Nvidia FX 5900 Benchmarks

[mattg1981]

I am not going to start a ATi vs Nvidia thread here ... but benchmarks from solid websites all point that the FX 5900 Ultra outperforms even the 9800 256 card.

However, I am not going to spend that much for a 5900 Ultra ... however, I can get my hands on a 5900 non-ultra (128 MB) and I just want to know how that stacks up compared to the 9800 non Pro and Pro. All the comparisons and benchmarks I have seen are for the Ultra version ... has anybody worked with the non Ultra or seen Non Ultra benchmarks?

 

[T9D]

..

 

[LesPaul]

yeah, I noticed that too.. In all the standard time demos and synthetic benchmarks, the 5900u smoked.. but when sites started using custom demos the 9800 pros started beating them in just about everything. I'm not sure how the 5900 non-ultra stacks up, sorry.

 

[Regs]

http://www.anandtech.com/#19834

The 5900s we saw today are pretty standard performers. That's their problem, being standard. When you pay in excess of £300 you want more than just standard, you expect to open the box, install the card and be blown away by it's speed when you turn the graphical detail up. Out of all the FX 5900s we saw today, none really did that. In fact it made us think how good the 9700 PRO still is, considering it's coming up to it's first birthday.

In a way they are correct. The new 5900 ultra is what should of been the 5800 ultra when it first came out. Why charge 399 dollars? I hate that. These over priced cards are simply not worth it right now. I am not going to pay 500 dollars to play half-life 2.

 

[FishTankX]

Originally posted by: Regs
http://www.anandtech.com/#19834

The 5900s we saw today are pretty standard performers. That's their problem, being standard. When you pay in excess of £300 you want more than just standard, you expect to open the box, install the card and be blown away by it's speed when you turn the graphical detail up. Out of all the FX 5900s we saw today, none really did that. In fact it made us think how good the 9700 PRO still is, considering it's coming up to it's first birthday.

In a way they are correct. The new 5900 ultra is what should of been the 5800 ultra when it first came out. Why charge 399 dollars? I hate that. These over priced cards are simply not worth it right now. I am not going to pay 500 dollars to play half-life 2.

FX5900 and FX5900 ultra are not the same thing. The FX5900 is a cut down version (clock wise) from the fX5900 ultra. The FX5900 ultra smokes the 9800 pro alot of the time, in the firingsquad benches. But the FX5900 ultra is pretty much between the 9800 and the 9700

 

[Schadenfroh]

Originally posted by: Regs
http://www.anandtech.com/#19834

The 5900s we saw today are pretty standard performers. That's their problem, being standard. When you pay in excess of £300 you want more than just standard, you expect to open the box, install the card and be blown away by it's speed when you turn the graphical detail up. Out of all the FX 5900s we saw today, none really did that. In fact it made us think how good the 9700 PRO still is, considering it's coming up to it's first birthday.

In a way they are correct. The new 5900 ultra is what should of been the 5800 ultra when it first came out. Why charge 399 dollars? I hate that. These over priced cards are simply not worth it right now. I am not going to pay 500 dollars to play half-life 2.

ValVe stated that a $4 tnt2 card will run HL2

 

[FishTankX]

Originally posted by: Regs
http://www.anandtech.com/#19834

The 5900s we saw today are pretty standard performers. That's their problem, being standard. When you pay in excess of £300 you want more than just standard, you expect to open the box, install the card and be blown away by it's speed when you turn the graphical detail up. Out of all the FX 5900s we saw today, none really did that. In fact it made us think how good the 9700 PRO still is, considering it's coming up to it's first birthday.

In a way they are correct. The new 5900 ultra is what should of been the 5800 ultra when it first came out. Why charge 399 dollars? I hate that. These over priced cards are simply not worth it right now. I am not going to pay 500 dollars to play half-life 2.

Because just the memory costs 250$. Don't believe me? Look up the price of 256 megabit DDR 480MHZ Chips. Not CHEAP! Then you've got the board, the chip itself (Which probably yields like $hit) and then Nvidia and the board makers have to make a buck and you quickly breach the 400$ barrier.

If you want, you can wait for a version with less memory. But 256MB of graphics RAM costs a *bunch*. Plus, remember now.. the 256MB Radeon 9800 pro costs just as much as the F5900 ultra. And underperforms it, too.

 

[Pete]

Both Firingsquad and UKGamer reviewed the MSI 5900 recently.

 

[mattg1981]

This was posted just recently ....

5900 Non Ultra/Ultra Benches

 

[DaveSimmons]

If you want, you can wait for a version with less memory. But 256MB of graphics RAM costs a *bunch*. Plus, remember now.. the 256MB Radeon 9800 pro costs just as much as the F5900 ultra. And underperforms it, too.

...though apparently mostly on standardized game timedemos that nvidia has "optimized" it for.

But yes, 256MB of ultrafast DDR is the budget killer. Given that 128MB is enough for 1600x1200 and games for the next couple of years, going with 256MB is, in hindsight, yet another bad decision by nvidia.

Since the 5900 is not really faster than the 9800 Pro 128 MB, selling it for $100 more is not going to lure many savvy people away from buying ATI instead.

I'm using a geforce3 now, but my next card is probably a 9800 Pro.

 

[Regs]

My mistake for not seeing they were only reveiwing the 5900 non-ultra.

 

[GTaudiophile]

Originally posted by: mattg1981
This was posted just recently ....

5900 Non Ultra/Ultra Benches

Aside from the 3DMark2003 Shader bench, I'd say that no card really spanks the other (5900 Ultra vs. 9800 PRO). Both solid cards, both having their pluses and minuses (though the 9700 and 9800 PRO 128MBs are cheaper and more widely available).

Again, big kudos to ATi. Their almost one-year-old R3XX core is still going strong. Few graphics processors have stood so well against the test of time.

 

[Regs]

I think I'll save the 100 dollars, get the 9700 pro and OC the bastid. If Nvidia made a good card like the 5900 ultra for a lower price I'm there. The 9700 pro will be more than enough for half life 2 , 4x FSAA @ 1280x1024.

 

[codehack2]

Originally posted by: FishTankX

Originally posted by: Regs
http://www.anandtech.com/#19834

The 5900s we saw today are pretty standard performers. That's their problem, being standard. When you pay in excess of £300 you want more than just standard, you expect to open the box, install the card and be blown away by it's speed when you turn the graphical detail up. Out of all the FX 5900s we saw today, none really did that. In fact it made us think how good the 9700 PRO still is, considering it's coming up to it's first birthday.

In a way they are correct. The new 5900 ultra is what should of been the 5800 ultra when it first came out. Why charge 399 dollars? I hate that. These over priced cards are simply not worth it right now. I am not going to pay 500 dollars to play half-life 2.

Because just the memory costs 250$. Don't believe me? Look up the price of 256 megabit DDR 480MHZ Chips. Not CHEAP! Then you've got the board, the chip itself (Which probably yields like $hit) and then Nvidia and the board makers have to make a buck and you quickly breach the 400$ barrier.

If you want, you can wait for a version with less memory. But 256MB of graphics RAM costs a *bunch*. Plus, remember now.. the 256MB Radeon 9800 pro costs just as much as the F5900 ultra. And underperforms it, too.

Err.... I have no idea on what is actually driving up the cost of these 256mb boards, but you are wrong on the memory... it's not 256bit... it's standard 64bit DDR running on quad channel controller.

CH2

 

[NesuD]

Nvidia "optimized" the hell out of their drivers for established benchmarks and test demos. If you want to see an example of this run the 3dmark 2003 benchmark then rename the 3dmark2003.exe to something different and run it with the new name. As I understand it renaming the EXE tricks the nvidia drivers so that the specially "optimized" settings in Nvidias driver for 3dmark2003 are not used. Performance I'm told drops about 20%.

 

[GaryShandling]

Hehe, such naive people believing the 5900 ultra outperforms the 9800pro 256.

After a review on a website is concluded and states that the 5900 ultra is better then the 9800 pro 256. The editors go on about "We would like to thank Nvidia in participating (Altering) in this test"

I wouldn't trust nvidia and there scores with what will be the "biggest selling" games. It's a marketing tactic, and most of you have fallen for it.

 

[FishTankX]

Originally posted by: codehack2

Originally posted by: FishTankX

Originally posted by: Regs
http://www.anandtech.com/#19834

The 5900s we saw today are pretty standard performers. That's their problem, being standard. When you pay in excess of £300 you want more than just standard, you expect to open the box, install the card and be blown away by it's speed when you turn the graphical detail up. Out of all the FX 5900s we saw today, none really did that. In fact it made us think how good the 9700 PRO still is, considering it's coming up to it's first birthday.

In a way they are correct. The new 5900 ultra is what should of been the 5800 ultra when it first came out. Why charge 399 dollars? I hate that. These over priced cards are simply not worth it right now. I am not going to pay 500 dollars to play half-life 2.

Because just the memory costs 250$. Don't believe me? Look up the price of 256 megabit DDR 480MHZ Chips. Not CHEAP! Then you've got the board, the chip itself (Which probably yields like $hit) and then Nvidia and the board makers have to make a buck and you quickly breach the 400$ barrier.

If you want, you can wait for a version with less memory. But 256MB of graphics RAM costs a *bunch*. Plus, remember now.. the 256MB Radeon 9800 pro costs just as much as the F5900 ultra. And underperforms it, too.

Err.... I have no idea on what is actually driving up the cost of these 256mb boards, but you are wrong on the memory... it's not 256bit... it's standard 64bit DDR running on quad channel controller.

CH2

Excuse me, but I was reffering to the capacity of the chips.

256 megabit chips hold 32 megabytes each. They are used in a 256 megabit X 8 chip configuration to bring you 256 megabytes.

Generally speaking in the memory market 'Bits' are what is used to describe the capacity of individual chips, while 'Bytes' are used to describe complete memory sticks or modules.

 

[FishTankX]

Originally posted by: GaryShandling
Hehe, such naive people believing the 5900 ultra outperforms the 9800pro 256.

After a review on a website is concluded and states that the 5900 ultra is better then the 9800 pro 256. The editors go on about "We would like to thank Nvidia in participating (Altering) in this test"

I wouldn't trust nvidia and there scores with what will be the "biggest selling" games. It's a marketing tactic, and most of you have fallen for it.

If you really want to believe that, then look at IL-2 sturmovik.

It's not exactly an industry standard benchmark, hell.. it's a flight sim. But the FX5900 ultra still manages to beat the 9800 pro in that benchmark. Ditto for all the other benchmarks firingsquad was using.

 

[tbates757]

Originally posted by: Regs
I think I'll save the 100 dollars, get the 9700 pro and OC the bastid. If Nvidia made a good card like the 5900 ultra for a lower price I'm there. The 9700 pro will be more than enough for half life 2 , 4x FSAA @ 1280x1024.

And who told you that a 9700 Pro will run HL2 at 1280x1024 with 4x FSAA? The wind? I'd be surprised if it ran 1024x768 max settings without any AA / AF at above 30fps on a ~2.5ghz cpu

 

[GaryShandling]

Originally posted by: FishTankX

Originally posted by: GaryShandling
Hehe, such naive people believing the 5900 ultra outperforms the 9800pro 256.

After a review on a website is concluded and states that the 5900 ultra is better then the 9800 pro 256. The editors go on about "We would like to thank Nvidia in participating (Altering) in this test"

I wouldn't trust nvidia and there scores with what will be the "biggest selling" games. It's a marketing tactic, and most of you have fallen for it.

If you really want to believe that, then look at IL-2 sturmovik.

It's not exactly an industry standard benchmark, hell.. it's a flight sim. But the FX5900 ultra still manages to beat the 9800 pro in that benchmark. Ditto for all the other benchmarks firingsquad was using.

If it's ment to be faster why does the 5900 ultra give out less frames per second then the 9800 256 in many new games? There altering benchmarks once more.

 

[FishTankX]

Originally posted by: GaryShandling

Originally posted by: FishTankX

Originally posted by: GaryShandling
Hehe, such naive people believing the 5900 ultra outperforms the 9800pro 256.

After a review on a website is concluded and states that the 5900 ultra is better then the 9800 pro 256. The editors go on about "We would like to thank Nvidia in participating (Altering) in this test"

I wouldn't trust nvidia and there scores with what will be the "biggest selling" games. It's a marketing tactic, and most of you have fallen for it.

If you really want to believe that, then look at IL-2 sturmovik.

It's not exactly an industry standard benchmark, hell.. it's a flight sim. But the FX5900 ultra still manages to beat the 9800 pro in that benchmark. Ditto for all the other benchmarks firingsquad was using.

If it's ment to be faster why does the 5900 ultra give out less frames per second then the 9800 256 in many new games? There altering benchmarks once more.

Can you give me a few links? I'd like to investigate that.

 

[GaryShandling]

Originally posted by: FishTankX

Originally posted by: GaryShandling

Originally posted by: FishTankX

Originally posted by: GaryShandling
Hehe, such naive people believing the 5900 ultra outperforms the 9800pro 256.

After a review on a website is concluded and states that the 5900 ultra is better then the 9800 pro 256. The editors go on about "We would like to thank Nvidia in participating (Altering) in this test"

I wouldn't trust nvidia and there scores with what will be the "biggest selling" games. It's a marketing tactic, and most of you have fallen for it.

If you really want to believe that, then look at IL-2 sturmovik.

It's not exactly an industry standard benchmark, hell.. it's a flight sim. But the FX5900 ultra still manages to beat the 9800 pro in that benchmark. Ditto for all the other benchmarks firingsquad was using.

If it's ment to be faster why does the 5900 ultra give out less frames per second then the 9800 256 in many new games? There altering benchmarks once more.

Can you give me a few links? I'd like to investigate that.

Im trying to find it, it was on a website showing unreal fps on both cards in various levels with different performance settings.

Edit: Here are the comparitive links in 1 scenario using basically equal but varied settings. Fps is in top right corner of the picture.


5900

9800 pro

 

[WhyteRyce]

Originally posted by: GTaudiophile

Originally posted by: mattg1981
This was posted just recently ....

5900 Non Ultra/Ultra Benches

Aside from the 3DMark2003 Shader bench, I'd say that no card really spanks the other (5900 Ultra vs. 9800 PRO). Both solid cards, both having their pluses and minuses (though the 9700 and 9800 PRO 128MBs are cheaper and more widely available).

Again, big kudos to ATi. Their almost one-year-old R3XX core is still going strong. Few graphics processors have stood so well against the test of time.

As opposed to previous cores that lasted well over a year?

 

[MangoTBG]

Originally posted by: WhyteRyce

As opposed to previous cores that lasted well over a year?

Wow. You registered in march, and wait for this to be your first post? It's not about "lasting" as in, in use. Of course people are still using GeForce 2s that came out how many years ago? It's about longevity in the topspot. The 9700pro is still up there killing all the games that come out, and with some overclocking I'm sure it will still be eating up all the games for another year or so. But, hey! Thanks for your input!!!!

 

[gramboh]

Pretty sure the machine using to demo HL2 at E3 was 9800 Pro 256meg in a P4 3.06 (dunno about RAM). It didn't look like there was any AA/AF and probably around 1024x768 maybe 1192x864. Also you could see it slow down (below 30fps) for a few frames here and there (the street combat scene, when there are large explosions). Then again the engine probably isn't fully optimized, and I'm sure ATi/Nvidia will be optimizing (hopefully not cheating) their drivers for HL2 since it will be a huge game for the next several years.

I don't think you will be able to do any AA/AF on HL2 with a 9700/9800/5900 series card though, maybe at 800x600.