AMD mulling break, spinoff

[Abwx]

AMD also denied it would sell its fabs.

http://www.eweek.com/c/a/Desktops-and-Notebooks/AMD-Denies-Fab-Sell-Off

Of course you have to rely on straws, guess that anything is good to spread the usual fud.

Next time i ll remind you this exemple if ever you say that someone denied something, i ll use a previous exemple and you ll have to abide by your own non sense...

That said article from Computerbase.de has been updated..

 

[JDG1980]

A neat idea would be for them to spin off their utterly pointless board of misdirectors.

They could throw in their marketing department while they're at it.

 

[ShintaiDK]

Of course you have to rely on straws, guess that anything is good to spread the usual fud.

Next time i ll remind you this exemple if ever you say that someone denied something, i ll use a previous exemple and you ll have to abide by your own non sense...

That said article from Computerbase.de has been updated..

The point is AMD have to deny it if they speak about it, truth or not.

 

[myocardia]

They probably could bankrupt an oil well in Saudi Arabia if they tried hard enough.

Mubadala is in the UAE, not Saudi Arabia.

 

[sm625]

I still say buying ATi was the biggest mistake in their history (they could have bought S3 or PowerVR for pocket change instead) -- the fiasco like Bulldozer pales by comparison.

If they could have matched intel on CPU performance, then buying ATI would have been a smart move. They obviously assumed they could do that at the time of acquisition. Their executives total ignored Conroe while making their deal with ATI.

 

[Genx87]

A spinoff is an interesting idea. But I am wondering if AMD has the capital to do it at this point.

 

[mrmt]

A spinoff is an interesting idea. But I am wondering if AMD has the capital to do it at this point.

Capital would be found if the deal was a good thing, but the real issue is that AMD doesn't have one, let alone two viable businesses needed to sustain the spin off.

 

[DrMrLordX]

For a chip (4C/8T Zen) that is meant to be a high end replacement for Cat Core SoCs, 24mm2 on 14nm is a lot of space.

Quad core Zen APUs are not meant to replace Puma. Basilisk will be the replacement.

 

[MiddleOfTheRoad]

If they could have matched intel on CPU performance, then buying ATI would have been a smart move. They obviously assumed they could do that at the time of acquisition. Their executives total ignored Conroe while making their deal with ATI.

Even if they had matched Intel CPU performance -- they still overpayed for ATi by a huge margin. That high debt level really did negatively impact future CPU development -- becoming a vicious cycle.

 

[looncraz]

AMD basically got a shot on the high end, low volume, high priced desktop SKUs and lost the entire high volume desktop *and* notebook SKUs. Unless HBM can make wonders for GPGPU, they will lose even more market share this year and revenue.

HBM is so far proving to be a foolish venture, as the technology is not ready to go mainstream and is confined to the very high end.

I think it is waay too early to be making any judgement as to HBM. We already see it transforming the R9 Fury in terms of physical size and power efficiency.

Not only that, but a single 1GB HBM1 chip with a 1024-bit bus on an APU would provide 128GB/s memory for the graphics portion of the chip, which could make for some very significant improvements in performance.

Personally, though, I want APUs with HBM as system AND graphics RAM. I think this would be a killer on laptops. Install 4/8GB of HBM shared on the interposer with an APU in the middle and call it a day.

 

[cbn]

Quad core Zen APUs are not meant to replace Puma. Basilisk will be the replacement.

The only place I saw Basilisk was from this leaked Roadmap:

AMD has since denied this Roadmap and even stated that there will be no Zen APUs till 2017.

 

[DrMrLordX]

AMD said what they said because the roadmap had Zen-based APUs launching in 2016. Basilisk won't show up until 2017.

Personally, though, I want APUs with HBM as system AND graphics RAM. I think this would be a killer on laptops. Install 4/8GB of HBM shared on the interposer with an APU in the middle and call it a day.

That's the only configuration that makes any sense (L3/L4 cache) unless AMD is going to use HBM for those APUs as a replacement for system memory.

 

[cbn]

AMD said what they said because the roadmap had Zen-based APUs launching in 2016. Basilisk won't show up until 2017.

We have no knowledge Basilisk even exists.

 

[DrMrLordX]

http://forums.anandtech.com/showpost.php?p=35853432&postcount=19

It's been around for awhile. One of those names that AMD keeps recycling . . . now it is apparently applied to a 2C Zen APU for 2017. All those products on the "fake" roadmap were real. They just got bumped up by 2-3 quarters, which is why AMD denied its authenticity. The timing was off, not the product lineup.

I repeat, AMD will not be pushing a 4C Zen APU in low-power devices. That is not their Puma replacement.

 

[cbn]

http://forums.anandtech.com/showpost.php?p=35853432&postcount=19

It's been around for awhile. One of those names that AMD keeps recycling . . . now it is apparently applied to a 2C Zen APU for 2017. All those products on the "fake" roadmap were real. They just got bumped up by 2-3 quarters, which is why AMD denied its authenticity. The timing was off, not the product lineup.

I repeat, AMD will not be pushing a 4C Zen APU in low-power devices. That is not their Puma replacement.

Here are the specs I wanted for a 4C/8T Zen with small iGPU:

http://forums.anandtech.com/showpost.php?p=37493921&postcount=46

For specs on a quad core Zen APU "Mini" here is what I would want to see:

TDP: 15W (2C/4T) to 65W
CPU cores: Zen in 4C/8T, 4C/4T, 2C/4T configurations
L2 cache: 1 MB per core (later versions could have 2MB per core)
L3 cache; None
Memory controller: dual channel DDR4 (However, I think most OEMs will probably use with one 8GB DDR4 3200 stick at the high end and one 4GB DDR4 2400 stick at the low end)
iGPU: 128sp (later versions might have 256sp)
8 to 16 PCIe lanes
two SATA 6 Gbps (integrated, so chip is a SoC)

Basically a replacement for cat core SoCs, but instead of adding more iGPU to the die (via increased xtor budget from die shrink) the CPU part gets beefed up instead (quad Cat core --> quad Zen Core)

So a replacement for Cat Cores (but only the higher powered ones). And yes, the power does extend past the where the Cat cores left off.

 

[ShintaiDK]

1M2T Carrizo-L parts are taking over. I think its quite safe to say there is no cat cores and wont be again.

 

[VirtualLarry]

1M2T Carrizo-L parts are taking over. I think its quite safe to say there is no cat cores and wont be again.

I thought Carrizo-L was a cat core? It's not Carrizo.

 

[ShintaiDK]

I thought Carrizo-L was a cat core? It's not Carrizo.

Sorry yes, its still Puma+.

The next one however isnt. (Unless this slide is fake)

 

[DrMrLordX]

Correct. Stoney Ridge looks like it's bumping off Carrizo-L which is just fine by me. AMD has done some exceptional work repurposing Excavator for low-power devices. A custom 1M Excavator die for low-power is just a good idea, period, especially considering how stagnant are the cat cores. It will be fairly short-lived, though, since Basilisk will knock it off (eventually), which will be another step forward for AMD.

 

[el etro]

The only place I saw Basilisk was from this leaked Roadmap:

AMD has since denied this Roadmap and even stated that there will be no Zen APUs till 2017.

Whoa, Nolan and Basilisk were rumors circulating have some time. IIRC Nolan was a 20nm shrink of Carrizo and Bazilisk was other core i don't remember.

 

[soccerballtux]

they're already broke, why are they mulling a break?

 

[cbn]

If Stoney Ridge exists, how much compatibility will it have with Bristol Ridge for laptops?

According to that chart, it will share the same socket (FP4 BGA), but I wonder where the similarity will end?

Lower amount of PCIe lanes, single channel memory controller? Obviously the TDP is lower on Stoney Ridge, but there is some overlap with Bristol Ridge.

 

[DrMrLordX]

Whoa, Nolan and Basilisk were rumors circulating have some time. IIRC Nolan was a 20nm shrink of Carrizo and Bazilisk was other core i don't remember.

Nolan was meant to be the "true" Puma+ (Puma++?), which would have been an incremental upgrade to the Puma/Puma+ cores that we have in Mullins and Carrizo-L. It was not Excavator-based. But yeah, it was supposed to be 20nm. AMD's 20nm plans got scrapped, so Nolan is either a). dead or b). destined to be a Carrizo-L variant on the low end of the TDP spectrum (still 28nm).

If Stoney Ridge exists, how much compatibility will it have with Bristol Ridge for laptops?

According to that chart, it will share the same socket (FP4 BGA), but I wonder where the similarity will end?

Lower amount of PCIe lanes, single channel memory controller? Obviously the TDP is lower on Stoney Ridge, but there is some overlap with Bristol Ridge.

Remember BGA = soldered onto the board. The board will be designed from the ground-up to accomodate the capabilities of the CPU's SoC-like features. Technically they'll be pin-compatible, but you won't see Stoney Ridge offered on the same boards as Bristol Ridge (in all probability). You also probably won't see Stoney Ridge in AM4, while you will see AM4 Bristol Ridge.

 

[mrmt]

they're already broke, why are they mulling a break?

In order to save at least some parts of the company.

Btw, the company is too tight-lipped since mid-quarter. I think the numbers coming in three or four weeks won't be pretty.

 

[dark zero]

After the Fury dissaster, definately AMD are in real danger. Seems that their end is near, along the x86 we know... What a dissaster... And the next is nVIDIA.