AMD's Tonga - R9 285 (Specs) and R9 285X (Partial Specs)

[raghu78]

There is a ton of reviews showing the R9 280X beating the 285 by a great margin.
Forget I said up to 25%.

Here is one showing 280X beating the 285 by 45% D:

u seem to be an expert at cherrypicking and exaggerating. shall I do the same.

faster in Bioshock Infinite

http://www.guru3d.com/articles_pages/amd_radeon_r9_285_review,13.html

http://www.bit-tech.net/hardware/graphics/2014/09/02/amd-radeon-r9-285-review/4

http://www.techpowerup.com/reviews/Sapphire/R9_285_Dual-X_OC/10.html

http://www.computerbase.de/2014-09/...marks/6/#diagramm-bioshock-infinite-1920-1080

faster in AC4 (without MSAA)

http://www.tomshardware.com/reviews/amd-radeon-r9-285-tonga,3925-8.html

http://ht4u.net/reviews/2014/amds_tonga-gpu_-_radeon_r9_285_im_test/index25.php

faster in Watch dogs

http://www.techpowerup.com/reviews/Sapphire/R9_285_Dual-X_OC/20.html

http://pclab.pl/art59409-9.html

faster in thief

http://www.computerbase.de/2014-09/amd-radeon-r9-285-test-benchmarks/6/#diagramm-thief-1920-1080

http://techreport.com/review/26997/amd-radeon-r9-285-graphics-card-reviewed/7

stock for stock on average the R9 280X is 10 - 15% faster than R9 285 . clock for clock that gap is even lesser.

http://www.hardware.fr/articles/926-21/recapitulatif-performances.html

http://ht4u.net/reviews/2014/amds_tonga-gpu_-_radeon_r9_285_im_test/index41.php

http://www.techpowerup.com/reviews/Sapphire/R9_285_Dual-X_OC/25.html

http://www.tomshardware.com/reviews/amd-radeon-r9-285-tonga,3925-14.html

But if you look at the list of advantages of Tonga over Tahiti -
1. FreeSync support
2. 4k H.264 hardware decode which means much lower CPU utilization in those tasks and a more responsive system
3. Much better tesselation and pixel fill rate.

http://techreport.com/review/26997/amd-radeon-r9-285-graphics-card-reviewed/2

Anyway the Tahiti cards are still great value for money as they go EOL and sell for clearance prices. :thumbsup:

 

[3DVagabond]

... My next card is Nvidia. AMD is no longer a relevant company.

Over react much?

As we speak AMD has the fastest card made. They have the best perf/$ at every price point that an enthusiast would consider. nVidia's only response to AMD's cards for about the last year or so has been to reduce prices. It doesn't look like AMD is going to have something to compete directly with Maxwell here with Tonga. They'll have to do as nVida has done and reduce prices

 

[Techhog]

Over react much?

As we speak AMD has the fastest card made. They have the best perf/$ at every price point that an enthusiast would consider. nVidia's only response to AMD's cards for about the last year or so has been to reduce prices. It doesn't look like AMD is going to have something to compete directly with Maxwell here with Tonga. They'll have to do as nVida has done and reduce prices

I don't think AMD will have a true competitor to Maxwell until it's time for the next thing

 

[ocre]

I think things are really muddied because of still being on 28nm. I really don't see how this chip could be Tahiti's replacement. I think this was supposed to be Pitcairn's replacement on 20nm.

+9000

I was thinking this when the specs first started getting leaked. I was just apprehensive to post anything here about it. Mainly because how fast and well people disregarded Hawaii being originally conceived with 20nm in mind. I was suggesting Hawaii was a chip that got pulled in to 28nm even before it launched. It was only once it came out that few people started to fancy the thought.

But since then, nvidia launched the gm107 and the majority of people were able to accept this was probably originally planned for 20nm. And now the gm204 is coming and I really really doubt that there is a single person who will reject the notion of it being pulled in from 20nm plans.

So the idea isnt so strange or fantastic anymore.

But wait, dont get me wrong. Just because a chip might have been originally conceived for one node doesnt mean that it was a last minute decision to pull it in for 28nm. Nvidia and AMD both have known for some time how things were going at TSMC. It is just the most obvious route forward.

So it was my belief that Hawaii was originally intended as the 20nm replacement of Tahiti. Sometime it was decided to become a 28nm chip. And tonga, it was originally the Pitcairn replacement but was pulled in as a 28nm chip.

These decisions could have been made long enough in advance that the chips could have been further tweaked. I believe there is a strong possibility that this is the case. I wanted to say it but didnt want to cause any stink. Now that the card has launched and to see that it is cut down and not very efficient, i think people might not be offended by the suggestion as much

 

[gamervivek]

You know its bad when all one can do is throw up a random sub chart of a synthetic benchmark. "Hey look how awesome score is on 3d Mark Color Fill" Yeah never heard that one. What about this "Post your 3d Mark Color Fill benchmark thread, suicide runs welcome" Yeah, haven't seen that one.......yet! I am not saying the card is bad, it is what it is. Just highlighting how ridiculous his attempts are to counter your claims. I have never seen anything like it.

The 256-bit card is brutally beating the 384-bit card at a memory bandwidth limited test. So much for more memory bandwidth at high resolutions.

Read the link.

Whoa

 

[ocre]

techreport review, quite positive
http://techreport.com/review/26997/amd-radeon-r9-285-graphics-card-reviewed/2

So they used an overclocked MSI card on all their charts?

I ask because several people have been posting charts up from this review and in the charts it doesnt specify. This could be why techreport felt so positive about the r9 285. But when i seen this, it blew my mind....

It seemed awfully high.
So i went back and looked. They were sent an MSI overclocked model. So then there is the reality, the card has the highest power consumption out of the pack. So with that and being a brand new architecture, surely it should outperform them all. It makes the overwhelmingly positive article seem a little strange. And with a handful of games? Is he really using an overclocked model during all the test and not identifying it in the charts?

Right now there are overclocked 760s for less than $239. Heck, the ACX is $229 with a 10$ mail in rebate. A highly overclocked MSI 760 for 219 after mir. There are crazy deals on the R9 280s, like nice OC custom cards for 199 after rebate. So i simply comprehend his excitement. Or his performance per dollar charts? There is no way thats right when you look at what is actually out there and their current price.

The one got me the most was the "Theoretical peak bandwidth vs average FPS" chart. Huh?

There are many reasons to be concerned with the techreport review. I dont hate the card. Its just no anything that is exciting. Its not a leap forward.

 

[ShintaiDK]

So 285 at 190W TDP uses more power than a 250W TDP 280X. So much for any efficiency. Its rather the other way around.... :/

 

[Cloudfire777]

u seem to be an expert at cherrypicking and exaggerating. shall I do the same.

bla bla bla bla

stock for stock on average the R9 280X is 10 - 15% faster than R9 285 . clock for clock that gap is even lesser.

bla bla bla bla

Anyway the Tahiti cards are still great value for money as they go EOL and sell for clearance prices. :thumbsup:

1. There is simply no way a 256 bit 28CU Tonga card is beating a 384bit 32CU Tahiti card.
You can try to twist this all you want, but the fact is that the R9 285 comes with less VRAM, less shaders, and less memory bandwidth. ALL of this is against the card.
If Tonga does beat Tahiti in a review, which it did but your examples by measly 5% in a couple of isolated scenarios with this and that settings turned off. Its because the Tonga card is tested with newer drivers or the Tonga card is tested with a faster CPU than the Tahiti card.

2. If your examples was presenting the truth, which I find it very hard to believe, would you rather buy A) A card that beats the other one in a few isolated scenarios by 5% and is performing less than the other one in the clear majority of games or B) A card that beats the other one from 5% all the way up to 20-25%, and up to 45% in a few of the newest titles.

I think the choice is very easy

 

[caswow]

So 285 at 190W TDP uses more power than a 250W TDP 280X. So much for any efficiency. Its rather the other way around.... :/

i can show you where the 285 only consumes 160watts...

http://ht4u.net/reviews/2014/amds_tonga-gpu_-_radeon_r9_285_im_test/index17.php

 

[el etro]

Bad things is bad things, no deny, but good things can be got of Tonga too.

Tonga packs(more or less) the performance of a standard 7970(year 2011) with 4CUs less(its minus 256 stream processors), much less memory bandwidth, and sightly less clock(918 versus 925Mhz).
Power consumption is greater than both 7950B/280 and 7970(Standard), efficiency don't exist here, but anyway there's some progress here. The two charts(texture fill test at 3Dv and Tessmark at maximum setting) confirms that there's is little progress here.

I have a bet that AMD is using the Intel tick-tock scheme for launching GPU archs by now, and this 285 is still the tock. I bet real competition by AMD for the big Maxwells still not came.

 

[AtenRa]

So they used an overclocked MSI card on all their charts?

If im not mistaken all the reviews are with custom OC cards, i haven't seen a reference card review so far.

 

[tviceman]

Over react much?

As we speak AMD has the fastest card made. They have the best perf/$ at every price point that an enthusiast would consider. nVidia's only response to AMD's cards for about the last year or so has been to reduce prices. It doesn't look like AMD is going to have something to compete directly with Maxwell here with Tonga. They'll have to do as nVida has done and reduce prices

The "fastest card" is a talking point and nothing more. No one in their right mind is going to buy that "fastest card" when dual card solutions are cheaper and just as fast. AMD having the best perf/$ is a matter of what the market dictates. Nvidia obviously commands a premium and nearly 2/3 consumers are willing to pay that premium. Nvidia and AMD have both reduced prices in response to each other. That is nothing new at all.

AMD is going to be in a world of hurt if Maxwell cards truly are coming next month and Nvidia scales up their TDP to the > 200 watt category. Maxwell will be faster, smaller, and more efficient than their direct competitors and Nvidia can price premium all day long while AMD simultaneously has to drop prices to stay competitive in frames per second only.

 

[el etro]

So they used an overclocked MSI card on all their charts?

It seemed awfully high.
So i went back and looked. They were sent an MSI overclocked model. So then there is the reality, the card has the highest power consumption out of the pack.

Power consumption is higher, but the clock is too, and a tad higher(97x Mhz vs 918Mhz for the original r9-285). Other sites have readings with less watts consumed(i.e. Computerbase review)

So with that and being a brand new architecture, surely it should outperform them all. It makes the overwhelmingly positive article seem a little strange. And with a handful of games?

Overwhelmingly is the perf/$ chart, not the review. Other sites review was positive too, like .

Is he really using an overclocked model during all the test and not identifying it in the charts?

He identifies on page 3. I think too is not nice to don't specify the clock of the card in the other pages.

The one got me the most was the "Theoretical peak bandwidth vs average FPS" chart. Huh?

In Kepler GTX680 review he made this kind of comparison too.

 

[artivix]

i can show you where the 285 only consumes 160watts...

http://ht4u.net/reviews/2014/amds_tonga-gpu_-_radeon_r9_285_im_test/index17.php

Test measures AMD R9 285 with special pcie card to find true load power. Only testers and reviewers outside USA can find true Furmark load power of R9 285!

Here is top list:

1) R9 270x 144w

2) R9 285 159w

3) R9 270x Toxic 164w

4) GTX 760 168w

5) XFX 270x 178w

6) R9 280 Dualx 180w

7) GTX 770 195w

8) GTX Titanium 198w

9) GTX 780 198w

10) GTX 770 Lightning 199w

11) R9 280x Vapor 211w

 

[Gloomy]

Tom's german site got 176 watts avg power consumption during gaming. They also test the card in isolation.

All these reviews with recent drivers just make me like the 280X more and more, instead of this new card. :\ Wheres the 285X at?

I don't think it's a bad card, it just has a crisis of identity. This should have been called the 370. I'm sure people would have liked it more that way.

 

[Cloudfire777]

Why would people want 285X?
Wasnt 285 good enough reminder that a 285X will barely be better than R9 280X?

The only good thing AMD could do now is release a 384bit R9 280 (28CU Tonga chip, just renamed) and 384bit R9 280X (32CU Tonga chip, just renamed), forget that R9 285 ever existed, and hurry the hell up with the 300 series to catch up with Nvidia`s Maxwell. It looks like the 900 series will get an entire quarter alone on the market before AMD is able to respond

But a 384bit Tonga chip may not exist. Its just a rumor at this point

 

[itsmydamnation]

I don't think it's a bad card, it just has a crisis of identity. This should have been called the 370. I'm sure people would have liked it more that way.

we all now thats exactly where its going to get renamed to.



edit: remember the 9600pro, this is the same thing.

 

[AtenRa]

As i have said before, its the same situation like AMDs Barts HD6870 vs HD5870 and HD6850 vs HD5850 back in October 2010. People were complaining about the same things at the time, about the name of the cards, prices vs discounted HD5870/5850 etc etc.

 

[JDG1980]

AMD seems to have an uncanny knack for always showing their products in the worst possible light.

If they had released the full Tonga chip (2048 SPs, 384-bit memory bus), it would have beaten Tahiti in most benchmarks and tied at worst in the others. With a price of $279, that would be a clear win - not worth upgrading for existing 280X/7970 owners, but at least a step forward in the state of the art. It would have made even more sense to pioneer GCN 1.2 technology on their APUs - the high-end Kaveri is really starved for memory bandwidth, and delta-based texture compression would be a big win here. But instead they choose about the worst possible slot in their lineup for the new card - one where it can't decisively beat the prior generation in cost, framerate, or power consumption. What were they thinking?

I agree with the other posters who said that this was really meant to be a 20nm card. Had they known from the start they'd be stuck on 28nm, I think they would have done a better job of optimizing the transistor budget. I'd be interested to see how the full-fledged 285X fares against its predecessor in the power usage arena.

 

[camelNotation]

If they had released the full Tonga chip (2048 SPs, 384-bit memory bus), it would have beaten Tahiti in most benchmarks and tied at worst in the others. With a price of $279, that would be a clear win - not worth upgrading for existing 280X/7970 owners, but at least a step forward in the state of the art. It would have made even more sense to pioneer GCN 1.2 technology on their APUs - the high-end Kaveri is really starved for memory bandwidth, and delta-based texture compression would be a big win here. But instead they choose about the worst possible slot in their lineup for the new card - one where it can't decisively beat the prior generation in cost, framerate, or power consumption. What were they thinking?

Given that Nvidia is releasing GM204 soon, AMD might be taking an incremental approach of not showing its best card first. With the R9-285, their strategy may have been "beat the comparable Nvidia card and replace the 280, but not the 280X." Then when Nvidia releases the GTX 980/970, AMD can price the R9-285X appropriately for the market.

Normally we see the fully enabled card first. This did not happen, and the fact that we're seeing a version with a cut down memory bus (assuming 285X is 384-bit) shows that AMD did not want the 285 to be overpowered for its market segment.

 

[Keysplayr]

Does anyone else think that this GPU was built for the anticipation (wrongly so) of continuing ridiculousness in the mining community? Cheaper to make than a 7950 but still pulling down two and a half bucks. I dunno. It's just that this was so lackluster of a launch that I had to wonder what they were thinking. I mean, when Charlie Demerjian disses AMD, the apocolypse is nigh....

 

[alexruiz]

Well, after all the reviews, results are all over the place. On some of them the R9 285 is decisively beating the R9 280, while in others is struggling hard. Furthermore, the overclocked models used for the reviews help with performance but not with power consumption.

The value proposition, at this time, however, is clear, buy it only when the predecessor is no longer available. This reminds me of the R7 265 / R9 270 launches. Why buy a R9 270 when the HD7870 is still available, it is faster and quite often much cheaper.

I don't think the cards are bad, they have some cool features, they are overall faster than the model they are replacing, but given the current market situation the R9 285 is a poor value. If it was available at the $199 street price that the R9 280 is selling, we wouldn't be complaining.

 

[DarkKnightDude]

Does anyone else think that this GPU was built for the anticipation (wrongly so) of continuing ridiculousness in the mining community? Cheaper to make than a 7950 but still pulling down two and a half bucks. I dunno. It's just that this was so lackluster of a launch that I had to wonder what they were thinking. I mean, when Charlie Demerjian disses AMD, the apocolypse is nigh....

I kinda doubt it, it was fairly obvious by summer time mining was no longer where it was. AMD had plenty of time to appeal to different segments.

 

[tviceman]

As i have said before, its the same situation like AMDs Barts HD6870 vs HD5870 and HD6850 vs HD5850 back in October 2010. People were complaining about the same things at the time, about the name of the cards, prices vs discounted HD5870/5850 etc etc.

Except Barts was fundamentally a great chip. It was small and very power efficient. The only thing wrong with Barts at it's release was it's name. Tonga, on the other hand, is nearly the same size as Tahiti, is thus far slower, and isn't particularly more power efficient. What's the point? In fact, it's bigger than GK104, slower, and no more (if not less) power efficient. GK104 came out 32 months ago. What is not wrong with this scenario? A full-fledged GK104 is a $50 price drop and rebrand (GTX 960) away from making Tonga look like the worst release since the GTS 450.

 

[JDG1980]

Does anyone else think that this GPU was built for the anticipation (wrongly so) of continuing ridiculousness in the mining community?

I think it's actually the other way around: there never would have been a Tahiti-based R9 280 (non-X) in the first place if not for the mining craze. That card was released pretty much explicitly in reaction to the R9 280X being sold out or wildly overpriced everywhere.

But once the mining craze ended, this put AMD in a difficult spot. Yields on 28nm were almost certainly good enough that there weren't that many defective chips to harvest, so fully functional chips must have been sold at cut-rate prices. It's hard for AMD to make money on a chip that big when the cards are retailing for $199.

Well, after all the reviews, results are all over the place. On some of them the R9 285 is decisively beating the R9 280, while in others is struggling hard. Furthermore, the overclocked models used for the reviews help with performance but not with power consumption.

I'd like to see what the power consumption numbers are like for a R9 285 run at stock speed. There are no reference boards, and all the cards out there seem to be factory-overclocked to some degree, but it shouldn't be hard for someone to set the clocks back to normal and drop the voltage to whatever its stock figure is. It's possible that AMD's official figures are a sort of "sweet spot" and increasing the voltage/clocks much beyond that causes power to rise significantly for not much more performance. Or maybe I'm just grasping at straws.