Sapphire R9 285 Dual-X OC 2 GB

[GhostRyder]

aye, and there apparently also are some 290(x) cards suffering from very high power use accompanied by higher temperatures at the moment. Its only some cards, and with the latest driver that suffer from it, but it makes one wonder if not something similar is happening here..
would both explain why the poweruse is higher than what we were let to believe it should be, and why the card is actually drawing even more than the PCI spec allows.
I wonder if this is why anandtech hasnt posted its review yet.

Seems that is the case with 14.7 for some reason but the cards I hear referenced at times with the issue is the aftermarket OC editions. I run PowerColor OC editions but my temps did not change that I noticed (granted they are my own custom clocks under liquid). It's likely how much and how the voltage is set causing the issue (or somehow related to recent fixes for 4K if that could make sense lol).

I've seen the ZOTAC card twice in the last 2 weeks around $275, I hear other cards trade spots with it from time to time. So there is a single GTX 770 available for $275 all the time

Never really payed enough attention to Zotac which is part why I've probably missed this deal mostly after their flops on the 5XX series (specifically the LC editions) but that price is pretty legit if it's a common thing. Most if the 770s are minimum 300+ which makes that zotac a good deal. I find it funny though because most of the 760s cost around that price point so buying a 760 would be pointless as well because of the performance hop up to the 770.

Maybe I need to pay more attention to Zotac again is they are offering great deals like that!

 

[RealNeil]

Thanks for the review W1zzard. The market will adjust and this card may be a much better deal before long.

 

[Jeffredo]

If the GTX 750 Ti is even remotely an example of what Maxwell can do in performance vs. watt AMD may be in huge trouble in another six months or so.

 

[MikeMurphy]

The performance of the 285 is still in the realm of high-end. This is not "mid-range" performance and is certainly relevant beyond 1080p. Being an avid overclocker I surprised myself when I downclocked my 280x toxic given it has never once come even close to dipping below 60fps at 1200p. How is that mid-range?

Sure, it's not like the giant leap in power efficiency that Maxwell will bring us, but be reasonable. The overclocks aren't bad, either.

Having said that, I expect Maxwell will be the better overall product.

 

[Steevo]

You're in a habit of comparing overclocked cards with stock as a viable benchmark comparison?
Stock vs stock is 27.1 fps (R9 285) vs 27.0 fps (GTX 760) - which by my reckoning is 0.37%

Probably depends where you shop. NCIX has a couple of non-reference models at a quick glance: PNY GTX 770 ($265) and an EVGA ACX ($280 after MIR)

W1zz dropped the clocks as did other sites, perhaps that was the 15% AMD was talking about, overclocked VS stock. lol


Bang for the buck all around still goes to the 280X, it enters the cavities of competitive and same line models and performs acts of violence upon their physical selves and their high priced ways, figuratively speaking of course.

 

[ISI300]

I'm just starting to realize how shallow and empty TPU reviews are. No in-depth architecture analysis, and what looks like to be a template to reuse on all graphics card reviews. Meanwhile the rest of hardware sites dedicate at least 2 pages for architecture.
Also it's not clear exactly what settings you are using, that makes it hard to reproduce.
For instance, you've not mentioned AMD s new color compression.
In short, pile of bulls**t.

 

[HumanSmoke]

I'm just starting to realize how shallow and empty TPU reviews are. No in-depth architecture analysis, and what looks like to be a template to reuse on all graphics card reviews.....Also it's not clear exactly what settings you are using, that makes it hard to reproduce.

Probably the same settings that are used on the earlier reviews, like the 295X2 when you said "Such a great review."

For instance, you've not mentioned AMD s new color compression.

Probably because its secondary to the actual results. If it's so fantastic why is the 2GB/256-bit R9 285 basically equalling the performance of a two-and-a-half-year-old 2GB/256-bit GTX 670 ? Until the tech actually shows up as a tangible difference maker it's more academic interest. If you're highlighting future possibilities then why not ask why Tonga doesn't include H.265 transcode or HDMI 2.0. I'm pretty sure reviewers have their own rationale for what they deem important or not, but just because it's in the release kit doesn't make it a highlight-able feature - I seem to remember that just about every HD 2900XT review had some obligatory blurb about its tessellation engine. Until it shows results that set it apart from its peers, it means little.

 

[Steevo]

Don't feed the trolls, and obvious troll is obvious.

People complained about driver testing, W1zz retested with new driver, and it gained less than 1%.
People complain about how this site and reviews were AMD fanboyish.
People complain about how this site and reviews were Nvidia fanboyish.
People complain about what games are used.
People complain about the power consumption tests (and fail to realize two identical cards may differ by a significant percentage due to sensor, GPU, VRM, and other NORMAL variations)


The reason other sites use GPUZ and the reputable sites are linked together is how great of a job is being done, and if one or two people want to get the panties all knotted up and chafing their delicate asses, they can collectively and politely leave, go start your own site, we have had enough people try that.

 

[xorbe]

I'm just starting to realize how shallow and empty TPU reviews are. No in-depth architecture analysis, and what looks like to be a template to reuse on all graphics card reviews. Meanwhile the rest of hardware sites dedicate at least 2 pages for architecture.

Yet those are the very pages I skip over in other reviews. Then go read the other reviews. No need for them to all be the same.

Card is faster than 760, slower than 670. Somehow I will sleep soundly tonight!

 

[mroofie]

Yet those are the very pages I skip over in other reviews. Then go read the other reviews. No need for them to all be the same.

Card is faster than 760, slower than 670. Somehow I will sleep soundly tonight!

the fps doesn't agree with you ?
Gtx 760 has same performance

 

[HumanSmoke]

the fps doesn't agree with you ?
Gtx 760 has same performance

 

[ISI300]

Probably the same settings that are used on the earlier reviews, like the 295X2 when you said "Such a great review."

Probably because its secondary to the actual results. If it's so fantastic why is the 2GB/256-bit R9 285 basically equalling the performance of a two-and-a-half-year-old 2GB/256-bit GTX 670 ? Until the tech actually shows up as a tangible difference maker it's more academic interest. If you're highlighting future possibilities then why not ask why Tonga doesn't include H.265 transcode or HDMI 2.0. I'm pretty sure reviewers have their own rationale for what they deem important or not, but just because it's in the release kit doesn't make it a highlight-able feature - I seem to remember that just about every HD 2900XT review had some obligatory blurb about its tessellation engine. Until it shows results that set it apart from its peers, it means little.

I wasn't referring to this one particular review. As for the 295x2 review, It was known Hawaii GPU on a single card and I was just checking in for PCB shots and power consumption numbers.
Tonga has some very interesting new features that AMD is being coy about to discus to the press, Some websites such as techreport.com try their best to probe the manufacturers to get some details.
Although I will admit that in some cases TPR really shines, such as the range of tested products and it power consumption (esp. bluray blayback) tests. But having the same basic layout for reviews for year after year comes across as lazy.

 

[BiggieShady]

But having the same basic layout for reviews for year after year comes across as lazy.

No, it does not ... layout is something people get used to ... changing it often would come across as crazy.

 

[W1zzard]

But having the same basic layout for reviews for year after year comes across as lazy.

We do make small incremental updates to our reviews from time to time, if you feel certain sections could be improved let me know.

I'm with Humansmoke on features that don't matter to you. At least 99.99% of people buy discrete graphics cards for gaming, so that's what we are focusing on.

What non-gaming graphics card technology do you use? Right now, the only technology where I would like to bolster our testing is accelerated video rendering using advanced shaders i.e. MadVR. But it's not easy to come up with a testing methodology.

Oh, and we received our card from AMD on Thursday last week.

Let's see.
- Mantle: 3 games available, broken on R9 285.
- 4K H.264 decoder: irrelevant without 4K content, no software for testing available. Not H.265 decode
- Compute: zero real life applications that consumers use, accelerated video encode has horrible quality compared to CPU, double precision completely irrelevant in the consumer space, barely relevant elsewhere unless you buy Teslas
- Tessellation performance: covered by game testing, which actually matters to you
- DeltaColor compression: interesting, too bad all AMD said about it is on that slide: http://img.techpowerup.org/140903/Capture3957.jpg ie, nothing but mention it exists
- Architecture: no arch diagram provided by AMD, no die size, no transistor count. I asked for it but got no reply.
- DX12: AMD's fine print says "The DX12 specification and support for it are subject to change without notice".
- Freesync: vaporware
- TrueAudio: cmon
- Raptr: really?

 

[bubbleawsome]

I'll admit. I would like to see compute (see: F@H) performance. I remember anandtech used to do F@H but I'm not sure if they still test that. Not really sure how you bench F@H though. Do you just throw the card at a work unit for 3 or so days until the ppd evens out?

I'd like it because no one mentioned how much the 770 and 7970 even out in compute. I figured the 7970 would blow it away but it didn't.
¯\_(ツ)_/¯

 

[Frick]

Here it seems it'll be pretty much exactly as much as the 760's, which makes it a better deal I assume.

EDIT: BTW, something that has always annoyed me with the reviews is the lack of paragraphs in the conclusion. There are paragraphs, but there is no space between them of any kind.

Spoiler

AMD's new Tonga-based Radeon R9 285 leaves me with mixed feelings. Performance has really seen no significant improvements since Tahiti, i.e. the R9 280(X). Sapphire's overclocked board performs exactly as well as the HD 7970, which is almost three years old now. AMD did not send us a reference board, so I had to clock the Sapphire card down to reference clocks, which made the R9 285 3% slower than the Sapphire R9 285 Dual-X OC and 5% faster than the GTX 760. Compared to the AMD Radeon HD 7970 GHz Edition, Sapphire's overclocked card is over 10% slower. I'm looking at 1080p here, which I believe to be the ideal resolution for the R9 285. While AMD tries to promote 1440p gaming on the card, I think it's just too slow for a really good gaming experience at the resolution. Overall performance considered, I'd say the R9 285 is an excellent card for 1080p gaming at maximum details.
Feature-wise, not much that really affects you has changed. Tonga adds support for TrueAudio, which has been available on Bonaire and Hawaii before and really didn't impress anyone. You can now use FreeSync in games on Tonga, but there are no FreeSync monitors, so we'll have to see if it'll be worth it. Mantle has been on the market for a while now. With some adoptions into games, it will hopefully pick up in the future. Last but not least, an improved display controller now allows you to run a 3-monitor EyeFinity setup without an active DisplayPort adapter.
In light of NVIDIA's Maxwell architecture, I expected AMD to introduce massive power consumption optimizations, but Tonga doesn't bring much in that regard. Single-monitor idle power consumption is improved, reaching levels that are more appropriate for 2014. Multi-monitor and Blu-ray power consumption is still bad, much worse than with any NVIDIA product. Gaming efficiency has improved by roughly 10% over the old Tahiti GPU, but that's not enough. The R9 28x Series (including the R9 285) is still the least power-efficient series on the market as NVIDIA's new Maxwell architecture seems to be twice as efficient during gaming.
Sapphire's custom cooler does a good job keeping the card cool, delivering decent temperatures that match those we've seen from the R9 280(X). Idle noise levels are fine, but could be quieter. While fan noise during gaming has improved, it's still not where I'd like it to as it doesn't beat competing products. In gaming, the two fans will definitely be noticeable.
AMD's MSRP for the R9 285 is $249, and we expect the Sapphire R9 285 Dual-X OC to retail for around $260, which isn't too bad, but not good enough to take over the market. Right now, you can find GTX 770 cards discounted to an amazing $275, and these cards are significantly faster, quieter, and more power efficient. Another option would be to look at used HD 7970/R9 280X class cards that compete well with the R9 285 at better, used prices. This means AMD has to reduce their pricing to somewhere below the $230 mark to really get things moving. AMD's latest game bundle also includes loads of titles, old and new, you could pawn off to save some cost.

That's sort of tight.

Spoiler

AMD's new Tonga-based Radeon R9 285 leaves me with mixed feelings. Performance has really seen no significant improvements since Tahiti, i.e. the R9 280(X). Sapphire's overclocked board performs exactly as well as the HD 7970, which is almost three years old now. AMD did not send us a reference board, so I had to clock the Sapphire card down to reference clocks, which made the R9 285 3% slower than the Sapphire R9 285 Dual-X OC and 5% faster than the GTX 760. Compared to the AMD Radeon HD 7970 GHz Edition, Sapphire's overclocked card is over 10% slower. I'm looking at 1080p here, which I believe to be the ideal resolution for the R9 285. While AMD tries to promote 1440p gaming on the card, I think it's just too slow for a really good gaming experience at the resolution. Overall performance considered, I'd say the R9 285 is an excellent card for 1080p gaming at maximum details.

Feature-wise, not much that really affects you has changed. Tonga adds support for TrueAudio, which has been available on Bonaire and Hawaii before and really didn't impress anyone. You can now use FreeSync in games on Tonga, but there are no FreeSync monitors, so we'll have to see if it'll be worth it. Mantle has been on the market for a while now. With some adoptions into games, it will hopefully pick up in the future. Last but not least, an improved display controller now allows you to run a 3-monitor EyeFinity setup without an active DisplayPort adapter.

In light of NVIDIA's Maxwell architecture, I expected AMD to introduce massive power consumption optimizations, but Tonga doesn't bring much in that regard. Single-monitor idle power consumption is improved, reaching levels that are more appropriate for 2014. Multi-monitor and Blu-ray power consumption is still bad, much worse than with any NVIDIA product. Gaming efficiency has improved by roughly 10% over the old Tahiti GPU, but that's not enough. The R9 28x Series (including the R9 285) is still the least power-efficient series on the market as NVIDIA's new Maxwell architecture seems to be twice as efficient during gaming.
Sapphire's custom cooler does a good job keeping the card cool, delivering decent temperatures that match those we've seen from the R9 280(X). Idle noise levels are fine, but could be quieter. While fan noise during gaming has improved, it's still not where I'd like it to as it doesn't beat competing products. In gaming, the two fans will definitely be noticeable.

AMD's MSRP for the R9 285 is $249, and we expect the Sapphire R9 285 Dual-X OC to retail for around $260, which isn't too bad, but not good enough to take over the market. Right now, you can find GTX 770 cards discounted to an amazing $275, and these cards are significantly faster, quieter, and more power efficient. Another option would be to look at used HD 7970/R9 280X class cards that compete well with the R9 285 at better, used prices. This means AMD has to reduce their pricing to somewhere below the $230 mark to really get things moving. AMD's latest game bundle also includes loads of titles, old and new, you could pawn off to save some cost.

The space!

 

[ISI300]

We do make small incremental updates to our reviews from time to time, if you feel certain sections could be improved let me know.

I'm with Humansmoke on features that don't matter to you. At least 99.99% of people buy discrete graphics cards for gaming, so that's what we are focusing on.

What non-gaming graphics card technology do you use? Right now, the only technology where I would like to bolster our testing is accelerated video rendering using advanced shaders i.e. MadVR. But it's not easy to come up with a testing methodology.

Oh, and we received our card from AMD on Thursday last week.

Let's see.
- Mantle: 3 games available, broken on R9 285.
- 4K H.264 decoder: irrelevant without 4K content, no software for testing available. Not H.265 decode
- Compute: zero real life applications that consumers use, accelerated video encode has horrible quality compared to CPU, double precision completely irrelevant in the consumer space, barely relevant elsewhere unless you buy Teslas
- Tessellation performance: covered by game testing, which actually matters to you
- DeltaColor compression: interesting, too bad all AMD said about it is on that slide: http://img.techpowerup.org/140903/Capture3957.jpg ie, nothing but mention it exists
- Architecture: no arch diagram provided by AMD, no die size, no transistor count. I asked for it but got no reply.
- DX12: AMD's fine print says "The DX12 specification and support for it are subject to change without notice".
- Freesync: vaporware
- TrueAudio: cmon
- Raptr: really?

I do agree with you on many of the points you make (GPU compute being practically useless for consumers, Mantle not being worth the hype, 4K H.264 irrelevant for now, etc.)
I have checked in to other mainstream sites to see if there's is something extra floating about, but nothing except for the features you highlighted.
But, while I was reading the review, there where some questions that arose that weren't answered for me. Like, would the DeltaColor compression have any impact or reduce the amount of video memory usage? Without knowing whether this new architecture is purportedly any more efficient in VRAM usage compared to the prior or not, every review website (except some, TR for instance) are deducting points for it's 2 gigs of VRAM. Damage over at TR thinks that the 285 writes compressed data into the frame buffer. Seeing as how much more efficient the 285 is in 3DMarkV fillrate test according to this test:
http://techreport.com/review/26997/amd-radeon-r9-285-graphics-card-reviewed/2
would the 285 be more efficient in VRAM allocation?

Note that the frame buffer info stored in Tonga's DRAM is natively compressed, presumably giving it a smaller footprint. I'd like to test the 285 in 4K. I suspect it could get by almost as well as a 3GB Tahiti in some cases.

What I mean is it's good to know the base architecture (with as much info as one can gather) before giving a score and calling it a day.
BTW, respect for the extensive list of tested cards and resolutions!

 

[W1zzard]

DeltaColor compression

nobody really knows. I posted the slide above that has all info from AMD regarding that feature.

2 GB of VRAM is plenty for all games at all playable settings at 1080p

 

[DaedalusHelios]

http://www.newegg.com/Product/Product.aspx?Item=N82E16814500302

That's the problem for R9 285, and main reason why low score/no award

Or even http://www.newegg.com/Product/Product.aspx?Item=N82E16814127741 which is even cheaper.

 

[W1zzard]

Or even http://www.newegg.com/Product/Product.aspx?Item=N82E16814127741 which is even cheaper.

The Zotac was $275 until today, I guess AMD went to Newegg and cried, so they bump the price

 

[GhostRyder]

Don't feed the trolls, and obvious troll is obvious.

People complained about driver testing, W1zz retested with new driver, and it gained less than 1%.
People complain about how this site and reviews were AMD fanboyish.
People complain about how this site and reviews were Nvidia fanboyish.
People complain about what games are used.
People complain about the power consumption tests (and fail to realize two identical cards may differ by a significant percentage due to sensor, GPU, VRM, and other NORMAL variations)


The reason other sites use GPUZ and the reputable sites are linked together is how great of a job is being done, and if one or two people want to get the panties all knotted up and chafing their delicate asses, they can collectively and politely leave, go start your own site, we have had enough people try that.

Most people rant about one being significantly better in all situations or make fun of the other because of their own insecurities. If you have not noticed the same few people make fun of certain companies on a daily basis regardless of the topic at hand.

the fps doesn't agree with you ?
Gtx 760 has same performance

? It seens to be faster in most situations than a GTX 760...Which was the target of this card though its only relevant to people shopping in that price point and not to people looking for the king of GPUs.

Or even http://www.newegg.com/Product/Product.aspx?Item=N82E16814127741 which is even cheaper.

$274 when we were discussing yesterday.

I do agree with you on many of the points you make (GPU compute being practically useless for consumers, Mantle not being worth the hype, 4K H.264 irrelevant for now, etc.)
I have checked in to other mainstream sites to see if there's is something extra floating about, but nothing except for the features you highlighted.
But, while I was reading the review, there where some questions that arose that weren't answered for me. Like, would the DeltaColor compression have any impact or reduce the amount of video memory usage? Without knowing whether this new architecture is purportedly any more efficient in VRAM usage compared to the prior or not, every review website (except some, TR for instance) are deducting points for it's 2 gigs of VRAM. Damage over at TR thinks that the 285 writes compressed data into the frame buffer. Seeing as how much more efficient the 285 is in 3DMarkV fillrate test according to this test:
http://techreport.com/review/26997/amd-radeon-r9-285-graphics-card-reviewed/2
would the 285 be more efficient in VRAM allocation?

What I mean is it's good to know the base architecture (with as much info as one can gather) before giving a score and calling it a day.
BTW, respect for the extensive list of tested cards and resolutions!

It is his testing methodology...@W1zzard does it his way and the way he thinks is best and follows more what a majority of people are looking for in said price point and areas in general. If you are specifically referencing 4k being necessary, I doubt and of the sub 300 dollar cards even in tri-quad setups would be worthwhile at playing at those resolutions anyhow. This card was shot up to 1440p/1600p areas (Probably in the aftermarket variant 4gb versions with a second card) and mostly aimed in the 1080p area at that price point for gamers which is what the review focused on. There are plenty of reviews out there that check different things, not every review/reviewer can do every possible scenario for every single card. If people tried they would be stuck for weeks on end testing just one card...

 

[revin]

, 4K H.264 irrelevant for now, etc.)

@W1zzard Just to clarify, It's only because of lack of testing software, and for you to set a repetable standard
Correct ? I feel that that quote may be out of context to what you are trying to relay.

4K is relevant, and I know when you find how to approach it you will be just as diligent.

It's not fair to hear coment's negitave to w1zzards methodology, when he, and most all TPU reviewer's are far more diligent in a "Review", not passifying some Brand or Rep to get brownie points.

Never have even when the CPU TIM issue came up, it still revieled "some" flaw, even if it was "free", don't mean they{TPU} need to butter up................................

 

[W1zzard]

4K video: raise your hand if you have watched a real movie in 4K, not at the cinema. raise your second hand if you watched it via computer. nobody

I have no plans to test 4K movie playback in any way for the forseeable future.

My comment on video playback testing was for 1080p output with MadVR (if you don't know what it is, Google and start using it).

 

[DaedalusHelios]

The Zotac was $275 until today, I guess AMD went to Newegg and cried, so they bump the price

My mistake, sorry about that.

I don't see why Nvidia doesn't make 6GB or 4GB 780 ti cards. I guess every major player makes weird decisions. I believe W1zzard is correct in his assessment of the R9 285. If it was instead a more robust card than the 280X and featured more memory bandwidth instead of less with even higher core clocks. Or perhaps more shaders, then they would have been better off. Nobody looks to AMD for power efficiency given what we know Maxwell will bring to the table.

 

[W1zzard]

Rest of the 4K graphs (due to 10 attachment limit)

I dont have that for other cards in the R9 285 segment