AMD found itself in a tricky situation with its launch schedule for the DDR3-supportive AM3 socket platform: the revision guide document for 10h family of processors disclosed a rather serious erratum with the processors, with the company not recommending the use of more than two DDR3-1333 memory modules in all, or more than one module per memory channel, stating unreliable operation as consequence (more here). We rose doubts on a certain statement in the document circling around what we saw as a potential solution to the problem, and hence sought the company's comment.

AMD replied to us with a set of things, not necessarily interrelated, though in some way connected to the erratum, and the company's approach towards the DDR3 memory standard:

• AMD confirms the issue as stated by the revision guide document, and is indeed working on a solution. The issue does not affect, in any way, using one DDR3-1333 memory module per channel
• The company has already specified the safest workaround for the issue: specifying the memory modules to run at the speeds of PC3-8500, 533 MHz (DDR3-1066). One can refer to hardware literature on how to do that
• AMD has given guidance to motherboard vendors on this issue. Some motherboards could rectify the issue by themselves, by operating the memory modules at PC3-8500 specifications. The incentive of tightening DRAM timings stays
• Good news for overclockers: You will be able to override the memory frequency control by motherboards, and can attempt to tweak frequencies at will. AMD product warranties, as always, don't cover damages caused by overclocking, even when overclocking is enabled via AMD Software
• Better news: Indications are that AMD is doing a lot more than creating a workaround for this issue, it may be devising a new feature altogether that, in the end, upholds the company's initiative to be enthusiast-friendly. Furthermore, AMD iterates that it is its intention to provide higher DDR3 frequencies as DDR3 becomes more prevalent

Barely a week into the introduction of the DDR3-supportive AM3 socket CPUs, the processors seem to be having design flaws. This, as circulated by AMD in its revision guide document for the 10h family of processors (found here, page 80). The issue, as described by AMD, centers around the DDR3 memory sub-system. On machines with more than one PC3-10600 (1333 MHz) memory module populating a memory channel, the users may experience unreliable operation. The company does not get into the specifics of the symptoms. This issue however, does not affect systems with a module per channel (one or two modules installed in the motherboard), and only those with three to four modules installed.

The AMD K10 memory controllers on AM3-socket processor provide a 128-bit wide memory interface (with DRAM Ganged mode enabled), which amount to two 64-bit wide memory channels. On most motherboards, four DIMM slots with two slots sharing a memory channel are present. With this issue, one is not recommended to use more than one DDR3-1333 memory module per channel. AMD recommends a quick fix for the issue for systems using more than one DDR3-1333 module per memory channel: to manually specify the memory to run at 533 MHz (1066 MHz DDR), and accordingly set DRAM timings. As a little compensation, one can tighten DRAM timings with the drop in frequency. AMD will fix this issue in the next stepping (sub-version) of the CPUs. The "x-factor" with this erratum revolves around DRAM voltage, a significant factor. One might note AMD saying "the processor memory subsystem may exhibit unreliable operation over the allowable VDDIO voltage range", which leads us to think if there is a potential workaround with adjusting the DRAM voltage beyond the allowable range (read: over-volting the memory). We hope to hear more from AMD on this.

UPDATE (02/13): AMD replied to the report, saying that work is in progress toward fixing the issue. While not getting into the specifics, AMD indicated to us that apart from addressing the issue, the company is also working toward something "which will make you 'feel cozy' about the DDR3 support". The statement is ambiguous, and is best left uninterpreted at this point in time.

ACTICA Inc., manufacturer of mission critical server memory products, announces the next generation DDR2-800/PC2-6400 ECC Registered for AMD Opteron Shanghai Platforms.

ACTICA's new DDR2-800 ECC REG are compatible to AMD Opteron Shanghai sever platforms such as Supermicro H8DMT, H8DMU+, H8DM3, H8DMA, H8DA3, H8DAi-2, Tyan S4985-E, S3992-E, S2927-E, S2915-E, S2912-E, S2937, S2935, and Asus KFSN4-DRE/SAS/iKV, KFSN4-DRE/iKVM.

• ACT4GER72E4G800S-DDR2-800 4GB ECC REG Samsung-chip
• ACT2GER72F8G800S---DDR2-800 2GB ECC REG Samsung-chip
• ACT1GER72A8G800S---DDR2-800 1GB ECC REG Samsung-chip

BIOSTAR MICROTECH INT'L CORP., a professional manufacturer of motherboards today announced the TA790GX A3+, the first AMD "Dragon Platform" qualified motherboard with next generation AMD AM3 Phenom II processors support.

The Phenom AM3 processors are able to support up to DDR3 1600MHz memory speed (when equipped with AMD 790FX northbridge), the TA790GX A3+ offers Dual x16 PCI-E 2.0 graphics interfaces to deliver up the latest ATI CrossFireX support for the ultimate gaming platform.

AMD today extended its Phenom II line-up with five new processors, including the industry's only 45nm triple-core CPUs and three new AMD Phenom II quad-core processors. Part of the fire breathing AMD Dragon platform, these new Phenom II chips fit in either AM2+ or AM3 sockets and support DDR2 or next generation DDR3 memory. The new triple-core and quad-core AMD Phenom II processors are available immediately with the following model numbers:

• AMD Phenom II X4 910 2.6GHz, 6MB of L3 cache
• AMD Phenom II X4 810 2.6GHz, 4MB of L3 cache - $175
• AMD Phenom II X4 805 2.5GHz, 4MB of L3 cache
• AMD Phenom II X3 720 Black Edition 2.8GHz, 6MB of L3 cache - $145
• AMD Phenom II X3 710 2.6GHz, 6MB of L3 cache - $125

All five Phenom II chips have the same 95W TDP.

Sapphire, the single largest add-in board partner for AMD, announced its FireStream 9270 GPGPU accelerator. Devices of this class exploit the general purpose computing capabilities of graphics processors, back them with a software architecture, and a programming environment to make for a dedicated number-crunching machine that holds rated computational power much higher than that of CPUs. Product lines in this class includes the AMD FireStream, that finds competition in NVIDIA's Tesla.

The appearance of the FireStream 9270 suggests it has been derived from the ATI Radeon HD 4870 graphics accelerator, albeit different clock speeds, different drivers, four times the amount of memory, and connections. The core features 800 stream cores that churn out peak computational power of 1.2 TFLOPs (single precision floating-point) and 240 GFLOPs (double precision floating-point). The floating point format is in adherence to IEEE standards. It features 2 GB of GDDR5 memory running at 850 MHz, with a peak bandwidth of 108.8 GB/s. The FireStream 9270 is backed by AMD's Stream 1.3 SDK that can be currently downloaded for free. It supports OpenCL in Microsoft Windows and Linux environments. The accelerator will be released by Sapphire towards the end of February, at a price of $1499. Additionally, AMD is teaming up with server-builder Aprius to release a 4U rack chassis system that holds as many as eight FireStream 9270 accelerators that churn out a theoretical max computational power of 9.6 TFLOPs.

NVIDIA is planning a new high-end desktop chipset for the AMD processor platform: the nForce 980a SLI. The news validates some of the earliest roadmap slides of ASUS leaked to the media, showing a certain M4N82 Deluxe to feature the said chipset according to the specifications sheet. The chipset is built to feature on newer motherboards supporting the AM3 socket and DDR3 memory, though the one from ASUS supports DDR2 memory only, with no mention of a DDR3-supportive motherboard from the company.

VR-Zone juiced some more information about this new chipset, which suggests that the nForce 980a could be merely a re-branded nForce 780a SLI. Earlier reports also suggested the possibility of the 780a SLI successor to support the Advanced Clock Calibration (ACC) feature that enhances overclocking the CPU using AMD's Overdrive software. With the 980a, NVIDIA will replace the markings on the package from "NF780A-SLI-N-A2" to "NF980A-SLI-A2". The company has also reportedly issued a notice to its partners:

The NVIDIA nForce 980a SLI SPP is an enthusiast-class high performance interface to AMD AM3 CPUs. It is RoHS-compliant. It includes 2×16 PCI Express 2.0 lanes (also configurable as 1 x16, 2 x8) and 3×1 PCI Express 1.0 lanes, 12 USB ports, six SATA ports, and single GigE. To ensure timely availability and flexibility, we encourage customers to qualify and release products using the NVIDIA nForce 980a SLI SPP.

The chipset will be available to motherboard manufacturers starting from March.

In the weeks to come, AMD will introduce its triple-core processors based on the 45 nm Heka core, starting off with the Phenom II X3 700 series. One of the first SKUs in the series would be the overclocker-friendly Phenom II X3 720 Black Edition. Based on the Heka core, it holds three x86 processing cores, 512 KB of L2 cache per core and a shared 6 MB L3 cache. The chip is specified to run at 2.80 GHz (200 MHz x 14.0), and features an unlocked bus frequency multiplier. It comes with a rated TDP of 95W. Turkish website Donanim Haber put an engineering sample carrying its specs to test on both DDR3 and DDR2 platforms based on Gigabyte's GA-MA790FXT-UD5 and MA790FX-DQ6, to find out its mettle.

Conceived to be a successor to the RV770 graphics processor, the RV790 emerged out of various media-leaks as the next big thing for ATI, in its never-ending pursuit for market leadership. Information on this mysterious new GPU has been trickling in at leisure, and it has been shaping up to be a current-generation GPU built using newer manufacturing technologies that could make it cheaper to produce. What we do know about RV790XT based engineering samples so far, is that they carry faster memory chips that are rated to operate a higher frequencies, 1 GHz to be precise, but specified by AMD to run at 975 MHz.

Fresh details sourced by Hardware-Infos from internal sources point out to the possibility of AMD stepping up the core clock speeds by as much as 100 MHz, taking it to 850 MHz. The ramp in clock speeds however, come at the expense of increasing core voltage to 1.3 Volts. There have been several reports speculating the core voltage so far, on of which points at 1.23V for 750 MHz. Another grey-area for now is the RV790's shader domain, with a certain level of ambiguity surrounding stream processor count. All of this leads us to believe that AMD hasn't finalized the specifications of RV790 and could be racking its brains on this one. AMD is expected to make critical announcements on its new GPUs in March and might choose the CeBIT event as the ideal platform.

Beating its chest on the "massive headroom" (read: clock speed increment potential) its newest line of processors, the Phenom II have, it was about time the company utilized the said headroom to carve out new SKUs. With the AM2+ exclusive Phenom II X4 940 already showing impressive overclocking potential, while shipping with a clock speed of 3.00 GHz, the company is planning to increase stock clock speeds in increments of 100 MHz (0.5x multiplier) with a string of SKUs post Phenom II X4 950. This according to the insider sources a German website called AMD News caught up with.

All the upcoming processors will be built on the AM3 package, making them compatible with DDR3 and DDR2 memory standards. With the Phenom II X4 950 being clocked at 3.10 GHz, succeeding SKUs take 100 MHz steps, starting from a Phenom II X4 960 at 3.20 GHz to a Phenom II X4 990 at 3.50 GHz. Despite the high clock speeds, the chips will continue to maintain rated TDPs of 125W. At some point within the succession, AMD will crank up the HyperTransport interconnect speed from 1,800 MHz to 2,000 MHz. The launch-schedule for these processors will be spread throughout 2009.

Although the Radeon HD 4850 X2 is an official SKU by AMD, launched (rather paper-launched) by the company along with the HD 4870 X2, Sapphire ended up being the only company to come up with an HD 4850 X2 series product line, since AMD didn't specify a reference design of its own. Sapphire went on to launch 1GB (2x 512MB) and 2GB (2x 1GB) variants of the said accelerator.

In AMD's latest release of the Catalyst driver suite (version 9.1), the HD 4850 X2 was excluded from the supported hardware list. Instead the company now released a special release of Catalyst 9.1 for the Radeon HD 4850 X2 series. The driver provides the same feature-set and list of fixes as described in the original release notes document for Catalyst 9.1, albeit being specific for the said accelerator. The WHQL-signed drivers are available for Windows Vista and Windows XP operating systems from the AMD website here.

AMD today released its timely update to the ATI Catalyst system drivers package that provides drivers driver support for ATI Radeon graphics accelerators, AMD 7-series chipset with integrated graphics. The drivers are effictive for Windows XP, Windows Vista and Linux operating systems.

The new release expands the feature-set of the ATI Catalyst, along with a set of fixes as described in the release notes (PDF). The fixes mostly revolve around Catalyst Control Center and the video acceleration features of the driver. To begin with, the new driver provides full OpenGL 3.0 support, including a few new GL extensions. The release also favours Linux by providing support for Hybrid CrossFireX. More importantly, the Linux version of the driver, provides MultiView support, that enables using independent display-heads on setups with multiple ATI GPUs. It is supported by any combination of ATI Radeon GPUs, Radeon HD 2000 series and later.

DOWNLOAD: ATI Catalyst 9.1 for Windows XP (32bit) | Windows XP (64bit) | Windows Vista (32bit) | Windows Vista (64bit)

AMD is on course of releasing the industry's first GPU built on the 40nm silicon process: the RV740. Its specifications make it a cheap yet powerful GPU for the segment it is about to cater to. With the earliest pointers hinting at 640 stream processors and a 128-bit wide GDDR5 memory interface. The GPU also marks the industry's first mainstream implementation of the GDDR5 standard.

Fresh details emerging out of the inspection of the driver information files of ATI Catalyst version 9.1, as German website ATI Forum finds out, hints at the possibility of the RV740 being implemented in two SKUs: the Radeon HD 4730 and Radeon HD 4750. While in the driver file, the device string lists with "RV630", it can be seen throughout the file that several strings are assigned to the same RV630, and is merely a means to identifying a whole class or generation of graphics processors to the driver. Additionally, the string bears the device ID of the RV670, again, that's not relevant. The strings are what matter. The RV740 is not very far away from large scale production. It should become a reality by the upcoming CeBIT event, in March. From the view of being a mere postulation, RV740 being the GPU that goes into making the new SKUs, does end up sounding plausible. When released, these cards are intended to further strengthen AMD's market position in the sub-US $150 market.

After AMD embraced the DDR3 memory standard, the eventuality always existed where motherboard vendors could release motherboards that supported both DDR2 and DDR3 memory standards, to support the complete spectrum of processors built for the AM2, AM2+ and AM3 sockets with appropriate memory modules installed. The Jetway MA3-79GDG COMBO will be one of the first in its kind.

Based on the AMD 790GX + SB750 chipset, the MA3-79GDG COMBO holds two DDR2 DIMM slots for DDR2-1066 MHz memory, and two slots to support DDR3-1066 MHz memory. Memory support is subject to the processor being used. The motherboard comes in the micro-ATX form-factor and goes ahead with providing support for the ATI CrossFireX multi-GPU standard apart from its Radeon HD 3300-class IGP. The rest of this motherboard consists of a standard feature-set, with single PCI and PCI-Express x1 slots, 8-channel audio with optical and co-axial SPDIF audio connectivity, IGP video output through DVI, HDMI and D-Sub connectors, gigabit Ethernet and eSATA. This motherboard will reach stores next month and hopes to cash-in on its biggest USP: upgrade-friendliness. Its price isn't known at this point.

With the AMD K8 architecture, AMD had made it very simple for CPUs and motherboards to get along. Memory support being purely subjective to the memory controller on the CPU, the only thing the motherboard chipset was left to do was to connect the rest of the system to the CPU using the HyperTransport interface. With the introduction of the DDR2 supportive AM2 socket that was mostly pin-compatible with socket 939, motherboard vendors took to cost-cutting using older s939 and s754-"supportive" chipsets such as the NVIDIA nForce 3. ASRock is one of them. The company used the nForce 3 in the AM2NF3-VSTA motherboard.

Having bought that motherboard in its time, might just pay off now, with the company including it in its latest socket AM3 CPU compatibility list. The motherboard might go on to support the latest Phenom II X4, X3 series; Athlon X4, X3 series, and other processors in AMD's 45nm desktop CPU lineup. That will also add to the reputation of the NVIDIA nForce 3 chipset being the only platform core-logic to span across three generations of CPU sockets: socket 754, socket 939, and socket AM2. And supporting AMD processors from three generations of sockets: AM2, AM2+ and AM3. This is what AMD originally conceived when designing its socket model. Unfortunately, vested commercial interests and "sandbagging" have plagued many a motherboard manufacturer who continue to discriminate CPU support within the AM2/3 socket series, to create new product lines, and to keep their sales up.

AMD today announced the availability of Let the Games Begin: A Toolkit 4 Making Social Issue Games, which helps nonprofit organizations to create social issue games on such topics as the environment, energy consumption, poverty and health. Produced by Games for Change with support from the AMD Foundation's AMD Changing the Game initiative, the toolkit includes examples of successful games with social content as well as in-depth presentations by key game-design experts. The toolkit is free of charge and available here.

Having taped out late last year, samples based on the new RV790 graphics processor have been doing rounds in the industry. A few more details have surfaced about it. Earlier noted to have identical clock speeds to that of the RV770XT (Radeon HD 4870), the RV790 samples are now known to have higher memory clock speeds. While the Radeon HD 4870 has its memory frequency at 900 MHz GDDR5 (effective 3.60 GHz), the samples carry memory clocked at 975 MHz (effective 3.90 GHz). Interestingly the memory chips on the sample, labeled IDGV1G-05A1F1C-40X, made by Qimonda, are specified to run at 1.00 GHz, reaching the 4 GHz effective memory speed mark. The samples feature 1 GB of memory. The RV790 is AMD's new current-generation graphics processor built on the newer 40nm silicon fabrication process. The new process is expected to reduce the GPU's power consumption and thermal footprint. The RV790 is conceived to be an immediate successor to the RV770 GPU.

Addressing the number one challenge for data center managers, AMD today announced widespread availability of five new low-power 45nm Quad-Core AMD Opteron HE processors. At 55-watt ACP, these new highly efficient processors help meet the industry's rising demand for low power, cost-saving data center technology. Now available in eight server systems from HP, the new processors can help customers immediately recognize power savings and increased virtualization performance. Rackable Systems will also have Quad-Core AMD Opteron HE processor-based servers available immediately and additional systems from global OEMs including Dell, Sun, and other solution providers are expected later this quarter.

Better late than never, AMD is embracing the DDR3 memory standard. Its newest desktop CPU socket, the AM3, connects the processor to dual-channel DDR3 memory. Initial specifications about AMD's processors on the new socket suggest that DDR3 1333MHz (PC3-10600) will be held as the memory standard the integrated memory controllers on AMD's processors support. The far-east team of Tom's Hardware posted the first series of benchmarks of a Phenom II AM3 processor, when paired with DDR3-1333. The benchmarks featured Gigabyte's MA790FXT-UD5P motherboard along with Corsair's dual-channel DDR3-1333 kit running at DRAM timings of 9-9-9-24. The CPU-Z window also shows that the memory modules are running unganged, meaning that the two memory controllers of the Deneb core are independent, with each controller handling a 64-bit wide memory channel. This state is somewhat comparable to that of single-channel memory, except that multi-threaded applications will still be able to independently address memory on each channel, utilizing all the bandwidth on offer.

At the center of the test-bench was the AMD Phenom II X4 910, the company's first flagship quad-core AM3 processor. The X4 910 features a default clock speed of 2.60 GHz. The same chip was tested in two settings: default clock-speeds, and overclocked to 3.12 GHz (240 x 13.0 @ default vCore). The screenshots below show the test results for WPrime. The overclocked chip scored a crunch-time of 46.613 s (1M). The overclock did not affect SuperPi at all, with insignificant differences in the scores between the two. PC Mark 05 got a decent boost, while 3DMark06 didn't. Memory bandwidth and latency tests showed something strange, with the overclocked CPU (in effect memory) turning up with lesser latency (while normally, increase in clock speeds tend to step up latencies). The tests show that there are increments in performance with the use of DDR3 memory, though they are merely proportional to the clock speeds the memory is running at.

Silicon giant Intel flexed its IPR muscle by reportedly calling rival AMD to discussions on possible violations of past license agreements. AMD and Intel share license agreements dating back to 1976, which eventually transformed into the agreement of letting AMD use Intel's x86 micro-architecture in making its own products. Later in 2001, the two companies met again to share x86-related technologies. Such agreements bind licensees of the x86 architecture to confining their technologies to themselves as the license is non-transferable.

The two companies are expected to meet to discuss on how the past two major events in the recent history of AMD: the acquisition of ATI Technologies, and the spin-off of the manufacturing division to The Foundry Company respect the terms of the agreements. Earlier, AMD stated that the ATI acquisition does not violate any of its cross-agreements with Intel, while the company sounds optimistic about getting The Foundry Company to existence smoothly through one of its big hurdles: a clean-chit from Intel, after having crossed bigger hurdles such as government approvals and support from the various countries it holds its facilities in.

AMD today reported fourth quarter 2008 revenue from continuing operations1 of $1.162 billion. Fourth quarter 2008 revenue decreased 35 percent compared to the third quarter of 2008 and 33 percent compared to the fourth quarter of 2007. Fourth quarter 2008 revenue was down 28 percent sequentially, excluding third quarter 2008 process technology license revenue of $191 million.
In the fourth quarter of 2008, AMD reported a net loss of $1.424 billion, or $2.34 per share. For continuing operations, fourth quarter 2008 loss was $1.414 billion, or $2.32 per share, and the operating loss was $1.274 billion. The results for continuing operations include an unfavorable impact of $996 million, or $1.64 per share as described in the table below. Loss from discontinued operations was $10 million, or $0.02 a share.
For the year ended December 27, 2008, AMD achieved revenue of $5.808 billion. Fiscal 2008 net loss was $3.098 billion. AMD reported revenue of $5.858 billion and a net loss of $3.379 billion for fiscal 2007.

AMD started its marketing offensive against NVIDIA's new dual-GPU GeForce GTX 295, and single-GPU GeForce GTX 285 accelerators, in an attempt to put the Radeon HD 4800 X2 series accelerators ahead of its competition in terms of performance on a "broad scale", and used the cards' availability in non-reference designs as a USP against NVIDIA's offerings using a uniform company-specified design. All this, in an internal presentation leaked to Donanim Haber. As for those wondering why, ATI and NVIDIA have a history of picking on each others' flagship products by means of such presentations, with which they intend to influence OEMs and channel vendors, though it always happens so, that these presentations reach public domains. Viral marketing or something plausible? Find out for yourself:

Qualcomm Incorporated and AMD today announced that Qualcomm has acquired certain graphics and multimedia technology assets, intellectual property and resources that were formerly the basis of AMD's handheld business. The acquisition enhances Qualcomm's multimedia capabilities, allowing Qualcomm to strengthen its leadership position in delivering more advanced products that redefine next-generation mobile user experiences.
"This acquisition of assets from AMD's handheld business brings us strong multimedia technologies, including graphics cores that we have been licensing for several years," said Steve Mollenkopf, executive vice president of Qualcomm and president of Qualcomm CDMA Technologies. "Bringing this technology in-house creates even greater synergy as we seamlessly integrate the best-in-class multimedia performance AMD offers into our system-on-chip (SoC) products."

AMD today launched the first official version of its Fusion Gaming Utility. Working on AMD systems only, the Fusion Gaming Utility turns off various Windows processes and thus gains some additional performance when needed. AMD Fusion works with both Microsoft Windows XP 32/64-bit and Windows Vista 32/64-bit.

DOWNLOAD: AMD Fusion Gaming Utility Version 1.0

Barely a month into its launch, AMD's newest processors, the Phenom II X4 920 and Phenom II X4 940 Black Edition, are already in for global price-cuts, of as much as 20%, in the days to come. Sources close to the company indicate that the price-cuts are in view of the recent announcement by Intel to significantly lower the prices of certain models of its Core 2 Quad lineup, which pose direct competition to AMD's new chips.

Once implemented, the Phenom II X4 920 would sell for US $195, down from its launch-price of $235, and a whole $50 would be waived off the $275 price-tag of the Phenom II X4 940 Black Edition, sending it down to $225. These price-cuts are not yet officially confirmed or announced by AMD, and so it isn't known as to when exactly they get officially implemented. All prices in 1000 unit tray quantities.