Where is AMD/ATI in the mobile hardware sector?

[OOBradm]

I'm reading Anand's article here: http://www.anandtech.com/show/4144/...gra-2-review-the-first-dual-core-smartphone/5

and there is all of this talk of Nvidia really becoming a leader in this new mobile race.... from the article:

there are three major GPUs you’ll see crop up in devices this year: Imagination Technologies’ PowerVR SGX Series5 and Series5XT, Qualcomm’s Adreno 205/220 and NVIDIA’s GeForce ULV."

Where is AMD here? Shouldn't they be trying to capitalize on this new market? Isn't the x86 ISA a deadend? Shouldn't they be trying to at least get a foot in the door of the future?

 

[Patranus]

The same place they are in the desktop sector, no where.

 

[zokudu]

They sold off their Mobile Division to Qualcomm years ago. They actually recently fired Dirk Meyers allegedly because he was against reinvesting the mobile market preferring to focus on improving their CPU and GPU products.

 

[simonizor]

Intel and AMD pretty much stay out of the cellular market. Their chips aren't designed for that kind of thing, and it would be a huge investment for them to get it going. Better off for them to just leave it to the companies that already specialize in it.

 

[kaerflog]

Oh Intel is going there, no doubt about it.
Intel has the money and resources.
Mobile is the future and it would stupid of Intel not to dedicate the resources for it.

 

[LostPassword]

intel is teamed up with nokia doing the meego thing. they got a joint phone coming out powered by atom.

good news is amd just finished their atom killer chips for netbooks.
bad news is power draw still too high. probably take another generation or two before they enter the market.

 

[Genx87]

They had the brilliant plan to sell off their mobile division in 2009 for a whole 65 million dollars. If brains were dynamite. AMD couldnt blow their nose.

 

[ChAoTiCpInOy]

ARM has such a legup on AMD/ATI and Intel, I don't know how they would catch up.

 

[MrX8503]

Intel and AMD pretty much stay out of the cellular market. Their chips aren't designed for that kind of thing, and it would be a huge investment for them to get it going. Better off for them to just leave it to the companies that already specialize in it.

Staying out of the cellular market would be a huge mistake on their part. The only reason why they're not in it yet is because the large investment it would take to penetrate a market dominated by ARM. ATI/Intel are planning behind closed doors most likely.

intel is teamed up with nokia doing the meego thing. they got a joint phone coming out powered by atom.

good news is amd just finished their atom killer chips for netbooks.
bad news is power draw still too high. probably take another generation or two before they enter the market.

The Atom chip is not fit for cellphones, the power draw is way too high. As for AMD Zacate, I was under the impression that it was more power efficient than the Atom.

 

[Bateluer]

ARM has such a legup on AMD/ATI and Intel, I don't know how they would catch up.

Intel has billions in cash. If they were serious, they could license ARM and purchase some smaller ARM designer for immediate talent and move into the sector in force.

Don't understand Patranus' comment about them going no where in the desktop sector though. The Radeon 5xx0 and 6xx0 are at the top of the heap right now.

 

[cbn]

good news is amd just finished their atom killer chips for netbooks.
bad news is power draw still too high. probably take another generation or two before they enter the market.

I've read posts about how great a die shrinked bobcat would be for phones, but I would hope AMD would go much lower and more stripped down than that (if at all possible).

No legacy, no floating point on the CPU part (for cost and power savings). Some kind of Bare minimum Micro APU with maybe ARM Cortex A5 (to undercut the original Tegra) and work their way up from that point on?

With Hardware acceleration starting to appear on Android (and hopefully Meego as well) maybe something like this could work for them? Or does someone with a programming/hardware background have a differing opinion on this? (I am not a tech guy myself)

 

[Patranus]

Don't understand Patranus' comment about them going no where in the desktop sector though. The Radeon 5xx0 and 6xx0 are at the top of the heap right now.

AMDs market share in terms of desktop hardware is 6% to 8%.
That is going nowhere fast.

 

[Genx87]

Intel has billions in cash. If they were serious, they could license ARM and purchase some smaller ARM designer for immediate talent and move into the sector in force.

Don't understand Patranus' comment about them going no where in the desktop sector though. The Radeon 5xx0 and 6xx0 are at the top of the heap right now.

Intel still owns a license for ARM processors from when they had XScale.

 

[zebrax2]

AMDs market share in terms of desktop hardware is 6% to 8%.
That is going nowhere fast.

IIRC their market share is ~20% for CPUs while 50/50 with nvidia on the GPU market.
AMD selling their mobile division might look pretty stupid now but as IDC said in another thread selling that division at that time might be the best choice considering their company is in the brink of a collapse and every dollar they could get matters plus IIRC the division has a history of not being profitable when AMD owned it.

 

[Puddle Jumper]

I've read posts about how great a die shrinked bobcat would be for phones, but I would hope AMD would go much lower and more stripped down than that (if at all possible).

No legacy, no floating point on the CPU part (for cost and power savings). Some kind of Bare minimum Micro APU with maybe ARM Cortex A5 (to undercut the original Tegra) and work their way up from that point on?

With Hardware acceleration starting to appear on Android (and hopefully Meego as well) maybe something like this could work for them? Or does someone with a programming/hardware background have a differing opinion on this? (I am not a tech guy myself)

Why would you do away with floating point on the cpu ? That would be a massive step backwards in cpu design and goes against the current trend of increasing integration.

 

[cbn]

Why would you do away with floating point on the cpu ? That would be a massive step backwards in cpu design and goes against the current trend of increasing integration.

I was thinking floating point on the GPU. (However, like I said I am not a programmer or Engineer so I don't know how feasible a move like that would be).

 

[Puddle Jumper]

I was thinking floating point on the GPU. (However, like I said I am not a programmer or Engineer so I don't know how feasible a move like that would be).

Ah, I've heard that option discussed as the potential future of APU's but we are nowhere close to there yet. Even on desktops gpgpu is in it's infancy, on smartphones it's all but non existent.

I wouldn't be surprised if Bobcat showed up in smartphones after a die shrink or two since it has enough of a performance lead that they could still compete in the high end even after waiting a year or more for newer manufacturing processes.

 

[Genx87]

Two die shrinks would put bobcat into the 2014 time frame. By then I am sure there will be plenty of ARM based processors that will do more for less energy consumption. x86 is a dead end in the ultra mobile space due to the complexity of the parts for backwards compatibility.

 

[simonizor]

Staying out of the cellular market would be a huge mistake on their part. The only reason why they're not in it yet is because the large investment it would take to penetrate a market dominated by ARM. ATI/Intel are planning behind closed doors most likely.



The Atom chip is not fit for cellphones, the power draw is way too high. As for AMD Zacate, I was under the impression that it was more power efficient than the Atom.

I just don't think it's a good move on their part yet. They should focus on desktop processors for the time being. I think that down the road, the time will come that they make desktop processors so small that they will fit in pretty much any device.

 

[Puddle Jumper]

Intel has billions in cash. If they were serious, they could license ARM and purchase some smaller ARM designer for immediate talent and move into the sector in force.

Actually Intel could probably just license the Cortex A9 design from ARM and implement it on their 32nm process which should give them a huge advantage over Qualcomm/TI/Samsung/Nvidia who are all building their Cortex A9 parts on 40nm or 45nm processes. They wouldn't actually need to design their own ARM implementation to have a upper hand over their competitors.