AMD GPU14 Tech Event Sept 25 - AMD Hawiian Islands

[Jaydip]

Anyone who keeps blaming AMD for introducing API as an unfair way to compete, AMD was asked directly by developers for this!

AMD's head of GPU developer relations, Richard Huddy, says games developers are asking AMD to 'make the [DX] API go away'

AMD pleases enthusiasts with dual bios switches, voltage control overclocking, price/performance and game bundles.
AMD now is listening to developer feedback.
While not a Titan killer per say, we now should have a competitive AMD that forces NV to innovate, release faster cards, drop prices.

Response to all of this? This is so unfair, negative feedback, backlash! Brilliant. Didn't Titan owners already say that even if AMD releases a card with Titan's performance for $500, it won't impress them since they had Titan for 10 months? So what's the problem here exactly? AMD upped the performance which is better for gamers.

I find it surprising RS that people who bash Physx because it is proprietary is happy about mantle.This is way too funny.

 

[RussianSensation]

@ RS "NV's cards are just slower because Kepler is an inferior architecture for compute" is simply untrue.In the professional space kepler is much faster than tahiti where it really matters.

I think what you meant to say is that GK110 SKU is much faster than Tahiti XT SKU. Yet in your post you are comparing Kepler architecture to Tahiti SKU. Not apples-to-apples comparison. What happens if you bring a 550mm2 GCN vs. a 550mm2 Kepler GK110? Game over in compute. It took NV a 561mm2 die to beat a measly 365mm2 Tahiti XT at compute. Wow, big deal. In OpenCL and games that use DirectCompute for say SSAA, ambient occlusion, global illumination, AMD provides way superior performance per mm2. How is that possible? Only if GCN is a more efficient compute architecture per mm2 than Kepler is for compute shaders in games. In games like Sleeping Dogs, Hitman, Dirt 3, TR, COH2, 7970GE laid the smack down on the 680.

I find it surprising RS that people who bash Physx because it is proprietary is happy about mantle.This is way too funny.

Because it's not the same. People bash PhysX not only because it's locked to NV, but because for 5-6 years NV didn't produce a single killer game that uses PhysX the way Ageia intended it. No one stops NV from releasing its own API. If Mantle fails to take off, it'll also be ridiculed by AMD/NV owners. Don't worry. :thumbsup:

 

[SiliconWars]

Anyone who keeps blaming AMD for introducing API as an unfair way to compete, AMD was asked directly by developers for this!

AMD's head of GPU developer relations, Richard Huddy, says games developers are asking AMD to 'make the [DX] API go away'

AMD pleases enthusiasts with dual bios switches, voltage control overclocking, price/performance and game bundles.
AMD now is listening to developer feedback.
While not a Titan killer per say, we now should have a competitive AMD that forces NV to innovate, release faster cards, drop prices.

Response to all of this? This is so unfair, negative feedback, backlash! Brilliant. Didn't Titan owners already say that even if AMD releases a card with Titan's performance for $500, it won't impress them since they had Titan for 10 months? So what's the problem here exactly? AMD upped the performance which is better for gamers, while GTX780/Titan owners enjoyed a fast card for months. Everyone wins.

The reason is simple, they are scared that this is the beginning of the end for Nvidia. They should be scared. AMD has so completely and utterly outmaneuvered Nvidia on all fronts it's scary. We could be looking at situations where $200 AMD cards outperform Nvidia's $500 cards (which won't be $500 for long). Their APU's are going to perform like crazy as well, can't wait to see how a 6800K does vs some discrete Nvidia cards.

 

[SiliconWars]

I find it surprising RS that people who bash Physx because it is proprietary is happy about mantle.This is way too funny.

Even if Mantle was proprietary it would still be completely fair because of the constant stream of proprietary garbage Nvidia attempts in order to corner the market. If AMD had to beat them at their own game, all the better. In this case sure call me a hypocrite - once Nvidia is dead and buried AMD can get back to open standards.

Oh btw, Nvidia's big secret is coming soon. You can bet your last dollar it'll be proprietary in the real sense. We'll see if it's enough to save them or not.

 

[Carfax83]

And? Guess what NV and Intel threw $ at developers for a decade for architecture/brand specific/compiler/driver optimizations. AMD fought fair and square for years, offered nearly unbeatable price/performance (4850/4870/5850/5870/6950 unlocked) beat NV to market (HD5000), kept features open (TressFX), offered dual bioses for free, and their profitability & market share hardly improved. If you want to blame someone for this development, blame NV and consumers that keep rewarding NV. Now, AMD said forget this, we are going to go from working closely with developers to having their babies.

Well if AMD has offered unbeatable price/performance as you put it, and enthusiasts and gamers still rejected them, what does that say?

Does that mean we are stupid because we don't realize how much better AMD is?

I bought AMD years ago, two HD 4870s to be precise. After a few days, I ended up returning them due to a host of issues. After years of using NVidia, I gave AMD a shot and they blew it....big time. The experience left such a bad taste in my mouth, and it confirmed everything bad I'd read and heard about AMD from other people; especially concerning their driver quality.

Fast forward to years later and Crossfire still hasn't been fixed. D:

And btw, TressFX uses DirectCompute, which is vendor agnostic and made by Microsoft.

R9 280X will have unlocked voltage and support for DX11.2. See, AMD again continues to please enthusiasts by offering them bios switches for flashing and full voltage control vs. NV that allows almost no voltage control and charges $100-150 premiums for cards that are 7-10% faster ($280-300 1Ghz 7970 vs. GTX770 2/4GB).

If people continue to pay those prices, you can hardly blame NVidia for trying to make a greater profit.

Worth and value is all about perception. Rightly or wrongly, many people believe that NVidia video cards are inherently superior to AMD video cards..

I don't feel sorry for NV even for a second. They pulled so many stunts with gimped AC performance, Batman games, etc.

The Batman AC performance issues was due to the developer's shoddy DX11 implementation.. I'm pretty sure NVidia didn't have anything to do with that.. At any rate, the problems were patched out.

AMD doesn't stop NV from running DirectCompute in AMD GE titles, doesn't prevent NV from using features like TressFX.

How could AMD stop NVidia from running DirectCompute, when they don't even own it? DirectCompute is owned by Microsoft, and is vendor agnostic.

NV's cards are just slower because Kepler is an inferior architecture for compute

For compute in general, or double precision? If Kepler is slower for compute generally speaking, then I'd like to know how they caught up with AMD in Tomb Raider.

NV ends up making chips 20-30% larger to compete. Not to worry since Maxwell should blow GCN out of the water since NV has been working on it for 4-5 years. If anything we want a more competitive AMD so that NV doesn't keep charging $1,000 for flagship single GPUs. The only way for this to happen is for the cash starved AMD to figure out a way to compete better. This is going to bring more competition, not less since it'll force NV to make even faster GPUs if they are going to be behind on the software optimization level. They can make up for it on the hardware level with 550mm2 die. At the rate AMD was going, they are lucky to break even or make measly $20-30 million a quarter on the entire GPU division.

I want AMD to be successful with both their GPUs and CPUs. Weak competitors are never good for consumers.

 

[Final8ty]

I find it surprising RS that people who bash Physx because it is proprietary is happy about mantle.This is way too funny.

Most people bash Physx because of its potential is not being used not because its proprietary. Its just in the case of Physx its lockout is getting in the way of its potential.

Now this may not be the case for Mantle and even if Mantle was locked out, it has no potential gaming changing properties like GPU Physx could if used to the full extent but then would unplayable on anything but NV cards, Mantle does not have the same issues as it has no potential game play changing properties.

 

[RussianSensation]

We could be looking at situations where $200 AMD cards outperform Nvidia's $500 cards (which won't be $500 for long). Their APU's are going to perform like crazy as well, can't wait to see how a 6800K does vs some discrete Nvidia cards.

It's already happening in games "no one plays" like Amnesia where a $180 HD7870 beats a $400 GTX770 and a $320-350 HD7970GE is only 2 fps slower than a $650 Titan.

No one says anything about this being unfair but god forbid if a major developer like DICE had their game running faster on AMD hardware because AMD decided to work even closer on Frostbite game engine optimization. I guess when NV had a 10 months lead in BF3, it was fair but now if BF4 is a title where AMD leads, it's totally unfair! :hmm:

 

[Jaydip]

I think what you meant to say is that GK110 SKU is much faster than Tahiti XT SKU. Yet in your post you are comparing Kepler architecture to Tahiti SKU. Not apples-to-apples comparison. What happens if you bring a 550mm2 GCN vs. a 550mm2 Kepler GK110? Game over in compute. It took NV a 561mm2 die to beat a measly 365mm2 Tahiti XT at compute. Wow, big deal. In OpenCL and games that use DirectCompute for say SSAA, ambient occlusion, global illumination, AMD provides way superior performance per mm2. How is that possible? Only if GCN is a more efficient compute architecture per mm2 than Kepler is for compute shaders in games. In games like Sleeping Dogs, Hitman, Dirt 3, TR, COH2, 7970GE laid the smack down on the 680.



Because it's not the same. People bash PhysX not only because it's locked to NV, but because for 5-6 years NV didn't produce a single killer game that uses PhysX the way Ageia intended it. No one stops NV from releasing its own API. If Mantle fails to take off, it'll also be ridiculed by AMD/NV owners. Don't worry. :thumbsup:

No I meant GK104 only.Quadro K5000 is much faster than it's tahiti counterpart.NV didn't release a single killer game because if it did it would alienate many pc gamers.That is why Physx is only eye candy.

 

[Final8ty]

Because it's not the same. People bash PhysX not only because it's locked to NV, but because for 5-6 years NV didn't produce a single killer game that uses PhysX the way Ageia intended it. No one stops NV from releasing its own API. If Mantle fails to take off, it'll also be ridiculed by AMD/NV owners. Don't worry. :thumbsup:

Indeed.

 

[FredNotFound404]

The Presentation was one of the worse I'v ever seen, yet apart from the audio bit AMD seems to be coming strong for 2014, I still want some proper reviews on the 290x but if mantle shows its potential, we might be heading for an interesting future, specially since MS seems to have lost interest in DirectX API.

 

[Jaydip]

Even if Mantle was proprietary it would still be completely fair because of the constant stream of proprietary garbage Nvidia attempts in order to corner the market. If AMD had to beat them at their own game, all the better. In this case sure call me a hypocrite - once Nvidia is dead and buried AMD can get back to open standards.

Oh btw, Nvidia's big secret is coming soon. You can bet your last dollar it'll be proprietary in the real sense. We'll see if it's enough to save them or not.

Dude I don't give a damn what is proprietary or not but I find this simply hilarious. AMD couldn't beat Intel or NV not because of proprietary stuff but because the opponent had better products in the shelf.

 

[Final8ty]

The reason is simple, they are scared that this is the beginning of the end for Nvidia. They should be scared. AMD has so completely and utterly outmaneuvered Nvidia on all fronts it's scary. We could be looking at situations where $200 AMD cards outperform Nvidia's $500 cards (which won't be $500 for long). Their APU's are going to perform like crazy as well, can't wait to see how a 6800K does vs some discrete Nvidia cards.

NV has the Professional market and other things and its gaming market share has a longways to fall for them to end which is very unlikely.

 

[Jaydip]

Most people bash Physx because of its potential is not being used not because its proprietary. Its just in the case of Physx its lockout is getting in the way of its potential.

Now this may not be the case for Mantle and even if Mantle was locked out, it has no potential gaming changing properties like GPU Physx could if used to the full extent but then would unplayable on anything but NV cards, Mantle does not have the same issues as it has no potential game play changing properties.

Source? we have no information to go on here.I replied to RS regarding the first part.

 

[RussianSensation]

For compute in general, or double precision? If Kepler is slower for compute generally speaking, then I'd like to know how they caught up with AMD in Tomb Raider.

$1,300 of overclocked GPUs to beat a $650 card doesn't count as catching up in my book.

21% faster than 770 and 23% faster than 680 doesn't seem like catching up to me. 770 costs $380-450 for crying out loud. The Tahiti XT chip is nearly 2 years old too! :sneaky:

 

[Final8ty]

Source? we have no information to go on here.I replied to RS regarding the first part.

Logic! just like GPU Physx only adds more gfx and does not change the game for non NV owners.

Intel have some CPU exclusive stuff in Dirt2 and again it didt not change the game, only added more gfx.

So Mantle will only add more performance and maybe more gfx if its locked out to others.

 

[RussianSensation]

No I meant GK104 only.Quadro K5000 is much faster than it's tahiti counterpart.NV didn't release a single killer game because if it did it would alienate many pc gamers.That is why Physx is only eye candy.

In what benchmarks that are "compute"? The comparison of Quadro K5000 vs. FirePro isn't even relevant since most of the performance in that market segment is based on driver/optimization. If you have a program that used CUDA for 5-6 years, of course it's going to run faster on Quadro. And the discussion was for compute in games or otherwise I would have specified FirePro vs. Quadro. We should see R9 290X beat GTX780 in nearly every compute heavy title where 7970GE beat 680.

If you look at distributed computing forum, AMD cards annihilate NV in nearly every modern computing program. Folding@Home used to run faster on NV cards because it was compiled with an old inefficient code and never took advantage of GPUs with modern features. Hashing, Milkyway@home, CollatzConjecture, etc. the more modern DC programs -- NV is not even on the map. Those programs were not designed specifically to run faster on AMD hardware while NV worked for more than half a decade to optimize CUDA + developer framework for Quadro cards.

 

[cmdrdredd]

If you crap up the thread much more you're going to need to get yourself a diaper to wear.

Good grief man, let it go. So AMD will own one benchmark, Battlefield 4 - in December - not even when the game releases...

We don't even know if the performance increase will put it on par with titan and gtx 780 numbers either. For all we know it may be a nice gain but just not enough to make a difference when declaring a "winner".

 

[sontin]

Anyone who keeps blaming AMD for introducing API as an unfair way to compete, AMD was asked directly by developers for this!

AMD's head of GPU developer relations, Richard Huddy, says games developers are asking AMD to 'make the [DX] API go away'

And developers asked for Tessellation and GPU-PhysX. So why did people complain about Crysis 2 and all these GPU-PhysX games? :sneaky:

 

[DarkKnightDude]

And developers asked for Tessellation and GPU-PhysX. So why did people complain about Crysis 2 and all these GPU-PhysX games? :sneaky:

Nobody complained about that tessellation was added to Crysis 2. They objected to how it was added.
http://techreport.com/review/21404/crysis-2-tessellation-too-much-of-a-good-thing

 

[ShintaiDK]

Asking for DX to go away. Is the same of removing competition in the game developing segment.

 

[Jaydip]

Logic! just like GPU Physx only adds more gfx and does not change the game for non NV owners.

Intel have some CPU exclusive stuff in Dirt2 and again it didt not change the game, only added more gfx.

So Mantle will only add more performance and maybe more gfx if its locked out to others.

Then by your own logic how can you call it progress ?

 

[RussianSensation]

And developers asked for Tessellation and GPU-PhysX. So why did people complain about Crysis 2 and all these GPU-PhysX games? :sneaky:

DX API affects all GPU vendors since it's inherently inefficient according to developers. Using Crysis 2 to justify its unoptimized use of tessellation is not related in any way to Mantle since Mantle is supposed to improve performance, not degrade it. If Crysis 2 looked great with a 50-100% performance hit that tessellation brought, no one would have said a word. Instead, you had tessellated water that you couldn't see on a screen and inferior graphics to Crysis 1.

---


All 119 slides can be found here:

http://www.techpowerup.com/191453/amd-gpu14-event-detailed-announces-radeon-r9-290x.html

Asking for OpenGL to go away with a new DX API is the same of removing competition in the game developing segment.

Sound familiar? Industry desires a more efficient way to utilize hardware resources. If there is a superior alternative to DX, let it be an option. Worst case it fails, best case, DX goes the way of OpenGL and we get better looking games that run faster.

---

http://www.techradar.com/news/compu...u-the-radeon-r9-290x-1183950?src=rss&attr=all
Most notable it seems is that gone are the Crossfire connectors. Crossfire will now be handled via PCI Express. Additionally, the Radeon R9 290X will sport two DVI outputs, a DisplayPort output, and an HDMI output.

 

[flopper]

We don't even know if the performance increase will put it on par with titan and gtx 780 numbers either. For all we know it may be a nice gain but just not enough to make a difference when declaring a "winner".

True, just that the amd guy said the Titan wont be near them when mantle is out with Bf4.
they have the data to make such statement means it will perform big with Mantle and BF4.

 

[cmdrdredd]

Asking for DX to go away. Is the same of removing competition in the game developing segment.

Pretty much my thought here. I have my doubts that developers will alienate potential customers by using an api that only works on specific hardware. At best we may see a few titles using this through gaming evolved while everything else runs off directx and you brute force your way to performance.

 

[sontin]

DX API affects all GPU vendors since it's inherently inefficient according to developers. Using Crysis 2 to justify its unoptimized use of tessellation is not related in any way to Mantle since Mantle is supposed to improve performance, not degrade it. If Crysis 2 looked great with a 50-100% performance hit that tessellation brought, no one would have said a word. Instead, you had tessellated water that you couldn't see on a screen and inferior graphics to Crysis 1.

Tessellation in Crysis 2 costs 15% on a Fermi card. And Tessellation is providing more FPS than the normal way over the CPU.

And yet people complained. Mantle is no different, only its working on a specific hardware configuration. Unless Tessellation which people can activate with Fermi, Cypress, Cayman, Kepler and GCN gpus.