Slashdot Mirror
Workstations For Poor 3D-artists
Peter writes: "Ace's hardware has written an 'article for the creative people, who are searching to build or buy an affordable number cruncher to run their favorite workstation application. Maybe you already have an Athlon Thunderbird/XP and you are wondering if a dual Thunderbird/Athlon XP workstation might make sense for you. Or you might be interested in an affordable dual Athlon MP 1800+ workstation.' Included are benchmarks based on almost all available 3D-animation packages."
Cheap box for 3D artists?
What about cheap software for 3D artists?
(BTW - IANA3DA, but I'm pretty sure that all 3D software for modelling and such is mucho dinero)
Good quote, too many chars. Seriously, the slashdot 120 char limit sucks!
You probably haven't checked the price of those 3d tools lately ... the only one in my pricerange is povray ...
;-)
... ;-)
We're not gonna copy those tools illegaly now are we ?
And btw, 3D studio 4 (the dos version) has a lot of possibilities and renders nearly real-time on an athlon 700
Out of all of the 3D Animation packages they got their hands on, they forgot 2 of the most important ones out there.
Softimage|3D and Softimage|XSI.
Those two give Lightwave, Maya, and 3D Studio MAX a run for their money, considering they're the modeling environment used by most all major CG Effects studios out there (coupled with either Mental Ray or RenderMan).
Considering the parent to this post was marked "intersting", I think "paranoid" would be a more accurate description.
... if not then your PC manufacturers are shafting you.
... it's the same FUD that holds people back.
If you are blinkered enough to follow the mighty chipzilla instead of AMD in 2001 for desktop performance then you need to smell the coffee or at least try a fair comparison.
Yes, your P4-optomised build of the kernel will scream, but when I go out and buy 3d tools to run on top of a micro$haft operating system I can't just go recompiling the application to fit the specific hardware it's running on, and that usually means it's much faster on an Athlon by default.
And Athlon-based systems should be *much* cheaper than their Intel counterparts
The sooner people start realising the desktop processor market is about more than Intel then the sooner people may be ready to consider more than one desktop operating system
I'm not sure about you guys, but I noticed a really subtle pro-AMD bias in this article. For instance, the banner ad on the top of the page was for the new Athlon XPs and linked to AMD's page. And the author gave Intel a few token references, and then completely ignored them in the benchmarks.
The author performed benchmarks in a number of major 3D applications, and in all the AMD chips absolutely rocked: If you have a problem with the methodology, or feel that it isn't telling the whole story, then post your own site (that's the beauty of the net). I think it's fairly obvious that putting an ad for a Xeon chip on an article where it was pummeled probably doesn't make an awful lot of sense.
Well, as a very satisfied Pentium 4 owner and a somewhat satisfied Athlon owner, I can tell you that if you're serious about getting work done (not just overclocking your Unreal box), you'd be best off going with a P4
Let me get this straight: You refute an article that is packed full of actual metrics by saying that it's biased, and then you say that people should get a Pentium 4 if they're not going to "overclock their unreal box" (again an absolutely absurd supposition given that we're talking about an article where the AMD trounced the Intel chips in something much more serious than "overclocking their unreal box").
My Athlon had some heat and manufacturing issues (this is my second chip because the first one was DOA), and really isn't any faster in the real world than my P4.
The absolute definition of FUD. "Uh, sure the AMDs are faster, but they have heat and manufacturing issues!". Whatever. Metrics are all that matter, and the metrics in the industry say that the power consumption of upper end Intel's and AMDs are very similar (hence similar heat), and that major manufacturers have roughly equal DOA rates with both chips. The metrics also say time and time again that the "real world performance" of the AMDs are often faster than the Intels.
In fact, having rebuilt my kernel with the new Intel compiler, the P4 just screams and leaves the Athlon in the dust.
I see. Again please tell us when you've put up a site and posted some benchmarks with your platform and methodologies, because as it sits it sure sounds like a bunch of bullshit.
I'm going to shoot in the dark here and make a wild guess: You ran out and bought yourself a fancy new Pentium 4, spending top dollar to be the top dog in the tech arena (of course not doing any research), but now that you have your new purchase you're a little more sensitive whenever you see performance benchmarks, and everytime you see another review that shows the Athlon dominating it just burns at you, so here you are with your "real world" experience. Bullshit. I highly doubt you have an Athlon whatsoever.
I am not biased whatsoever, and if Intel comes out with something that is competitive with the XPs at a similar price then damnit, I'll be there. But I owe nothing to Intel, nor do I owe anything to AMD, so I lack "brand loyalty" and simply go for what is proven the best at the best $. If only more consumers were that way.
My first experience with PoV (and dkbtrace) was on a 25MHz 386. Complex scenes (due to objects, textures, solids and such) could take an entire weekend to render, which now take a few minutes on a 933MHz PIII (at work, but can't be doing that here) I'd run thumbnails, which took 10-15 minutes until I was pretty sure of what I was getting then launch it on Friday evening before heading home, one scene finished about an hour after I got in on Monday morning (had to busy myself shuffling paper or something ;) Yeah, you could do this on a 386/486, but why would you want to, when people are throwing away Pentium 133 machines?
A feeling of having made the same mistake before: Deja Foobar
Codewarrior benefits from SMP, as do typical "make -j " project builds under unices.
Well, these "metrics" you claim to trust so much also indicate that the IBM 75GXP drives have a "normal" failure rate. And we all know the truth about that, don't we?
You see this really is humorous: You see a Slashdot story with a couple of people saying that their 75GXP failed, and you're sold (obviously just like the AMD issue). I actually HAVE a 75GXP that hasn't failed, and I am prone to believing IBM that the failure rate is normal. Let me put it another way: I know lots of people who are sure that Honda cars are the biggest POS out there because they had a lemon that had 27 faults, but the industry statistics say that they're the exception, not the rule. If there was more than anecdotal evidence (or biased polling) that the IBM drives were unreliable then I would be extremely happy to listen and take action based on it.
which, by the way, you haven't even tried out
I remember back in the BBS days asking a sysop to remove a "CPU Speed Up" program that promised to "convert your 386 to a 486/66!". The Sysop refused claiming that lots of people claimed that it really did vastly improve the speed of their systems. It's called the placebo effect, and it's one of the biggest truisms about people: People are extremely unreliable metrics of anything, because most people go into an evaluation with preconceived notions. As such, I'll be a little more trustworthy of site after site after site after site giving methodologies and performance metrics that show the Athlon XP winning. Again when Intel comes out with a cost effective (meaning cost effective all around: Memory, MB, etc.) high performing chip then I'm there, but as it stands there is a clear winner.
It was nothing to do with affordable, cheap 3D workstations. It was about the latest and greatest x86 CPUs - and the latest and greatest, high-end software.
/. is a pretty sad statement about what has become of this once useful and interesting site.
There are better machines out there - SGI Indigo2s and Octanes with OpenGL (and more!) in the hardware, many decent Macs - all of which are more affordable 2nd hand than the wunderboxen on display here.
And if you're looking for a career in 3D animation/design, are you going to use some x86 toy, or would experience with what the rest of the industry uses be a bit more helpful in your career?
A vanishingly small amount of 3D work requires a fast CPU - it's about shoving large amounts of data around. It doesn't matter how many mhz your bus does - it's still a data bus, and it's inherently unsuited to this type of work.
Look at the Octane's Crossbar. Look at the O2s UMA architecture.
Apple continue to make the same mistake, and are going to cripple their G5 machines. Lovely fast processor, crap bus to the gfx, memory and disk.
And I just loved the way this compares the Shake results to a 4 year old Octane. Nice. I notice we don't see any playback information at high resolution - what's the point of fast render speeds if you can't *view* what you've just created?
Never mind that 4 year old Octane can be bought for a fraction of the price of the systems under discussion.
This was a truly laughable article that, while demostrating an understanding of consumer x86 toys, showed a clear lack of clue about 3D graphics needs.
That this article ever showed up on
I'm surprised we haven't see the Linux kiddies moaning about how the tests were run on Windows 2k.
wow you dont do much in computing power then do you.
:-)
povray renders 67% increase.
BMRT renders get a whopping $75% increase.
Heck I get a 50% speed increase on compiling anything on my lowly PIII850 SMP box. (I know I should trash it, it's almost 8 months old now.)
I see major increases by going SMP, but then I do things that take advantage of both processors. (BTW, make -j2 will speed things up nicely
Do not look at laser with remaining good eye.
I figure I will put this out there because it needs to be said. For anyone getting into 3D, this is the process that you need to take.
You need a computer, make it a x86 PC running windows 2000 Professional, this is the best way to go right now. Linux, Mac, SGI, are not options for you in terms of money and ease of use. A PC will be low cost and dynamic. If you already have a computer, there is nothing wrong with using that, unless you can't put more than 128 MB of RAM in it.
Put as much RAM in it as you can 128 will work, 256 will be comfortable and let you get into more complex projects, 512 will probably be more than you will use, but it isn't a bad thing.
Your processor speed matters, but if it can run windows 2000 you will be fine. Renders may turn out to be slow on a slow computer, but with enough RAM they will be slow and steady, and still allow you to get work done. If you can get a fast processor, good, if not, don't sweat it.
Get a good gaming graphics card. Go for a Geforce 2 MX or Radeon or a Geforce 3 if you can afford it. They will all work very very well. It will increase interactivity and minimize frustration.
Get access to broadband and use morpheus to pirate all the goddamned software you can find. Look for Lightwave 6.5b or 7.0, 3DS Max R4, Maya 4, Softimage 3D (rare), Softimage XSI 1.5 (rare), or Houdini (super mega rare). - (The magic five, 95% of studios will own at least on of these programs) Finding good 3D software for the Mac is very difficult, Lightwave and Maya are the two programs you should be concerned with, and Maya for MacOSX was just recently released and will be extremely difficult to find, if not impossible.
Look for Photoshop 6.0, After effects 5.0, and Painter (rare) to compliment your 3D software.
Get Sound Forge 5.0 and Cool Edit to mess around with any sound you might want.
Take the time to click every button in every program you have and figure out what it does. After you know the features pretty well start a project, if you are enthusiastic about 3D you will certainly have something you want to achieve.
Try to make it look good, but don't get frustrated if it doesn't. Completing something is much better than keeping your standards so high, you freak out and don't progress.
Reading is good, experience is better, make sure you have both read about animation and do as much as possible.
While you are doing all this, save up to actually buy the educational, or full version of the software, it is worth it. I am not just saying this so I don't look like an ungrateful pirate, I truly mean it, all of that software is worth every penny.
Don't believe any nay-sayers or egotists, this is the way to go. I know about Blender and other free projects, just avoid them, pirate, and save up for the real version of what you like best. The free projects won't be ready for at least 2.5 years, probably more. Blue Moon Rendering Tools is a very good renderer and is free, but works off of the Renderman standard, and it will be very difficult to get anything to interface with it.
And lastly, remember, take it further, take it further, take it further!
If you want to get into 3D, save this comment and make it a check list. Flame me if you like, but I know that this is the best path to take to enter the world of 3D and computer animation, it will take you where you want to go.
This Wiki Feeds You TV and Anime - vidwiki.org