Sign in with
Sign up | Sign in
Your question

Gtx295 X2 vs ATI 5970??

Last response: in Graphics & Displays
Share
December 8, 2009 12:44:12 PM

Hi guys,

I saw that the ATI 5970 is now the most powerful graphic card currently. I looked at a lot of reviews and it is very clear that 1 5970 can easily beat 2 Gtx285 and 1 Gtx295. However, I am wondering if the 5970 will be able to knock-out 2 Gtx 295, or maybe 3 Gtx285?

Also, it seems that 2 5870 is slightly more powerful than a 5970. Is this true?

Thanks

More about : gtx295 ati 5970

December 8, 2009 12:53:56 PM

A Quad-SLI GTX295 is obviously more powerful than a single 5970 (4 vs 2).

The 5870 is clocked slightly higher than the 5970, that's why 2x5870 Cross-fired appears to be faster than the 5970. Although Sapphire has a factory-overclocked 5970 which solves that "problem". But you can always overclock it yourself...
a c 189 U Graphics card
December 8, 2009 2:20:18 PM

In my opinion, i would prefer only 1 card on my system, it doesn't consumes more power if u compared with 2 cards, in this case: 1 HD5970 vs 2 GTX295 X2.
More cards=more performance=more power needed=more heat=more bills every month. :) 
Anyway, that's just my opinion.
Anyone?
Related resources
a b U Graphics card
December 8, 2009 2:58:40 PM

2 5870's are more powerful than 2 gtx295's because sli/crossfire scaling drops significantly on the 3rd card, and hugely on the 4th.
a b U Graphics card
December 8, 2009 3:15:38 PM

The 5970 has a internal pci-e bridge connecting the two gpu's. So their bandwith is slightly limited by having 8 pci lanes each , on the 16x slot. Only noticeable going up against a 5870 crossfire on a platform with 16x X 16x. A lot of this is all theoretical derived by the few sites that have done these tests and also software is probably not completely tweaked for these all out set ups.
a c 130 U Graphics card
December 8, 2009 3:41:22 PM

As Jennyh said, SLI or even Crossfire drops performance scaling massively on the third and fourth card. And considering that the HD5970 is $400 cheaper than two GTX295's...you'd be better off Tri-Firing an HD5970 with an HD5870, which should easily beat the GTX295x2.
December 8, 2009 3:43:18 PM

wa1 said:
In my opinion, i would prefer only 1 card on my system, it doesn't consumes more power if u compared with 2 cards, in this case: 1 HD5970 vs 2 GTX295 X2.
More cards=more performance=more power needed=more heat=more bills every month. :) 
Anyway, that's just my opinion.
Anyone?

I agree 100%!

More heat also means = more noise. :pfff:  I don't want a jet engine in my room...
Scalling is also a problem, that's why I always try to avoid SLI/Crossfire when buying a new system. I try to get the most powerful single GPU availabe, then I don't have to worry about adding a new card later... Keeping it simple always, my motherboards usually don't even have a second PCI-E 16x slot.

Cheers.
December 8, 2009 3:46:51 PM

So, can u put 2 5970 together to run Quad cross fire?
a c 130 U Graphics card
December 8, 2009 3:47:47 PM

Four cards in Crossfire is called CrossfireX, and yes, it is possible.
December 8, 2009 4:26:36 PM

overclocker said:
So, can u put 2 5970 together to run Quad cross fire?



Yes, but unless you have several 30'' monitors, you're basically flushing money down the toilet...
a b U Graphics card
December 8, 2009 4:30:08 PM

At this point it would be ubsurd to even think about the 295, the GPU is just waaaay to overpriced and it's last-gen harware. Spend your money wisely don't just look at the numbers....

Why would anyone need dual 5970's??? E-PEEN???
December 8, 2009 4:33:41 PM

jennyh said:
2 5870's are more powerful than 2 gtx295's because sli/crossfire scaling drops significantly on the 3rd card, and hugely on the 4th.


2 5870s will not be more powerful than 2 295 GTXs (especially on paper), but 2 295s are very unstable, not all games are going to take advantage of the 3rd or 4th gpu (especially 4th).

Plus whats the point of having a high max and avg, if your min frame rates are killed.

I had 3 285s, which is more stable in 2 295s, and I had terrible min frame rates. Whats funny is that I switched to 1 5870, and while the avg frame rate wasn't as high (although not much lower) the min frame rate was MUCH higher

3 285s min was around 21-23
1 5870 min frame rate was 35 fps (yes amazing)

3 285s avg was around 50-80 (Honestly it was very picky, never was it the same all around)
1 5870 was 40-50

I would say alot morestable if you grab 1 5970, although they are downclocked u can OC them to stock:) 
December 8, 2009 4:38:25 PM

shadow187 said:
As Jennyh said, SLI or even Crossfire drops performance scaling massively on the third and fourth card. And considering that the HD5970 is $400 cheaper than two GTX295's...you'd be better off Tri-Firing an HD5970 with an HD5870, which should easily beat the GTX295x2.


Do you mean I can run 1 5970 and 1 5870 in crossfire mode at the same time?
December 8, 2009 4:40:19 PM

yes they are both the exact same card, only difference is different clocks.
December 8, 2009 4:44:03 PM

What if I really get 1 5970 and 1 5870, will the 5970 overclock to the 5870 clock speed?
December 8, 2009 4:48:46 PM

Does it matter? You can use ati overdrive to set it to w.e you want.

From my experience, I can prob put more money on it downclocking to match the slower card. But when I did my 285s, they each ran w.e they were clocked at. Not sure, when i was in game because I never checked.
December 8, 2009 4:57:56 PM

So, a conclusion:( from most powerful to less powerful)
2 5970> 1 5970+1 5870> 3 gtx285>2 gtx295>1 5970>2 5870>2 gtx285

I hope I'm right
December 8, 2009 5:03:40 PM

Sorry, mistype

it should be a conclusion: (from most powerful to least powerful)
a b U Graphics card
December 8, 2009 11:01:50 PM

^^ switch the 1 5970 with 2 5870 and i can agree with that lineup.
a b U Graphics card
December 8, 2009 11:14:07 PM

overclocker said:
So, a conclusion:( from most powerful to less powerful)
2 5970> 1 5970+1 5870> 3 gtx285>2 gtx295>1 5970>2 5870>2 gtx285

I hope I'm right


Its 2 5970> 1 5970+1 5870> 3 gtx285>1 5970>2 gtx295>2 5870>2 gtx285

A 5970 is faster than 2 GTX 295s, slightly, and far faster when overclocked.
December 8, 2009 11:36:12 PM

^ ummm???

No

2 295s exchanged blows with the 5970 depending on the game

Here is the chart


2 5970 > 1 5970 + 5870 > 3 285 gtx ~ 2 5870s > 2 295 GTXs ~ 5970 ~ 2 5850 ~> 2 285 GTX

for this 2 295 GTXs ~ 3 285 gtx ~ 2 5870s

Here are my calculations

Example:

5870 = 100 fps
285 GTX = 70 fps
2 5870 = 180 fps (80% scaling)
2 285 GTX = 126 fps (80% scaling)
3 285 GTX = 126 fps + 30% scaling (which is very generous) = 163.8 fps

180 fps vs 163.8 fps

So now the 5970 is a little weaker than the 2 5870s I would say 15-20% (judging by the clocks, lets say 15% to be fair)

1 5970 = 153 fps (vs 180 fps the 5870 has)

1 295 GTX should be around the same differnce between 2 285s, 15%

1 295 GTX = 107.1 fps (vs the 126 the 285 GTX has)
2 295 GTX = 107.1 + 30% scaling (which again is even more generous at times, since quad generall has a handful of games that like the scaling, COD being 1 of em)

2 295 GTX = 139.1 fps

Now lets give higher scaling

3 285s @ 80% scaling = 126 + 80% = 226 fps
2 295s @ 80% scaling = 107.1 + 80% = 192.78 fps


giving that 80% scaling is VERY slim, I'm sorry If I say that 295 GTX ~ 3 285s ~ 2 5870s

And to be frank, I don't even want to point out the min frame rates, such as my incident with the 3 285s being 21 fps min and the 5870 being 35, thats 66% higher.

Hopefully this came out well, I'm in class now so yeah give me a break:D 

Edit:

The reason I gave the 2 295 GTX a spot with the 5970 is because the calculations I did was for general power, and alot of games now are Nvidia loving, so on avg the 295 GTX should perform better on those (exampl COD games)

I should do a 50% scaling as well, since I do see alot of games scaling at 50% (but they are Nvidia meant to be played games)
a b U Graphics card
December 9, 2009 12:00:43 AM

L1quid we agree that 2 5870s are faster than 2 GTX 295s right? I mean every benchmark basically shows so. Well the 5970 is just an underclocked pair of 5870s and tie 2 5850s with immature drivers. So lets say that the 5970 is in between dual 5850s and dual 5870s. Meaning it is very close to dual GTX 295s and has far better minimum framerates which means it plays far better. So in that context the 5970 > 2 GTX 295. Do you argue that? You made the same point before hand. Now lets add in that a 5970 can be clocked, and is meant to be really, to 5870 speeds which then would make it faster even in benchmarks. So 5970 OC is >> than 2 GTX 295s...

I'm not seeing how what you are saying contradicts this, I'm going by benchmark results and an admitted semi-accurate assumption on minimum framerates to get mine. Your going on what seems to be random numbers and theoretical results.
December 9, 2009 12:12:02 AM

I'm going on what happens in a lot of games, whether the card can oc or not isn't the point.

Yes the min frame rates square it a winner, but like you said for theoretical results and avg results they should be about the same.
Whether it can oc or not is beside the point. Such as "FAR FASTER", I wouldn't say its far faster. I would say its faster even with the OC.

A 5850 can OC very high enough to pretty much match a 5870 @ stock, and wouldn't be suprised if it passes it too, that doesn't mean its a better card.

Also the random numbers are taken from power percentages.

Is the 5970 a winner over the 2 295 GTX? In my book yes, because its alot more stable with better min frame rates.

But please ppl need to stop exaggerating the percentages

Here this should explain it better than me, and might actually prove me wrong in parts.

Average summary @ 2560x1600 resolution.

a b U Graphics card
December 9, 2009 12:20:21 AM

L1qu1d said:
I'm going on what happens in a lot of games, whether the card can oc or not isn't the point.

Yes the min frame rates square it a winner, but like you said for theoretical results and avg results they should be about the same.
Whether it can oc or not is beside the point. Such as "FAR FASTER", I wouldn't say its far faster. I would say its faster even with the OC.

A 5850 can OC very high enough to pretty much match a 5870 @ stock, and wouldn't be suprised if it passes it too, that doesn't mean its a better card.

Also the random numbers are taken from power percentages.

Is the 5970 a winner over the 2 295 GTX? In my book yes, because its alot more stable with better min frame rates.

But please ppl need to stop exaggerating the percentages

Here this should explain it better than me, and might actually prove me wrong in parts.


Maybe far faster is more opinion than fact, I admit. But common, the difference between the playability is pretty significant atleast. So, as I said, the 5970 > 2 GTX 295, slightly, with a 5970 OC being "noticeably" faster. I only had problems with the theoretical numbers, they don't account for variables such as bottlenecks, game design, benchmarking technique, drivers, the individual cards, ext.
December 9, 2009 12:29:16 AM

nothing does, the theoretical numbers, is an example of how it works for power. I mean what bottlenecks 2 295s will bottleneck others around its power.

I mean if you start saying game design and bottlenecking comesinto play, then you won't be able to compare anything anymore.

I mean look at Crysis, it runs faster on 2 285s (avg frame) than 2 5870s, are you going to say that 2 285s are faster? No its an Nv game.

So what I did was take percentages for general games.


Again look at the chart, the 295 GTX is just 23% slower than 1 5970... I think 2 295s can scale atleast 20% don't u?

Also overclock the 5970 to match 2 5870s and thats what 23%+ 15%? 38%, I don't think its Going to be NOTICEABLY different

Please use more appropriate words, especially in most games (excluding Crysis and Stalker) it won't be so easy to notice 130 vs 160 fps.

So conclusion I would say the 2 295 GTX ~ 2 5870s ~ 5970. Min frame aside, for over all power your paying for.

Min frames might be different, it prob will favor the 5870/5970 just like I said, but come on, lets not sell a card short.

This ATI love has to go away, we're all enthusiasts and should go for the company we need not the one we have a highschool crush on (NOT talking about you, I'm just in general speaking to some users).

I switch from Company to company to get what I want.

I hope Jaydee or TGGA will enlighten us both, I'm sure we're both missing something, and they seem to catch it.
a b U Graphics card
December 9, 2009 12:49:55 AM

L1qu1d said:
nothing does, the theoretical numbers, is an example of how it works for power. I mean what bottlenecks 2 295s will bottleneck others around its power.

I mean if you start saying game design and bottlenecking comesinto play, then you won't be able to compare anything anymore.

I mean look at Crysis, it runs faster on 2 285s (avg frame) than 2 5870s, are you going to say that 2 285s are faster? No its an Nv game.

So what I did was take percentages for general games.


Again look at the chart, the 295 GTX is just 23% slower than 1 5970... I think 2 295s can scale atleast 20% don't u?

Also overclock the 5970 to match 2 5870s and thats what 23%+ 15%? 38%, I don't think its Going to be NOTICEABLY different

Please use more appropriate words, especially in most games (excluding Crysis and Stalker) it won't be so easy to notice 130 vs 160 fps.

So conclusion I would say the 2 295 GTX ~ 2 5870s ~ 5970. Min frame aside, for over all power your paying for.

Min frames might be different, it prob will favor the 5870/5970 just like I said, but come on, lets not sell a card short.

This ATI love has to go away, we're all enthusiasts and should go for the company we need not the one we have a highschool crush on (NOT talking about you, I'm just in general speaking to some users).

I switch from Company to company to get what I want.

I hope Jaydee or TGGA will enlighten us both, I'm sure we're both missing something, and they seem to catch it.


Look in the sig. bro no ATI favoritism here. The point is that the 5970 will ATLEAST match 2 GTX 295s and when you pay $600-$1000 for these set ups then, yes, 130 vs 160 FPS is important because the difference will be important when a game comes along that makes that 40 vs 50 FPS. Lets say that my words are too strong, thats fine. Is this too strong? For almost twice the price, buying 2 GTX 295s instead of a 5970 is stupid. I think we can both agree on that. I'm not arguing that a 5970 is to a pair of GTX 295s as a 5870 is to a 4870 or even a 4890, I'm saying that a 5970 is > than 2 GTX 295s, no matter how insignificant the difference, and an overclocked 5970 is > than 2 GTX 295s by a larger amount, no matter how insignificant.

Oh and as an aside, I'm pretty sure Crysis runs faster on dual 5870s than dual GTX 285s, but the difference isn't that large. I am only looking at Crysis Warhead benchmarks, if you have some to the contrary I would like to see them. I hate how they substituted Warhead for the original Crysis even though the original Crysis taxed hardware even more so than Warhead.
a b U Graphics card
December 9, 2009 12:51:46 AM

Dual 5870's thrash dual gtx285's in everything. I saw one review where dual 5870's were beating tri gtx285's in everything as well, I'll try to dig that up.

There we go - http://benchmarkextreme.com/Articles/HD%205870%20TriFir...

Everthing except Crysis Warhead I think, and that has to be drivers. It seems from that 5870 crossfire = gtx285 tri-sli but only because of Crysis Warhead bringing the 285's much closer than they really are.
December 9, 2009 1:35:28 AM

well thats what I was trying to prove^ that in everything else the 285s loose, yet in Crysis it doesn't so we can't say its stronger than the 5870s just because of that.

@ AMW1011 re read my post about the fanboysim, it wasn't directed at u.

As for buying it I agree, I wouldn't get 2 295s, but I'll tell the truth about performance so this isn't about which to buy what I'm talking it

If it were which to buy I'd say 5970 any day, but I would like to point out to not exaggerate.
!