Richland & Kabini rumours

[Torn Mind]

AMD's marketing department wasn't just a few guys on forums. They were being very well-paid to spend time on forums and do their marketing there. That said, trying to pimp up one company's ethicalness over another is silly because nearly every corporation is a hypocrite. There's no point because there's always going to be some crime.

However, what I cannot figure out is how was AMD unaware of their own processor's failings. Surely they could have put together a rig and benchmarked it?

 

[Olikan]

It turns out that at the same clocks and core-count, Temash will offer NO power-draw advantage whatsoever, over Bobcat! Unless AMD itself posted invalid info.

not surprising...

jaguar have more IPC, uses less manufacturing specific tools...

...both uses kinda low clocks, making dinamic power use "less" important... and static power comsuption increases at node shrinks

 

[AtenRa]

Just to prove the point about too high expectations...

I personally hoped that, Temash will offer at least SOME power savings compared to Bobcat.

Seems like a very modest hope right? After all, it's a SOC, built on a smaller process and is heavily optimized to power gate better. We also all remember the Hot Chips talk about faster CC6 and C6 times and about L2 running on half the frequency and being able to be power gated separately, etc, etc ... We also had articles like these:

http://www.eetimes.com/design/eda-d...rocessor-core-with-RTL-clock-gating-analysis-


It turns out that at the same clocks and core-count, Temash will offer NO power-draw advantage whatsoever, over Bobcat! Unless AMD itself posted invalid info.

Here is a slide, from the Bobcat based Z-60 we probably remember:

And here is the information about Temash:

http://www.amd.com/us/products/notebook/tablets/Pages/tablets.aspx#4


So if it's a typo on AMD's part, it's quite a typo (they specify "Larne" platform 2 times as well as the A4 branding and GPU) and they'd better fix it fast.

So even with really modest expectations you can get burned

You get 25%+ more CPU performance (15% IPC + 10%+ higher frequency) PLUS double the iGPU performance for 20% lower power.

I believe thats great

 

[vampirr]

I bought an Athlon in the same time frame. No body forced me to buy a pentium.

I get it, but I wasnt buying the PC... My dad this with help of his friend, so we probavly get fucked up. Audio editing worked better on AMD than on Intel in 2004

I had, at the time with the complete PC build with monitor, keyboard, mouse, speakers and Windows XP Professional it costed in my country almost 2500$ but since my dad had a connection/friend he got it for 2000$ so it was kinda of a sweet deal even thought I would get more performance with AMD in 2004:
Pentium 4 2.4 or 2.7Ghz
768mb DDR2 RAM
ATI Radeon 9800 Pro 128mb

 

[AtenRa]

However, what I cannot figure out is how was AMD unaware of their own processor's failings. Surely they could have put together a rig and benchmarked it?

http://atenra. com/2011/10/12/amd-fx-processor-launch/

-Aten-Ra: What IPC improvements will we see against Deneb or Thuban in single Thread and Multi thread applications?

-AMD : Bulldozer is a new and radically different core architecture making it impossible to draw IPC parallels to any previous design. With the design of the AMD FX, we focused on offering the highest possible instructions per watt. AMD FX is designed with current and future workloads in mind. At the same time, we have added the new instruction sets and as improvements are realized in applications and operating systems (like the Windows 8 scheduler improvements), we believe performance will increase.

Note: In older apps, either compiled with an older compiler, or those using older instructions will see FX at a disadvantage. Again this was a design decision. Design for today’s apps and the future was the engineering mantra.

We beginning to witness this with new games and applications.

 

[monstercameron]

^when you say software is being designed for newer hardware, you mean better compilers, use of hardware accel or better design?

 

[inf64]

http://atenra. com/2011/10/12/amd-fx-processor-launch/


We beginning to witness this with new games and applications.

Don't forget that a *new* Bulldozer core is just around the corner . They didn't just sit and wait for software to catch up (it did catch up,as we can see PD does great in modern workloads,games included) ,they worked on hardware part too. I expect that SR will fix the shortcomings of BD in legacy software and we will finally have a core that is well rounded and well suited for wide range of workloads(ST/MT/mixed).

 

[grimpr]

Don't forget that a *new* Bulldozer core is just around the corner . They didn't just sit and wait for software to catch up (it did catch up,as we can see PD does great in modern workloads,games included) ,they worked on hardware part too. I expect that SR will fix the shortcomings of BD in legacy software and we will finally have a core that is well rounded and well suited for wide range of workloads(ST/MT/mixed).

Fingers crossed they release an 8core SR FX for AM3+...

 

[Torn Mind]

^when you say software is being designed for newer hardware, you mean better compilers, use of hardware accel or better design?

It's pretty much all about better core-scaling and perhaps including some instruction sets. But since the so-called cores on the FX-8350 are slower than the Intel cores, hence, the FX-8350's lead in current practically perfectly multithreaded apps is only about 1.2 times what the i5-3570K provides.

Or in other words, in general, everything is in the same ballpark. For specific apps, Intel does better in some and AMD does better in others.

 

[Gideon]

You get 25%+ more CPU performance (15% IPC + 10%+ higher frequency) PLUS double the iGPU performance for 20% lower power.

I believe thats great

Yeah, your statements are actually about what I was hoping for and that would have been very good. But did you even read my post ?

A4-1200 is a Dual Core APU @ 1.0GHz, therefore no gain from additional frequency. More agressive trubo might help, but given that AMD has said absolutely nothing about Turbo, I'd be ultra careful about that.

More importantly, power draw is NOT 20% lower. Even if we round all Bobcat number up, we get 12.5% less in S3, 3,45% in Idle, 5,13% while Web Browsing. During video playback, Temash uses 10,42% MORE power!

 

[monstercameron]

Yeah, your statements are actually about what I was hoping for and that would have been very good. But did you even read my post ?

A4-1200 is a Dual Core APU @ 1.0GHz, therefore no gain from additional frequency. More agressive trubo might help, but given that AMD has said absolutely nothing about Turbo, I'd be ultra careful about that.

More importantly, power draw is NOT 20% lower. Even if we round all Bobcat number up, we get 12.5% less in S3, 3,45% in Idle, 5,13% while Web Browsing. During video playback, Temash uses 10,42% MORE power!

is it that temash uses more power or that the tested platform(larne) number are a bit higher?

 

[AtenRa]

Yeah, your statements are actually about what I was hoping for and that would have been very good. But did you even read my post ?

A4-1200 is a Dual Core APU @ 1.0GHz, therefore no gain from additional frequency. More agressive trubo might help, but given that AMD has said absolutely nothing about Turbo, I'd be ultra careful about that.

More importantly, power draw is NOT 20% lower. Even if we round all Bobcat number up, we get 12.5% less in S3, 3,45% in Idle, 5,13% while Web Browsing. During video playback, Temash uses 10,42% MORE power!

You are right the A4-1200 is at 1GHz, but those numbers are for the Brazos 2.0 Platform (Z60) and the Temash (Larne) Platform and not for the SoC alone. Larne Platform has a 13.3" display when Brazos 2.0 has 11.6".
We will have to wait a little bit longer to know the actual numbers.

 

[vampirr]

http://www.fudzilla.com/home/item/30711-three-new-kabini-parts-leaked
http://www.fudzilla.com/home/item/30710-amd-temash-tablet-looks-good
http://www.fudzilla.com/home/item/30708-temash-chip-pictured

http://www.fudzilla.com/home/item/30709-amd-richland-notebooks-to-have-face-unlock
http://www.fudzilla.com/home/item/30662-amd-to-launch-richland-on-12th-of-march

http://www.fudzilla.com/home/item/30572-amds-john-taylor-talks-about-ps4-apu

 

[Phynaz]

You don't need to spam the Fudzilla links. We kinda know about that website here. And there's a reason why we generally just point and laugh at it.

 

[pablo87]

Well obviously the standard work is done this time. After all AMD is growing up and starting to realize its business is not designing and producing APUs but providing profit for the OEM /Sony. I doubt the cpu designer, Dirk ever understanded - to the bone - the business he was in.

If Intel succeded this time in keeping kabini out of the 13.3 and sub notebook size with its kickback, bonus, and half leagal grease, RR needs to have implanted new balls. If he can not prevent it, with his history, he is absolutely useless. We all know he doesnt really know what a transistor is, but who cares, but he have to deliver right here. Right now. This should be his prime time.

Who cant sell a full windows experience, in a fanless tablet, less than 10mm?
http://www.youtube.com/watch?v=8N7LNHRl2jE

The problem is, AMD is always driving for show and not putting for $$$. They do 3 tablet windows8 reference designs, but no Android? And where are the 11, 12, 13 notebooks? Also, they are good in doing what OEM'S want but cannot come up with a final product idea of their own (to take to an exclusive customer).

For example, they could come up with a $100 ready to run (!!), kabini based, ssd based, 1lb, chrome or Linux PC reference design, why don't they? Because Intel hasn't done it, and OEMs haven't asked..

Remember that RR is a salesman...

 

[Vesku]

pablo87, AMD is crippled by it's failure to make friends and influence people among the major computer OEMs. With Llano and Trinity they demoed decent reference notebooks. They don't have the clout to make anyone actually produce those models. Intel has the influence and money to do so "hello ultrabook program". AMD, not so much.

http://www.anandtech.com/show/4444/amd-llano-notebook-review-a-series-fusion-apu-a8-3500m/6

http://www.legitreviews.com/article/1931/3/

Note for Trinity they used a debranded 14 inch Dell platform. Dell doesn't even offer AMD notebook options anymore as far as I can tell using their website.

Lenovo, the company the current CEO came from offers very few AMD options and only in 11.6" or 15.6" screen sizes. Their selection of AMD options has shrunk from when I last was looking at Lenovo (1 year or so ago).

 

[pablo87]

pablo87, AMD is crippled by it's failure to make friends and influence people among the major computer OEMs. With Llano and Trinity they demoed decent reference notebooks. They don't have the clout to make anyone actually produce those models. Intel has the influence and money to do so "hello ultrabook program". AMD, not so much.

While I agree this is the case, it doesn't mean they should throw their arms up in the air and give up. They're #2, they have to try harder as Townsend said. To me they have to look beyond the traditional channels (Intel is not federal reserve - they can't print money and give everyone who might one day become customer), regardless how small initially (see problem with RR's philosophy? Typical IBM sales guy).

So go ahead open 1x 12" tooling with Compal or Wistron, be innovative (SSD only, external optical/HDD, better screen) and look for new customers with production/logistics/service ready solutions to overcome objections.

Truth is, a big part of why they don't do this is they are beholden to a few OEM's and fear retribution: "oh you're only doing Intel in this form factor? That's fine, we understand, let us know if you change your mind, we have a good product you know (voice trailing as RR and team leave OEM premises and door closes)"

Blaming it on Intel is a copout. and, as you've correctly identified, is a slippery slope.

 

[AnandThenMan]

Just throwing this out there, maybe AMD should come out with their own ATI branded tablets and notebooks. I like the ATI logo I think it looks really good, plus it may be the only way they can get around the Intel grip on OEMs. Also ATI has at least some brand recognition.

I know it has a lot of logistical problems, but I think it could work.

 

[Vesku]

While I agree this is the case, it doesn't mean they should throw their arms up in the air and give up. They're #2, they have to try harder as Townsend said. To me they have to look beyond the traditional channels (Intel is not federal reserve - they can't print money and give everyone who might one day become customer), regardless how small initially (see problem with RR's philosophy? Typical IBM sales guy).

So go ahead open 1x 12" tooling with Compal or Wistron, be innovative (SSD only, external optical/HDD, better screen) and look for new customers with production/logistics/service ready solutions to overcome objections.

Truth is, a big part of why they don't do this is they are beholden to a few OEM's and fear retribution: "oh you're only doing Intel in this form factor? That's fine, we understand, let us know if you change your mind, we have a good product you know (voice trailing as RR and team leave OEM premises and door closes)"

Blaming it on Intel is a copout. and, as you've correctly identified, is a slippery slope.

Where is the "blame" being heaped on Intel? I'm simply pointing out AMD has neither money nor influence with which to push OEMs. AMD is in a precarious financial position meaning even less leeway in bypassing traditional channels. I generally agree with you, ever since Llano I've been of the opinion AMD should take a more independent strategy with smaller partners as a way to overcome their traditional place in the x86 market. However, the window of opportunity closed while Dirk was still in charge and the AMD board chose as his replacement someone squarely focused in the traditional x86 market.

 

[Phynaz]

Intel is not federal reserve - they can't print money and give everyone who might one day become customer

I'll just leave this here

http://www.intelcapital.com/

 

[Phynaz]

I know it has a lot of logistical problems

You said it. With the execution problems AMD has you really don't want them distracted running an entirely different operation. There's really very little synergy when you think about microprocessor design and CE manufacturing.

And you want to go up against Apple? Jeez man, it would put them in the exact same position they are with Intel right now!

 

[AnandThenMan]

I'm not saying my suggestion is perfect. ^_^ But still if I was AMD I would try it on a small scale at first, heck they have to try something. They would not really be competing with Apple, I don't know many people that look at the iPad and then think hmmm maybe I should buy a Windows tablet. People that want an iPad buy an iPad because they think they need an iPad and that's it. At least that's what I've seen.

On that note, why doesn't Apple just buy AMD? For pocket change they would get some pretty good tech for their products. Maybe Apple can't afford to anger Intel....

 

[Vesku]

Main reason Apple wouldn't want to buy AMD is x86 encumbrances. No telling what they'd actually get the legal right to use once the lawyers were done. Instead they've been pursuing a strong in house development model with a few twists and turns (Samsung -> TSMC).

 

[krumme]

As we can see from the bobcat apearance on laptops Intel will try the folowing by kickbacks, offers, scaring/motivating ...:

Dont let kabini into anything smaller than an 15.6
Dont let kabini use its faster models

For AMD the response this time should be very simple. Only two variants of kabini, no scaled down versions:

Quad core 15w
Quad core 25w

Simple as that. Take it or leave it. For the few kabinies that down meet the Ghz, burn them. Better than burn your brand, even though there is hardly any left. Notebookcheck will fx. take whatever low performance variant comes out and compare it to an high-perf ulv IB straigth away. And we will see tons of stupid comparison also here on anandtech - can you say single threaded performance ? - compared to the core line, and low end discrete Nivida/AMD cards - even though its completely different markets. Thats the name of the game when you are the small player, and therefore, you have to be more selective what you put on the market. AMD havnt been that.

Use the same tactics for Temash. No discount, only high performance/efficiency parts. For the embedded market - tv streamer and so on - its a different story. But they need to man up here, otherwise they will automaticly be positioned by Intel marketing/sale to the lower side. The strategy here, i think is in line with Pablos more push approach.

 

[mrmt]

As we can see from the bobcat apearance on laptops Intel will try the folowing by kickbacks, offers, scaring/motivating ...:

Dont let kabini into anything smaller than an 15.6
Dont let kabini use its faster models

Sorry Krumme, I know it sounds easier to blame Intel for all AMD's woes but in this case wouldn't the most plausible explanation for not finding Brazos in models smaller than 15.6 be AMD's own restriction?

Llano and Trinity weren't exactly stars in sales performance. By putting Brazos in 13/14 notebooks it would cannibalize not only Intel sales, but also AMD sales in only of the few segments where it could get descent margins with Piledriver chips.

Remember that until they are manufacturing Jaguar or successors in GLF, they must sell at least 265 million (COGS) in Piledriver chips per quarter.