worth the money.

I can see some scenario where it's useful for you to buy & build a $700 machine now, and then a super expensive one later once your busienss picks up and justifies the cost. With video editing, the upcoming Sandy Bridge/Bulldozer architectures may provide a big boost over the current machines. And in the future, if your business is dependent on working hardware, it's not a bad thing if have a ~$2500 primary computer and a $700 backup computer.

If time really is money for you in the future, I can appreciate your previous argument for buying a prebuilt system with a warranty.

Using a $700 build as your "dry run" to practice your system assembly skills is also a lot less risky than doing your first build on expensive high-end hardware.

well I've used the MacPro 8-core, I've had Celeron machines, P4s...laptops, etc. All of them pretty much suck for image/video editing and 3D modeling.

I know it's better to build a $700 machine first but once it's built it'll be useless to me. I have backup laptops, I don't need another slow machine. Of course I'll benefit from the learning experience, but I'd really rather not spend money on something I won't need in the end.

When you come down to it how hard is it to put a system together?

The hard part seems to be making sure your parts will work together. Besides applying the thermal paste there really aren't any other demanding, you-better-know-what-you're-doing tasks, are there? The rest looks like snapping in parts, routing cables, airflow considerations...installing an OS.

The way I look at it right now - and I'm open to suggestion - I have the money to buy the best proc out there - but that money won't last long if I wait days and days trying to decide. And it really won't last if I have to build a practice machine.

I'll do it if you guys think I have practically no chance on my first build - I'll take the $700 and build an econo box. But I'd really rather not, and if there's at least a 50/50 chance - with help online - that I can build a workstation - that's what I want, and need, to do.

canoli wrote:

I have backup laptops, I don't need another slow machine.

If your primary computer gets buggy for a week, are you sure your backup laptop will be strong enough to buy you time to fix it? I'm a little skeptical, if your tasks benefit from the $2500 hardware.

canoli wrote:

I have the money to buy the best proc out there - but that money won't last long if I wait days and days trying to decide. And it really won't last if I have to build a practice machine.

If that's the case, and you won't benefit from saving the money or spending it on other stuff (like a 30" IPS), then ignore my previous strategy.

No you're right, I said it wrong - I don't have a backup for image/video editing - I've "made do" on laptops - good ones like the Qosmio and a MBP for 3 years now. But no matter how good they are for laptops, they're still laptops. They were okay to learn on but now my ideas are getting too complicated for notebooks and rendering times are ridiculously long.

That's why I'm so hyped up to build a real workstation finally. I've got the money - I just need the nerve to plunk down 2+ grand on parts...

oh, btw, I've got a good S-IPS 24" and a still-going-strong (if you can believe it) CRT. So for displays and storage I'm okay.

What's at stake if your primary machine breaks and isn't functioning for a week? Is this an inconvienence, or a time-is-money catastrophe where you lose customers and lots of money?

DO you ever have long tasks that run several hours overnight?

Oh yes, lots of overnight renders.

Right now my income isn't dependent on my graphic skills, so being down a day or a week isn't the end of the world. But eventually I want to offer my photo clients a wide range of services - some are corporate, some are mom & pop & the kids - everything from wedding albums for couples/families to advertising production pieces for ...

I'm trying to future-proof myself as much as I can with this build. I don't have any clients waiting on me right now but once my portfolio includes interesting, cutting-edge title sequences / lower third, various 3D models, photo, slide show and DVD templates, web animations, flash games, whatever I can think of... I'll be offering my services everywhere.

But I can't get any of those ideas (or most of them) to the screen on these slow machines. I can't work for 3 minutes and wait 20 for a scene to render.

It may turn out I'm not very talented and all this will be a waste. On the other hand I may be able to produce some cool stuff. Do I need the top-of-the-line proc to find out? Probably not. I'm putting together a system on newegg right now and it keeps coming out to $3K with that 980X. So I'll prob swap that for the 950. I have to buy windows too unfortunately...so that's a few hundred right there. All the other software I have.

I'm also putting together an 875K build...much less money! and it gives me the chance to learn some overclocking.

At some point though soon -today, tomorrow- I want to stop thinking and start DOING!

I can't wait to get going on this...!!

canoli wrote:

I've got the money - I just need the nerve to plunk down 2+ grand on parts...

Oh, statements like this made me think this was some grant or corporate budget that had money that you had to spend before the end of the month. If you're spending personal money, I think a $800 + OS build for now makes even more sense. Especially if you too anxious right now to investigate benchmarks and relate them to you likely tasks. If the old/backup laptop renders a scene in 20 seconds, the 875k or Phenom X6 should complete quicker than it takes to sip some water.

Edit: whoops. see SNM's correction of my confusion about 20 seconds.

If you happen to live near a Microcenter, you may want to look at their in store pricing on CPUs -- they don't ship them but they seem to have great prices on certain models and you could save some decent money that way. (unfortunately the 980x is the msrp, but I believe they had a good price on the 875k and the 675k if you go that route...)

The TR guide to building is really excellent, just take your time and you will be fine

wibeasley wrote:

canoli wrote:

I've got the money - I just need the nerve to plunk down 2+ grand on parts...

Oh, statements like this made me think this was some grant or corporate budget that had money that you had to spend before the end of the month. If you're spending personal money, I think a $800 + OS build for now makes even more sense. Especially if you too anxious right now to investigate benchmarks and relate them to you likely tasks. If the old/backup laptop renders a scene in 20 seconds, the 875k or Phenom X6 should complete quicker than it takes to sip some water.

Pretty sure that was 3 minutes work, 20 minutes render...

But even a regular $200 i7 is going to kick the pants off your laptops in both speed and cores. At this point you could probably build a $1200 machine that'll lost you effectively as long as a $3000 one will. It'll be slower, but it'll still be in the same ballpark. Unlike your laptops.

Yes I know you guys are right about setting my sights lower. Compared to a laptop almost anything will feel swift and efficient. But don't think that's all I've ever used. My comparison - when I'm sitting in front of my new machine - won't be to a laptop, it'll be to my friend's 8-core MacPro that cost him his right arm and half his left leg.

If I came off as impatient and half-cocked I regret that impression; it isn't true. This is no "money's-burning-a-hole-in-my-pocket" thing; I've been reading and taking notes, posting in forums and asking questions for nearly a year.

I do still have a couple questions about GPUs in a workstation - whether I really need the Fermi architecture (or AMD's FirePro equivalent) or if I'll be fine with a $150 5770, but I'm modeling this build on the DoubleStuff. Running 2 monitors, I'd like to be able to calibrate both but the software I have right now (iOne display) only does one per GPU - it's a little twitchy doing two. I'm just not sure I need two GTX 460s, esp at that price.

I live in NYC and there's a MicroCenter in Long Island. It's a couple train rides, not crazy far but it'd take most of a day, there and back. I've looked at their proc prices and you're right -they had the i7 930 for $199 for the whole summer - don't know whether it went back up or not now. I wish the 980X would come down a bit but there's no point in waiting for that to happen I know.

The money for this build comes from a bonus that I earned (if a bonus is ever really "earned" but you know what I mean) but it was a very large bonus because the giver is a very generous person and because he and his wife want to see me create some cool stuff. I'd hate to blow it on Just Another Ho-Hum Quad. This feels like my one chance to get a really great processor. I'll probably never have a thousand dollars to drop on a processor again, certainly not anytime soon. I really want to make this money count, make something special...I figure either a dual processor setup or a 6-core. The whole thing is about ease of use, creating animations, SD video editing, HDR images, animated 3D model rendering - making it all as easy to do as possible. With that kind of work there's a whole lot of trial and error.

If the graphics folks had better things to say about the 1090XT I'd jump on it for the price. Unfortunately from what I've read it's nothing to rave about in the Creative Suite apps, or in Cinema 4D. It's very good, it's just not even close to "great." That's the impression I got anyway...

I know I'll regret buying a standard quad and feel it bog down when I open one too many images, or when I try to render a QuickTime movie out of After Effects and it seems only slightly faster than the Qosmio laptop I used last year. I want to be thrilled with this machine when it's done. I don't mind spending, I don't mind taking my time. I just want to be able to put some cool stuff together and not curse the machine while I work.

Sorry for writing a book here - after this is all about Parts, Specs, Compatibility. THANK YOU for bearing with me, and thank you for your replies.

If the graphics folks had better things to say about the 1090XT I'd jump on it for the price. Unfortunately from what I've read it's nothing to rave about in the Creative Suite apps, or in Cinema 4D. It's very good, it's just not even close to "great." That's the impression I got anyway...

doesn't seem to do too badly in Cinema 4D in TR's review:
http://www.techreport.com/articles.x/18799/12

I seem to remember Phenoms don't do so well in photoshop etc though.

While the cpu itself isn't that much cheaper than the i7s that it is bracketed by (in that cinema benchmark) you'd probably save a bit on the motherboard too.

This said you're almost certainly right to be going with intel.

Personally I'd consider building a cheap interim machine so I could get started and then build the monster when I'd got a bit of work coming in. That way you'd have a spare if your main machine failed and you'd have some parts to help with fault finding if anything did go wrong or was DOA. Plus you'd have had some practice before getting your hands on the really expensive stuff.

cheesyking wrote:

Personally I'd consider building a cheap interim machine so I could get started and then build the monster when I'd got a bit of work coming in.

Well that would be the smartest thing to do I'm sure because your point about learning from building a cheaper box is a good one. If I could do it for a few hundred maybe I would. But I can't drop 500, 600 on a practice machine. I'd much rather build the backup at my leisure once I've finished a few jobs, got my portfolio in order, etc.

Plus - the whole reason I'm doing this is because the work I want to do is too frustrating on slow machines. I can't stomach the thought of trying to work on a C2D or even one of the newer Clarkdales. Waiting till I get some jobs sounds good but I won't get any jobs if I can't produce a smart, interesting portfolio. I can do all my photography stuff on a slow machine but it's nothing a million other artists aren't already doing. Until I get some serious horsepower I can't really set myself apart. I just don't have the patience to work that way, waiting on an After Effects preview every time I make a minor change to a scene.

I think I'm ready to tackle building what I want minus a few questions. One big question - am I going with the 980X or should I be looking at server/workstation chips? Not necessarily a dual processor setup, but maybe one of the (expensive) 6-core Westmeres. Their TDP is quite a bit lower than the Gulftown. Not sure if the savings in electricity are worth an extra 4 or 500 initial expense though...for a lower-clocked processor. Then again I understand the server/workstation processors are built under tighter QC...although maybe not better than Intel's flagship 980x...

I guess I'll start a new thread for this - I've got a couple newegg wish lists, maybe some folks will take a look, see where I'm going, how I can improve on it.

Thanks for your thoughts - I appreciate your input very much.

personally I'd spend less on the machine and replace it sooner. Computers follow laws of diminishing returns so spending twice the money is usually only going to get you something in the order of 20% more speed (numbers pull out of the air).

Traditionally this years bleeding edge top of the range etc is only as fast as next years midrange and the following years low end. (OK so this is slightly less true in this age of atom powered nettops etc but you get the point)

It's my view that spending a lot of money on a machine that won't earning its keep for a year or two is a complete waste. When the time comes for it to do real work it'll be a dinosaur that could be brought for a fraction of what you paid for it.

Anyhow, I'd forget about server/workstation hardware. You'll have to spend a hell of a lot more to get the same performance and it'll make servicing the machine a lot harder in the long term as server/workstation parts aren't as easy to get hold of as standard desktop kit. It'll be worth investing in once you've got plenty of cash coming in but given that the life time of a machine like this is only going to be a few years it's a luxury.

just my opinion of course

Thanks Ch - I appreciate your advice. I have a feeling the 980x bucks the trend when it comes to planned obsolescence but still I get your point.

And while I agree with

cheesyking wrote:

It's my view that spending a lot of money on a machine that won't earning its keep for a year or two is a complete waste.

I'm not sure where you got the idea that it'd be "a year or two." If it was something I said let me correct myself. I must admit it's possible it'll never "pay for itself" because I'm just gambling that I have the talent to create marketable ideas - though I've done enough in the past to make the odds pretty good. It's conceivable, if not likely, that by the end of the year I'll have recouped most of the investment. If it takes a little longer so be it.

and of course anyone considering the 980x who isn't 100% sure it's because business demands it is operating under the "Spell of Intel." It's lust, it's ego, it's - who knows. But there's no denying that proc rocks. Maybe I'm being ridiculous, maybe any of the Bloomfields, or even the Lynnfields would be perfectly fine. But I don't want to be just satisfied with the machine, I want to be blown away.

I'll buck the trend here and recommend building "the killer rig."

Practice machine - bah!
You've got a one-time-chance to reach for the stars, so do it.

Personally, I think you should go balls out and buy the machine you want, 980x and all. The reason why people buy i7-980s isn't for longetivity. Its a simple business proposition. Consider if you're an engineer making $200,000 a year working with autocad or something else that requires a PC to render, calculate, or do something before you can do something. Imagine you can buy a i7-860 for 200 bucks and have the calculation job done in an hour. Or you can buy an i7-980x and be done in 20 mins. That time saved translates to dollars and that engineer who uses for their PCs for work (or his employer) easily make up the 800 dollars difference between the processors very quickly (I'd say honestly within the first day of use to be honest) simply in wages not lost to idle time, let alone increases in productivity and your final profit margin.

People on this forum usually speak bad about buying intel's top of the line processors because they aren't coming from the point of view where even a 10 percent gain in performance in an App can easily save someone thousands of dollars over the course of a year. Instead, they (correctly) speak of what those 10 percent gains mean to people who use PCs for leisure (gains, which to be honest are negligible in that regard because most people's apps aren't cpu intensive and can be efficiently run on much lower power machines).

You sound like you seriously need a top of the line PC for work, so build one. Your most valuable assets are your time and expertise, NOT YOUR PC, and once you understand that, you'll understand how its worth paying 800 dollars more for a top of the line processor and 12gb of ram and a SSD, if that prevents your most valuable assets from sitting around and being wasted. I've worked in environments where heavy duty processing is important and the more power the better always. Having 3 or 4 research labs, and up to 18 people sitting around idle, backed up because there'e only one workstation to render, measure all sorts of things, adjust, and tune all the images (thousands of them) from all the high powered microscopes we use is proof that even a 10 percent performance gain can make avery noticeable difference in your bottom line.

I can't advise you on the server parts though. As far as I know, intel server parts allow you to use more than one processor on a motherboard for even more processing power, and server parts generally are a higher grade. However, I would go with non server parts since you're going to have to maintain this PC yourself. Plus the kind of servers those parts are built for literally never turn off and run for years at very high loads. You just need a lot of processing power for the hours of 8-5.

Building a PC is very simple. The hardest part for me is usually getting all the cabling from the case to the mobo via those tiny headers done.

If you are afraid, you can always order the parts and pay someone (ie fry's, best buy, a local PC service store, etc) to put the parts together. Fry's and best buy of course charge the most and local PC stores charge the least. I'd expect fry's to charge about 100 bucks and your local PC store to charge 50 bucks. The good thing is they are responsible for the parts and so if they screw up (which they won't) they'll cover the parts.

You can also try craiglist and put out an ad for a university student in computer science or something to come over and do it for a reasonable price. Its up to you really. If you're going to spend 2500 bucks on a PC, whats 50 bucks to have someone assemble it?

Sunburn74 wrote:

Personally, I think you should go balls out and buy the machine you want, 980x and all. The reason why people buy i7-980s isn't for longetivity.

no the reason most people buy a 980x is for the epeeeeen, your paying 400% more for maybe 30% more performance (based on an i7 870 vs 980x on cinbench). Sure, if you've got unlimited money why not but this guy is saying he hasn't got unlimited money!

Also if rendering a scene takes an hour on a slower machine then the 980x will only bring it down to 40 minutes so it's not like that investment will change how you're working. You'll still have to start the render and then go off and do something else until it completes.

So having a second machine allows you to both queue up jobs to run while you get on with other work and have a spare machine so you can keep working if the main machine fails for some reason.

Obviously I've not counted software costs here

Sunburn74 wrote:

...Its a simple business proposition. Consider if you're an engineer making $200,000 a year...

I agree it makes sense when they have (1) a constant supply of work and (2) a decent backup computer. I'm doubtful that a 100% premium makes sense in this case, as he's starting a new business with all those unexpected expenses.

We don't have hard numbers, but let's take his estimate that his old laptop took 20 minutes to render after he worked for 3 minutes. (However I bet on average, it takes longer to setup the scene). Let's say a 875k is 8x as fast, so it will render in 2.5 minutes. Let's say the 980x is 50% faster (6 vs 4 cores), so it will render in 1.6 minutes. If his labor of arranging the scene is 3 minutes, his cycle shortens from 5.5 minutes to 4.6 minutes.

At most, that saves him a few hours per week. If his primary $2500 computer goes down, he loses many hours per day. And his budget now doesn't have the extra money to buy a new one, or pay somone to fix it.

I know he didn't originally ask for business advice, but these issues seem just as important as the hardware issues. If the business becomes established/profitable in 12-18 months, and he finds he likes doing it 40+ hours/week, I think he has strong argument to by the top of the line computer then. (And like cheesyking said, the 980x will be outperformed by something that costs half as much.)

wow that's an interesting thought - I hadn't even considered someone else assembling it for me. I'd never trust BestBuy - the ones around here (Manhattan) are full of kids that speak and act like they haven't yet graduated 7th grade - for the most part anyway, with a few exceptions. We don't have Frye's in NYC but there is a MicroCenter out on Long Island.

Still - I better do this myself. There's no reason I can't - the only thing I'm not good at is soldering and new builds don't require any of that, so what the heck.

Thanks for your advice and suggestions. I think I'll be fine. If I were to let someone else do it...well where's the fun in that?! just kidding, mostly. But seriously, I want / need to learn the process once and for all - might as well be now.

Thanks again!

Also if rendering a scene takes an hour on a slower machine then the 980x will only bring it down to 40 minutes so it's not like that investment will change how you're working.

Yeah but look at what you're saying! Imagine you're paying a guy 100 dollars an hour to do what he does and assume the equipment he must use is the limiting factor to how fast he works (which very often is the case. I've worked with a company where I worked 12 hours shifts but only actually worked for maybe 1.5-2 hours. The rest of the time was spend literally sitting on my butt waiting for liquid concrete to cook in massive pressure/heat chambers. The equipment was by far the limiting factor. My input and knowledge, whilst crucial, was very minimal once I had the results). That i7-980 is reducing his wait time by 33% best case scenario. Literally if he's spending 8 hour a day working on his PC, he now only has to work 5.34 hours a day. You'll make up that 800 dollar difference in 3 days best case scenario.

Now I realize there's more to work than just PCs rendering away, but geez in terms of the wait time alone per day, the time you free up for the guy to spend on other things and other projects, the i7-980x is a no brainer, even for people without unlimited money.

@wibeasel
I see what you're saying and agree somewhat, but again at $100 dollars per hour, he only has to save 8 hours to make up the difference in cost between the processors. Even if he takes that 8 hours as vacation, its still a tangible benefit. It's a judgment call really. If I had to pick, I'd go with the time saver because I consider my expertise to be my most valuable asset and the thing I need to most be in action, not my PC which anyone can put togehter.

@canoli
All you need is a screwdriver and some thermal paste. Everything else should come with the parts you order. If you're worried, ask a friend to help you. Two brains are better than one.

canoli wrote:

ONE - is it stupid to use super-expensive parts on my first build?
I'm really only worried about applying the thermal grease correctly and getting the cooler attached.

IMNSHO, it is stupid to use super-expensive parts for any build, first or otherwise. The $1000 CPUs are for the people who have more money than common sense, or for those that need the absolutely fastest machine and cost is completely irrelevant. When you go from a $500 processor to an $1000 processor you don't get double the performance; you get a 10% increase in performance for double the price. Have you heard of the law of diminishing returns? It applies pretty well here.

canoli wrote:

TWO - should I go for 2 Xeon or that 980X? If I do a server board, what else does that commit me to? Special RAM? Special power requirements? Restrictions on audio?

For the love of god, don't get server parts. They're not worth the pain. Most of the times you need special motherboards, special power supplies, special memory sometimes. All these add to the cost significantly, while severely limiting your options at the same time.

Let me tell you a little story: In 2004, my then flatmate decided that the computer he was about to buy had to be a multiprocessor rig. So he got two Xeons, one of the two motherboard models on the market at the time that had both two sockets *and* an x16 PCIe slot, a special EPS12V power supply, and a huge case that would actually fit that EATX motherboard. That computer was pretty sweet in the end, no doubt, but a year later he could have bought a single-socket dual-core processor along with a huge selection of standard desktop motherboards / PSUs / cases / etc, for probably half the price.

Anyway, having read your requirements, I'd say that overall the components of the Sweeter Spot (and/or its alternatives) would serve you pretty well. But hey, if you're determined to exhaust that $2k that you feel burning in your pocket, don't let me get in your way. :^)

i appreciate your advice V - I've realized there are basically 2 camps when it comes to the 980X and other premium products. One group sees no point in the latest and greatest because Moore's law is always in effect. The other side places a higher value on what those products bring to the table now, today. I see you belong to the former group. Me I see merit to both sides and I'm far from the "money burning a hole in my pocket" syndrome. I'm being painfully deliberate about my choices. And I think the poster before you made an excellent "time is money" point. Even if time isn't money it's still awfully precious...

My idea behind using server parts was to increase the number of cores. For 3D rendering the more cores the better. These days you can find server boards that are priced similarly to the higher-end enthusiast boards. Unfortunately I can't find any good benchmark comparisons between dual-processor setups and the 6-cores. Anantech is great but even they don't benchmark dual-proc workstations.

I know you're right V when you say the Sweeter Spot is "good enough" - I know because I've worked on machines that were "good enough." Now I'm looking for something better, something that is closer to "holy $%@ that was fast!" The things I want to do, the ideas I have for video combined with 3D - or really any visual presentation in the design stage - I need to see them happen on the screen immediately, or at least as quickly possible. I lose the flow of the work, of the creation process when I'm staring at the green bar in After Effects' comp window, or watching C4D's picture viewer render frame...by...frame...by...frame...

Maybe that sounds artsy-fartsy or whatever but that's how it is. Waiting disrupts that flow immensely and AFAIK the best medicine costs a thousand dollars. It's not a cure, nothing ever will be until we're all just pure energy beings exchanging thought-speak and materializing anywhere on the planet in an instant (now that's artsy-fartsy!). And certainly in a few months - maybe even sooner - there will be hexacore offerings that are just as fast (maybe) for a lot less money. But Sandy Bridge and the rest of the future pipeline can't help me now, today.

Thanks again for your thoughts on the subject, you've helped me come closer to deciding. And thanks to everyone who replied because you've all helped me very much. Later today or tomorrow I will submit for your approval 1 or 2 prospective builds, and I hope you'll be as helpful as you've already been. Thanks!

I may have missed similar commentary along the way but if not:
I'd suggest doing one solid machine (~$1500 or so) and then get two smaller machines: and then use the network render functions of Cinema 4D and AE (NetRender and BackBurner). We keep 6 machines on and running even if we only have 1 or 2 guys running the software during the day so that they can take extra advantage of that. And if something should happen to the main machine, you can still go back to a Quad machine easily (maybe just slide your video card over). But a $500 network render node isn't that hard to do (basically get a quad CPU, at least 4GB RAM, smallest/cheapest HD, and a motherboard with an upgrade future so you can use a 6 core when they drop in price [so probably an AMD]). Those other machines can be easily bumped later, and it also gives you the infrastructure to upgrade later (moving your main machine to a 3rd render node).

Something like this would be a solid foundation for a render node: http://www.newegg.com/Product/ComboDealDetails.aspx?ItemList=Combo.500546 I'm sure there are others, but as a starting point it's pretty good.

Also, remember AE like at least 1GB RAM per thread/core.

Also, there is a MicroCenter on the Deegan heading north around Yonkers- right around the racetrack area I believe (I keep meaning to stop, but haven't had the out-and-out need to).