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

[Keysplayr]

So you conveniently missed the part where on average it beats 760 by 12-15% in AT's review?

On the whole the 285 is a lot closer to the 770 than 760.

NV has no answer right now to a $200 R9 280, $240-$260 R9 280X and a $340 R9 290 and $450 R9 290X. The $180-450 desktop dGPU market is all AMD right now until 970/980 change the situation.

With 280X outperforming the 760 by 26% at 1080P, when R9 280X is going for $240-260, the 760 hardly makes sense even at $220.

The review you linked to did in fact show the 760 ahead of the 285X in BF4 with or without Mantle at all resolutions tested at Ultra quality. Did he say something untrue?

 

[realibrad]

The review you linked to did in fact show the 760 ahead of the 285X in BF4 with or without Mantle at all resolutions tested at Ultra quality. Did he say something untrue?

Yes. The poster is implying that the 285 is slower than a 760, even on BF4 with Mantle. Mantle currently is slowing down the 285, so the assumption that mantle makes the 285 faster, and that with mantle the card cannot keep up is wrong.

On avg the 285 is faster than the 760. while in some cases a 760 will beat a 285, the typical way the 2 cards will be used gives the lead to the 285.

 

[el etro]

The review you linked to did in fact show the 760 ahead of the 285X in BF4 with or without Mantle at all resolutions tested at Ultra quality. Did he say something untrue?

He did not specified if he was talking about this review or not. In AT review 285 won against GTX 760.

 

[RussianSensation]

The review you linked to did in fact show the 760 ahead of the 285X in BF4 with or without Mantle at all resolutions tested at Ultra quality. Did he say something untrue?

The way he phrased it sounds like: "Even with the help of Mantle, the 285 still couldn't beat 760", suggesting it lost overall and even in Mantle cases, while the opposite is true with 285 easily beating 760 overall.

 

[Phynaz]

Also expecting game developers to shoulder the responsibility of constantly patching their games each time AMD releases new hardware is absolutely ridiculous. That task should always fall on the driver team because it's the driver's job to program hardware directly.

I run decade old DirectX games and they get faster on each GPU I buy, and these games haven't been patched in years. That's the power of true abstraction.

Yeah, this is a concern. Hopefully it's just because Mantle is still beta and isn't out of the oven yet.

 

[Keysplayr]

The way he phrased it sounds like: "Even with the help of Mantle, the 285 still couldn't beat 760", suggesting it lost overall and even in Mantle cases, while the opposite is true with 285 easily beating 760 overall.

At BF4, more accurately, IS what he said. You can also say he alluded that the moon was made of cheese.

 

[RussianSensation]

At BF4, more accurately, IS what he said. You can also say he alluded that the moon was made of cheese.

Yes and it's explained in the article that because Mantle is GCN generation specific, it's up to AMD and the developer to create support for 285 which is why it's currently not working well in BF4. And the point of discussing 4-5 fps loss under Mantle while 285 is trading blows with 760 in the same game under DX while ignoring cases where 760 bombs proves what exactly? If you get the 285, you get to choose DX or Mantle path so it's not as if you are locked in to playing BF4 under Mantle. It more shows that Mantle has a lot of weaknesses rather than telling us much about 285 vs. 760.

 

[monstercameron]

Yes and it's explained in the article that because Mantle is GCN generation specific, it's up to AMD and the developer to create support for 285 which is why it's currently not working well in BF4. And the point of discussing 4-5 fps loss under Mantle while 285 is trading blows with 760 in the same game under DX while ignoring cases where 760 bombs proves what exactly? If you get the 285, you get to choose DX or Mantle path so it's not as if you are locked in to playing BF4 under Mantle. It more shows that Mantle has a lot of weaknesses rather than telling us much about 285 vs. 760.

aside from the nv v amd arguments, it gives us a taste pf what to expect of mantles forwards compatibility. depending on how amd solves this issue, this might instill more support for mantle...or the opposite.

 

[realibrad]

At BF4, more accurately, IS what he said. You can also say he alluded that the moon was made of cheese.

Incorrect. What he said was...

I saw reviews of this card even on mantle this card cannot beat GTX 760 on BF4.

He said "even" which means that the 285 loses to the 760 even with the advantage of mantle. The statement is misleading because mantle currently hurts the 285 and does not help it.

Its possible the poster did not realize this, but that is what his words mean. Its also possible that the 2+ reviews the poster saw had the 760 winning but I feel that is unlikely as all reviews I have seen show the 285 winning.

 

[96Firebird]

Yes. The poster is implying that the 285 is slower than a 760, even on BF4 with Mantle. Mantle currently is slowing down the 285, so the assumption that mantle makes the 285 faster, and that with mantle the card cannot keep up is wrong.

The way he phrased it sounds like: "Even with the help of Mantle, the 285 still couldn't beat 760", suggesting it lost overall and even in Mantle cases, while the opposite is true with 285 easily beating 760 overall.

Bolded above is the problem you guys have. Instead of reading exactly what he said, you try to interpret it in different ways. What he said was exactly true, no matter how much you hate to hear it.

Its also possible that the 2+ reviews the poster saw had the 760 winning but I feel that is unlikely as all reviews I have seen show the 285 winning.

http://www.anandtech.com/show/8460/amd-radeon-r9-285-review/10

http://hardocp.com/article/2014/09/02/msi_radeon_r9_285_gaming_oc_video_card_review/5#.VBH4KGM9Ml0

There you go, 2 reviews that show the 760 beating the 285 in Battlefield 4.

 

[lopri]

Reading AT's review there is a huge improvement in 2D processing power with GCN 1.2. I have no idea why it did not make it earlier when they were so keen to advertise 4K gaming with Hawaii - I guess it's better late than never for AMD but it kind of sucks for 290/290X owners.

 

[realibrad]

Bolded above is the problem you guys have. Instead of reading exactly what he said, you try to interpret it in different ways. What he said was exactly true, no matter how much you hate to hear it.



http://www.anandtech.com/show/8460/amd-radeon-r9-285-review/10

http://hardocp.com/article/2014/09/02/msi_radeon_r9_285_gaming_oc_video_card_review/5#.VBH4KGM9Ml0

There you go, 2 reviews that show the 760 beating the 285 in Battlefield 4.

So what you did is another great example of the misuse of a word. You just said. You gave 2 "reviews" that show the 760 beating the 285. You meant to say slides or select situations, but the word review is misused. The entire review says something quite different, such as the hardocp link in which the conclusion stated this...

This one is going to depend on your game title as we have found out. For the most part, the AMD Radeon R9 285 is going to be faster than the GeForce GTX 760. However, there may be some games and situations where these break even in terms of performance, and maybe even a game or two where the GTX 760 is still faster.

Language is a tricky thing, which is why I'm not making personal attacks, as its very hard to use.

Look at it like this. If I said "in reviews I read, car a is faster than car b even when car b uses rockets". The implication of of saying even when car b uses rockets is that car b us using an advantage. Mantle in this context is not a benefit even though the implication was that it helps.

The problem is that you are reading his statement to mean that the 760 beats the 285 in BF4 with mantle. That is not the literal meaning of that statement, even if that is what the poster meant to convey.

I cannot say what the intentions of the writer were, but I can say what his statement means. Intent does not = outcome.

 

[Keysplayr]

Incorrect. What he said was...

He said "even" which means that the 285 loses to the 760 even with the advantage of mantle. The statement is misleading because mantle currently hurts the 285 and does not help it.

Its possible the poster did not realize this, but that is what his words mean. Its also possible that the 2+ reviews the poster saw had the 760 winning but I feel that is unlikely as all reviews I have seen show the 285 winning.

Straw man much?

I said with or without mantle so what difference does it make?

 

[Keysplayr]

So what you did is another great example of the misuse of a word. You just said. You gave 2 "reviews" that show the 760 beating the 285. You meant to say slides or select situations, but the word review is misused. The entire review says something quite different, such as the hardocp link in which the conclusion stated this...



Language is a tricky thing, which is why I'm not making personal attacks, as its very hard to use.

Look at it like this. If I said "in reviews I read, car a is faster than car b even when car b uses rockets". The implication of of saying even when car b uses rockets is that car b us using an advantage. Mantle in this context is not a benefit even though the implication was that it helps.

The problem is that you are reading his statement to mean that the 760 beats the 285 in BF4 with mantle. That is not the literal meaning of that statement, even if that is what the poster meant to convey.

I cannot say what the intentions of the writer were, but I can say what his statement means. Intent does not = outcome.

Nah. Firebird had it right. Words like "implying" and "sounds like" were used to change the context of the original statement. No matter what you say.

 

[el etro]

Some info AT Gives us is very good to out Tonga analysis:

Power consumption on Stock Speeds is the same as Tahiti's. Bad efficiency is result of high clocks from overclocked samples that reviewers got.

On the graph non-oc 285 is at the reference speed and OC 285 is at 965Mhz clock.


The review claims a 3-4% performance uplift with Tonga versus Tahiti.

 

[96Firebird]

So what you did is another great example of the misuse of a word. You just said. You gave 2 "reviews" that show the 760 beating the 285. You meant to say slides or select situations, but the word review is misused. The entire review says something quite different, such as the hardocp link in which the conclusion stated this...

It's pretty clear what I said. 2 reviews that show the 760 beating the 285 in Battlefield 4. For some reason you are ommiting this part, not sure if it is out of shear ignorance or intentional...

Warning issued for personal atack.
-- stahlhart

 

[Paul98]

Maybe new drivers can get some more out of this. Though I do want to know more about how the memory compression affects usage and speed.

 

[ocre]

So you conveniently missed the part where on average it beats 760 by 12-15% in AT's review?

On the whole the 285 is a lot closer to the 770 than 760.

NV has no answer right now to a $200 R9 280, $240-$260 R9 280X and a $340 R9 290 and $450 R9 290X. The $180-450 desktop dGPU market is all AMD right now until 970/980 change the situation.

With 280X outperforming the 760 by 26% at 1080P, when R9 280X is going for $240-260, the 760 hardly makes sense even at $220.

But the r9 285 does?
This is the Tonga thread isnt it?

So I am trying to make sense of your post. Is it, Nvidia doesnt have an answer for AMD EOLed cards that are clearing stock? Or did AMD decide that they would just make their customers pay more for less performance?

Perhaps it has something to do with volume. Do you honestly think that AMD is charging these prices out of the kindness of their hearts? Or maybe, more rationally it has something to do with the fact they need to sell them.

So, its great and all you have given us another sales pitch (plug?) for the 280x and 280 but honestly I have no idea what they have to do with the tonga thread.

 

[desprado]

Maybe new drivers can get some more out of this. Though I do want to know more about how the memory compression affects usage and speed.

Sorry bro but u are taking about AMD drivers.It will take a lot of time to get this problem solved.

This is what Nvidia said in their interview that it is very hard to keep Mantle going new upcoming cards.

 

[PPB]

The way he phrased it sounds like: "Even with the help of Mantle, the 285 still couldn't beat 760", suggesting it lost overall and even in Mantle cases, while the opposite is true with 285 easily beating 760 overall.

I would wait for Thief on mantle with 285 to trust that claim in the review.

Because I feel this is one of the cases the reviewer failed to setup the game (Battlefield 4) correctly.

The very first time I activated Mantle on MP I lost performance too, I was like "WTF is going on". Then I realized, that somehow activating mantle forced the command renderdevice.renderaheadlimit to 0, which always incurs in a loss of performance (on mantle, it can actually halve your fps), while on DX I had it on 2 or 3.

 

[BFG10K]

It's the drivers job to talk with the hardware the best way possible, but it's not the drivers job to optimise the API calls.

It most certainly is the driver's job. The driver developers know the hardware and the API better than any developer does. It's no different to compilers or OoOE reordering instructions.

Besides, if it was all about API calls then the 285 wouldn't be falling over. There's a lot more going on here.

Don't forget that Mantle is a low level API and as such we want to give the control the the dev, not the driver.

All that's happening is the workload of driver optimization is being pushed onto the developers. The only people asking for this are people who don't remember programming under DOS. There's a reason why nobody programs in assembler anymore except outside of certain embedded systems.

If DICE thinks Mantle is so good, how come we don't have a pre-emptive patch for BF4 to support the 285? Where's their "we'll support Mantle for X years" roadmap?

Right now we have two Mantle games and one generation of hardware that runs it, and all it took was a single new card for it to completely fall over.

Meanwhile DirectX can run thousands of games across dozens of generations of hardware. Even if Mantle is faster it means squat if it breaks all existing games every 18 months when new hardware arrives.

Now, AMD will certainly release a new driver that will improve the performance, but if the problem lies in the change of the architecture, AMD can't do anything here.

The driver won't do anything if the application is targeting specific hardware, unless you start going behind the developer's back again, which defeats the purpose of Mantle.

Once you lose that abstraction you need to constantly patch the game, and that's a very bad place to be. It's one of the reasons why we need to run DOS games in emulators on today's systems.

 

[monstercameron]

Right now we have two Mantle games and one generation of hardware that runs it, and all it took was a single new card for it to completely fall over.

Meanwhile DirectX can run thousands of games across dozens of generations of hardware. Even if Mantle is faster it means squat if it breaks all existing games every 18 months when new hardware arrives.

-snip-

Once you lose that abstraction you need to constantly patch the game, and that's a very bad place to be. It's one of the reasons why we need to run DOS games in emulators on today's systems.

firstly, mantle does work but there is a performance regression. This is a first taste of mantle running on a newer uarch, it is a good trial.

secondly, You shouldn't have to constantly upgrade the game because it is still interfacing with a driver, just that alot of the automagic layers have been lifted.

 

[AtenRa]

If DICE thinks Mantle is so good, how come we don't have a pre-emptive patch for BF4 to support the 285? Where's their "we'll support Mantle for X years" roadmap?

You dont need BF4 patch to run Mantle on 285, the current beta driver is not optimized for the 285. Next drivers will address that so stay tuned.

Right now we have two Mantle games and one generation of hardware that runs it, and all it took was a single new card for it to completely fall over.

As of today, we have Three (BF4, Thief, Plants vs. Zombies: Garden Warfare) games that Mantle already works. Also there are Three generations or cards that Mantle already works, GCN 1.0 and GCN 1.1 and Tonga GCN 1.2. All of them are supported by current Mantle 1.0 version or later.

 

[BFG10K]

You dont need BF4 patch to run Mantle on 285, the current beta driver is not optimized for the 285. Next drivers will address that so stay tuned.

This is not reflected by AMD's statements from the article: In explaining the situation, AMD tells us that this is an application level issue due to these games not being familiar with Tonga, and that this can be fixed through further patches.

Nowhere does AMD state it will be resolved through driver updates. And again, doing so would completely defeat the purpose of Mantle given the driver would be doing per-game optimizations, as it happens now on DirectX.

Not to mention that in the earlier days, one of the things we were told about Mantle being superior to DirectX was that we wouldn't have to wait for driver optimizations to get good performance.

But according to you we have to wait for new drivers, and according to AMD we need to wait for application patches - all the while being locked to certain GCN hardware. I say certain because even now some versions don't work properly.

So exactly how is this "superior" to DirectX again?

As of today, we have Three (BF4, Thief, Plants vs. Zombies: Garden Warfare) games that Mantle already works. Also there are Three generations or cards that Mantle already works, GCN 1.0 and GCN 1.1 and Tonga GCN 1.2. All of them are supported by current Mantle 1.0 version or later.

Again, this is not reflected by reality: If you have ever used some of the lower end GCN products (e.g. Cape Verde) then you’ve seen first-hand that these games already are hit & miss depending on the GPU in use, so Tonga is an extension to that limitation.

Again, Mantle can't even do what it's supposed to on existing hardware. Basically only the flagship 290X has a chance, and even then the BF4 path still has issues such as memory leaks.

Eventually there's a point where saying "it's beta" no longer cuts it as an excuse. The reality is that the paradigm around the design of this API is extremely brittle from a software engineering standpoint.

 

[dacostafilipe]

...

You seem to forget something really important here. The 285 runs Mantle, it's just not optimised.

The benefit of something like Mantle, it that the devs can optimise the heck out of it. If AMD starts rewriting calls inside the driver, this would bring back the problem we have today with OGL/DX: having a blackbox to code against with possible undefined behaviours.

You also seem to forget that this problem (optimisation) also exists in OGL/DX. That's the reason why developers (try to) create multiple DX path for maximum performance on specific GPU architectures.