AMD looking to buy MIPS?

[ikachu]

http://www.brightsideofnews.com/news/2012/4/16/amd-wants-to-buy-mips2c-but-can-they-beat-google.aspx

Pretty interesting if true...

 

[BallaTheFeared]

I'm having a hard time getting through the spelling errors and missing words

stronge - really?

 

[ikachu]

Oh nm just saw it was that Theo Valich guy; I don't think he has much credibility...

 

[ikachu]

I guess MIPS hiring Goldman-Sachs is really the only piece of (semi-)concrete news and the rest is speculation.

http://www.bloomberg.com/news/2012-...o-have-hired-goldman-sachs-to-find-buyer.html

 

[sm625]

What AMD needs is to find a company like that Euclideon graphics company, AND have it be something real that actually lives up to its potential, AND they also have to buy it before Apple or Google does.

 

[thilanliyan]

If nV, Google, and ARM are involved in the bidding, how the heck can AMD win with such shallow pockets? Lol.

 

[Arachnotronic]

ARM doesn't have a lot of cash. In fact, AMD has significantly more cash on hand than ARM does.

I still don't get why people think ARM is anywhere near as big as the old-school tech giants. Oh, wait, it's that ridiculously over-inflated market cap XD

 

[thilanliyan]

My fault, I figured ARM has more cash than AMD, although I knew Google and nV have much more than ARM.

 

[soccerballtux]

ARM doesn't have a lot of cash. In fact, AMD has significantly more cash on hand than ARM does.

I still don't get why people think ARM is anywhere near as big as the old-school tech giants. Oh, wait, it's that ridiculously over-inflated market cap XD

they don't need to be big to take profit from Intel.
A15 at 1.5ghz is going to be enough for most people for their facebook browsing and internet surfing and youtube watching.
They can replace their desktop with the phone.

 

[Arzachel]

ARM doesn't have a lot of cash. In fact, AMD has significantly more cash on hand than ARM does.

I still don't get why people think ARM is anywhere near as big as the old-school tech giants. Oh, wait, it's that ridiculously over-inflated market cap XD

That said, ARM has magnitudes lower operational overhead.

 

[nenforcer]

With nVidia already committed to ARM, I don't see them as being a potential suitor.

Why AMD would want to encroach on their X86 APU business I don't know. The SeaMicro acquisition made sense.

I see Qualcomm or Google as a more likely suitor.

 

[nismotigerwvu]

With nVidia already committed to ARM, I don't see them as being a potential suitor.

Why AMD would want to encroach on their X86 APU business I don't know. The SeaMicro acquisition made sense.

I see Qualcomm or Google as a more likely suitor.

Agreed and between those two it seems like Qualcomm would be more logical. On the surface it seems Google would only be interested in the IP whereas Qualcomm would certainly take the engineers in with open arms.

 

[richaron]

Why AMD would want to encroach on their X86 APU business I don't know. The SeaMicro acquisition made sense. I see Qualcomm or Google as a more likely suitor.

Agreed and between those two it seems like Qualcomm would be more logical. On the surface it seems Google would only be interested in the IP whereas Qualcomm would certainly take the engineers in with open arms.

...what? Obviously AMD sees low power/portable as a priority, so how can you pretend buying a company with exceptional low power credentials/patents/engineers doesn't make sense?

 

[Tuna-Fish]

I still don't get why people think ARM is anywhere near as big as the old-school tech giants. Oh, wait, it's that ridiculously over-inflated market cap XD

ARM is mostly owned by it's customers. That's why they can afford to not make all that much money while the market cap stays up. The common stockholders don't exactly want them to raise prices...

Also, that is why they have much more financial weight to throw around than they have on the books. They make a presentation to shareholders outlining how some acquisition is strictly necessary for the continuation of their product lines, and they will be authorized to issue as much more stock as they like.

 

[nismotigerwvu]

...what? Obviously AMD sees low power/portable as a priority, so how can you pretend buying a company with exceptional low power credentials/patents/engineers doesn't make sense?

For the same reasons that the Sprint/Nextel merger tanked really. Sprint viewed the enterprise market as a golden opportunity and wanted to get a string of long term commitments. This vision was undone by the fact they operated on two fundamentally different and incompatible technologies. While they may both be shooting for performance in low power applications, MIPS and AMD are operating in two completely different ways. Not to say that the bright minds at MIPS wouldn't make an impact on making SoCs, X86 is just a completely different animal. Also, I seriously doubt AMD would want to produce anything on the MIPS ISA, much akin to Sprint having no use for an iDen network. While Qualcomm has made a nice living off their in-house ARM platforms, I could see them aggressively seeking an established ISA they could have more of an influence and would LOVE to have the MIPS team on board. So it isn't that AMD wouldn't need/want these guys, they just might not need/want them as badly as some other companies. That said, I feel whoever swoops in on MIPS is going to be extremely happy with their purchase (assuming it happens now while they are still undervalued).

 

[richaron]

For the same reasons that the Sprint/Nextel merger tanked really. Sprint viewed the enterprise market as a golden opportunity and wanted to get a string of long term commitments. This vision was undone by the fact they operated on two fundamentally different and incompatible technologies. While they may both be shooting for performance in low power applications, MIPS and AMD are operating in two completely different ways. Not to say that the bright minds at MIPS wouldn't make an impact on making SoCs, X86 is just a completely different animal. Also, I seriously doubt AMD would want to produce anything on the MIPS ISA, much akin to Sprint having no use for an iDen network. While Qualcomm has made a nice living off their in-house ARM platforms, I could see them aggressively seeking an established ISA they could have more of an influence and would LOVE to have the MIPS team on board. So it isn't that AMD wouldn't need/want these guys, they just might not need/want them as badly as some other companies. That said, I feel whoever swoops in on MIPS is going to be extremely happy with their purchase (assuming it happens now while they are still undervalued).

I'm not american so you're analogy means nothing to me... & I cbf researching sorry. Whilst ~think~ I understand the point you are trying to make, I would say you are vastly overplaying the differences in these technologies. They both need similar logic in many areas, a logical addition is a logical addition not matter what chip it's located.

There is plenty of logic which can be used, many patents which can be applicable, and a lot of engineering experience.

Edit: "...plenty of logic which can be used" not as is; but scaled up, duplicated, etc.

 

[soccerballtux]

With nVidia already committed to ARM, I don't see them as being a potential suitor.

Why AMD would want to encroach on their X86 APU business I don't know. The SeaMicro acquisition made sense.

I see Qualcomm or Google as a more likely suitor.

if amd wants to specialize in making slow processors, , I don't see any reason to buy another company to do so, when they can just continue business as normal with their current processors.

 

[Olikan]

if amd wants to specialize in making slow processors, , I don't see any reason to buy another company to do so, when they can just continue business as normal with their current processors.

ROLF!!!

it's actually a good strategy agains ARM
When they finnaly reach the performance of bobcat, AMD might just need a die shrink :biggrin:

anyway, i doubt than AMD can buy MIPS, they just bought sea micro, and an ARM license will be much cheaper with the same results

 

[sm625]

Well we know AMD believes heterogenous computing is the way of the future, as do I. Now how much would a cpu be worth if it had 2 ARM cores, 2 x86 cores, and 2 MIPS cores, as well as 256 stream processors? What if AMD found a way to simply combine the front ends of each uA so that they share 80% of the same core? I have little faith that AMD could actually do it given past blunders, but it seems possible. They were successful with x86-64. Who is to say they cant come up with an xAMD that could handle ARM, MIPS, GPGPU, and x86 all in the same breath?

 

[cytg111]

http://www.brightsideofnews.com/news/2012/4/16/amd-wants-to-buy-mips2c-but-can-they-beat-google.aspx

Pretty interesting if true...

http://semiaccurate.com/forums/member.php?u=1435

This dude apparantly works with rendering++ .. and he's on record stating that they're testing ARM cores with 'AVX2' like extentions for heavy-compute. (as alternative to gpgpu etc..)

 

[nismotigerwvu]

What if AMD found a way to simply combine the front ends of each uA so that they share 80% of the same core?

Actually someone did something similar to this back in the 90's. I want to say it was the IDT Winchip, but I'm not certain. I'm sure someone will look up which chip it was that I'm think of here, but basically it was a proprietary RISC ISA internally that had a frontend that converted X86 to this (undocumented) internal ISA on the fly. Granted this is how pretty much every X86 CPU from the K6 on has functioned (to a degree) but I remember these fellows boating that a quick firmware flash could add support for new instructions or extended capabilities. Then again we heard the same thing from Rendition on their FPGA based Verite as well.....

 

[Arachnotronic]

http://semiaccurate.com/forums/member.php?u=1435

This dude apparantly works with rendering++ .. and he's on record stating that they're testing ARM cores with 'AVX2' like extentions for heavy-compute. (as alternative to gpgpu etc..)

That guy is a huge, huge ARM fanboy. I very much doubt it. ARM's current FPU is a joke, so it'll be a while before I believe AVX2 type instruction sets.

 

[cytg111]

That guy is a huge, huge ARM fanboy. I very much doubt it. ARM's current FPU is a joke, so it'll be a while before I believe AVX2 type instruction sets.

- But isnt SIMD/AVX type implementations an alternative to a weak FPU? I have never gotten the impression that simd/avx utilizes the x87 FPU in these designs. My mistake?

Anyway and even if, the only point of reference to this information is AMD moving towards APUs, merging CPU and GPU on die. How a super fictional mips-avx2 "CPU" would fit in current AMD APU plans is also pretty much in the dark.

Still thought it was worth a mention .

(maybe it could just be patents ... sounds more likely)

 

[JustMe21]

Transmeta did on the fly software x86 decoding within their processors but they were more power efficient. Perhaps that idea might be approached again. I know MIPS has been around for quite a while but isn't Android geared more towards ARM or rather don't people develop more Android apps for ARM? It almost seems that the ones having compatibility issues are using non ARM processors.