I am somewhat skeptical.

A series AMD Llano motherboards coming out soon. Good for those who wants a PC without having to fork out extra money for a discrete level GPU.

http://www.ecs.com.tw/ECSWebSite/Media/New...=1408&MenuID=45

The first pictured Llano compatible Socket FM1 motherboard, ECS A75F-A


Asus Sabertooth 990X Socket AM3+ motherboard


This post has been edited by dma0991: May 19 2011, 03:55 PM

AMD Bulldozer and Llano estimated pricing.


Another A75 AMD Llano board out by ASRock. New heatsink bracket for all AMD AM3+/FM1 motherboards and 8 SATA 6GB/s ports natively supported by the AMD chipset.

It is not expensive if you consider the current price of a Core i7 2600K is $315. You cant just do a conversion from USD to MYR and expect that 8 core BD to cost that much. If you look here you can also see that Intel recommends a price of $317 for the Core i7 2600K.

Even if it is superior in terms of performance comparable to SB Extreme not a lot of people is going to pay 3k just for an AMD CPU. That kind of super high end market is just for Intel as only some rich enthusiasts which is <1% of the consumers are willing to pay more than 1k for the best CPU. If they get some performance out of the BD architecture and sell it at a similar price to the regular SB then it is just fine.

Simply put if a company can sell something that is cheap and in large quantities will be more profitable than a company that sells something that is expensive but in small quantities. That is why Toyota is a much bigger company than Ferarri.

I wouldn't say that AMD is crippled because they currently do not have any fab to manufacturer CPU. They are currently relying on Global Foundries to do their CPU unlike Intel which is rich enough to have their own fab. Not many know how good GloFlo is at manufacturing yet so if they are good at it then maybe it might perform better. As for efficiency also depends if the company has a lot of money to do R&D faster than their competitor in this case Intel would lead. As for the manufacturing size there is a limit and some have said that we're approaching that limit really soon and it gets harder to manufacture as it gets smaller due to current leakage.

No they are not. AMD got shares in GloFlo but not even close to 50% majority share so GloFlo is not owned by AMD. The difference is that Intel controls its fab so it can tell it to do whatever they want while GloFlo is independent like TSMC and they can do whatever they want and not be a customer to AMD only.

I don't see how a semiconductor company can lose its way to innovate when AMD designs the architecture and GF makes the wafers. There is no advantage or disadvantage to having a separation of a fab from the company as AMD and Nvidia makes their graphic cards at TSMC and they're doing pretty well.

And you think that AMD is not going for those as well? They are going for that but it is slightly slower due to the fact that they do not have enough money for R&D therefore they would be behind Intel in this. AMD will implement 3D FinFET when they are going for 20nm. HKMG was implemented long ago already and will be a part of every Intel and AMD processors.

Adding all that to their 32nm process along with a new architecture and new manufacturing process will just make yields lower. That is why Intel implements their Tick Tock because they think it is too risky to include die shrink and new architecture together as it will reduce yield. If there is no performance gain from a smaller process then they would usually stick to the cheaper and easier to produce bigger process. Take for example the H67/P67/Z68 chipset which is made on a bigger manufacturing process instead of following SB at 32nm because there is no performance increase for the chipset to be smaller.

Intel is the leader when it comes to smaller manufacturing process in the semiconductor industry. However this trend cannot continue for long as the limit for silicon is near, I won't mention numbers here even though I know because the future is still unknown but we will not be going beyond or near 1nm I can guarantee that. They have to find some other ways to improve other than reduce the manufacturing node.

If BD is released and it doesn't show a strong single threaded performance against SB then it is within my assumption as well because BD is a server based CPU and will be stronger in terms of multithreaded applications. AMD aims to reclaim the server market with BD so I am not surprised at all. But of course I hope there would be some kind of silicon fairy dust they can throw in that makes it an all rounder champion.

The trend would not continue long enough I think. Next year will be 22nm then 14nm and most probably 10nm and 6nm is the last from Intel. What comes after I am not so sure. Even if it requires a high initial investment in R&D in new materials the cost will be passed down to the consumer. One of the reasons why going smaller with silicon is not the best way as it gets more expensive and like I said cost to consumer. The material that replaces silicon should be cheaper or at least comparable in terms of price and manufacturing in order to maintain a price that many consumers can benefit from.

Although it is true that Intel is Chipzilla and they are very good in what they do but it won't help them if the market is trending towards the mobile and ARM is the dominant processor in that segment due to its low power consumption. When it comes to the mobile, less power consumption is more important than more performance. To put into perspective Intel has to put all kinds of special methods like the recent Tri Gate and 22nm or smaller just to compete with the plain 40nm ARM. If Intel doesn't improve Atom soon they will be left behind but since they have tons of cash that may not be true.

As for BD, AMD has their direction and I am sure they know what market they are targeting, I can elaborate but too lazy to type.

Fusion strategy


Llano APU for the mobile segment. 4 cores + 400 SP @ 444MHz with only a 45W TDP.

If they announce it June then most probably it will have retail availability by late July but if they announce it late July that could mean that it can be delayed till September like what was initially rumored. But then AMD denies the September delay launch so I'm not really sure what is going on. Then again it could be that because many know so little about Bulldozer at its possible announcement date (June 7) then all kinds of rumor is happening.

I'll give AMD that if by July they tell that they will announce it by September.

Intel delayed their Ivy Bridge till March-April 2012 so maybe AMD delayed it just because Intel does it too lol.

3D transistor or better put is called FinFET. Tri Gate is just what Intel calls its form of FinFET and Intel did not invent the idea. It is already according to research that it will be implemented around 2x nm for CPU. There are different types of FinFet and Tri Gate is only one type. Since Intel implemented Tri Gate doesn't mean that AMD can't. Same goes for HyperThreading which AMD can put it in Bulldozer and have a similar design like Intel but they decided to go with a different design instead. Congrats to Intel though for being the first to actually have a production ready CPU with FinFET.

Can buy the FX-8130P now for 1998 Yuan. link

Looks like Bulldozer is delayed and production will begin by August and on sale by September only. link


This post has been edited by dma0991: May 31 2011, 05:15 AM

Well it is never a good thing to have something delayed especially something that many are anticipating but some said that it was never a delay as AMD never specified officially when it would be launched. AMD did say 2011 but not an accurate date of release. It's going to hurt AMDs chance and their time frame to sell their CPU till Ivy comes out is cut short by 2-3 months.

Always take leaked benches with a grain of salt. They are most of the time Photoshop-ed images and even I could do it if I want to create FUD. AMD themselves are not silly to actually produce the chips in thousands just to realize that they are not meeting expectation. If that was the case then AMD would just sell the chips with the previous stepping as trashing thousands of sale worthy chips is a waste to AMD. Those stepping are running at low speed and most probably AMD now has the green light at this stepping to mass produce their chips.

I don't see how is that good news when it will only hinder Intel slightly from actually selling their 22nm node with Tri Gate. Intel is not a small corporation like AMD. If Intel wants something and they find that it benefits their pockets in the long run they can just buy out the Korean patent and AMD does not have the capacity to do that. Intel needs it so that they can compete with ARM as Tri Gate will benefit stuff like Atom the most instead of Ivy Bridge.

I am quite sure that AMD will have FinFET as well by the time they reach 20nm. GloFlo is currently R&D for the technology but will not be available till 2013 or later if there are no delays. As an example there is also a different variation of FinFET which is not Tri Gate. So in actual fact there is absolutely no reason why AMD can't have it on their CPU.

AMD will have to sell Bulldozer eventually so September is not that far. It's better that AMD delays a few months to work out the problems rather than having to recall their CPU or having many unsatisfied customers by selling a inferior product. By the time AMD releases their CPU with the latest stepping their 32nm should mature a bit more.

That and the fact that 1 billion in recalls might be a big loss to AMD in money and reputation.

Llano is out officially so its not all that bad actually. If AMD really did not release anything at all during Computex then they deserve an epic facepalm. Would be nice to see some A75 board + Llano based recommendations competing against the Core i3 2100 in the low price range while waiting for Bulldozer to be an alternative to the i5 and i7.

I think that motherboard manufacturers only shows their boards at Computex and not in full production yet. Even the X79 boards are out but SB-E is still not out yet. Most probably SB-E will be out at the same time as BD if it is certain that BD WILL launch at September. Now the problem with that is the competition is going to be hard on AMD at that point in time when SB-E appears.

Even without BD yet I'm sure many others with AM3 processors can drop in theirs into the AM3+ socket of the new board while they wait for BD to come out by September. Let's just hope that Llano is promising enough to cover up for Bulldozer's lateness.

What I mean by full production is that all the manufacturers are at full production capacity to make that particular motherboard. It doesn't happen at launch/announcement and would usually take a month. Also during launch/announcement there is no retail availability of the board, meaning it is not sold at your local shops at time of launch.

A prototype/concept will be stated by the manufacturer themselves. An example is the LGA2011 board, ASUS Danshui Bay.