AMD to Close Geode Facility, Hints at Embedded Opterons

According to several reports, AMD announced that it will be closing down its embedded microprocessor design facility in Longmont, Colorado. Back in 2003, AMD purchased the facility from National Semiconductor that was working on a system-on-chip processor families. These original Geode embedded processors were x86 compatible processors that were used in devices such as handhelds, competing with Intel’s line of ARM architecture XScale processors.

AMD will be cutting roughly 75 jobs from the Longmont location, while another 75 jobs from the same location will be relocated to Ft. Collins, Colorado. According to reports, AMD plans the Ft. Collins location close to where HP has its location.

Closing of the Geode facility marks AMD’s second closure of embedded microprocessors as the company previously announced that it sold the Alchemy division to Raza Microelectronics. Interestingly, AMD did not comment on whether or not it would outright sell its Geode division. The Geode processors were known to be more powerful than the Alchemy processors, although they are also generally more expensive. In fact, AMD’s Geode NX processor was based on a mobile Athlon processor running at 667MHz to 1.4GHz. According to Erick Salo, AMD’s director of marketing for embedded processors, closing the Geode facility was "a design focus decision, rather than a decision about product lines."

Intel this year also sold its XScale division to Marvell for $ 600 million. The transition Intel said, would take roughly four months to complete. AMD’s announcement of the Alchemy line separation came around roughly the same time as Intel’s announcement, and the new details about its Geode line seems to confirm AMD’s change of focus. Salo indicated that customers showed interest in embedded processors derived from its highly successful Opteron processors. "There is a lot of opportunity there," Salo said.


AMD to bring 6-core ‘Thuban’ Processor To Consumers

Looks like AMD is going all out in the War of Cores, while everyonemight be worried about the platform compatibility. No worries – it is backward compatible with AM3 and AM2+ boards

According to source, AMD’s chip is codenamed Thuban and will feature all six cores on single45nm die. Also, it will feature an integrated DDR3 controller. The chip will most probably have 3MB of L2 and 6MB of L3 cache.

No specific branding or clock speed was revealed by AMD to the source, until then, you may call it a “Phenom X6”. However, Thuban is derivative of the six-core Opteron chip which made its earlier this summer.

Getting excited ? Thuban is not due until sometime next year though.

Quoting source,

“We are all about platform longevity and long-lived upgrade paths,” andAMD spokesman said in a sideways ding at its competitor Intel which hasa penchant for requiring new sockets for its CPUs. Intel currently hasthree different socket infrastructures on its desktop computers – allincompatible. The confirmation also comes one day before Intel’sthree-day IDF conference which usually blots out all news fromcompetitors for days.

Backward compatibility is definitely good news here, lets see how AMD’s Thubian will contest against Intel’s Gulftown.

AMD Tips On Dual Core Performance

A two-processor dual-core system clocked 5 clock speeds down from AMD’sfastest part will generate performance of approximately 125 to 140 percentperformance of the dual-processor unicore system. A dual-core, dual-processorsystem clocked at only three grades slower than the fastest AMD chip willperform at between 130 to 160 percent of the performance of the base system.Future versions of the dual-core Opterons/Athlon 64s may contain two or morememory controllers to keep the processors fed with instructions. The problem isthat processor speeds have outpaced memory clock speeds, meaning that theprocessor can be forced to wait for the rest of the system to keep up. To solvethe problem, processors like Intel’s Montecito have added massive amounts oflevel -2 and level-3 cache to store frequently-used data close to the processor.

AMD TFE 2011 report

Yesterday we attended AMDs Technical Forum and Exhibition, or TFE as its simply called and this was the 7thtime AMD held the event in Taipei. Although AMD didnt talk too much about its future products, there were a couple of glimpses of new technology and there was also an NDA press briefing for the upcoming FX-processors from AMD, but sadly we cant tell you about it, yet.

At last years TFE AMD had invited media from all over the world for a press briefing on its then upcoming 6000-series of GPUs, but this year it was a fairly low key event for the media with only select media from a few Asian countries being on location. That said, the event isnt intended for the media, but rather for AMDs partners and we saw representatives for both notebook, motherboard and graphics card manufacturers at the show. The afternoon consisted of various presentations from both AMD and some of its components partners relating to how to make better components or systems based on AMD technology.

What the company did talk about was a little bit about the future of its CPUs, although as you can see, the roadmaps are hardly what youd call detailed. AMD mentioned its Piledriver core which should arrive next year, first in Trinity which replaces Llano and later on in its next generation desktop CPU. We should point out that AMD is not planning any future mobile CPUs, only APUs, although this is hardly surprising. AMD is expecting about 10 percent performance boost overall from the Piledriver core, although this is seemingly just an estimation as this point in time.

AMD was also keen to point out that its working hard to promote open standards, specifically OpenCL in this case as the company believes it offers the best cross-platform compatibility, scalability and of course the ability to take advantage of the GPU for computational tasks. To that extent AMD has launched a couple of OpenCL programing handbooks on top of going around and promoting OpenCL to its partners. AMD was mentioning a few products in its presentation such as Adobe Flash Player 11, Microsoft Silverlight 5, various DVD/Blu-ray playback applications and some video editing applications of which the most well-known one is Sony Vegas Pro.

AMD also talked about graphics and showed off one of its 28nm mobile Radeon HD 7000 chips and they also had a live demo of it playing Dragon Age 2, although there was no mention of which specific model it was, or the specifications of the test system but the game played and looked as youd expect it to. Apologies for the lack of a close-up of the chip, but it wasnt shown for very long. AMD did also show a block diagram of the compute unit of the 7000-series GPUs, which we did snap a shot of. AMD recons itll be first with 28nm GPUs, but as Nividia has as yet to announce anything with regards to its 28nm, well just have to wait and see wholl be first.

AMD was also keen to point out that its PowerTune technology allows its graphics card to automagically overclock while staying within the power and thermal envelope the cards have been designed for, although if an application or benchmark is run on the GPU that draws a lot of power and as such creates additional head, the GPU will throttle back to prevent its cards from failing.

Last but not least we also have a couple of videos from the event, the first one shows AMDs new five screen wide Eyefinity setup which is apparently able to display 10 million pixels if you have the right resolution displays. The second video shows MSIs AMD powered tablet running the Windows 8 preview code and AMD is of course talking about the benefits of Fusion here and how the GPU can help make for a smoother experience.

Thats all for now, but we will have some more details about the upcoming FX-processors next week, so check back with us then for some thoughts from AMD about its new platform.

AMD teases Cape Verde / HD 7700 with… Pills

In a most unique marketing stunt, AMD is sending out special pills to internet websites. Yes, you read that right, pills. They are no ordinary pills, however, they reveal several details about the upcoming AMD GPU along with a viral website for the same pills.

The "natural performance enhancement" pill is called "Verdetrol" which not only refers to Cape Verde but is also an obvious jab at NVIDIA. AMD sends in exactly 28 pills – as we know Cape Verde is a GCN based 28nm GPU.

HD 7770 is now all but confirmed as worlds first 1 GHz reference GPU. The "Never Settle" slogan for HD 7000 family continues, this time pushing "Never Settle for Average Performance". The viral website can be viewed at, where a promotional video will be uploaded soon. HD 7700 series is rumoured to release in February with performance close to HD 6850.

Source: Legit Reviews

AMD Takes On Intel’s Havok By Taking On Ageia

Rumours afloat say that AMD intends to acquire the manufacturer of Ageia Physics Accelerator in order to access physical acceleration technology. This could be AMD’s answer to Intel’s acquisition of physics acceleration engine development company, Havok. According to AMD graphics group’s European research and development relations manager Richard Huddy, AMD is indeed examining such an acquisition, but results would depend on whether Ageia’s acquisition price is “reasonable.”

It is understood that Intel in September acquired physics engine software maker, Havok. Havok’s modular suite of software development tools are widely used in games and digital animation. Films such as “Poseidon”, “Charlie and the Chocolate Factory”, as well as more than 150 world-renowned games, such as “Crackdown”, “Half Life2”, “Stranglehold” have used Havok’s physics engines.

Noteworthy is that Havok, Intel, AMD and ATI’s cooperative relationships has changed. This will affect the future Fusion processor AMD plans to release. Richard Huddy has publicly announced that AMD is now discussing merger with Ageia Physics Accelerator manufacturers to obtain physical acceleration technology.

According to Richard Huddy, AMD in the past few months has been discussing the merger. Ageia’s value also rose sharply after Intel’s September acquisition of Havok. AMD’s limited fundings caps investments over 100 million US dollars, hence the likelihood of acquisition mainly depends on the acquisition price.

Graphics industry watchers pointed out that the current Ageia PhysX is of low popularity compared with Havok. Coupled to AMD’s fund shortages, AMD’s acquisition of Ageia not probable.

AMD takes a Wait and See approach for UMPCs

Microsoft’s latest “Origami” hand held device was built together with Intel’s collaboration. Intel’s arch rival, AMD, says that they’ll be taking a “wait and see” approach to this new type of device. AMD Taiwan had commented recently that there’s still a couple of barriers of entry to this new platform. Taiwanese manufacturers are hinting that AMD might use their mobile CPUs like the Turion 64, or even their Geode – a CPU made for embedded devices, for these UMPC platforms.