Affordable Workstation Graphics Card Shoot-Out

MojoKid writes "While workstation graphics cards are generally much more expensive than their gaming-class brethren, it's absolutely possible to build a budget-minded system with a workstation-class graphics card to match. Both NVIDIA and ATI have workstation-class cards that scale down below $500, a fraction of the price of most high-end workstation cards. This round-up looks at three affordable workstation cards, two new FireGL cards from AMD/ATI and a QuadroFX card from NVIDIA, and offers an evaluation of their relative performance in applications like Cinema 4D, 3D StudioMax, and SpecViewperf, as well as their respective price points."

Comments

Workstation class??

[CranberryKing]

Um.. I use my 'workstation' for spreadsheets and web browsing. The dell integrated shares-sys-memory controller is fine and didn't cost me no 500 bucks.

Re:Workstation class??

[Broken scope]

Well i guess you really don't have a clue what they are referring to when they say "Workstation".

Re:Workstation class??

[merreborn]

applications like Cinema 4D, 3D StudioMax, and SpecViewperf

You may not use those applications on your "workstation", but there are thousands of professionals who do

Note that the term workstation usually means a high end system used for something a little more complex than web browsing and spreadsheets:

http://en.wikipedia.org/wiki/Workstation

I believe the progression, marketing-wise, goes:
Desktop -> Workstation -> Server

You're thinking of desktop hardware/software.

Re:Workstation class??

[bytesex]

And I daresay that he isn't the only one. The write-up is confusing, at best. Had me going for a bit, anyway. To ordinary people, even 'ordinary' slashdot-readers, a 'workstation' is some 'station' (a desk with a computer) that you do your 'work' on. That thing will usually contain a graphics controller that is on-board these days, the cost of which has been discounted in the price of the board, and certainly isn't expensive to an extent that a gaming-person's graphics controller will have a 'fraction' of the cost. Chagrin or no chagrin about lay (non-graphics) people reading topics that aren't meant for them, but to act as if this is logical, implicit or otherwise self-explanatory, is disingenuous and not much different from those slashdot-write-ups that start off describing some event in second life as if it happened in real life, and pretend that everybody knows what they're talking about. Clarity is king, and no man is an island and that sort of thing.

Re:Workstation class??

[TheLink]

"bus station is where the bus stops, a train station is where the train stops"

So workstation = where work stops?

But how about a playstation then?

Difference?

[AdeBaumann]

I'm sure I'm not the only one who doesn't know... so can anybody explain the difference between a high-end workstation card and a high-end gaming card?

Re:Difference?

[Broken scope]

I think it works like this.

Game cards are designed to render stuff as fast as possible, many times a second.

Workstation cards are designed to render everything in the desired quality, and take as long as it needs.

Re:Difference?

[kcbanner]

The workstation cards tend to have very low error tolerance, while the real time graphics cards allow for quite a bit of error in the name of speed. This is fine unless your rendering something.

Re:Difference?

[TheSpengo]

High end gaming cards specialize in pure speed while high-end workstation cards specialize in extreme accuracy and precision is the basic answer. They are incredibly accurate with FSAA and sub-pixel precision. Workstation graphic cards also have other features such as overlay plane support which really helps in things like 3dsmax.

Re:Difference?

[acidream]

Workstation cards typically have certain features enabled that their gaming counterparts do not. Some are just driver features, others are in silicon. Hardware overlay planes are a common example. This is required by some 3d applications like maya in order to display parts of the gui properly.

Re:Difference?

[Psychotria]

I am not sure "error tolerance" is the correct term; there is no "tolerance" (you are correct though; I am just debating the term), it's just that the high-end workstation cards sacrifice speed over accuracy. To say "error tolerance" implies that both types of card have errors (that they may or may not have and may or may not compensate for), and one tolerates them more than the other. This, strictly, isn't true. A better analogy would be something like high-end gaming cards have (for example... making the figures up) 24-bit precision and the high-end cards have 64-bit precision. There is no "tolerance" involved; just that one does the math better for accuracy and the other does the math better for speed.

Re:Difference?

[prefect42]

I'd say it's more complicated than that. Gamer cards push game graphics around fast. This often means high memory bandwidth for texturing, fast full screen anti-aliasing, and these days fast shader performance. Workstation cards often are better at line-antialiasing, much better with high polygon count work, much better working with mutiple windows. Quadros always used to support more clipping planes in hardware for example. How much of this is a real hardware difference, who knows.

We've got a home-grown application rendering a 4 million polygon model. Quadro 4500 is an order of magnitude faster than a 7800 GTX. You wouldn't guess that from the tech specs.

Re:Difference?

[citizenr]

>This is fine unless your rendering something.

you dont render anything on the GPU (at least not yet), video cards are only for visualisation, that measn your theory is not valid

Re:Difference?

[0xygen]

Error tolerance refers to pixel errors in the output image compared to a reference rendering.

eg, the fast texture sampling methods on gaming cards lead to aliasing errors, where the pixel is in error compared to a refernce rendering.

There are also a lot more factors to this than just floating point precision, for example how the edges of polys are treated, how part-transparent textures are treated and how textures are sampled and blended.

Re:Difference?

[White Flame]

Another difference that at least existed in the past, and probably still holds true today, is that workstation cards have more geometry pipelines, whereas gaming cards have more pixel pipelines. The gamer stuff puts out very pretty but a lower number of polygons, whereas workstations often just use kajillions of tiny untextured polygons. It's a tradeoff that affects silicon size and internal chip bandwidth, and explains why games and workstation apps run slowly on the wrong 'type' of card with their different demands.

Re:Difference?

[Molt]

This doesn't hold true any more, the latest generation of hardware are all using the Unified Shader Model. This removes the distinction between a pixel pipeline and a vertex pipeline as a unified pipeline is used which can be switched between pixel and vertex processing as the scene demands.

Re:Difference?

[prisoner-of-enigma]

"Workstation" generally means you're using some sort of 3D application, pushing hundreds of millions (or billions) of textured, lit triangles around. I have a 2S/2P Opteron workstation with 8GB of RAM and two Quadro FX 3500 cards, and I use it with 3DSMax.

The difference in cards is subtle. Most gaming cards are tuned for ultimate speed (framerate) but perhaps not as much accuracy or quality. Workstation cards have things like hardware anti-aliasing of wireframes, a great feature when you're working with a huge model in wireframe mode. Textures are handled differently as well. Gaming cards tend to have smaller textures (again, for speed) than high-end rendering for movies or video. It's all in how the card is tuned. That's why gaming cards tend to perform lower at workstation tasks and high-end workstation cards tend to perform badly (or just plain hideous) at games.

Note how an NVIDIA Quadro FX 5600 (a card costing nearly $3,000) gets spanked by an 8800GT (costing a little over $200) in games. Some people buy workstation cards for gaming thinking they are faster than gaming cards. It's the old "it costs more so it should run faster!" argument. A fool and his money are soon parted.

Anyway, there's one last dirty little secret in the workstation graphics card department: there is almost no hardware difference between a gaming card and a workstation card. In some cases there is no difference at all except the BIOS. There's a fellow called "Unwinder" over at www.guru3d.com who writes a program that will "softmod" a gaming card into a workstation card by strapping the BIOS. Benchmarks seem to show that a $200 gaming card softmodded into a $3,000 workstation card actually performs identically to the real $3,000 workstation card. This further bolsters the claim that NVIDIA and ATI are charging a ridiculous premium for their workstation cards.

I softmodded some gaming cards to workstation cards a few years back. Worked like a charm. However, it got to be more trouble than it was worth because NVIDIA kept trying to break the softmods with driver updates. You'd have to wait for Unwinder to update his program for the new driver before it would work again. For my next rig I bought real Quadros, just not the $3,000 ones :-).

It's a shame they don't test them against 'game ca

[TheSunborn]

It's a shame they don't test them against 'game cards'. It would be really interesting to find out how theese cards differ from the normal gaming cards, when doing realtime 3d.

Quadro FX5700 vs 8800 GTS OC?

[alwaystheretrading]

I'd really like to see a low end workstation card like one of these compared to a high end consumer card. When I'm working with half a million polys in 3DS Max 2008 is it really going to be worth the extra money to get the workstation card?

All I can say is...

[snl2587]

...if you're planning on using a Linux workstation, don't buy an ATI card. I don't mean this as flamebait, just practical advice. Even with the new proprietary drivers or even the open source drivers, there are still many, many problems. Of course, I prefer ATI on Windows, so it all depends on what you want to do.

Re:All I can say is...

[doombringerltx]

It depends. I bought a new ATI card after they opened up the 2D driver specs. When booted into Linux I haven't had any problems with my day to day activities. Its only when it tries to render anything in 3D that it shits bricks. To be fair there may be a problem besides the driver that I haven't found yet, but right now all signs are pointed to driver/card problems. Honestly its not a big deal to me. I just don't use any fancy compositing manager and I never played games in Linux anyways. While I'm on the subject, I know when they released the 2D specs they said the 3D specs were on their way, but then I never heard anything out of that again. Does anyone know if or when that will happen if it hasn't already?

We're next

[Hansele]

As soon as the shootout's over, they'll come gunning for us. I, for one, welcome our new graphical overlords.

Re:Gaming vs Workstation Cards

[SynapseLapse]

Not exactly.
Gaming grade video cards tend to be very fast at special types of pixel shaders and excel at polishing the image to look better. Where they tend to be inaccurate is how they clamp the textures and even then it's fuzzy estimates that only are ever issues at extreme angles.
This is only in the way it displays data and wouldn't cause a COD program to "fall over."

Workstation cards are primarily high polygon crunchers. Games are rendered entirely in Triangles, whereas rendering programs use Triangles, Quadrangles and honest to goodness polygons (5+ sides).

Superficial Market Creation

[ludomancer]

There was a time when you could purchase a 3D card that worked excellently for both work and play. These new "workstation" cards are a farce. They are an ostensible attempt at a solution where there is no problem. I am a professional 3D Artist and I can attest to this due to personal experience over the last 15 years. Don't buy into this crap. They DO perform better for workstations, but only due to the fact that gaming cards are intentionally crippled in this area in order to push this alternative product. Luckily most gaming cards currently on the market work well enough for 3D workstations, so I encourage everyone to ignore this attempt at desultory market generation as much as possible, because it's perfectly possible for you to get great performance out of a gaming card for both purposes.

Re:It's a shame they don't test them against 'game

[TheSunborn]

]In a rendering competition between the latest geforce and the latest quadro in maya or 3ds max or something, the quadro would completely obliterate the geforce I know that's how its suposed to be, but I have newer seen a benchmark between 'workstation cards' and 'gaming cards' which included example images from the different cards, that showed the difference.

This benchmark don't even include any example images, which I don't understand because it might be the biggest difference between the cards. Having a benchmark of 'workstation cards' that are suposed to look better then the gaming cards, and then not even including anything about the image quality is wierd.

Re:It's a shame they don't test them against 'game

[neumayr]

I think the OP meant that a test against consumer cards would be very interesting for 3D artists on a budget.
As in, do I stick to this GeForce and get that quadcore CPU in order to speed up my test renderings or does it make more sense to spend my money on a Quadro and stick to my slower CPU?

The Biggest Scam of the Graphics Industry!

[MindPrison]

The biggest curtain that have ever been pulled over the artists eyes is the "PRO"-Graphics card-Fad! Youre paying to feel "pro" - you dont get more "pro" for your money at all, you just get to "feel-like-pro" but very little extra to justify the real bucks youre spending on Quadro & FireGL series.

I know this, Im a "graphics pro" myself that makes a living of designing 3D-Models & prototyping every day and Ive used nearly every card known to mankind.

Heres my advice - take it or leave it:

Buy a Gaming-Nvidia card! The difference between the Gaming Series cards and the Quadro series card is just some extra driver software that is optimized for your "insert-favorite-3D-app-here", yes...there are some less pixel-flaws..but this will never ever affect your final-render unless youre using Nvidias Gelato (which has - by the way - proven in many cases to render less effectively than modern Multi-core-CPUs with software rendering)

You will save up to THOUSANDS of Dollars by not buying into the "PRO" hype, and youll be one happy puppy you didnt - and work just as efficiently (I know - we do) as the ones with the "PRO" cards, the game cards are actually using the same chipsets (remember the Quad-Mod you could perform on their cards, it aint fake you know!)...it would make absolutely NO SENSE for them business wise to produce 2 different cards when their cards can in fact do the same thing....and actually use the same chips.

Re:The Biggest Scam of the Graphics Industry!

[prefect42]

Problem is, your advice sounds reasonable even though it's not.

Looking at the hardware spec sheets, I'd agree with you. But when it came to it, and I compared what at the time were the top cards (Quadro 4500 vs 7800GTX) the difference was night and day. If you wanted to play games, but the 7800GTX, it was waaaay faster. Want to do your own OpenGL apps that are quite demanding (high polygon count, multiple clipping planes, lots of transparency) and it's clear that not only is the 4500 faster, but it gives almost twice the bang for buck. That's pretty impressive for a 1500 ukp card, where you're not expecting value for money...

What you need to see are benchmarks of a Quadro 1700 against a similarly priced 8800. I'd be tempted to call in favour of the Quadro for things that matter to me, but short of buying some to test, it's hard to get decent figures.

Re:The Biggest Scam of the Graphics Industry!

[Spy Hunter]

Clipping planes are one of the features NVIDIA cripples in the gaming drivers (as games hardly use them); it has nothing to do with hardware. Buy a GeForce and flash it with the Quadro firmware if you really care about clipping planes, but honestly features like clipping planes, hardware overlays, etc are better implemented in your application and in your shaders anyway, where they will run equally fast on gaming and workstation cards.

Re:The Biggest Scam of the Graphics Industry!

[Anonymous Coward]

You are exactly right. Many Geforce series cards can be made to function exactly the same as the Quadro series with RivaTuner and the NVStrap driver. I have actually done this myself on one of my cards.

The only people who buy Quadros are non-saavy artist types. Those of us who know better can have the exact same thing for a fraction of the cost.

Make your own Quadro at a fraction of the price

[nano2nd]

Have a look at this site - it is possible to flash an 8800 GTX to Quadro FX 5600:

http://aquamac.proboards106.com/index.cgi?board=hack2&action=display&thread=1178562617

Re:do we care?

[TimFenn]

They're specifically in the market for 3D CAD, 3DS, Maya, that sort of stuff, of which there really isn't a heavy weight open source equivalent.

I don't do 3D CAD, but being a biochemist type, I actually hang out with lots of folks that do work with all kinds of 3D data such as molecular models and volumetric MRI datasets. Workstation cards are especially useful for their stereo support, which many bio-folks find helpful when modelling. Most of the development is done on linux using stuff like VTK or VMD - its not just the engineering guys doing CAD in windows that want workstation cards.

As a scientist that uses linux daily for 3D applications, I would like to see an open source workstation card for 3D graphics, dangit.

in short... go for the CPU

[Animaether]

sorry, but a graphics card does not speed up your rendering unless your renderer can take advantage of the graphic card; hint: that's not very many, and those that do only do so for very limited tasks.

The only reason you should have for upgrading your graphics card within the 'consumer' market is if your viewport redraws are being sluggish; this will still allow you to play games properly* as well.
The only reason to upgrade to e.g. FireGL or a QuadroFX is if you're pushing really massive amounts of polys and want a dedicated support line; e.g. for 3ds Max, there's the MaxTreme drivers for the QuadroFX line - you don't get that for a consumer card.

* on the other hand, do *not* expect to play games with a QuadroFX properly. Do not expect frequent driver upgrades just to fix a glitch with some game. Do not expect the performance in games to be similar to, let alone better than, that of the consumer cards.

For 3D Artists dealing with rendering, the CPU should always be the primary concern (faster CPU / more cores = faster rendering**) followed by more RAM (more fits in a single render; consider a 64bit O/S and 3D Application), followed by a faster bus (tends to come with the CPU)/faster RAM, followed by a faster drive (if you -are- going to swap, or read in lots of data, or write out lots of data, you don't want to be doing that on a 4200RPM drive with little to no cache) followed by another machine to take over half the frames or half the image being rendered (** 'more cores' only scales up to a limited point. A second machine overtakes this limit in a snap), as long as you don't have something slow like a 10MBit network going (for data transfer).

The 'pro' cards may not be meant for you

[Anonymous Coward]

You may think that your 3d modeling and prototyping is professional work - and I'm sure it is.

However, you should be thinking of people using CATIA to build an entire car or even more exotic pieces of software for building entire airplanes. We're not talking the piddly few million of polies that the average Disney/Pixar movie ponders about in Maya/etc., even though those would benefit as well - we're talking a dew hundered million polies. Now we're talking 'pro'. Now we're talking the kind of people who used to buy SGI workstations at a couple $10k a piece, then switched to 'generic' workstations but fitted them with E&S (Evans & Sutherland) cards that were so big (similar in design to dual-GPU cards people are messing with now) they had to keep the casings off their machines or the things wouldn't even fit, and who are currently salivating at the nVidia QuadroPlex solutions in both desktop and rackmount form ( http://www.nvidia.com/page/quadroplex.html ) before crying as even they think that's just a might bit too pricy and go back to the suped-up PNY QuadroFX offerings ( http://www2.pny.com/category_buymulti.aspx?Category_ID=329 ).

Consumers, prosumers and small business need not apply. As you do say, it's not worth the extra money (and it -is- a good chunk of extra money) for those groups.

Free/Open Source workstation graphics card needed

[SST-206]

What we need for our audio workstations is a fanless (silent) graphics card that will do OpenGL nicely, using Free/Libre/Open Source drivers. Affordable is helpful, but not essential.

I've been watching the gradual progress of the Open Graphics Project (and now Open Hardware Foundation) with interest and hope they can release something good before the major manufacturers get a clue - quite likely considering their years of promises (ATI) and proprietary drivers (nVidia). It seems that Intel are doing good things, although IIUC those cards aren't so powerful; I know: power, silence, freedom (choose TWO only)... but progress? Is the ATI Radeon 8500 still the best fanless card with open drivers?

Please wake me up when we get to the 21st Century. I'd happily read a whole page of adverts for news on such a product.