AMD Reportedly Looking At Breaking Itself Up Or Other Options

[phoronix]

Phoronix: AMD Reportedly Looking At Breaking Itself Up Or Other Options

According to an exclusive report by Reuters, AMD is reportedly looking at splitting itself up or spinning off one of its business units...

http://www.phoronix.com/scan.php?page=news_item&px=AMD-Split-Possible-Report

[hajj_3]

damn, i hope they don't. I am just hoping that the zen cpu architecture will be very fast, within 20% ipc of intel instead of 60% ipc worse. I hope they at least wait until zen is out 6 months before deciding to do anything.

[schmidtbag]

I don't think they really need to break off anything. What they need to do is get Zen out the door yesterday and completely change their R9 300 series lineup, for example:
* What the Fury GPUs are now should be the 390 and 390X
* The updated 290 and 290X GPUs should be the 380 and 380X
* The updated 285 should be the 370
* A Tonga GPU with 4GB HBM should be the 370X
* Leave no GCN 1.0 models in any of the R9 and R7 series.

[sirdilznik]

I imagine the GPU part of AMD, formerly ATI, is what's keeping the company afloat. From a purely business standpoint, it might be wise to consider scrapping CPUs altogether and sticking to what's actually making them money. That would, of course, be terrible for consumers as it would leave Intel with essentially a monopoly in a significant part of the market, not that they have much competition from AMD as it is. I pray the successor to the disastrous Bulldozer architecture exceeds expectations and rights the ship.

[duby229]

I imagine the GPU part of AMD, formerly ATI, is what's keeping the company afloat. From a purely business standpoint, it might be wise to consider scrapping CPUs altogether and sticking to what's actually making them money. That would, of course, be terrible for consumers as it would leave Intel with essentially a monopoly in a significant part of the market, not that they have much competition from AMD as it is. I pray the successor to the disastrous Bulldozer architecture exceeds expectations and rights the ship.

That's not possible. The thing that sells GPUs is the platform. It would be the most retarded thing AMD could do to exit x86. AMD needs to buff their CPU. Theres nothing more or less to it.

[eydee]

AMD just needs to focus more on market research and start fulfilling the needs of customers more. They do a lot of things that are completely ignored by most of the customer base and cost a lot. (Think of 3D, Eyefinity etc.) In the meantime, basic stuff like drivers aren't the best, and they affect everyone.

Fix basic stuff first, then go for innovation, but think of something people actually use. Innovation is not a checklist, it needs to be based on what people want. (Even if they don't know it yet.)

[Kano]

HBM will show next week how much speed it can deliver. Nvidia wants to use HBM but not too soon. Broadwell and Skylake use 14 nm, Carrizo still 28 nm. Intel Ivy Bridge used 22 nm already in Q2/12. Globalfoundries was already split off 2009 and seems to lack innovation. If AMD splits up now they do nothing else than many companies:

A) (cross-)license a CPU core from Intel or ARM
B) license a GPU core
C) contract manufacturing at Globalfoundries or TSMC

Basically they could license other GPU cores and even produce at Intel. The main question is if that will lead to more or less innovation... Especially for production improvements you need lots of money. Intel can pay this directly but Globalfoundries itself wants to create revenue. Every step you need external increases total costs and can only be racked up with huge produced (and sold) units. Currently AMD seems to sit on mountains of unsold chips - why would you rename em so often?

[madjr]

AMD just needs to focus more on market research and start fulfilling the needs of customers more. They do a lot of things that are completely ignored by most of the customer base and cost a lot. (Think of 3D, Eyefinity etc.) In the meantime, basic stuff like drivers aren't the best, and they affect everyone.

Fix basic stuff first, then go for innovation, but think of something people actually use. Innovation is not a checklist, it needs to be based on what people want. (Even if they don't know it yet.)

I think that when Vulkan becomes widely used, drivers may not be that big of an issue anymore. That's probably why they gave away their mantle api,

[Kano]

Vulkan is nice but how many expected users they target in the near future? Maybe Sony could use it too for PS 4/5. MS uses DirectX 12...

[gens]

are big companies ever really not split up ?