Richland & Kabini rumours

[NostaSeronx]

I believe and agree with all that, apart from one thing- phones. That's just not happening with Temash.

There are already phones with Temash in them. If they get past the showroom or not to the production line is not up to me.

 

[Nemesis 1]

I would vary much like to see a link .

http://www.computerworld.com/s/article/9235483/AMD_passing_on_phones_focusing_on_tablets_and_hybrids

 

[CHADBOGA]

There are already phones with Temash in them. If they get past the showroom or not to the production line is not up to me.

Phones with 6hr battery life?

 

[NostaSeronx]

Phones with 6hr battery life?

okay...

Hondo is the 'Z-60' <-- The Predecessor
Temash is the 'A6-1450' <-- The Successor

Last time I checked:
A10 -> A8 -> A6 -> A4 -> E2 -> E -> C -> Z
wait... let me make it in a way you'll get it.
Z -> C -> E -> E2 -> A4 -> A6 -> A8 -> A10

The Z-60 isn't a phone APU nor is the A6-1450.

The AMD Z-60 APU comes loaded with features designed to enhance the tablet experience, including AMD Start Now technology that allows fast boot and resume from sleep times, eight hours of battery life while Web browsing, and up to six hours of HD video playback*.

*The AMD Z-60 based reference platform is projected to measure .75 W at idle, 1.1 W during web browsing, 1.6 W during video playback and .02 W during a system S3 &#8220;sleep&#8221; state. Total system power for the reference platform is projected at 2.9 W at idle, 3.9 W during web browsing, 4.8 W during video playback and .08 W during a system S3 &#8220;sleep&#8221; state.

There are C/Z APUs from the Temash line up that are native(not die-harvested) dual-cores built for phones.

 

[itsmydamnation]

Phones with 6hr battery life?

Thats rather meaningless, its very easy to make any phone have only 6 hours of standby with it completely stock. reality is you have no idea just like the rest of us how Temash performs powerwise in idle/power/clock gating senerios.


But i guess you were do nothing other then having a troll, correct?

 

[Homeles]

Thats rather meaningless, its very easy to make any phone have only 6 hours of standby with it completely stock. reality is you have no idea just like the rest of us how Temash performs powerwise in idle/power/clock gating senerios.


But i guess you were do nothing other then having a troll, correct?

http://en.wikipedia.org/wiki/Hyperbole

 

[itsmydamnation]

http://en.wikipedia.org/wiki/Hyperbole

im well aware of the mechanisum im talking about the underlying point. But when your Hyperbole can be perfectly normal operating conditions you really haven't hit the mark.

 

[Homeles]

A 6 hour idle standby on a phone would be awful.

 

[itsmydamnation]

A 6 hour idle standby on a phone would be awful.

And yet could be perfectly normal operations of any phone. The single biggest impact on 3G/LTE phone standby time is Cellular network design.

 

[Homeles]

Any phone that can't last a work day in my pocket, sitting idle, is not a phone worth buying.

 

[itsmydamnation]

Any phone that can't last a work day in my pocket, sitting idle, is not a phone worth buying.

well not many people will disagree with that, but your ignoring the point, it can largely be out of the phones control.

 

[AtenRa]

IF AMD wanted, they could have a dual Core 1GHz + 64 Radeon Cores at almost half the power consumption of Hondo + FCH. I dont believe Intels ATOM Z2460 has less power consumption than that.

 

[piesquared]

When John was an active member here your statement is simply not true. Countless members were sanctioned for their unnecessary personal attacks of John.

To say differently means you are either ignorant of the reality of what was going on at the time or it is just another sleazy attempt by you to impugn the integrity of the moderators of this forum. Given that you have documented history of doing the latter, I am going to be surprised if you claim your err is because of the former

Since John left the forums, as he indicated by virtue of pm'ing me and saying "I'm leaving the forums, delete my account if you wish", that prevents the moderators from penalizing active members for insulting John all they like after the fact. No different than the libelous tripe you come here and spread about Anand Lal Shimpi (who by virtue of not being an active member is himself not covered by the forum guidelines and is fair game to be insulted if members want to).

All that you should be amazed by is the fact that despite your general nasty character and trolling behavior the mods haven't decided to permaban you yet. I chalk this up to the fact that you are so absurdly ridiculous in the vast majority of your posts that the mods have simply concluded you are harmless and your own worst enemy. Why stop the entertainment when you so willingly provide it for free?

WhoTF am I? What I say doesn't carry any weight. How many forum visitors do you have monthly viewing those sensationalist headline thread titles dragging the guys name through the mud? Why not move such off topic controversial posts to the correct forums and out of high traffic areas where they don't belong? That might solve a lot of your problems.

[edit]

Anyway, I didn't mean to make a big issue out of JF..

It's extremely obvious that he didn't agree with it. You have failed to comprehend what he wrote &#8212; try again.

Well, I am Swede if that explains anything. 😛

It's extremely obvious that he didn't agree with it. You have failed to comprehend what he wrote &#8212; try again.

Maybe he does, he just doesn't know it yet. 😉

 

[piesquared]

IF AMD wanted, they could have a dual Core 1GHz + 64 Radeon Cores at almost half the power consumption of Hondo + FCH. I dont believe Intels ATOM Z2460 has less power consumption than that.

This is the slide i'm most interested in. 🙂

perf/watt goes up 2x in some cases. And Hondo is no slouch.

 

[pablo87]

I still dont quite understand - or accept - why AMD can be number 2 in x86 and number 2 in GPU,- the only player that have good technology in both areas, and still not make a profit. And dont do it for so many years. So many excellent people, excellent technology in relevant business. We know the reasons, but its still kind of weird.

So from an economic perspective its just plain bad. But compared to the AMD of the 80 and 90ties its just another world. They were copying Intel, now its both ways. AMD have had en enormous positive impact on the innovation and development of the industry, and benefit for the consumers. Everyone who had to pay for their own pc in the 80ties knows from hard earned experience what difference a strom AMD have meant.

Good question to which I would give two answers.

First, profit and good business aren't synonymous. There are some good businesses that are unprofitable and there are some really bad businesses that are very profitable. There are also - and this is very common - good businesses that are profitable that are taken down by other divisions that are losing money, or saddled by some structural problem - too much debt, defined benefit pension shortfall, bad contract, environmental liability, etc... In the US, they have a very good mechanism for shedding these "burdens" - its called Chapter 11.

Western Digital went through that - they were making hard drive controllers (had their own fab in Irvine, Ca. and factory in Puerto Rico) then once IDE (Integrated Drive Electronics) hard drives (that integrated the controllers onto the HDD PCB) came out, that business was gone. WD started losing a lot of money, filed for Chapter 11, bought Tandon's fledgling hard drive business and the rest is history.

AMD is a classic case for Chapter 11, has been for years. Why not?

That would be answered in the second point. But I can't write about it.

 

[mrmt]

I still dont quite understand - or accept - why AMD can be number 2 in x86 and number 2 in GPU,- the only player that have good technology in both areas, and still not make a profit. And dont do it for so many years. So many excellent people, excellent technology in relevant business. We know the reasons, but its still kind of weird.

Management can explain a lot here. Management decided to go with Bulldozer, an architecture that placed AMD in a very bad position, Management gave a go for APU, a concept that simply did not pan out for AMD, Management decided to adopt a "develop and let them come" approach for professional market, effectively killing whatever chances they had to grab a share of the profitable Nvidia PSB business, Management decided to sell their ARM division, Management decided to delay bobcat shrink...

While AMD is resource starved in some levels, there are an awful lot of things that management could do in order to make better use of the existing resources, and no matter how good your engineering dept is and how much resources you have, there is no point on them if you don't have strong management giving out sound guidance on product conception, allocating resources and managing time frames.

When a management team can't get right the next quarter' sales forecast, a very basic management task, do you really think they have the capacity establish sound design goals and allocate adequate resources for an architecture that will debut in a 3-4 years time frame?

 

[itsmydamnation]

How did the bulldozer architecture place AMD in a bad place? The execution and implementation certainly did but i have seen very little in the way of justification of "bulldozer architecture" is bad that appears so prevalent. I guess people just choose not to differentiate?

 

[mrmt]

How did the bulldozer architecture place AMD in a bad place? The execution and implementation certainly did but i have seen very little in the way of justification of "bulldozer architecture" is bad that appears so prevalent. I guess people just choose not to differentiate?

AMD share in servers is almost 0, AMD share in desktops is shrinking and AMD share in desktop is only doing more or less fine because of bobcat. This is Bulldozer legacy to AMD.

Regardless of what you think, Bulldozer is a failure from a commercial standpoint, and commercial standpoint is the ultimate measure to judge success or failure of a given product. Commercial victory is the only victory that matters.

 

[ShintaiDK]

How did the bulldozer architecture place AMD in a bad place? The execution and implementation certainly did but i have seen very little in the way of justification of "bulldozer architecture" is bad that appears so prevalent. I guess people just choose not to differentiate?

I think sales says it all. Utter disaster.

And its a CPU mainly designed for server/workstation workloads, something AMD is nonexistant in today.

Intel sells more 3960X than AMD sells "8 core" CPUs.

 

[Idontcare]

How did the bulldozer architecture place AMD in a bad place? The execution and implementation certainly did but i have seen very little in the way of justification of "bulldozer architecture" is bad that appears so prevalent. I guess people just choose not to differentiate?

Compared to their other microarchitecture, present in Llano, AMD found the one way to create a microarchitecture that had all of the negatives that one must accept in going with a CMT-based microarchitecture while simultaneously managing to capture none of the benefits of going with a CMT-based microarchitecture.

Bulldozer really was a tour de farce. Higher power, lower performance, bigger die...they literally managed to give up what little they had going for them with their Stars core and thuban-styled product lineup.

An 8-core Llano (sans the iGPU) would have had the same die size as the 8-core bulldozer but much higher performance...and would have cost them almost nothing to develop given that they had already put in the time and money to develop the quad-core shrunk logic to make Llano in the first place...so they invested their limited resources into something that only served to further limit their ROI and cashflow needed to fund their next-gen (post bulldozer) projects.

Bulldozer put AMD in a bad place, their marketshare reflects this reality, but even worse is that bulldozer is going to keep AMD from ever getting out of their bad place thanks to AMD now being in crisis mode (layoffs, project cancellations, etc) because bulldozer can't pull its weight and bring in the cashflow needed for AMD to move past bulldozer.

An 8-core llano w/o iGPU would not have been the ideal product, but it for sure would have done better in servers and on the desktop to keep sales going and cashflow going while AMD funded the development of something that made sense for their revenue model.

Bulldozer makes no sense for AMD, it was a pipedream and it went up in smoke.

 

[nehalem256]

Management can explain a lot here. Management decided to go with Bulldozer, an architecture that placed AMD in a very bad position, Management gave a go for APU, a concept that simply did not pan out for AMD, Management decided to adopt a "develop and let them come" approach for professional market, effectively killing whatever chances they had to grab a share of the profitable Nvidia PSB business, Management decided to sell their ARM division, Management decided to delay bobcat shrink...

Management didn't decide this last one. It got cancelled because GF botched the 28nm process and there wasn't time to profitably port it to TSMC.

 

[ShintaiDK]

Management didn't decide this last one. It got cancelled because GF botched the 28nm process and there wasn't time to profitably port it to TSMC.

Aint that abit of a rumour story? Bobcat was already a TSMC product. Why move it to GloFo?

 

[NTMBK]

Compared to their other microarchitecture, present in Llano, AMD found the one way to create a microarchitecture that had all of the negatives that one must accept in going with a CMT-based microarchitecture while simultaneously managing to capture none of the benefits of going with a CMT-based microarchitecture.

Bulldozer really was a tour de farce. Higher power, lower performance, bigger die...they literally managed to give up what little they had going for them with their Stars core and thuban-styled product lineup.

An 8-core Llano (sans the iGPU) would have had the same die size as the 8-core bulldozer but much higher performance...and would have cost them almost nothing to develop given that they had already put in the time and money to develop the quad-core shrunk logic to make Llano in the first place...so they invested their limited resources into something that only served to further limit their ROI and cashflow needed to fund their next-gen (post bulldozer) projects.

Bulldozer put AMD in a bad place, their marketshare reflects this reality, but even worse is that bulldozer is going to keep AMD from ever getting out of their bad place thanks to AMD now being in crisis mode (layoffs, project cancellations, etc) because bulldozer can't pull its weight and bring in the cashflow needed for AMD to move past bulldozer.

An 8-core llano w/o iGPU would not have been the ideal product, but it for sure would have done better in servers and on the desktop to keep sales going and cashflow going while AMD funded the development of something that made sense for their revenue model.

Bulldozer makes no sense for AMD, it was a pipedream and it went up in smoke.

A good indication of how thoroughly botched Bulldozer was is how much improvement AMD managed to get out of Piledriver. There wasn't any major changing of the architecture and no die-shrink, but they managed to seriously improve single threaded performance over Bulldozer at the same clock speed:

And those tests are comparing with Trinity, which has no L3$ to help it out.

If they managed to improve the core so much in less than a year, without a major redesign, how the heck did those changes not get made in the original Bulldozer core? Piledriver was improved enough to be an improvement over Llano in all but the worst case scenario (multithreaded FP heavy), whereas Llano pretty clearly beat Bulldozer as you rightly point out.

 

[NTMBK]

Aint that abit of a rumour story? Bobcat was already a TSMC product. Why move it to GloFo?

The WSA, probably. If they can get their most popular line onto GloFo it covers a lot of wafers. Besides, if you're porting from one process to another anyway, moving it to a different foundry shouldn't be much worse- especially for a design which was made to be portable, like Bobcat was.

 

[mrmt]

Management didn't decide this last one. It got cancelled because GF botched the 28nm process and there wasn't time to profitably port it to TSMC.

Ok, they didn't decide to delay, but to bring Brazos shrink to GLF was yet another management decision that backfired.