5870 TRIFIRE vs GTX 285 TRI SLI....review

[CH33TAH]

http://benchmarkextreme.com/Articles/HD%205870%20TriFire/P1.html

5870 Rocks !!

[Utnorris]

Sorry, but I cannot take this review seriously. At what point does a 850 watt PSU produce 128 amps?

"The 850 W might seem like a low wattage but it boasts a whopping 128 Amps on the combined 12V rails!"

Seriously? So I guess that 1000 watt PSU's are just underrated. Oh well, interesting review with no real surprises. I would assume that if he used 9.10 drivers he would have had better CF results, but hey, it's a review.

[CH33TAH]

Some PSUs are rated as continuous power while others at peak power. The cooler master 850 watt used for the review is rated at 850 W continuous / 1000 Watt peak power with 6 virtual 12 V rails. Two have max 28 Amps each on them while the rest have 18 amps. Cooler master could have easily rated this as 1000 W but they just played it safe and labelled it as 850 W. The "128 Amps" should not have been there since it is inaccurate. It has been removed now.

[flopper]

its basically 3x4870x2 vs 3x280gtx.

I wanna know how crossfire/trifire/quadfire scale in eyefinity, anything else is not that interesting.

[Blkout]

its basically 3x4870x2 vs 3x280gtx.

I wanna know how crossfire/trifire/quadfire scale in eyefinity, anything else is not that interesting.

No, it's not basically 3 4870x2's and you won't have the same scaling issues with 3 GPU's vs 6 GPU's. I guarantee that you can compare 3 5870's to 3 4870x2's and the 5870's will win everytime. Also for the record, the 285's are faster than the 280's so that's not the same either. Considering that the 5870 is in the same price category as the 285, I think this is a good comparison that I haven't seen done anywhere else.

[spajdr]

its basically 3x4870x2 vs 3x280gtx.

I wanna know how crossfire/trifire/quadfire scale in eyefinity, anything else is not that interesting.

Crossfire dont work with eyefinity yet.
http://www.guru3d.com/news/radeon-hd-5870-quadfire-user-video/

Anyway nice test, but i want to see how Crossfire improves min.fps in games.

[Blkout]

http://benchmarkextreme.com/Articles/HD%205870%20TriFire/P1.html

5870 Rocks !!

I notce you made a note was made that the Warhead benchmarks are being redone which is good because those aren't very accurate. I won't deny that Nvidia's SLI scales better than ATI's crossfire, much more in some games, but that kinds of difference it not accurate. Correct numbers should look like the 5870's pulling about identical numbers as the 285's, even though the 5870's are more powerful. Crysis and Warhead are just two games where Nvidia has a slight advantage over ATI.

The problem with the Vantage benchmark is that the default performance preset was used instead of the extreme preset. Who would use 3 video cards at 1280x1024 resolution with no AA? If the benchmarks were done at extreme settings, the 5870's would walk all over the 285's. A 285 is just as fast if not faster than a 5870 at low resolutions, but again, who would be using either of these cards at that resolution?

[Bodar]

Thanks you guys for checking it out!! :)

[dan7777]

wow! crysis war 285 tri 75fps against 5870 tri 55fps lol good old NV

[Blkout]

wow! crysis war 285 tri 75fps against 5870 tri 55fps lol good old NV

I've already stated those results aren't accurate. I get 62fps with my dual 5850's at stock speed and my i7 at 3.8GHz vs the test setup at 4.2GHz.

Sure the 5850 is about 15% more powerful than the 285, but in Crysis and Warhead they are very close due to the fact that Nvidia has a slight advantage, in the end, the 5850 is about equal to a 285 in Crysis, and the 5870 is slightly faster which is why these test results for Crysis are garbage.

[texasreefer]

Yeah, but this is no new phenomenon....crysis has always scaled better with nVidia cards. Anything 'nVidia - the way it's meant to be played' is not going to scale as well. I don't think that the whole review should be debunked just because he/or she, used a 850W power supply. I use a 750W PC power on 4 cores, that are less efficient than the 5870, and I have had no problems. I'm OCed higher than 4.2, and have both cards OCed as well. THat is just the edge that nVidia has always had, and now that people are starting to jump on board with ATI....I bet future games will not have that 'Crytek' advantage.

[Blkout]

Yeah, but this is no new phenomenon....crysis has always scaled better with nVidia cards. Anything 'nVidia - the way it's meant to be played' is not going to scale as well. I don't think that the whole review should be debunked just because he/or she, used a 850W power supply. I use a 750W PC power on 4 cores, that are less efficient than the 5870, and I have had no problems. I'm OCed higher than 4.2, and have both cards OCed as well. THat is just the edge that nVidia has always had, and now that people are starting to jump on board with ATI....I bet future games will not have that 'Crytek' advantage.

No doubt about that, I use a Seasonic M12 750w PS with my i7 at 3.8GHz and dual 5850's overclocked to 850/1300 with watercooling on the CPU. My power supply holds up just fine.

[texasreefer]

For some reason I have always had problems with EA games....they're the best IMO, but i always get minor tears in the post level vids. The intros give me some tearing also. For Crysis (and warhead) I have to enable vsync and triple buffering, and turn off motion blur. Thats the only way to get smooth intro and video playback at enthusiest at 1080p in my 47" westinghouse.

[Utnorris]

Yeah, but this is no new phenomenon....crysis has always scaled better with nVidia cards. Anything 'nVidia - the way it's meant to be played' is not going to scale as well. I don't think that the whole review should be debunked just because he/or she, used a 850W power supply. I use a 750W PC power on 4 cores, that are less efficient than the 5870, and I have had no problems. I'm OCed higher than 4.2, and have both cards OCed as well. THat is just the edge that nVidia has always had, and now that people are starting to jump on board with ATI....I bet future games will not have that 'Crytek' advantage.


No doubt about that, I use a Seasonic M12 750w PS with my i7 at 3.8GHz and dual 5850's overclocked to 850/1300 with watercooling on the CPU. My power supply holds up just fine.

No one said it was because he used a 850watt PSU, it was that he stated it would deliver 128amps, which is incorrect. If he didn't understand the basics of PSU's, how can he be taken seriously to give a proper review of GPU's. He has also been critizised for using older drivers and running various benchmarks at low rez, kinda defeats the point of CF or SLI.

[texasreefer]

I just don't understand where you're getting 128amps, this is what he says:

"It is rated 850 W continuous and 1000 Watt Peak with 6 12 V rails. Two of the rails have 28 Amps each on them while the rest have 18 Amps."

Correct me if i'm wrong....not trying to be a dip stick or anything :shrug:

From what I see he did list an err with the 28amp, because on the egg it lists all 6 rails at 18A. Still wouldn't disregard the review.

http://www.newegg.com/Product/Product.aspx?Item=N82E16817171017&Tpk=coolermaster%20850

EDIT - On second thought he may be using a different model

EDIT 2 - WHen you use these suckers " quotations " you're supposed to enter your quote verbatim. Now i'm being a dip stick. :stick:

[cstkl1]

http://benchmarkextreme.com/Articles/HD%205870%20TriFire/P1.html

5870 Rocks !!

+1
dont care about the psu

but the results seems to be right
based on gtx285 sli/tri-sli setups that i have tested.

[Smartidiot89]

I want to see a Catalyst 9.9 vs. 9.10 review ^^

[Dezmen]

Asked in other thread but..

Is there any point to have 5870CF + GTX285. Would GTX285 add some performance or just PshyX support(with cracked drivers) ?

[Utnorris]

I just don't understand where you're getting 128amps, this is what he says:

"It is rated 850 W continuous and 1000 Watt Peak with 6 12 V rails. Two of the rails have 28 Amps each on them while the rest have 18 Amps."

Correct me if i'm wrong....not trying to be a dip stick or anything :shrug:

From what I see he did list an err with the 28amp, because on the egg it lists all 6 rails at 18A. Still wouldn't disregard the review.

http://www.newegg.com/Product/Product.aspx?Item=N82E16817171017&Tpk=coolermaster%20850

EDIT - On second thought he may be using a different model

EDIT 2 - WHen you use these suckers " quotations " you're supposed to enter your quote verbatim. Now i'm being a dip stick. :stick:

When the review first came out the author stated the 128amps, after several people pointed his mistake and explained how it was impossible he corrected the review. As I stated before, this review has been discussed on several forums already.

[Blkout]

He has also been critizised for using older drivers and running various benchmarks at low rez, kinda defeats the point of CF or SLI.

I stated this early in the thread, but thanks for pointing it out again, Captain Obvious.

[Blkout]

Asked in other thread but..

Is there any point to have 5870CF + GTX285. Would GTX285 add some performance or just PshyX support(with cracked drivers) ?

You can only use a secondary Nvidia card for PhysX support until the Lucid Hydra chip is available.

[CH33TAH]

I stated this early in the thread, but thanks for pointing it out again, Captain Obvious.

Besides 3D mark vantage all games had 2560 X 1600.. what more do you want? As for drivers, besided Crysis, they scaled pretty well with all other games, which is on par with the latest drivers. I think the review is being over-bashed.

[texasreefer]

Also, 3dmark Vantage has always been tested using the default settings for comparative purposes. I understand that it is kinda pointless to use that much power on a low resolution, but a run on extreme could hardly look great with ATI cards. nVidia would've taken that benchmark also, Vantage is a horrible bench for ATI. It's been long debated with the whole physx test and even getting a score had been an issue with extreme settings.



You can only use a secondary Nvidia card for PhysX support until the Lucid Hydra chip is available.

Blkout, considering this, would a secondary card for physX increase a Vantage benchmark? Can you explain breifly the Lucid Hydra chip?

[Behemot]

There are mistakes in graphs on page 4 (Crysis Warheads Scaling) - HD 4870 instead of HD 5870 :)

But the result are nice. I have to learn a lot, so I can work for ČEZ (czech energetic gigant). Than I'd have electricity for free and I'd buy three pcs of HD 7870 by that time :ROTF::surf::rocker:

[texasreefer]

You can only use a secondary Nvidia card for PhysX support until the Lucid Hydra chip is available.

Dezmen....You don't need the lucid hydra chip to run physx with an ATI setup. A gtx 285 would just be considered massive Overkill.

EDIT - I don't think ATI/nVidia are going to join forces like the Lucid chip would entail, I think the x58 platform was their answer to that. 100% scaling with multiple GPUs would be nice.....way too nice.

[Blkout]

Besides 3D mark vantage all games had 2560 X 1600.. what more do you want? As for drivers, besided Crysis, they scaled pretty well with all other games, which is on par with the latest drivers. I think the review is being over-bashed.

I don't want anything, I've already stated my concerns with this comparison earlier in the thread. I stated that the Crysis benchmarks are just flat out wrong, maybe due to a driver issue, but obviously not correct as I pointed out and that the Vantage benchmark was a waste at the default setting. Extreme would have been the only way to test such a setup.

And with that being said, I didn't bash the comparison, in fact I believe I was the first one to defend it. I simply stated my concerns with Crysis and Vantage, which are completely valid.

[Blkout]

Blkout, considering this, would a secondary card for physX increase a Vantage benchmark? Can you explain breifly the Lucid Hydra chip?

Yes it would increase the score if PhysX was enabled. I don't test Vantage with PhysX enabled, even when I had an Nvidia card, it's a cheat and nothing more.


http://www.anandtech.com/video/showdoc.aspx?i=3646

[CH33TAH]

I don't want anything, I've already stated my concerns with this comparison earlier in the thread. I stated that the Crysis benchmarks are just flat out wrong, maybe due to a driver issue, but obviously not correct as I pointed out and that the Vantage benchmark was a waste at the default setting. Extreme would have been the only way to test such a setup.



And with that being said, I didn't bash the comparison, in fact I believe I was the first one to defend it. I simply stated my concerns with Crysis and Vantage, which are completely valid.

I hear you. Crysis, yea something wierd is going on there. But it is not out of line with the information available on the internet. There is discrepencies in scaling but you never see Crysis scaling really well in any of the reviews out there. I am not saying a driver release cant fix that, but I highly doubt it. Running Crysis with 8.66 vs 9.11 yielded gains fro me but not a whole lot of improvements. As a reference, here is the scaling comparison from various websites at their highest settings tested. (scaling = FPS in multi GPU / FPS in single GPU). This is for Crysis / Crysis Warhead.

Driverheaven ---> 1.21
Guru3d ---------> 1.28
BenchEx --------> 1.36
Anandtech------> 1.40
Toms-----------> 1.43
Overclock3d-----> 1.51
Tweaktown------> 1.57


As you can see Benchex is about the same as Guru3d, Anandtech, and Toms.

As for Vantage, I agree with you to some extent. Having said that, I think you wont see much of an ATI lead even at extreme preset. As someone else stated earlier, chances are that nvidia's lead would increase and running at lower resolution might actually be going in ATI's favor.

[Blkout]

I hear you. Crysis, yea something wierd is going on there. But it is not out of line with the information available on the internet. There is discrepencies in scaling but you never see Crysis scaling really well in any of the reviews out there. I am not saying a driver release cant fix that, but I highly doubt it. Running Crysis with 8.66 vs 9.11 yielded gains fro me but not a whole lot of improvements. As a reference, here is the scaling comparison from various websites at their highest settings tested. (scaling = FPS in multi GPU / FPS in single GPU). This is for Crysis / Crysis Warhead.

Driverheaven ---> 1.21
Guru3d ---------> 1.28
BenchEx --------> 1.36
Anandtech------> 1.40
Toms-----------> 1.43
Overclock3d-----> 1.51
Tweaktown------> 1.57


As you can see Benchex is about the same as Guru3d, Anandtech, and Toms.

As for Vantage, I agree with you to some extent. Having said that, I think you wont see much of an ATI lead even at extreme preset. As someone else stated earlier, chances are that nvidia's lead would increase and running at lower resolution might actually be going in ATI's favor.



Well, it's not just the ATI scaling issue in Crysis, the numbers are just low for any 5870 configuration. My dual 5850's score much higher than what that comparison showed and I'm using my i7 at 3.8GHz instead of 4.2GHz and of course dual 5850's aren't quite as fast as dual 5870's in stock form. Those numbers just don't jive with my benchmark numbers and I've done a lot of benching with this setup.

[KingOfsorroW]

Thanks for the link! That was the info I was looking for...

[texasreefer]

Yes it would increase the score if PhysX was enabled. I don't test Vantage with PhysX enabled, even when I had an Nvidia card, it's a cheat and nothing more.


http://www.anandtech.com/video/showdoc.aspx?i=3646

THats' pretty awesome stuff there... sounds too good to be true.

[Unbornchild]

I notce you made a note was made that the Warhead benchmarks are being redone which is good because those aren't very accurate. I won't deny that Nvidia's SLI scales better than ATI's crossfire, much more in some games, but that kinds of difference it not accurate. Correct numbers should look like the 5870's pulling about identical numbers as the 285's, even though the 5870's are more powerful. Crysis and Warhead are just two games where Nvidia has a slight advantage over ATI.

(...)

If the benchmarks were done at extreme settings, the 5870's would walk all over the 285's. A 285 is just as fast if not faster than a 5870 at low resolutions, but again, who would be using either of these cards at that resolution?
What makes you think that?
Look:
HD 5870 dual and triple crossfire (Crysis Warhead):
http://www.hardwarezone.com/articles/view.php?id=3032&cid=3&pg=6 hardwarezone
(HD 5870 - Driver 8.66 Beta)
(HD 4870 X2 - Catalyst 9.8)
(GTX 285 / 295 - ForceWare 190.62)
http://www.blindnero.com/pictures/hd5870pictures/hardwz-5870-4870x2-1.jpg

[Blkout]

What makes you think that?

Think what?