Powercolor HD 7990 Devil 13 6 GB Review

[RussianSensation]

The drivers problems are exactly the same that plagued PowerColor HD6870 X2. While HD6870 CF ran perfectly fine, the special 6870 X2 card continued to have problems. It's definitely GPU dependent. We've seen this before too when HD5970 had driver problems but HD5870 CF didn't.

HD7970 CF doesn't have these cross-fire scaling issues. HD7970 1050mhz in CF easily beats a 690 in games:
http://www.xbitlabs.com/picture/?src=/images/graphics/geforce-gtx-690-sli/zfulltable.png

Even with the old drivers, 925mhz HD7970 CF is just 6% behind GTX690: http://www.computerbase.de/artikel/grafikkarten/2012/test-nvidia-geforce-gtx-690/4/

That means a 1000mhz 7970 should have been at least as fast if not faster and in their test it's 10% slower. :hmm:

I'd rather get 2x 7970 cards, guaranteed better CF scaling, more overclocking headroom, cheaper and quieter cards. Vapor-X 7970s are $450 each, or 2 HIS IceQ 7970s for $800! Better yet, just get HD7950 CF or catch a sale on GTX670 SLI for $350 a piece.

 

[cmdrdredd]

The drivers problems are exactly the same that plagued PowerColor HD6870 X2. While HD6870 CF ran perfectly fine, the special 6870 X2 card continued to have problems. It's definitely GPU dependent. We've seen this before too when HD5970 had driver problems but HD5870 CF didn't.

HD7970 CF doesn't have these cross-fire scaling issues. HD7970 1050mhz in CF easily beats a 690 in games:
http://www.xbitlabs.com/picture/?src=/images/graphics/geforce-gtx-690-sli/zfulltable.png

I'd rather get 2x 7970 cards, guaranteed better CF scaling, more overclocking headroom, cheaper and quieter cards. Vapor X 7970s are $450 each. Better yet, just get HD7950 CF or catch a sale on GTX670 SLI for $350 a piece.

Micro stutter etc aside this is so true. Better off going with cost savings and then you can buy some games and a case of beer and make a weekend out of it lol.

Doesn't look as bad as some of you are making it out to be. Looks good enough to run my eyefinity setup.The looks and branding alone of the card is worth it for me. I'm going to return my XFX 7970 ghz edition (which keeps shutting off and fans going 100% and screen going black with no signal) to amazon and just pay double for this awesome looking card.

The only thing stopping me is if AMD surprises us all and plans to release their own 7990 with the ghz edition xt2 chips... Which from what I have been hearing is unlikely... but still.

Hopefully amazon gets them in stock soon so I don't have to order from new egg or ncix and pay sales tax and shipping....

Did you look at the performance at 5760x1080? no? Sometimes a single Ghz edition 7970 is just as fast. For $1000 that's beyond pathetic.

Can't even run BF3 at 5760x1080

 

[chimaxi83]

Seriously, 7950 CF for ~$600-650 with sales and/or rebate, or 7970 CF for ~$800-900, make this card irrelevant. Nice looking card though :thumbsup:

 

[piesquared]

Uh...no...this is a card with a bridge chip. It works just the same as the GTX 690 works.

In the most basic terms it is two GPUs in crossfire mode on a single card. Nothing in the review indicates it needs special drivers, Wizzard even used the standard 12.7 beta that everyone uses. The bridge chip adds latency which may affect the performance some, but the larger issue in my eyes is that the card just doesn't work sometimes. It is not likely something drivers can fix because as I said before the drivers see two GPUs in crossfire mode and don't care or know whether it's on one card or two cards. There is likely a reason AMD never made a 7990 and probably won't. The results speak for themselves.

Lol what do you mean um no. Your claiming this is a hardware problem?

 

[Olikan]

Uh...no...this is a card with a bridge chip. It works just the same as the GTX 690 works.

no....that's the diference with SLI and crossfire...

crossfire bridge don't give enought bandwidth, so they have to use pci-e links

nvidia bridge is WAY better, because it gives enought bandwidth....actually that's one reason than SLI is better than x-fire, it avoid the mobo

(you know... less chance to shit happen)

 

[Jionix]

This is not the 7990 you are looking for... waves hand

This isn't an official spec card. It's another cobbled together PowerColor card.

I would wait for.. Well, if AMD actually releases a true spec 7990 with their own engineering and "tricks".

 

[moonbogg]

Dual GPU cards from both camps are ridiculous this round. How can they be tempting to anyone?

 

[Jeff007245]

Here is another review: http://www.hardwareheaven.com/revie...-7990-graphics-card-review-battlefield-3.html

 

[ViRGE]

Normally I wouldn't say this, but in this case I'm seriously wondering whether PowerColor is buying reviews and/or that reviewers intentionally softballed it to get the cards.

I mean seriously, TPU gives a "Highly Recommended" award to a defective card? I could see publishing the review and withholding final judgement until they can get another card, but to recommend a card that would appear to have a systematic flaw is mind-boggling.

And HH isn't much better. The fact that they include 19x10 but not 25x14 is weird; the fact that they're the only site I've ever seen an AMD card win at BF3 is even weirder. They didn't even throw in any other configurations either (e.g. 7970GE CF), yet they still gave it an award.

 

[AnandThenMan]

Why would any AIB bother trying to influence these sites to "cook the books" anyway? Such a low volume part, hardly seems worth it. Maybe they were just happy to have the card to test and went a little easy on it. :sneaky:

Ugh what a pointless piece of hardware, AMD is dumb to even allow this thing to exist IMO.

 

[ViRGE]

Why would any AIB bother trying to influence these sites to "cook the books" anyway? Such a low volume part, hardly seems worth it. Maybe they were just happy to have the card to test and went a little easy on it. :sneaky:

That's a good point. The best answer I can come up with is that a product like the Devil13 is an overbuilt card created for bragging rights, so it gets a lot of attention relative to its low sales. I'd think you'd want your braggart part to look good at all costs. Otherwise if you can't brag about it, why make it?

 

[RussianSensation]

This card will sell out anyway, only 500 or so units and bitcoin mining is still going strong. For gamers 7950 OC CF, 7970 GE CF, GTX670 SLI 2/4GB and 690 SLI are all better and more elegant choices.

 

[blanketyblank]

Is everyone missing the point of this card? This is essentially the new 5970, a bitcoin miner with the ability to play games occaisonally. Put 4 of these into a mobo that can support them and you could potentially be doing 6.4 Ghash/s. Not sure if the cost is worth it, considering all the ASICs coming out but someone might try.

 

[Final8ty]

These are the same drivers that someone running 2x HD 7950s would use and to the drivers this card is just simply two HD 7950 or 7970s. The drivers don't care if it's on one card or two, it sees it as two GPUs. Powercolor doesn't make the drivers and no, AMD is not going to make special drivers for this card.



I didn't actually read the review, but is that what the picture was trying to say with the washers to shim out the space in the screw?

Actually that is not as clear cut as you may think, i remember when the 5970 first came out A review showed some games would not run on it at all while running perfectly on 5870 in CF.

 

[3DVagabond]

Normally I wouldn't say this, but in this case I'm seriously wondering whether PowerColor is buying reviews and/or that reviewers intentionally softballed it to get the cards.

I mean seriously, TPU gives a "Highly Recommended" award to a defective card? I could see publishing the review and withholding final judgement until they can get another card, but to recommend a card that would appear to have a systematic flaw is mind-boggling.

And HH isn't much better. The fact that they include 19x10 but not 25x14 is weird; the fact that they're the only site I've ever seen an AMD card win at BF3 is even weirder. They didn't even throw in any other configurations either (e.g. 7970GE CF), yet they still gave it an award.

Ouch! on the bold part. I think W1zzard is very reputable and it's a bit out of line to state something like that without any evidence. I realize it's only your opinion, and you are of course entitled to it. I'm not trying to make this sound like a harsh criticism of you either. You are generally very even handed. I think it's just a bit too harsh of a statement.

He does very plainly state right at the beginning that he got a non functional card. I agree that unless he's received something to make him certain that this was a one card fluke, or assurance that production models are/will be fixed, he shouldn't recommend it as is.

Looks like the HH review was run on a Eyefinity set up and they only ran one screen and 3 screen. Seems reasonable, to me. It would have been good if they had compared more cards, but they at least compared it against the directly competing card.

HH used newer drivers (12.8 instead of 12.7) That might be why they had more consistent scaling than TPU. HH ran some of the newer releases that AMD is ahead on. It's not like we've never seen that done with nVidia before, though. How many reviewers run HAWX 2 and Lost Planet 2? What's fair for one is fair for the other. That or do they have to only run games that they have similar performance with to be fair, which really wouldn't accomplish anything.

EDIT: Sorry. Rechecked the review and W1zzard does use 12.8 for the 7990.

 

[-Slacker-]

Here is another review: http://www.hardwareheaven.com/revie...-7990-graphics-card-review-battlefield-3.html

Yeah, so nobody seems to give a damn about this post, so I'll just quote it to see if it's ignored again.

Cliffs: The card does quite well in that review.

 

[Silverforce11]

Yeah, so nobody seems to give a damn about this post, so I'll just quote it to see if it's ignored again.

Cliffs: The card does quite well in that review.

Why is it HH got great scaling and TPU had defective CF (its operating less than a single 7970).

HH's results are excellent, faster than gtx690 and it still had a good OC on top of it. 74W extra vs 690.

 

[-Slacker-]

Happens sometimes, I guess.

 

[Veridian3]

Looks like the HH review was run on a Eyefinity set up and they only ran one screen and 3 screen. Seems reasonable, to me. It would have been good if they had compared more cards, but they at least compared it against the directly competing card.

Just to clear up the thought process here (the HH review is mine) when creating this review a few things were considered but here are some of the more important...

1) The 690 comparison is a must have.
2) Given the performance of this card, and others at 19x10 the inclusion of Eyefinity was a must have (we always test it anyway)
3) It is almost a no-brainer that this card and 2x 7970 would give near identical performance that it was wasted resource to include both, resource that could be applied elsewhere. (We test every product in every review for that individual article. This means the latest patches, drivers, OS reinstall etc etc. All real world testing, no old results are re-used but I feel it is the best way to do it. A lot of other sites reuse older testing making it look like they tested a bunch of cards but the results, in my opinion, are not directly comparable).

Using the resource issue as an example no 7970 CF (other than 3DMark) means we could spend time ensuring we tested in the latest games (e.g. Sleeping Dogs) as well as checking a few others that would have had potential but ended up not included for various reasons (e.g. Dark Souls).

There is always a lot more thought goes into reviews (everywhere) than is immediately obvious so I hope that clarifies a little about ours.

I should also say that our card worked fine out of the box, no thermal/heatsink issues and that when comparing results take into account that review systems have an impact. We went for an i7-3960X at 5GHz, 16GB, games running off SSDs, the OC Devil BIOS only, and so on, others wont.

Its always best to read a spread of articles and form your opinion on those rather than one, even two.

 

[Silverforce11]

Great thanks for the insight.

Just suprised TPU would be sent a defective product... kinda stupid of Powercolor since TPU is one of the major review sites.

 

[3DVagabond]

Just to clear up the thought process here (the HH review is mine) when creating this review a few things were considered but here are some of the more important...

1) The 690 comparison is a must have.
2) Given the performance of this card, and others at 19x10 the inclusion of Eyefinity was a must have (we always test it anyway)
3) It is almost a no-brainer that this card and 2x 7970 would give near identical performance that it was wasted resource to include both, resource that could be applied elsewhere. (We test every product in every review for that individual article. This means the latest patches, drivers, OS reinstall etc etc. All real world testing, no old results are re-used but I feel it is the best way to do it. A lot of other sites reuse older testing making it look like they tested a bunch of cards but the results, in my opinion, are not directly comparable).

Using the resource issue as an example no 7970 CF (other than 3DMark) means we could spend time ensuring we tested in the latest games (e.g. Sleeping Dogs) as well as checking a few others that would have had potential but ended up not included for various reasons (e.g. Dark Souls).

There is always a lot more thought goes into reviews (everywhere) than is immediately obvious so I hope that clarifies a little about ours.

I should also say that our card worked fine out of the box, no thermal/heatsink issues and that when comparing results take into account that review systems have an impact. We went for an i7-3960X at 5GHz, 16GB, games running off SSDs, the OC Devil BIOS only, and so on, others wont.

Its always best to read a spread of articles and form your opinion on those rather than one, even two.

Reading your review you mention actually playing the games, multiple runs, etc. I'm assuming that if there were any anomalies you would have mentioned it. just to be clear though, did you see anything like micro-stutter, screen tearing, crashes, artifacts, features not working or driver settings not available? Anything like that?

P.S. Thanks for taking the time for posting in response to the concerns voiced. :thumbsup:

 

[cmdrdredd]

Lol what do you mean um no. Your claiming this is a hardware problem?

Yes same as their 6970x2. It doesnt work right based on the tpu review. If you look at HH it works. So yeah it is a hardware issue.

 

[96Firebird]

My guess is the card is throttling itself because of the iffy heatsink contact with the GPU causing temperatures to skyrocket. No?

 

[Veridian3]

Reading your review you mention actually playing the games, multiple runs, etc. I'm assuming that if there were any anomalies you would have mentioned it. just to be clear though, did you see anything like micro-stutter, screen tearing, crashes, artifacts, features not working or driver settings not available? Anything like that?

P.S. Thanks for taking the time for posting in response to the concerns voiced. :thumbsup:

Yes absolutely, we/I play through sections of the game which are pre-chosen multiple times. Essentially when reviewing the games themselves we note the demanding sections, save a game there and use it for benching the cards at a later date.

The real world testing is the only way to properly identify all/any issues and while it would be sooooo much easier and quicker to do timedemos, inbuilt benches or just 3DMark and similar they rarely give results which are the same as users will see... or thorough enough to pick up any flickering textures, micro stuttering and the like. (Shogun 2/Total War is a great example, it's inbuilt benchmark mode doesn't give results even close to the framerates when playing.)

Nothing is 100% perfect obviously but for this particular product there were no noticeable issues.

 

[Silverforce11]

My guess is the card is throttling itself because of the iffy heatsink contact with the GPU causing temperatures to skyrocket. No?

Makes a lot of sense. Throttle = crap performance even if CF is working.