Quadro 4, Quadro FX1000 or Quadro FX2000

[TAKA]

Hey guys, greg.

I'm going for a new video card, but which one to choose?

I read a review, more of a benchmark about these 3 cards along with the FireGL. And although the FX2000 outperforms, it didn't seem much like a noticeable thing to me.

I mean like at most tests for instance the Quadro 4 would score a 22, the FX1000 a 23 and the FX2000 a 25. Or at some other benchmark with wireframe mode and animation, the quadro 4 would score 90 FPX the FX1000= 120 FPS and the FX2000 like 160FPS.

 

90 FPS is more than enough for a scene with 400K polygons to me.

 

My main concern in getting a new board is to be able to navigate the model reasonably without any lags and display textures correctly (plus i want the dual output)

PS: my models are usually very big (1000K polygons, and so are my textures)

 

So what are really the main differences between these 3 cards in litteral english and which one would u recommend me??

[Greg Hess]

FX 3000 is also out now

 

I'll answer other questions when I have some time. (Preparing for siggy).

 

Your best bet would be to just wait until everyone at siggy gets to use the cards first hand your talking about...then reports back.

[garethace]

http://www.aceshardware.com/forum?read=105030728#TEXT

 

check out the review linked here, at 3d chips.com.

[garethace]

For my money the 3Dlabs Wildcat VP990 Pro, offered the best middle of the road performance and consistency. It is really the only card the quite manages to struggle on with 10million and 20 million vertices, with its 512MB of memory, when all the other runners have departed from the race. Even the 128MB 3Dlabs Wildcat VP870 looks to be a very solid and respectable card too.

 

Next up in terms of merit IMO, would have to be the QuadroFX cards. I think the Quadro FX cards may not offer as high a performance as the FireGL X1 and X2 cards, but the consistency of the Quadro FX cards going from 1 million up to 20 million vertices says a great deal for the cards stability i think.

 

If you compare the profiles of the Geforce 4, compared to the Radeon - the nVidia card is a straight line completely, and never, once drops below that base level. While the Radeon 9700 Pro starts very high, and completely ends up in the dustbin, in a very dramatic fashion indeed. While the nVidia Geforce 4 manages to lend some kind of performance at 10 and 20 million vertices, the Radeon 9700 Pro, doesn't even want to know and takes an early night cap!

 

The high end FireGL X1 and X2 cards exhibit a very similar eratic behaviour i think, but not a serious as the Radeon 9700 Pro. I think i could manage to use workstation applications with a Geforce 4Ti card, because of its consistency - but i would hateto have to use any of the FireGL or Radeon Product owing to the eratic nature of the driver behaviour.

 

The Radeon 9700 Pro sports a memory clock speed of 650MHz, which is almost as high as the FireGL X2's 700Mhz memory clock speed - so they are similarly spec-ed out products, but the speed of the memory doesn't do one bit of good towards using the Radeon as a CAD card does it? If it cannot even finish the Spring Mark Open GL benchmark.

 

This failure of the Radeon 9700 Pro to finish benchmarks is very worrying IMO - at least the entry level nVidia card, doesn't hang you out to dry! Dell workstations don't even have a smell of the Radeon, but offer at least some kind of Quadro product (for the certified drivers mostly) on all their workstations now - which is a good sign of a workstation product IMO.

[Greg Hess]

Gare,

 

The wildcat cards have a variety of issues in cross-application swapping. Users have reported problems with large video playing, as well as bad d3d support, and problems in some applications.

 

I wouldn't consider the current generation of VP cards a solution for 3d graphics due to issues which are currently floating around with them.

 

Hopefully the next iteration will address the previous ones flaws.

 

I'd go for either a Quadro 900/980XL, or 1000 FX if you had the budget for it.

 

The nicest thing about the quadro's is that they work extremely well in an extremely wide variety of applications.

 

Be it maya, max, lightwave, xsi, cinema4, games, combustion..etc.

 

That tends to be their greatest strength. They may not be the performance leader in every single benchmark...but they will work in virtually every application...and if they don't, there will be a driver set that allows them to.

 

Try a Radeon gpu and XSI and you'll see what I mean.

 

Btw Gare, the FireGL cards have a bug in Max6 which greatly reduces their performance. Don't know if this will be addressed by the time max6 ships...but it was a 120 fps to .5 fps drop.

 

(The Quadro's were 120 to around 15-20 fps)

[garethace]

My own believe, is that a FireGL card is using alot of cpu power to gain the performance with about 1-5million vertices, and after that crumbles away, at 10-20 million vertices, because of lack of support from the cpu.

 

Whereas the 3dlabs wildcats tell a different story, as do the Quadro cards to a certain extent. The 3dslabs drivers are not calling out to the cpu, for so much help. This means, when the cpu isn't there to help - like with an aging x86 box, that has been working away content enough for a couple of years, without very many service packs or driver upgrades - i.e. if the system is not broken, don't fix it. Or if the model just becomes so large, that the cpu can no longer help out the graphics card drivers, calling for help - then the 3dslabs card isn't effected so much.

 

I mean, for solidworks, or any general cad program, where manipulating large models is a priority - most of those softwares are ancient anyhow - because for engineering your whole patent, design and work is stored in a .stl universal file format or something, to send straight to the manufacturer's machines - the lack of some 'fal-dals' in the 3dlabs cards, and the odd driver issue, isn't a real problem.

 

I had a bad driver issue with a Synergy card once i remember, if i installed a good MicroStation J driver from nVidia, the mouse would always turn into a square on the screen. If i installed the older driver, then performance went to crap in MicroStation J, but was fine in most other applications. Anyhow the Synergy card was built using a TNT2! So i upgraded to a Quadro2Pro, and since then can use most softwares, without any glitches whatsoever, because of the certified Quadro drivers. The latest 4x.xx series are great even with this old vintage Quadro card.

 

That is what you really need, a card to put in their and leave it for the engineer/product designer to work with for years. I would not upgrade the service pack in w2k anymore now, especially after that last sp4 scare with MAX. I think sp2 did a bit for dual processing alright, but thats about it.

 

What is the Quadro 4 based around? Is it Geforce 4 chip or something?

 

Anyhow, that Quadro 4 chip would still be a good investment, i think, over a gaming card, for the ability to have stable drivers in so many applications like you have said.

[Geoff Alder]

Always the dilemma. Which card to get.

 

How deep is your wallet? I would certainly go for the FX series, I am using the FX 2000 and it flies. Our clients are finding the lack of D3D support and driver install problems of the Wildcats a big turn off, particularly for max and combustion.

[TAKA]

ok, i just orderd an FX1000....it'll be here in a couple of days......i can't waitttt.....yiiha

 

thx greg

[Greg Hess]

Taka,

 

Make sure to download the latest maxtreme and drivers from www.nvidia.com.

 

Hope your new card install goes well!

[Chad Warner]

Realize this is an old thread, but has anyone had the chance to use the quadrofx 3000? and if so, how does it compare to the 1000? Thanks for any insight.

 

-Chad

[xgarcia]

Beware fellas, Maxtreme isnt released for Max6 yet (as of today, Feb. 03/2004). Really really stupid of Nvidia...