With the advent of the Radeon HD 4000 series, AMD is regaining lost ground in the consumer PC graphics industry. They have been 'third time lucky' in their DirectX 10(.1) conquest after HD 2000 and HD 3000 series didn't quite make a mark against the might of NVIDIA. Apparently AMD is through with preparing the RV710 and RV730 graphics processors (GPUs) which will drive video-cards with several different configurations of memory and GPU parameters. AMD looks to compete in the sub-US $150 market with these cards for which it will phase-out the HD 3000 series cards within this quarter (Q3 2008). The products will reach End-Of-Life (EOL).

As we inch closer to the R700 release, which previews prove to have dethroned the GeForce GTX 280 convincingly, taking the crown for the fastest graphics card there is, NordicHardware reports AMD could be giving 'final-touches' to a fresh-wave of GPUs for early 2009. The timing is a critical factor here as that's the time-range the proposed NVIDIA G300 carrying 384 SP's and 512bit GDDR5 memory bus is marked for. The R800 cards could carry 2,000 stream processors, with its unit processors the RV870 holding 1,000 stream processors each. They could enter the realm of either 40nm or 45nm fabrication process.

With the launch of AMD's Radeon HD 4870 X2 flagship knocking at the front door, a couple of hardware sites have managed to get their hands on early engineering samples of the card and post preliminary results, dropping a hint at what's coming next month. From what I saw today, AnandTech and [H]ard|OCP are the first two web sites to drop unofficial benchmark results. PC Perspective and The Tech Report are also among these sites. To make their previews even more interesting they have added performance results from GeForce GTX 280 in SLI and other popular cards that can work in pairs.

Please let us know what you think about ATI's fastest single card.

Yes, you read that right. You can use upto four HD4870 X2 accelerators in tandem for an 8-GPU, 9.6 TFLOP, 8 GB graphics crunching monster. You need a 4-slot motherboard though. There are 4-slot solutions available on the AMD 790FX platform, for Intel though X48 did support 4 slots and images of prototype X48 boards with 4 slots did surface months back, there isn't such a board out yet. You do have the Skulltrail platform and upcoming X58 Bloomfield-supportive boards do promise to come in 4 slot flavours.

In a press-conference, Raja Koduri, worldwide CTO (Products Group), AMD talked to Indian website TechTree, among minor revelations such as "Fusion in 2009", here's a shocker:

"AMD has already built a computer that has four 4870X2s in it. So it has eight GPUs; drivers will not be supporting eight GPUs at this point of time."

-Implies that AMD is fully geared up to go head on against GT200b whenever it comes up. It's all a matter of them releasing a supportive driver. Again, unreliable sources point towards the possibility that also in the pipeline could be a R700+ which could be based on 'Super-RV770XT' processors. If you thought you're witnessing the peak of the GPU battle for supremacy, hold on, you're only getting appetised.

Sure, the graphics division brings some cheer for the company but seriously, to what extent is it helping the company alongside a marginally increased market-share with processors? Not much. AMD struggles to survive as stocks plummet below the $6 mark at NASDAQ index, which was priced at $15 /share only a year ago; this is the lowest value for the AMD stock since 2002.

AMD's survival is crucial for the entire computing industry as it keeps check on inflating prices by major players such as Intel and NVIDIA (who themselves are seeing bad days at the stock-markets these days). It has immense engineering potential to take on major players and force them to slash their prices. There are talks already doing rounds of CEO Hector Ruiz planning to quit.

Market forces and mal-informed consumerism are also to blame. A person chooses competitive brands over AMD products mainly because they're supplied and marketed better, sure Intel and NVIDIA do make better products in many categories but 'better' is a very relative value, how much better and for how much more (price) is something that keeps fluctuating, again fluctuations are mainly triggered by competition that AMD brings into the market. In other words, thank AMD for making NVIDIA sell GeForce 8800 GT for as low as $120 or better still, giving rise to a whole new SKU, the GeForce 9800 GTX+, with the '+' matching the red cross on first-aid kits.

Pre-release Tests Conducted on AMD Deneb

AMD Deneb is the code-name for the 45nm quad-core CPU which AMD plans to release soon. Chinese website ITOCP got their hands on two engineering samples. They used these samples at various clock-speeds set by altering the FSB multiplier and Vcore voltage. These chips were then subjected to rounds of Super Pi 1M benchmark. The results look rather luke-warm compared to what we saw of the Intel Bloomfield chips recently. The Deneb CPUs were supported by an AMD RD790 motherboard and 2 GB of DDR2 800 MHz unganged memory, running at timings of 5-5-5-18. The Phenom X4 Deneb 45nm will feature 6 MB of L3 cache apart from the usual 512 KB L2 caches dedicated to the cores.

The Taiwanese team of Tom's Hardware revealed several confidential (no more) slides pertaining to the AMD R700 product line, exposing details of the product codenamed "Spartan". It's now confirmed that it comes with a total of 2 GB of GDDR5 memory spanning across sixteen 1 Gb chips.

The picture is that of the HD4870 X2 itself in an angle never pictured before:The first slide shows certain details about the R700 "Spartan", the GPU core speed isn't disclosed yet. Crucial bits: 32Mx32 GDDR5, driver version 8.52 or later, deviceID: 0x9441.

Although it's known that the HD 4870 X2 (R700) is slated for a July thru August launch, reports from GPU Café suggest that AMD could subject the R700 to an early launch, and that's as early as four days from now, July 14. That's more than two weeks ahead of its launch schedule. It's not clear as to what's the nature of the launch, presumably a 'paper-launch', since there already is concrete information from industry observers in Taiwan as to the time-line of the R700 launch we covered here. Some high-profile overclockers are already to have their sample of the R700. Unreliable sources already point at the possibility of a HD 4850 X2 which comes with 2 GB of GDDR3 memory.

Update by W1zzard: Sorry guys but my sources tell me "middle of August" launch.

Presenting the fastest IGP to date. The upcoming AMD 790GX chipset, unlike preliminary reports suggesting it was going to be an extreme performance competitor to the GeForce 8200 IGP incorporated NForce 780a, is a high-end chipset armed to compete with the likes of NForce 750a, Intel G35. It supports ATI Crossfire multi-GPU technology.

In two new company slides published by the Chilean website CHW.net, early performance evaluations of the 790GX IGP were conducted, where the AMD chip beat the GeForce 8200 IGP by a very significant margin. It outperformed the Intel G35 by nearly 300 per cent margins in some tests. Apart from this ferocious little IGP, the 790GX supports ATI Crossfire with the energy-saving Hybrid-Crossfire feature. When not gaming, the display is handled by the IGP, depending on the load, the graphics card(s) are powered and speeds are adjusted. So when playing a game that the graphics card(s) can churn out 300 fps from, the GPU parameters can be lowered so you still get say, 100 fps while the GPU power is reduced significantly.

Eran Badit of the NGOHQ ATI PhysX driver fame now joins the NVIDIA developer program. In a more assertive statement, he believes that NVIDIA wants to "take on Intel with CUDA and to deal with the 'latest' Havok threat from both AMD and Intel.".

In a conversation with TG Daily, Roy Taylor, vice president of developer relations for NVIDIA says: "Eran and I have been talking via email and we have invited him to join NVIDIA's registered developer program. We are delighted at his interest in CUDA and in GPU accelerated physics using PhysX. Eran joins a long line of developers who are now working on using the GPU to run physics and who are doing so with the world's leading physics software - PhysX. "

Derek Perez, in charge of Nvidia's PR department, joined Taylor with this statement:
"We'll help any and all developers are using CUDA. That includes tools…documentation…and hands on help. We're delighted with the interest in CUDA and PhysX; and that includes the news on www.ngohq.com."

Once feared by many for aggressively defending its intellectual property, NVIDIA seems to have changed its approach towards things. In a bid to popularize their technology, it is willing to give the 'PhysX advantage' to even its competitor, AMD. But it still remains to be seen if PhysX performs on Radeon "the way it's meant to perform", now with Eran Badit working in close collaboration with NVIDIA. He says he still awaits samples from AMD so he could devise PhysX drivers for them. Surely NVIDIA has brought in good days for small developers with big aspirations.

Amidst this whole 'nGOHQ' saga, and us getting astonished at the way things turned out, here's what we missed:

NVIDIA is in talks with AMD over a possible technology-transfer agreement over PhysX, say industry sources in Hong Kong. In a not-so-recent article by HKEPC, it is said:

As of today, AMD is close to over a year and a half behind Intel with the implementation of the DDR3 system memory standard, and it doesn't look like we are going to see a DDR3 AMD platform only until late this year or early next year. Chilian website CHW.net published slides of the roadmaps for AMD chipsets in the months to come, also published are slides refering to the details of the next generation southbridge by AMD, even though the latest entry, the SB700 is only teething and only the latest motherboards with 7-Series chipsets feature this.

In a dramatic turnaround of events, NGOHQ.com, the creators of a special system software that allowed users of the ATI Radeon graphics accelerators to use proprietary features of NVIDIA graphics accelerators such as GPU-accelerated version of the NVIDIA PhysX game physics API, claim that in fact NVIDIA wanted to help them with this effort. On June, the 26th we had covered reports of the said outfit improvising a driver after proving that NVIDIA's proprietary GPGPU architecture, CUDA was flexible enough to work on a ATI RV670 graphics processor.

AMD Introduces New Phenom X4 Processors

AMD Introduces three new Phenom X4 quad-core desktop processors, the flagship Phenom X4 9950 Black Edition along with two low-power parts, the Phenom X4 9350e and Phenom X4 9150e.

The Phenom X4 9950 Black Edition is targeted at the performance/enthusiast segment, it features a 2.60 GHz clock speed and comes with an unlocked FSB multiplier among standard AMD Phenom specifications. This part however has a rated TDP of 140W and not all motherboards will be able to run it owing to its thermal and power characteristics. ASUS have already released a partial list of their motherboards that are ready for these 140W CPUs we covered here. You have to check with your motherboard vendor for compatibility. This chip attains its speed at a voltage of upto 1.3 V.

The RV770 is perhaps the best thing that happened to AMD in a long while. But more than AMD, it's perhaps the best thing that happened to us, the consumers. But general product launches seem to be just the tip of the ice-berg. The new PCB's designed by ATI for RV770 cards are actually running at well below the clock speed they can support and there is every reason to believe that these cards will be challenging NVIDIA's very best.

The HD4870 PCB with two 6-pin power connectors can support a maximum TDP of 225W (2x 75W from the power connectors + 75W from the PCI-Express interface). While at stock parameters, the HD4870 will not consume over 170W, it implies that with a fair bit of binning for high-performing parts, there is a serious lot of room for overclocking way beyond what the ordinary HD4870 cards can take.

AMD Fusion, the latest buzzword in the industry, simply put is a complex chip called Accelerated Processing Unit (APU), with a dual-core CPU, graphics processor, a PCI-Express switch and a DDR3 memory controller all rolled into one. This level of integration helps reduce system-level latencies, thereby improving performance and more importantly, reducing overall power-consumption and heat output, bringing it into the ultra-low power niche. AMD claims performance increments scaling up to 20 per cent for the CPU and up to 35 per cent for the GPU. Its compact package allows manufacturers to use it in building less than 1 inch thick notebooks.

The platform codenamed 'Shrike' is the first implementation of this design methodology. This industry schematic shows the various components of it. Shrike is slated for a H2, 2009 release. While it's not aimed to compete directly with the Intel Centrino Atom or VIA Isiah, it could just become a competitive platform for portable computing. Shrike consists of an APU codenamed "Swift" that connects to an external southbridge chip.

With the cost of manufacture for a standard G200 die reaching up to US $110, thanks to yields as low as 40 per cent, NVIDIA seems to be in a rush for a 55nm revamp of its current GPUs. While nothing revolutionary is on the cards, and with 55nm G92b already in the making, NVIDIA plans to revamp its G200 graphics processors to the 55nm fab process, increasing yields up to 50 per cent. At 55nm, the G200 die will be effectively reduced to 470 sq. mm, implies 120 dice on a 300mm wafer.

The pace at which things are moving is having the partners red-eyed. NVIDIA's new Unilateral Minimum Advertised Price Policy (UMAP) has limited partners' playing field and minimizes competition between them. When NVIDIA at the same time decides to launch new cards based on existing cores, at lower prices, partners get upset over diminishing earnings. Add to that AMD's new RV770 chip is looking very tempting to some of these partners.

DigiTimes reports that with AMD's ATI Radeon HD 4800-series graphics cards successfully cutting into the US$200-300 mainstream market forcing Nvidia to cut the price of its previous generation GeForce 9800 GTX down to US$199, sources at graphics card makers see AMD's discrete graphics card market share as having a chance to increase to 40% in the third quarter this year from around 30% at the beginning of 2008.

Since Nvidia's GeForce 9800 GTX does not provide advantages in power consumption or performance over the same priced Radeon HD 4850, graphics card makers are more optimistic over AMD's upcoming performance.

During AMD's recent Cinema 2.0 event, TG Daily caught up with representatives of JulesWorld. They are associated with the video production process of an unnamed movie studio and adopted components that are standard in the FX industry (AMD Opteron + Nvidia Quadro). Ray-tracing was an early focus of the company and they were told that JulesWorld started experimenting with a 100% ray-traced pipeline on a GPU with the arrival of ATI's R600 (2900XT) chip. And the results were impressive.

JulesWorld will be releasing OTOY and LightScape, two distinctive technologies that could shape the future of movie and games production. The company developed a ray-tracer that uses elements of the DirectX 9 API as well as its own high-level code that uses Tessellation and anti-aliasing algorithms.

AMD Releases ATI All-In-Wonder HD Series Card

AMD today expanded The Ultimate Visual Experience for North America with the ATI All-In-Wonder HD, combining award-winning ATI Radeon Premium graphics and ATI TV Wonder HD tuner technology card in one PCI Express 2.0 solution. As the newest multimedia powerhouse to join the long line of All-In-Wonder offerings, ATI All-in-Wonder HD transforms the PC into a highly immersive digital video recorder for HDTV and analog TV, plus expands the realm of exceptional gaming with cinematic HD graphics for mainstream PCs. Blu-ray disc playback can be enjoyed in full HD glory (1080p) thanks to ATI All-in-Wonder HD's unified video decoder (UVD) technology, ensuring movies play back smoothly and with incredible detail.

The graphics core is derived from the Radeon HD 3650, 55 nm, clocked at 725 MHz (memory: 600 MHz DDR2). AMD claims that the graphics performance of the card will beat a standard Radeon HD 3650 by 5 to 15% in popular games. The card incorporates the ATI Theater Pro 650 chip. Other prominent features include ClearQAM, a hardware MPEG-2 encoder and full hardware acceleration for H.264 and VC-1. The Adaptive Comb-filter provides better image quality.

For more information, please visit this page.

Asetek's industry-leading, LCLC liquid cooling solution is now available for AMD's latest ATI Radeon HD 4870 family of graphics cards. The unique, single-slot liquid cooling approach provides OEMs with an extremely high overclocking capability enabling them to design and produce outstanding gaming and entertainment products with virtually no noise.

To meet the demands of advanced gaming or graphics-intensive applications, Asetek's highly integrated solution lowers the GPU temperatures found on the 4870 by as much as 26 degrees, enabling it to run significantly cooler in extreme performance modes of operation. And, since the Asetek solution is totally liquid cooled, the graphics card can run completely silent, except for the system heat exchanger fan that typically runs at a quiet 30 dB(A).

Well-known graphics card maker GECUBE, released today with great fanfare, the new HD 4870 series of high-end graphics cards, the successor to its popular HD 4850 series. The GECUBE HD 4870 series comes with exclusive GDDR5 memory and the latest AMD graphics technology. Built on the 2nd generation 55nm chip process, it incorporates the PCI-Express 2.0 interface, as well as support for DirectX10.1/SM4.1 and ATI CrossFireX. Also included, are built-in HDMI and HDCP together with support for Blu-ray and HD-DVD decoding technology. Extreme processing power, coupled with high-speed memory performance, delivers world-class video visuals that will introduce gamers to the most realistic audio and video experience to date.

TUL Corporation - A leading manufacturer of AMD graphics cards, today announced the new generation of enthusiast graphics cards - The PowerColor HD 4870 and HD 4850. Based on the RV770 core chip, the PowerColor HD 4870/4850 support GDDR3/GDDR5 high-speed memory and aim to deliver superior visual performance.

The PowerColor HD 4800 series has adopted 55nm technology and is powered by ATI's Radeon HD 4800 graphics processing units. The HD 4870 GPU and memory run at 750 MHz and 900 MHz respectively, with 512MB of GDDR5 on a 256-bit memory bus design and memory data transmission rates up to 3.6 Gbps. As for the HD 4850, it's equipped with 512 MB of GDDR3 on a 256-bit memory bus design, whilst the GPU and memory run at 625 MHz and 1986 MHz respectively.

Two 45nm Phenom X4 and an Ultra-Mobile CPU Slated for Q4

NordicHardware reports that in Q4 2008, Advanced Micro Devices is slating two models of 45nm Phenom X4 quad-core chips along with a cheap ultra-mobile CPU aimed to compete with the Intel Atom and VIA Nano processors for release. The two new Phenom parts would be based on the new Denreb core to be succeeded by Propus based CPU's in Q1 2009.

Apart from retail models, AMD plans to release a CPU targeted for the ultra-mobile consumer electronics market. The CPU would be derived from the tested K8 architecture that is supposed to give it a design advantage over competing products from Intel and VIA. A major design advantage is of the memory controller being integrated on die that effectively eliminates the northbridge, saving on power consumption. The CPU has a rated TDP of 8W.

TG Daily is reporting that AMD is preparing to stoke up its GPU war with NVIDIA by unleashing overclocked HD 4850 powered solutions, in order to parry against rival NVIDIA's 9800GTX+. Apparently, by removing overclocking restrictions, AIB's will be able to offer a range of overclocked SKU's, with higher core and memory clock speeds, together with up-rated cooling solutions not being a stretch of the imagination. It is claimed that senior ATI officials anticipate overclocked SKU's to offer an increase of 50 MHz to the core clock speed, compared to currently available HD 4850 parts. Company names on the proverbial grapevine and, set to offer these solutions are, unsurprisingly, HIS, together with Sapphire Technology and Diamond. The upcoming HD 4870 isn't expected to escape pre-overclocking treatment either.