AMD on track for launch of Kaveri in February 2014

[Erenhardt]

Not really, considering the cost of an entire system, it is only about ten percent more for double the performance.

I don't know how it is in you country, but here if you want to upgrade your PC you dont trash the whole box and buy new one. You get the new mobo and APU and call it a day.

I will be upgrading my sister's PC. Old Athlon 64 x2 and 4650 512MB DDR2. I will buy cheap asrock fm2+ mobo, cheap and fast RAM (2400MHz cost alomost the same as 1600MHz), and an APU.
Mobo $80
2x4GB RAM $100
APU= $150?
$350 should be enough (depends on kaveri apu pricing) for a quite a good system for her needs.

The rumor is, there will be no FX kaveri CPUs, at least at release.
Kaveri being GCN based could possibly yield great results in combination with 7750 as a possible upgrade later on.

 

[frozentundra123456]

I don't know how it is in you country, but here if you want to upgrade your PC you dont trash the whole box and buy new one. You get the new mobo and APU and call it a day.

I will be upgrading my sister's PC. Old Athlon 64 x2 and 4650 512MB DDR2. I will buy cheap asrock fm2+ mobo, cheap and fast RAM (2400MHz cost alomost the same as 1600MHz), and an APU.
Mobo $80
2x4GB RAM $100
APU= $150?
$350 should be enough (depends on kaveri apu pricing) for a quite a good system for her needs.

The rumor is, there will be no FX kaveri CPUs, at least at release.
Kaveri being GCN based could possibly yield great results in combination with 7750 as a possible upgrade later on.

Obviously I was talking about building from scratch, because it is impossible to calculate a price for every person's individual situation. Even if you assume 350.00 for the upgrade, I still would argue that even if the discrete is 20% more, that 20% more for double the performance is a very good compromise. But of course, the decision is up to each person, and you seem to have decided on an APU no matter what.

The performance and value could change if you are willing to wait for Kaveri. That is pretty much unknown on every front though: price, performance, efficiency of pairing with a dgpu, and even when it will be readily available.

 

[Erenhardt]

Obviously I was talking about building from scratch, because it is impossible to calculate a price for every person's individual situation. Even if you assume 350.00 for the upgrade, I still would argue that even if the discrete is 20% more, that 20% more for double the performance is a very good compromise. But of course, the decision is up to each person, and you seem to have decided on an APU no matter what.

The performance and value could change if you are willing to wait for Kaveri. That is pretty much unknown on every front though: price, performance, efficiency of pairing with a dgpu, and even when it will be readily available.

That is a bit tricky situation because it all depends on how many broken APUs AMD have. It is all up to AMD to make or break the deal with Athlon. It is hard to argue when APU is 50% more expensive than the cut-down version (5800k vs 750k).

I don't want to put discrete card in there because I have 7750 in other PC, which I'm planing to upgrade to 20nm entry level card.

 

[Blitzvogel]

But that is what AMD needs to do. Occasionally outperforming intel's $600 chip with a $200 chip is what has kept AMD around all these years. Besides, iris pro is nothing anyway. Look at this:

Granted, that's an i7 that is probably only running at 2.6GHz during that benchmark. But really, cpu speed doesnt have that much impact.

I doubt iris pro @ 100W would score higher than 32 at that benchmark. It is really not a good gpu. I dont think its too much to ask that AMD be able to beat it by 50% or more.

Then AMD has to either go quad channel, or beat Intel at their own game in having an ultra robust cache system.

 

[Idontcare]

Then AMD has to either go quad channel, or beat Intel at their own game in having an ultra robust cache system.

Yep.

Thing is, AMD already knows how to design and implement high-bandwidth memory controllers...they already have them in their discrete GPU products.

Why isn't AMD attacking this problem a little more aggressively?

 

[jpiniero]

Something tells me the typical AMD OEM isn't interested in GDDR5.

 

[Blitzvogel]

Something tells me the typical AMD OEM isn't interested in GDDR5.

Or tri/quad channel lol.

 

[SiliconWars]

Yep.

Thing is, AMD already knows how to design and implement high-bandwidth memory controllers...they already have them in their discrete GPU products.

Why isn't AMD attacking this problem a little more aggressively?

Scenario 1 - "AMD can't afford to do blah blah"

Scenario 2 - "Why isn't AMD doing blah blah"

I have to hand it to you IDC, you sure know how to get the full mileage out of the same basic issue. :thumbsup:

 

[tweakboy]

On track for another disaster CPU.

 

[Blitzvogel]

Scenario 1 - "AMD can't afford to do blah blah"

Scenario 2 - "Why isn't AMD doing blah blah"

I have to hand it to you IDC, you sure know how to get the full mileage out of the same basic issue. :thumbsup:

AMD's probably banking on hUMA handling a good chunk of the "memory problem". While it's not in the known spec so far, you never know, AMD might have upped the cache. While I certainly wouldn't expect 128 MB or 64 MB, AMD could go with 16 or 32 MB of eDRAM L3 cache. They have this experience with the Wii U GPU die and the Xbox One.

 

[SiliconWars]

AMD's probably banking on hUMA handling a good chunk of the "memory problem". While it's not in the known spec so far, you never know, AMD might have upped the cache. While I certainly wouldn't expect 128 MB or 64 MB, AMD could go with 16 or 32 MB of eDRAM L3 cache. They have this experience with the Wii U GPU die and the Xbox One.

There is no obvious fix to the bandwidth issue that doesn't require massive sums of money or the industry moving in that direction. AMD will innovate the hell out of what they can do, but they can't compete against countless $billions.

Just trying to keep it real - innovation can only get AMD so far. If they had the money then they'd be a different company.

 

[Blitzvogel]

There is no obvious fix to the bandwidth issue that doesn't require massive sums of money or the industry moving in that direction. AMD will innovate the hell out of what they can do, but they can't compete against countless $billions.

Just trying to keep it real - innovation can only get AMD so far. If they had the money then they'd be a different company.

True, but the work AMD has done for Sony, MS, and Nintendo clearly and certainly possibly could be used for the PC space, just like it's their PC work that is powering the two real next gen consoles (whether you think they qualify as nextgen or not lol).

What is bound to happen is that even if AMD manages to pull out a winner this time with the massively high benches non-reputable sources have been saying, they will still get shut out some how by the OEMs and Intel. We'll also continue to see the huge amount of binned products with a quarter or half of the graphics array disabled that we have been seeing.

I want AMD to do well like everyone else here seems to. Competition is good, but past business decisions really screwed them over in the long run. In ten years, there will likely be a book or two chronicling the demise of the only company that could stand up to Intel in the x86 market at any point in history. If it wasn't for our huge backlog of x86 Windows software and games, I would be more than happy to welcome ARM with an open.....arm lol. Maybe we should just bow down to our new master, while still honoring our past, looking to the future where all our devices, mobile, desktop, and server run the same software.

 

[SiliconWars]

What is bound to happen is that even if AMD manages to pull out a winner this time with the massively high benches non-reputable sources have been saying, they will still get shut out some how by the OEMs and Intel. We'll also continue to see the huge amount of binned products with a quarter or half of the graphics array disabled that we have been seeing.

I wish it had this ending, but sadly both Intel and AMD decided it wasn't worth fighting over a couple of years ago.

AMD might have one last surprise up their sleeve, but it's "last chance saloon" and then some. I really can't emphasise enough that if AMD doesn't win with Kaveri, they'll never "win" again. Mostly because Intel stopped competing years ago.

I want AMD to do well like everyone else here seems to. Competition is good, but past business decisions really screwed them over in the long run. In ten years, there will likely be a book or two chronicling the demise of the only company that could stand up to Intel in the x86 market at any point in history. If it wasn't for our huge backlog of x86 Windows software and games, I would be more than happy to welcome ARM with an open.....arm lol. Maybe we should just bow down to our new master, while still honoring our past, looking to the future where all our devices, mobile, desktop, and server run the same software.

Don't worry about that, AMD has a core of brilliance that has never been removed even after all the trials they have been put through.

I doubt any other company in history has been so abused by a much larger company (and their friend Nvidia) yet still stands. AMD spent the last few years getting out of Intel's way. Now that Intel has abandoned Nvidia because they are no longer useful, AMD will take Nvidia apart while Intel fights companies their own size. This will happen, I guarantee you that.

 

[Shivansps]

Define "win".

AMD sabotages itselft, right now, Kabini and Quad Temash cant compite to last gen Intel Ivy ULV on performance because someone trought OEMs did not wanted to put DC on low end devices, and we are now seeing BT devices implementing dual channel even on 2GB of ram. Right now, a A6-5200 provides better CPU performance than a I3-3227U, but worse, by a considerable margin on IGP... I cant even belive im saying that!

 

[Abwx]

I wish it had this ending, but sadly both Intel and AMD decided it wasn't worth fighting over a couple of years ago.

I dont think so because Intel current CEO was part of this firm
board of direction when they engaged in the illegal campaign
against AMD and people with such unscrupulous behaviour
never change , that is , who did drink will drink again safe
that next time they ll be far more cautious to not be caught ,
the dirty tricks policy never really ended it just morphed
its shape...

AMD might have one last surprise up their sleeve, but it's "last chance saloon" and then some. I really can't emphasise enough that if AMD doesn't win with Kaveri, they'll never "win" again. Mostly because Intel stopped competing years ago.

Don't worry about that, AMD has a core of brilliance that has never been removed even after all the trials they have been put through.

They have indeed a great creativity but they lack a process
that would be up to their enginering skills.

I doubt any other company in history has been so abused by a much larger company (and their friend Nvidia) yet still stands. AMD spent the last few years getting out of Intel's way. Now that Intel has abandoned Nvidia because they are no longer useful, AMD will take Nvidia apart while Intel fights companies their own size. This will happen, I guarantee you that.

Nvidia is playing more dirty than ever given the recent radeons
excellent perfs , i repost here what Tridam , the guy who tested
the radeon 290X at hardware.fr when he asked AMD about the
odd behaviour in Batman with perfs litteraly collapsing when
resolution was increased , here the response he got from AMD :

We are aware that the performance of AMD Radeon™ graphics cards in Batman: Arkham Origins is not up to user expectations. Despite our effort to improve this situation within the timeframe offered by Warner Brothers, our efforts were refused. Further, the design of the game fundamentally and deliberately prohibits AMD’s software engineers from collaborating with the developers to design performance optimizations that can be integrated into the game’s codebase.

AMD strategy is to crush Nvidia s bread and butter offering
thanks to a significantly cheaper die , Mr Read has quite
an agressive strategy of market share conquest...

 

[monstercameron]

We are aware that the performance of AMD Radeon™ graphics cards in Batman: Arkham Origins is not up to user expectations. Despite our effort to improve this situation within the timeframe offered by Warner Brothers, our efforts were refused. Further, the design of the game fundamentally and deliberately prohibits AMD’s software engineers from collaborating with the developers to design performance optimizations that can be integrated into the game’s codebase.

i refuse to believe that nvidia would allow such a thing, especially on the same franchise.Are they not afraid of bad publicity or are they now immune due to the rampant fanboyism surrending such a thing?

 

[USER8000]

i refuse to believe that nvidia would allow such a thing, especially on the same franchise.Are they not afraid of bad publicity or are they now immune due to the rampant fanboyism surrending such a thing?

This is the post:

http://translate.google.com/transla...ier-radeon-hawaii-sujet_942544_7.htm#t8918976

Hardware.fr is the main computer hardware review website in France AFAIK.

 

[blackened23]

When AMD or Nvidia have an agreement with a developer they both pay for exclusivity during development. AMD does the same thing. Now that the development period is largely over for Batman : AO, AMD simply has to release a driver with optimizations. If you'll remember Tomb Raider on GTX had the same issue - AMD had exclusivity with Square Enix so NV was not able to interact with the developers during the creation of the game. This is nothing different. NV had exclusivity during development of Batman: AO but AMD now has the ability to release an optimized driver.

It's called competition. This is how things work - AMD and NV both do this. I don't like it, but it's how things are for better or worse. Certainly AMD does the same thing.

 

[CHADBOGA]

When AMD or Nvidia have an agreement with a developer they both pay for exclusivity during development. AMD does the same thing. Now that the development period is largely over for Batman : AO, AMD simply has to release a driver with optimizations. If you'll remember Tomb Raider on GTX had the same issue - AMD had exclusivity with Square Enix so NV was not able to interact with the developers during the creation of the game. This is nothing different. NV had exclusivity during development of Batman: AO but AMD now has the ability to release an optimized driver.

It's called competition. This is how things work - AMD and NV both do this. I don't like it, but it's how things are for better or worse. Certainly AMD does the same thing.

Yeah but AMD are still pure of heart when they do this and every one of its competitors are inherently evil.

 

[frozentundra123456]

I dont think so because Intel current CEO was part of this firm
board of direction when they engaged in the illegal campaign
against AMD and people with such unscrupulous behaviour
never change , that is , who did drink will drink again safe
that next time they ll be far more cautious to not be caught ,
the dirty tricks policy never really ended it just morphed
its shape...




They have indeed a great creativity but they lack a process
that would be up to their enginering skills.

Get out the tinfoil hats, it is a conspiracy, everyone is out to get AMD.


Nvidia is playing more dirty than ever given the recent radeons
excellent perfs , i repost here what Tridam , the guy who tested
the radeon 290X at hardware.fr when he asked AMD about the
odd behaviour in Batman with perfs litteraly collapsing when
resolution was increased , here the response he got from AMD :




AMD strategy is to crush Nvidia s bread and butter offering
thanks to a significantly cheaper die , Mr Read has quite
an agressive strategy of market share conquest...

Get out the tinfoil hats, everything not favorable to AMD is a conspiracy.

 

[Ajay]

I doubt any other company in history has been so abused by a much larger company (and their friend Nvidia) yet still stands. AMD spent the last few years getting out of Intel's way. Now that Intel has abandoned Nvidia because they are no longer useful, AMD will take Nvidia apart while Intel fights companies their own size. This will happen, I guarantee you that.

It's been well established that the most significant damage done to AMD was done by AMD itself; specifically, Hector Ruiz, the BOD and, very sadly, even Dirk Meyer).
1) Delaying the transition to 65 nm (duh, since AMD was already capacity constrained and the shrink would have boosted capacity).
2) Buying ATi for $5B!!!
3) Selling their fabs and then agreeing to the restrictive WSA, including partially funding node development (so, now we are free of your old fabs, but not really).
4) BD (WTH, really! - we're going to innovate by tossing out already well proven architectures, like SMT, which is cheaper in terms of xtors, and go with CMT instead - oh, and we'll screw it up the on the first shot and then restart from scratch and then screw it up again).

The fact that AMD is still kicking after these and many more failure is practically a miracle. If Kaveri can't out-perform Richland/Trinity - then, IMHO, that's the final nail in the AMD big core x86 line coffin.

Looking forward to December and hoping Kaveri turns out to be a surprise despite some sketchy info on the state of GFL's 28nm (visa vi high performance CPUs).

 

[Idontcare]

Scenario 1 - "AMD can't afford to do blah blah"

Scenario 2 - "Why isn't AMD doing blah blah"

I have to hand it to you IDC, you sure know how to get the full mileage out of the same basic issue. :thumbsup:

I don't really understand your response.

AMD obviously does have the resources, they had enough resources to develop them for their GPUs, right?

So if they can find the resources to develop them (higher bandwidth IMCs) for lower revenue volume SKUs like their discrete GPUs, then why can't they find the resources to do similar for their higher revenue volume APUs?

These aren't facetious questions, something obviously doesn't add up.

I wonder if somewhere internal at AMD the decision was made for business reasons, not technology or accounting reasons, to not make the APUs too beefy for fear of cannibalizing their discrete GPU market?

 

[ShintaiDK]

not make the APUs too beefy for fear of cannibalizing their discrete GPU market?

A classic problem. It was obvious ever since they bought ATI that they would run into this problem with conflict of interest. IGPs keep eating dGPU share from below. It seems AMD dont want to do too much, and keep looking on what Intel does. That however never turns out good.

Result is both dGPU and CPU markets are crumbling for them, again due to the mismanagement of the company.

 

[NTMBK]

A classic problem. It was obvious ever since they bought ATI that they would run into this problem with conflict of interest. IGPs keep eating dGPU share from below. It seems AMD dont want to do too much, and keep looking on what Intel does. That however never turns out good.

Result is both dGPU and CPU markets are crumbling for them, again due to the mismanagement of the company.

It's pretty daft logic, agreed. If AMD really pushed the boundaries on APU performance then they could shut NVidia out of certain markets entirely. Instead they dithered, and now Intel is beating them to it with the Iris Pro.

AMD should accept that the dGPU market is doomed, and try as hard as they possibly can to hold on to their IGP advantage and push it as fast as they can.

 

[mrmt]

I wonder if somewhere internal at AMD the decision was made for business reasons, not technology or accounting reasons, to not make the APUs too beefy for fear of cannibalizing their discrete GPU market?

I don't think business is the real problem here. AMD cannot make money on the GPU line anyway, they barely break even on that segment and the only thing about its results is that it has been consistently shrinking since I can remember. What was a 500MM business is today roughly 300MM. I think they would gladly sacrifice volumes in the shrinking GPU segment to beef up their ailing CPU segment, which used to be more than twice the size by the time Trinity arrived.

One of the weakness of AMD APU is that the CPU is too weak and too power hungry, to the point that even Intel HD4000 beats it in some games. The design is far unbalanced, they could use a better CPU. By beefing up the GPU part AMD would make the APU even more unbalanced, and at 246mm^2 I don't think they can afford more die space that is not devoted to improve the CPU.