How will AMD answer the challenge posed by Haswell?

[Nemesis 1]

Intel has a huge advantage even tho there behind in the tablet /smartphone arenea. Intel(samsung) unlike any other arm company creating smartphones Intell fabs have to show a profit from the dies. Thats all. they do not have to mark up the price anyhigher like NV and apple and others. Samsung has fabs and they also have their own branded smartphone intel needs to do the same./.. Intel needs to compete with samsung and Apple . Apple is a powerful name brand but so is intel as is NV. Intel needs to become more Apple like or Apple will end up owning Intel.. Intell investing in sharp is a sign intel may move in this direction . If intel does this Apple is who intel needs to price with . Intel is a better brandname and likely has a bigger fanclub.

 

[pelov]

Nemsis, I'm finding it difficult to follow you and I highly doubt I'm the only one. What I will leave you with is perhaps a heavy dose of reality:

[Qualcomm] Makes 48 percent of all the money spent on mobile chips

Strategy Analytics beancounters say that just under half the money spent on mobile chips ends up in Qualcomm's back pocket. The next four popular chip companies are Samsung, MediaTek, Broadcom, and Texas Instruments.

The same TI that's pulling out their now defunct OMAP chips is ahead of Intel.

I do think AMD has a unique opportunity here. They're not as tied to x86 as Intel is, thus the flexibility that allows could perhaps present them a unique niche in the market where they can provide both, and maybe even both on the same chip. They're IP is quite weak though, and the companies making money in mobile are the ones with very good wireless/LTE IP -- something AMD lacks entirely.

On one side you've got the 500lb blue gorilla and on the other you've got cutthroat competition and a blistering pace and an ever-changing market, players and their respective positions. If they can net something in the middle they might provide something interesting, but I wouldn't count on a hybrid chip as a long term solution but rather a stopgap.

 

[NTMBK]

Intel is a better brandname and likely has a bigger fanclub.

People who do research into brand name recognition would disagree with you there.

1 Apple
2 Google
3 IBM
4 Microsoft
5 HP
6 Oracle
7 SAP
8 BlackBerry
9 Baidu
10 Facebook
11 Cisco
12 Accenture
13 Tencent/QQ
14 Intel
15 Samsung
16 Siemens
17 Nokia
18 Sony*
19 Infosys
20 Canon

Taken from here. Intel's not got nearly the same brand power it did back in the Pentium heydays.

EDIT: Actually, I'm a bit dubious of Intel being quite so low in that list. The one here seems a bit more grounded in reality.

 

[MisterMac]

People who do research into brand name recognition would disagree with you there.



Taken from here. Intel's not got nearly the same brand power it did back in the Pentium heydays.

You can't really bring mainstream brand value into this discussion.


All Intel needs is brandvalue at the right decision makers - not neccesarily so much as the blind public which listen to what the clueless sales teen @ best buy.


Case and point: Any OEM Desktop\Laptop maker.


They could benefit from the exposure they had in the pentium days sure - but... they have too much control over the production at any rate for it to hurt them - if you claim they're brand value is hurting them.

 

[pelov]

They could benefit from the exposure they had in the pentium days sure - but... they have too much control over the production at any rate for it to hurt them - if you claim they're brand value is hurting them.

At which segment of the market? Intel has less say as to what goes on in mobile than does TI, and they've admitted in pulling themselves out of that same market.

OEMs don't just sell desktops and laptops, they sell mobile devices as well. Where Intel dictates what goes on in Ultrabooks, conversely they have nearly zero say in OEMs mobile products.

 

[NTMBK]

They could benefit from the exposure they had in the pentium days sure - but... they have too much control over the production at any rate for it to hurt them - if you claim they're brand value is hurting them.

I'm not claiming their brand value would hurt them, no- I agree most people don't care what's inside their phone. I was simply using those stats to disagree with nemesis' claim that Intel's brand power gave some sort of edge over Apple.

 

[ShintaiDK]

I do think AMD has a unique opportunity here. They're not as tied to x86 as Intel is, thus the flexibility that allows could perhaps present them a unique niche in the market where they can provide both, and maybe even both on the same chip. They're IP is quite weak though, and the companies making money in mobile are the ones with very good wireless/LTE IP -- something AMD lacks entirely.

AMD is even more tied with x86 than Intel if any. Plus hybrid chips belong in the same area as reverse hyperthreading.

 

[pelov]

AMD is even more tied with x86 than Intel if any. Plus hybrid chips belong in the same area as reverse hyperthreading.

Idunno. I think Intel's success rides on x86, whichever end of the market you pick. On the other hand, AMD has a pretty strong graphics portfolio that's independent of ISAs, whereas Intel's MIC is just a bunch of little x86 cores. Other than Itanium, everything Intel makes is tied down to x86 (as far as processing goes). Then there's the fact that AMD has already integrated ARM cores into their processors, granted it's for security purposes.

I think at this point, AMD will have to do whatever it can to survive. If that means leaving x86 behind, if only partially, then they'll do that. The bigger question is what happens to Intel if AMD does that? I think Intel needs AMD in x86 only more than AMD does.

 

[NTMBK]

AMD is even more tied with x86 than Intel if any. Plus hybrid chips belong in the same area as reverse hyperthreading.

They're already putting a (teeny teeny tiny) ARM core into their new APUs, to let them access the ARM security platform, so clearly having ARM and x86 cores co-existing on one die isn't a problem. Whether it's a good idea is another matter, of course, but it's certainly a possibility. Very limited in scope, though- they'd essentially need someone like Apple going "We want to transition from x86 to ARM, but we want you to put a Jaguar core on the die for us so we can run backwards compatibility". And there's only a couple of generations of business they can get out of that.

 

[pelov]

Very limited in scope, though- they'd essentially need someone like Apple going "We want to transition from x86 to ARM, but we want you to put a Jaguar core on the die for us so we can run backwards compatibility". And there's only a couple of generations of business they can get out of that.

Yea, that really would be a stopgap solution rather than a niche worth filling. Temporary cash cow, yes, but not anything long term.

That's why I'm wondering just what AMD has brewing as far as SoCs go. They're not winning anyone over with Hondo, a 1ghz x86 netbook processor with 4.5W TDP not including the FCH, WiFi, etc.

 

[tipoo]

AMD COULD use the ARM ISA, but what would be the benefit? Android and Windows 8 can both run on either.

 

[pelov]

AMD COULD use the ARM ISA, but what would be the benefit? Android and Windows 8 can both run on either.

That's a pretty big benefit, no? Along with access to over 1.5m applications. Granted, AMD has teamed up with Bluestacks to port over Android applications to x86. That really goes to show how far ahead ARM/Apple are as far as software goes. Whereas Windows has a few content creation applications and the business crowd (that doesn't upgrade every often), ARM has active developers making quite a bit of money.

BTW, another issue hurting Intel is Microsoft's approach. I think Intel really needs to cut off ties with them. Within the next 5 years I'd predict Microsoft will be a shell of its former self. Just a couple of weeks ago, even facing the fact that Win8 has barely any applications that it's shipping with (in the thousands compared to the over 700k+ for Android and iOS each), Microsoft has kept to its initial decision in not sending out developer kits to smaller and medium sized software devs. I mean, you can't even get one if you wanted to pay for it out of pocket. Talk about an idiotic approach to entering the mobile space ^_^

Not enough applications? It's ok, we'll just limit them even further

 

[NTMBK]

BTW, another issue hurting Intel is Microsoft's approach. I think Intel really needs to cut off ties with them. Within the next 5 years I'd predict Microsoft will be a shell of its former self. Just a couple of weeks ago, even facing the fact that Win8 has barely any applications that it's shipping with (in the thousands compared to the over 700k+ for Android and iOS each), Microsoft has kept to its initial decision in not sending out developer kits to smaller and medium sized software devs. I mean, you can't even get one if you wanted to pay for it out of pocket. Talk about an idiotic approach to entering the mobile space ^_^

Not enough applications? It's ok, we'll just limit them even further

You can develop for Windows 8 using the free preview of the OS that Microsoft distributed, on any PC capable of Windows 7.

(Also, I presume your "in the thousands" refers to "Windows 8 Store" applications, as including all existing Windows software there is a LOT more than that on Windows.)

 

[pelov]

(Also, I presume your "in the thousands" refers to "Windows 8 Store" applications, as including all existing Windows software there is a LOT more than that on Windows.)

I was referring to both. This means the hardware that the applications run on depends on the version of the tablet. .NET or RT? Both? What about the ARM-based versions?

 

[NTMBK]

I was referring to both. This means the hardware that the applications run on depends on the version of the tablet. .NET or RT? Both? What about the ARM-based versions?

Ah, Windows 8 specifically on refers to the x86 machines- the ARM version's been branded Windows RT now.

 

[pelov]

Ah, Windows 8 specifically on refers to the x86 machines- the ARM version's been branded Windows RT now.

Yet you have access to both via the Win8 Metro app store.

This is what I mean by fragmentation. While ARM's entire ecosystem is hardware fragmented, the software, for the most part (looking at you Tegra 3/nVidia), is pretty similar. Install an application and it runs. On Win8 you've got 3 different sets of applications: Metro ARM, Metro x86, and traditional desktop. What's worse is that they behave differently depending on where you launch them. Your Metro applications act differently and have different options than your traditional desktop apps. You'd figure at least MS would set the example and get it right, yet IE isn't the same in Metro or desktop mode. A schizophrenic operating system

I think Intel invested in the MeeGo/Tizen OS jointly with Samsung in order to have something to fall back on in case Microsoft decides to ditch them entirely or dun' goofs it.

 

[ShintaiDK]

They're already putting a (teeny teeny tiny) ARM core into their new APUs, to let them access the ARM security platform, so clearly having ARM and x86 cores co-existing on one die isn't a problem. Whether it's a good idea is another matter, of course, but it's certainly a possibility. Very limited in scope, though- they'd essentially need someone like Apple going "We want to transition from x86 to ARM, but we want you to put a Jaguar core on the die for us so we can run backwards compatibility". And there's only a couple of generations of business they can get out of that.

That ARM core is in the chipset and not the CPU isnt it.

Not to mention they share no resources.

Your SATA controller, NIC etc also got their own "core".

 

[NTMBK]

That ARM core is in the chipset and not the CPU isnt it.

Not to mention they share no resources.

Your SATA controller, NIC etc also got their own "core".

Nope, it's in the APU, not the chipset, as far as I can tell.

 

[pelov]

That ARM core is in the chipset and not the CPU isnt it.

Not to mention they share no resources.

Your SATA controller, NIC etc also got their own "core".

I believe it's on-die

AMD executives said they will integrate ARM€™s Cortex-A5 processor with TrustZone security technology into future accelerated processing units (APUs), a move that will help AMD€™s efforts grow its presence in mobile devices, particularly tablets. TrustZone, first introduced in 2004, makes financial transactions, mobile payments and streaming content safer by creating secure zones on the chip that is protected from attackers. TrustZone is in all of ARM€™s Cortex-A chips, which are used in mobile devices.

AMD is somehow less committed to x86 [than Intel],€ he said in an interview, adding that it didn€™t mean that AMD was giving up on x86. Instead, €œphilosophically they don€™t have as many barriers to€ partnering with ARM.

When they're willing to integrate an ARM core despite having their own implementation, it tells you they're more willing than Intel to stray from x86.

 

[ShintaiDK]

I believe it's on-die





When they're willing to integrate an ARM core despite having their own implementation, it tells you they're more willing than Intel to stray from x86.

Well lets see, first product will _maybe_ come in 2013 or 2014.

Intel got other uarchs besides x86, and developed quite a few non x86 chips. AMD developed what again?

 

[pelov]

Intel got other uarchs besides x86, and developed quite a few non x86 chips. AMD developed what again?

Don't be an idiot. You know I'm no AMD fanboy, I was just stating that you were wrong...which you are. Don't get pissy.

Secondly, Intel's other uArch is the Itanium. I don't have to remind you the only reason that's still alive is because HP is paying Intel money to keep it alive, upwards of 80million plus. There have been articles noting that Intel wanted to cut the cord and let that red-headed stepchild starve to death but because of contractual obligations they can't. It was also developed to be a more "natural and seamless" approach to 64-bit, yet AMD won that battle and Intel followed suit with their own tweaked version of x86_64

Intel introduced x86 GPUs, ffs. How much more x86-centric can you possibly get?

Does anyone remember what the gaming devs and even people who worked on Larrabee said? Let's say they weren't exactly enthusiastic about the product

 

[ShintaiDK]

Don't be an idiot. You know I'm no AMD fanboy, I was just stating that you were wrong...which you are. Don't get pissy.

Secondly, Intel's other uArch is the Itanium. I don't have to remind you the only reason that's still alive is because HP is paying Intel money to keep it alive, upwards of 80million plus. There have been articles noting that Intel wanted to cut the cord and let that red-headed stepchild starve to death but because of contractual obligations they can't. It was also developed to be a more "natural and seamless" approach to 64-bit, yet AMD won that battle and Intel followed suit with their own tweaked version of x86_64

Intel introduced x86 GPUs, ffs. How much more x86-centric can you possibly get?

Does anyone remember what the gaming devs and even people who worked on Larrabee said? Let's say they weren't exactly enthusiastic about the product

One could also mention i960, xScale and so on. Other uarchs aint exactly new to Intel. Intel knows exactly what potential for example ARM got.

As you point out with IA64, x86 already won. This also includes against ARM for the very same reasons. Plus a few more in that case.

And for HP. Look at this:

Numbers are in billions.

 

[gammaray]

Simple, they can't.

AMD revenue: $6.3B
Intel R&D: $8.3B

Intel's R&D budget is 30% greater than AMD's revenue. AMD cannot compete with that.

That's why you don't see AMD doing platform level stuff, such as taking what used to be analog circuits and making them digital (Intel's presentation on making a digital WiFi radio).

AMD profit: 0$

in fact they are losing money each quarter.
I don't see how they can survive at all.

 

[2is]

Intel simply has too much money, too much engineering potential, and too much fab capacity for them to simply fade away, now, or in the distant future. Even if x86 eventually dies, Intel will still be around, making non-x86 CPUs, and plenty of them.

 

[sm625]

Even if AMD did everything I suggested, they would still have a tough time. To recap:

1. Move one or more GPU SIMDS into each cpu FPU core. Rewrite the drivers from the ground up to accommodate and optimize.

2. Put a quality SSD controller on the cpu die. Duh...

3. Modify the DDR3 bus to be able to share the bus with NAND. Provide a home for multichip multiplexing heterogenous memory (flash + DRAM) and they will come. This would of course also provide a compatible home for PCM when it arrives.

4. Place one or more ARM cores on the same die as the x86 cores.

5. Add custom DSP quicksync type logic to hardware accelerate java, adobe flash, and/or anything else that tends to peg a low power cpu. Ideally the gates inside the gpu should be reconfigurable to some degree, to allow custom reprogramming of the gpu similar to that of an fpga. Of course I dont think AMD is big enough to see something like this through, but keep an eye on xilinx and altera! If one of them buy AMD, watch out. This could be big.