Emergent Technologies Interview: Part One	NVIDIA G80: Architecture and GPU Analysis	R580: ATI Radeon X1950 XTX Review	G71: GeForce 7900 GT, GeForce 7900 GTX, GeForce 7950 GX2 Preview

Press Releases
AMD Close to Metal Technology Unleashes the Power of Stream Compu... AMD Introduces World's First Dedicated Enterprise Stream Processor AMD Brings Next-Generation Graphics to the Apple Power Mac G5 Quad... New NVIDIA Products Transform the PC Into the Definitive Gaming Plat...

ATI chipsets rebranded as AMD; ATI replaced by SiS on Intel motherboards
16-Nov-2006, 19:23.19	Reporter : Uttar
As a logical next step following the completion of the acquisition process, AMD is rebranding most of their chipsets and have removed the ATI brand name from these products; except for IGPs, where a red "ATI Radeon Graphics" logo will remain in addition to the AMD one. Motherboards with Multi-GPU support will also have an extra red CrossFire logo on the box. The new names follow a relatively simple naming scheme: "AMD " + Codename + Segment. The segment is refered to as a single letter; X for Crossfire, T for high-end IGPs, G for midrange IGPs and V for low-end IGPs. Two examples of this are the AMD 580X CrossFire chipset and the AMD 690G chipset. The high-resolution logos are available at HKEPC (see link above), and the ATI Radeon Graphics one is also likely to be featured on the boxes of upcoming discrete GPUs. In related news, it seems that since Intel is no longer going to brand ATI chipsets as its own in order to meet demand, they're going to do so with SiS ones instead. Dailytech reports that the D103GGV Gem Valley, which was based on an ATI chipset and slated for H2 2006, has been completely cancelled, while the D201GLY Little Valley will be released in Q2 2007 and be based on the SiS 662 chipset released in July.
(Comments(2)) | (Add Comment)	Back To Top

Catalyst 6.11 hits the streets
15-Nov-2006, 19:23.22	Reporter : geo
The guys and gals at AMD have released ATI Catalyst 6.11 into the wild. Mothers with small children and pet owners are advised to not allow their charges to be unsupervised out of doors until trained experts in the community have sounded the All Clear siren. More seriously, the release notes claim CrossFire support for X1900 and X1950 combos, for those who've been hoping such would make an appearance. On a more personal note, graphics news writers the world over may pass the hat to buy a bottle of something good for whoever at AMD decided to use a sensible URL for such releases!
(Comments(0)) | (Add Comment)	Back To Top

AMD release free HLSL-to-GLSL converter
15-Nov-2006, 09:06.32	Reporter : Rys
Seemingly in an effort to encourage Mac/OpenGL ports of upcoming PC games, AMD has released a developer tool that lets developers easily and painlessly convert shaders from HLSL to GLSL, which are DirectX's and OpenGL's standard shading languages respectively. Cg also supports converting shaders into GLSL (although it is mostly used at runtime), which is in fact what it does to support modern ATI Radeon cards. But the code's output quality is quite subpar sadly, at least in the latest build we've tried; it does this conversion after transforming the Cg code into its own native interpretation, but before much of the optimisation process. This results in sometimes strange outputs that use more registers than the original code and is significantly less readable - although, at least, it compiles fine and either of the IHV's compilers should optimise most of the problems away. HLSL2GLSL, existing currently as compiled Win32 and OS X binaries and released under the unrestrictive BSD license with full source code included, also compiles and interprets the HLSL code but does so at a higher level, with minimal changes, which results in cleaner (although slightly reordered) code that is more easily modified even after the conversion. On the minus side of things, it doesn't properly convert most Cg code (example: declaring uniforms in functions will output an error), but then again it doesn't claim to be able to do that either, and the code is available for anyone who'd want to hack such functionality in it. Based on our initial testing, HLSL2GLSL will certainly be appreciated by anyone working on OpenGL or Mac/Linux ports of DirectX video games, or even OpenGL developers in general in order to be able to rapidly and easily convert any public HLSL shader they may wish to use. Again, find HLSL2GLSL at its SF.net website.
(Comments(1)) | (Add Comment)	Back To Top

AMD opens up Close to the Metal to developers
15-Nov-2006, 01:34.00	Reporter : The Baron
AMD announced that beginning today developers can gain access to their Close to the Metal (CTM) GPGPU "thin hardware interface" for stream applications by applying here. CTM has been eagerly anticipated in GPGPU circles for some time, and heralds another significant step forward in the movement towards turning GPUs into general computing devices. To go along with the CTM announcement, AMD also is introducing the "AMD Stream Processor", discussed here. It is interesting to note that even though it is from the "AMD Graphics Product Group", this product has received "AMD" branding rather than "ATI", and is specifically identified by AMD as being in line with Torrenza. Since CTM is limited to a low-level instruction set and application binary interface, one of the biggest questions regarding its adoption is what high-level languages that use CTM will be available. We asked Mike Houston, one of the primary developers of BrookGPU, about their progress with a CTM backend for BrookGPU: "The compiler is set up, but the runtime isn't ready. When CTM is fully public, we should be ready to go, but it may be awhile until things get stable. CTM is a really good match as a Brook backend. However, Brook won't be exposing some of the more advanced CTM features like scatter and some of the extended memory formats."
(Comments(0)) | (Add Comment)	Back To Top

NVIDIA sampling G8x notebook chips for Intel's Santa Rosa platform
14-Nov-2006, 13:55.42	Reporter : Uttar
X-bit labs is reporting a tidbit based on NVIDIA's Q3 conference call that we seem to have missed, which is that NVIDIA is already sampling GeForce 8 Go GPUs for laptop computers. This timeframe is similar to last year's GeForce 7 Go's sampling schedule, as far as we are aware, which would hint at at an announcement in the February timeframe. Jen-Hsun Huang says to be expecting even higher success with this new platform than they had with Intel's previous "Napa" one; indeed, recent Mercury Research numbers are showing massively increased marketshare for NVIDIA in the notebook segment, although we believe these numbers might be inflated or exagerated based on certain indications. Other sites, especially so Dailytech, have been hinting at a large G8x desktop line-up also launching in February, which would imply NVIDIA is preparing a top-to-bottom, multimarket hardlaunch, should everything go as planned. R600 seems to be slated for availability in January, but we are unaware of the exact details and timeframes of the upcoming launches of ATI-branded midrange cards. We can also safely assume that NVIDIA's upcoming line-up will be exclusively 80nm-based, and that it will be produced on TSMC's high-performance (GT) process. It is uncertain at this point whether either NVIDIA or ATI are going to use the higher-end and possibly more expensive 80HS (high-speed) process for some upcoming products. G8x notebook chips probably are the same as the desktop ones, but manufactured on a slightly different process. Even on such a process however, it is unlikely that any G8x-based notebook GPU with more than 96 "stream processors" will see the light of the day.
(Comments(3)) | (Add Comment)	Back To Top

Emergent Technologies Interview: Part One
13-Nov-2006, 13:18.43	Reporter : Rys
For quite a number of you reading this, Gamebryo will be a familiar name or term. Members of the gaming crowd will notice that its the engine used to build a fair chunk of the two latest Elder Scrolls games: Morrowind and Oblivion. Thing is, for the average chap on the street itll probably conjur up a faintly disturbing image. Well leave the foetal visions to you guys, if thats OK. However despite Gamebryo becoming more memorable in the minds of gamers and 3D enthusiasts alike, the company responsible for Gamebryo remains an obscure, behind-the-scenes phenomenon, quietly plugging away at the technology and working with its implementers. In this Beyond3D interview, we check out Emergent as a company and Gamebryo as its flagship product, by interviewing Mike Steele, Emergent's VP and Technology Evangelist. Vysez took the reins for our Emergent Technologies Interview: Part One
(Comments(17)) | (Add Comment)	Back To Top

Exclusive: PlayStation 3's RSX is 240mm and bigger than the PC equivalent
11-Nov-2006, 16:56.08	Reporter : Uttar
Based on the RSX die shots released by websites following the PlayStation 3's launch in Japan yesterday, we have successfully estimated the chip's die size by comparing it to the nearby Samsung GDDR3 memory chips. The result is approximately 240mm and, even with the imperfect precision of our measurement method, we can safely claim it's between 230 and 245mm. That makes RSX roughly the same size as the CELL revision used in the PS3, which is known to measure 235mm, and it's a fair bit bigger than NVIDIA's PC equivalent to the RSX: the 90nm NVIDIA G71, which has a die size of 196mm. While TSMC's 90nm transistor density might be slightly different from that of Sony's fabs, it is unlikely to be a major factor here. The chip is also rumoured to run at 500MHz internally. Based on our analysis and information, it seems that the RSX has 28 pixel shaders internally, with 24 active. G71, on the other hand, has only 24 and NVIDIA sells the defective parts as the 20 pixel shader GeForce 7900 GS. Such an opportunity for multiple SKUs based on the same die does not exist with consoles, and therefore higher redundancy is desirable. It's also likely that other parts of the RSX are therefore disabled for yeilding, such as the vertex shaders and possibly the raster operations units (ROPs). This is similar to the strategy used by Sony with the CELL, where 1 of the 8 SPEs is disabled. While the RSX might have some parts removed compared to the G71, such as its PureVideo technology, transistor count is likely balanced back by having more redundancy and larger caches: indeed, we believe that most of the RSX's internal caches are larger compared to any G7x part, including the post-vertex transform cache and compression caches. In addition to its normal 48KiB of texture cache to its local memory, it also has an extra 96KiB of cache dedicated to communication with the XDR memory pool, in order to improve bandwidth utilization and average latency, and make it possible to use that memory pool for texturing operations - resulting in greater overall system bandwidth utilization. Xbox 360's Xenos, on the other hand, is a 180mm chip (plus the daughter die, with the eDRAM and ROPs) and does not seem to have this level of redundant parts disabled for yield reasons. Instead of coarse-grained yield management (entire functional blocks duplicated), it is rumoured to be duplicating smaller elements in each block (likely 17 shader ALUs per 16 required working - some patents, including this one, hint at this), in order to minimize the chip's die size while keeping yields in check. Both approaches are perfectly acceptable and industry standard nowadays. Based on all of this information, RSX does not seem significantly more expensive to produce than Xenos, especially so when counting Xenos' daughter die, which is estimated at more than 70mm. It should be taken into consideration, however, that the PlayStation 3's memory costs are most likely higher than the Xbox 360's, and that the lack of eDRAM on the RSX is a contributing factor to this difference, as this requires the system to have higher overall memory bandwidth. The memory chips being "nearer" to the graphics core, as visible on the die shots that can be found on the web, will however not give any performance advantage, and while it may increase costs very slightly, it will help for spreading the memory's heat. It should finally be noted that while the GDDR3 chips people have found next to RSX are rated at 1400MHz effective (700MHz internally), they will apparently only run at 1300MHz effective. This has most likely been done in order to also be able to use chips rated at 1333MHz effective by Samsung and possibly other manufacturers, a step which Sony might have deemed necessary to meet its long-term production goals and lifetime cost reductions.
(Comments(15)) | (Add Comment)	Back To Top

NVIDIA GeForce 8-series coverage @ HEXUS
10-Nov-2006, 09:06.30	Reporter : Rys
As a prelude to doing a bigger at-launch roundup from across the web, here's what my fellow Hexans had in the way of related coverage on launch day. Scan International's 3XS Triad, featuring 8800 GTX ASUS EN8800GTX Review XFX GeForce 8800 GTX GPU IQ and Performance Eval
(Comments(0)) | (Add Comment)	Back To Top

Confirmed: NVIDIA entering the Intel IGP market, preparing large G8x lineup
10-Nov-2006, 00:59.02	Reporter : Uttar
NVIDIA has just held its quarterly earnings conference call, discussing its financial results and answering questions from analysts. The quarter is strong overall with record revenue and rapidly increasing market share, although and slightly higher net profit than expected by analysts. On a GAAP basis, it was however lower than some expected, and this was partially due to a confidential patent licensing charge of $17.5M related primarily to old products. But the real shocker is that Jen-Hsun Huang, President and CEO of NVIDIA, announced in the conference call that they're working on Intel IGP solutions, as customer demand is increasing and the merged AMD-ATI is leaving the market, thus creating a void to be filled. They're currently aiming at an introduction date as early as Spring 2007. Jen-Hsun cited high-definition video and DirectX10 as key reasons for the increased demand and necessity, but it is unclear whether he implied by that this IGP is G8x-based or not. NVIDIA is certainly boasting its architecture's scalability and increased performance per watt, and some presentations slides are in fact hinting at CUDA, their new GPGPU framework with direct support for C-based programming, extending all the way to embedded markets. It could also be that NVIDIA is working on a G80 derivate for the GoForce product line, potentially with only some parts shared (like the shader core), or a completely different architecture that boasts similar programmability. No matter what, they're working on at least nine more G8x-based products, that is to say, ones with unique codenames or brand names. This is substantially more than the historical average, although if an IGP was included in there and the notebook parts had separate codenames, that'd be roughly four chips for the desktop lineup - the same number ATI's lineup currently sports. It remains to be seen, of course, whether these upcoming products will have the same ALU-TEX ratios as the GeForce 8800GTX, for example. While the fundamental G80 architecture we've detailed in our in-depth piece on the subject will certainly still apply, it is not out of the question that some things might also be different, and we look forward to repeating part of our analysis process on these upcoming parts.
(Comments(18)) | (Add Comment)	Back To Top

NVIDIA G80: Architectural Overview
08-Nov-2006, 18:59.52	Reporter : geo
Four years and 400 million dollars in the making, NVIDIA G80 represents for the company their first brand new architecture with arguably no strong ties to anything they've ever built before. Almost entirely brand new as far as 3D functions are concerned, and designed as the flagship of their 8-series GeForce product line, their new architecture is squarely a D3D10 part but with serious D3D9 performance and image quality considerations. Click here to read our architectural overview of NVIDIA's G80-based GeForce 8800GTX.
(Comments(44)) | (Add Comment)	Back To Top

NVIDIA buys PortalPlayer for $161M net in cash
07-Nov-2006, 08:50.17	Reporter : Uttar
NVIDIA has just announced that they're acquiring PortalPlayer, Inc. for $357 million (or $161 million net, once you take into consideration the money they'll get access to on the acquired company's balance sheet), a company mostly known for its audio SoCs and firmware, which it sold to numerous companies and primarily for Apple's iPod product line. In early 2006, they lost the audio socket on most of Apple's upcoming products, and their stock price plunged instantly. Their profits were also going down rapidly, and while they were trying to diversify aggressively, most analysts agreed that they'd begin losing money soon enough. By acquiring PortalPlayer, NVIDIA increased its number of employees by approximately 300, among which more than 125 employees in Hyderabad, India and roughly 200 engineers. NVIDIA has been trying to expand in India lately, but what is of even greater interest is that PortalPlayer's headquarters was in San Jose, where NVIDIA recently decided not to rent the Sobrato building and make it its new HQ for "accounting reasons". It very well could be that the large amounts of cash involved in this deal had something to do with that, and perhaps NVIDIA did not want to begin using two buildings in San Jose at the same time. PortalPlayer's chips fundamentally are one or more licensed ARM7 cores on a single core, with a number of related systems, inputs, outputs and firmware. As such, they provided not only the iPod's sound chip, but also its CPU - although "just" a licensed one. This SoC expertise and focus on embedded markets where NVIDIA is desperately trying to grow as quickly as possible certainly were the primary reasons for the firm's acquisition, in addition to being relatively cheap due to the recently lost iPod contracts. There have been no recent leaks of what's coming from NVIDIA after the GoForce 5500 in the handheld market, and how near it is, so it is unclear at this point whether this purchase will influence either it or its refresh.
(Comments(0)) | (Add Comment)	Back To Top

VIA and AMD preparing single-chip IGPs
06-Nov-2006, 09:12.36	Reporter : Uttar
HKEPC is reporting that VIA is working on a single-chip core logic solution (KM960) slated for Q4 2007 that will combine all chipset and IGP functionality for AMD platforms. The article also speaks of SC780 from AMD, which integrates RS690 and SB600 on a single 65nm chip. Both approaches are similar to that of NVIDIA's MCP61S and MCP61P chips, which have been released a short time ago on 90nm. Unlike previous VIA IGPs, this one also supports SM2.0. vertex shaders, and not only pixel shaders. It is interesting to note that VIA is also focusing its efforts on the AMD platform, as their Intel bus license will expire in 2007. But considering AMD's Fusion initiative is slated for (late?) 2008, these efforts might prove worthless in the long term. It is also questionable, however, whether AMD might also consider eventually integrating all (or part of) the MCP (traditionally the southbridge) on the same, unique chip, which would effectively make that IC become a system-on-a-chip (SoC). This would clearly go against their open platform strategy, but might definitely result in some further cost reductions. Otherwise, there remains two dice, just like there would be with a single-chip IGP, although it might still be more cost effective and efficient for other reasons, including interconnects and memory latency.
(Comments(1)) | (Add Comment)	Back To Top

NVIDIA abandons moving its headquarters to the Sobrato building
03-Nov-2006, 18:39.14	Reporter : Uttar
It has been rumored for a while that NVIDIA was going to move its headquarters to downtown San Jose by leasing the Sobrato building, a 380,000-square-foot building that is currently vacant and represents 6% of the current vacancy rate of the area. The Mercury News is indicating,