First Working NVIDIA GF100 Video Card Pictured

Category: Video Cards
Posted: November 19, 2009 12:37AM
Author: Daryn Govender

NVIDIA's Senior PR Manager Brian Burke has uploaded the first picture of a working GF100 video card on the company's Facebook page. For those of you in the dark, GF100 is the codename for the first GeForce GPU based on NVIDIA's Fermi architecture. The card is running Unigine Corp's DX 11 "Heaven" benchmark on a Dell monitor. It seems as though NVIDIA is trying to reassure consumers that Fermi cards are on track for release early next year in light of last month's revelation that the Fermi Card used in Jen-Hsun Huang's presentation was a mock-up. Whatever stage of development Fermi is in, I think NVIDIA should be concerned with the recent launch of ATI's top-end behemoth, the 5970.

Hardware Roundup: AMD Radeon HD 5970 Edition

Category: Video Cards
Posted: November 17, 2009 10:58PM
Author: Dale Shuck

In the past two months AMD has been busy launching new videocards in it 5000-series lineup and today marks the fifth entry in the hit parade with the official release of the Radeon HD 5970. As suspected, the HD5970 is going to be priced around $600 so it's definitely a top of the line card in terms of price. So what are you getting for that price? Well, as you probably know, the HD 5970 packs dual GPUs onto a single PCB that combined offer 3200 stream processors and 2GB of frame buffer with a standard core clock of 725MHz. If you want to see what you're getting for your $600 in terms of performance and real-life game play, be sure to check out all the reviews we have lined up for you.

AMD ATI Radeon HD 5970 Video Card @ [H]ardOCP
Powercolor Radeon HD 5970 2GB Graphics Card @ Driverheaven
AMD Radeon HD 5970 2GB Review - AMDomination @ PC Perspective
ATI Radeon HD 5970 Dual-GPU Powerhouse @ HotHardware
Sapphire Radeon HD 5970 2GB Video Card @ TweakTown
ATI Radeon HD 5970 Review: Dual-GPU Graphics @ TechSpot
ATI Radeon HD 5970 Dual-GPU Video Card @ Legit Reviews
Radeon HD5970 @ Bjorn3d

ATI 5970 Price Revealed?

Category: Video Cards
Posted: November 16, 2009 11:49PM
Author: ScapeGoat

It looks like the computer parts website ZipZoomFly slipped some info regarding the HIS H597F1GDG graphics card. The price is set as $599.99 for that particular model, with the latter featuring 2GB of GDDR5 memory and support of all the other features in the actual DX11 generation of ATI cards, including Eyefinity making this the perfect choice for those who want to experience three monitor gaming. With the official launch set for the upcoming weekend I assume that most retail 5970's will have a similar price.

Matrox's M9188 Supports 8 Displays

Category: Video Cards
Posted: November 10, 2009 03:53PM
Author: Rpbert Bergem

Matrox has just announced the newest addition to their video card line-up, the M9188 PCIe x16 Octal graphics card. The card is labeled as Octal because it can drive 8 displays at once. Yes, I said 8 diplays. The card has 2GB of memory and can run each display at up to 2560x1600 for Display Port monitors or up to 1920x1200 for DVI Single-Link monitors. It is a single-slot design so that it will not block other slots on the motherboard and allows room for additional cards. Two of these cards can be combined in one system to drive a total of 16 displays! The connected displays may be arranged in two ways: One arrangement is a single desktop stretched across all connected displays, while another option provides an independent desktop on each. This allows for either more space for one program, or multiple desktops to display multiple programs at once. Two keys areas that Matrox mentions the card would be beneficial in are electric grid management and train dispatch management. Both of these areas require displaying large visualizations of the grid or train tracks to effectively manage them. The M9188 will retail for $1995 and supports Windows 7, Windows Vista, Windows XP, Windows Server 2003/2008, and Linux.

If $1995 is too much or you only need to drive 2 monitors, Matrox is offering the M9128 LP PCIe x16 that can support 2 monitors. This card will retail for $259.

ATI Radeon HD 5970 Announcement Date

Category: Video Cards, General News
Posted: November 10, 2009 12:20PM
Author: MrAlex

The Radeon HD 5970 will be the new flagship GPU coming from ATI. They have decided to drop the X2 moniker, and instead of branding their dual-GPU cards as X2 cards, they will be branded within the Radeon HD 5900 series. Officially, we know that the 5970 will be two 5870’s, but we are yet to know if there will officially be a 5950 (two 5850’s). The NDA lifts on the 16th of November, but it is not known if this will be the official date, but we can hope that it is. The specifications of the 5970 are as follows: Built on the 40 nm fabrication process, 334 mm² die size, 2154M transistors and PCI Express 2.1 X16 Interface. From various images leaked from the internet we can see that it is in fact two 5870 GPUs on a single PCB, destroying the hope of having a MCM GPU. This means that there will be two separate GPU cores clocked at 750 MHz, 2x 1600 Stream Processors, 2x 256-Bit Bus Widths (512-Bit Total), 2x Memory Sizes of 1024 MB with 1000 MHz clocks (4000 MHz effective), with support for DirectX 11 and Shader Model 5.0. Price hasn’t been announced, although due to stock shortages because of problems with TSMC and lack of competition, it could come in around $500+.

First DX11 Passive Cooling Card on the Way

Category: Video Cards
Posted: November 8, 2009 02:53PM
Author: ScapeGoat

PowerColor has just announced the SCS3 HD5750, the first passive cooling DX11 card. It will run at stock speeds, being 700MHz and 4600MHz for the GPU and memory respectively while also having all the features of the other cards, but instead of the typical fan cooler, it will have a 4-heatpipe passive cooling solution providing absolute silence. With its silence and good performance, the SCS3 HD5750 is a card to have in mind for silent computing and especially for the use in an HTPC. We can only expect similar solutions for its bigger sister, the HD5770, in the near future. The card is set to be launched in late November with a price around the 150 dollar mark.

ATI Radeon HD 5870 2GB 'Eyefinity' Edition Detailed

Category: Video Cards, General News
Posted: November 7, 2009 04:25AM
Author: MrAlex

As you might or might not already know, ATI plan to release a 2GB variant of their already popular Radeon HD 5870. What separates this card though from its siblings is that it comes with six Mini Display Port outputs, which shows that this card is specialised for multi-monitor setups. The new card also requires an 8-Pin and 6-Pin power connector instead of two 6-Pin connectors like the Radeon HD 5870 1GB variant. Apart from the ATI Eyefinity Technology and having no back plate, the cards specifications remain the same with 1600 Stream processors, a core clock of 850MHz and a 4800MHz effective memory clock. Price and release date has not been announced, neither has whether ATI's AIB Partners will be branding and selling the 2GB variant, or if it will only be only available from ATI themselves.

AXLE Releases GeForce GT 220 and GeForce 210 Cards

Category: Video Cards
Posted: October 29, 2009 05:30PM
Author: Dale Shuck

AXLE International Holding Limited may not be a name you are familiar with, but the Hong Kong company has been around for more than 20 years. The company's product lineup consists of a full range of Nvidia- and ATI-based graphics cards along with a series of power supplies. AXLE has just announced the addition of two new cards with the release of its GeForce GT220 and GeForce 210 videocards based on the latest 40nm process technology from Nvidia. Designed as a low-cost way to bring more powerful video processing to your desktop when integrated graphics just isn't enough, the cards also offer Nvidia Unified Architecture, CUDA support and Nvidia PhysX technology. AXLE did not diclose pricing or availability on the new cards.

ATI Catalyst 9.10 Driver Released

Category: Video Cards, Software
Posted: October 25, 2009 01:36AM
Author: Daryn Govender

ATI has released the newest version of its Catalyst video card drivers: ATI Catalyst 9.10. The Catalyst 9.10 driver supports the ATI Radeon HD 2000, 3000, 4000 and 5800 series cards. It is the first ATI Catalyst driver which is WHQL certified for the newly released ATI Radeon HD 5800 series of graphics cards. This new driver also supports "super sample anti-aliasing" which aims to bring high image quality whilst maintaining good performance. To see the full list of fixes and updates you can read the release notes here. ATI Catalyst 9.10 is available to download over at AMD's website here.

NVIDIA SLI Profile Update 3 Released

Category: Video Cards, Software
Posted: October 15, 2009 12:25AM
Author: Daryn Govender

NVIDIA has launched its third SLI Profile update. The new update bring SLI support to the following games:

• Borderlands
• Cloudy with a Chance of Meatballs
• G-Force
• FIFA Soccer 2010
• League of Legends: Clash of Fates
• NHL 2009
• Order of War
• Pro Evolution Soccer 2010
• Race On
• Star Trek: D-A-C

The update also increases performance in two games which are already compatible with SLI: Champions Online and Dragon-Age: Origins. NVIDIA recommends that the SLI Profile Update 3 is used in conjunction with the v191.07 WHQL display driver. You can download the new SLI profile update here.

ATI Radeon HD 5900 Sighted in AMD OpenCL Driver

Category: Video Cards, Software
Posted: October 15, 2009 12:11AM
Author: Daryn Govender

TechConnect Magazine has spotted some interesting information in AMD's latest group of OpenCL graphics drivers for Windows and Linux. The OpenCL drivers list the unannounced "ATI Radon HD 5900 Series" as part of the supported cards list under the Evergreen family of GPUs. What you make of this revelation is up to you, but I think that ATI is working on a high-end rival (most likely X2) to NVIDIA's Fermi GPUs. It's always good to see some competition between manufacturers as it helps products improve and stimulates lower prices for the end consumer, you and me!

So What's all this Talk about NVIDIA? - Rumor Roundup

Category: Video Cards
Posted: October 14, 2009 04:52PM
Author: Ben Grantham

By now, we are all used to rumors circulating about what big companies in the world of technology might be up to. With the constant break neck pace of hardware development, there is always something new to talk about, and with that comes plenty of ways for things to go wrong. When it comes to the two big names in GPUs (for desktops at least), rumor mongering is intense, with PR for each side very happy to highlight weaknesses in the competition. Add to this the fact that consumers often like to pick sides (you can witness this by dropping into any forum discussion where the names of the two companies crop up), talk can get quite heated. Since ATI/AMD kicked off the most recent generation of competition with the release of its 5800 series cards, stories have been appearing with regard to how NVIDIA might respond, with some doubting they were in much of a position to do so. Rather than try and negotiate all the conflicting information and attempt to come up with an answer, I will attempt to lay out all the varying reports and let you try and make sense of them.

Of course, we know about NVIDIA's upcoming Fermi architecture already, following the announcement made at the company's GPU technology conference. Now, the presentation in itself lead to some chatter after NVIDIA showed a 'Fermi' card without making it too clear that the board was a mock up (well at least not until afterwards). Whether you want to make much of that is up to you, as you could explain it in favorable or less favorable terms, but it certainly isn't the only thing that has got people talking recently. Now we have moved on to debating the state of NVIDIA's current generation products, the GTX 200 series. The same source that first highlighted the Fermi board mock up, claimed that NVIDIA was planning on ditching it GTX 260, GTX 275 and GTX 285 parts, citing that they simply couldn't be made and sold at a profit. While NVIDIA PR has denied this, there is still further information being published that suggests that it is/was NVIDIA's board partners opting to discontinue the parts. There have certainly been claims of shortages to go along with this, but you can still go out and buy say, a GTX 260 right now if you want. Of course, other reports say that this isn't the case at all, and that NVIDIA's partners claim mass production of GT200 based parts is continuing as normal. You are welcome to join our forum members in the debate (though as it is with these things, it easily gets sidetracked).

During this time, NVIDIA has also announced what many had suspected for a while, that it would be halting development for Intel CPU chipsets going forward. That means chipsets for Core i7 and i5 and anything else that uses Intel's new QPI/DMI bus architecture (that's going to be more and more in the not too distant future). The two companies have disagreed over the chipset license that is the main cause of this problem since early in the year (Intel believes the license doesn't cover products with integrated memory controllers, while NVIDIA believes it does). That certainly doesn't help matters.

As for Fermi, it is obviously going to be an interesting architecture and a move further towards using the GPU as a computing tool, rather than something we use just to play the latest games. Again, there have been some concerns raised about when NVIDIA will be ready to get products out to consumers, but similar concerns are always aired when next generation parts from another company have already hit the shelves. According to the latest reports, partners are actually pretty confident about GF100 and have already put in large orders. It's still unsure when we may see these launch, but it could be by the end of the year. Mainstream products based on the architecture would seem to be further away than that, which would give ATI a reasonable buffer to build up sales of its own cards.

Personally, I find all of this a little tiring and that is without even touching on issues surrounding recent PhysX happenings. I have always been one to make judgements once products actually become unavailable, or once I can get new hardware in my hands, but not before. Maybe that is a little strange now?

Hardware Roundup: AMD Radeon HD 5770 and HD 5750 Edition

Category: Video Cards
Posted: October 13, 2009 03:29AM
Author: Dale Shuck

It's been barely three weeks since AMD released it latest and greatest card, the HD 5870, which was quickly followed by the HD 5850. Now, ATI is bringing the 5000 series cards to the mainstream market with the release of the HD 5770 and HD 5750 cards. Our roundup provides plenty of coverage so dig in and enjoy!

Sapphire HD 5770 and 5750 @ OCC
PowerColor HD5770 @ OCC
XFX HD 5750 @ OCC
AMD Radeon HD 5770 and HD 5750 Review - Juniper and DX11 for all @ PC Perspective
AMD ATI Radeon HD 5770 and HD 5750 @ [H]ardOCP
HD 5750 and HD 5770 @ NeoSeeker
ATI Radeon HD 5770 and 5750 Mainstream DX11 GPUs @ HotHardware
ATI Radeon HD 5770 @ TechSpot
Sapphire Radeon HD 5770 1GB Video Card and Sapphire Radeon HD 5750 1GB Video Card @ TweakTown
ATI HD 5770 and HD 5750 Video Card Launch Roundup @ Motherboards.org
ATI Radeon HD 5770 - DirectX 11 for the Masses @ Techgage
Sapphire HD 5770 and Asus EAH5770 Voltage Tweak @ Bjorn3D

ATI 5700 Series Prices to be Lower than Expected?

Category: Video Cards
Posted: October 9, 2009 04:07PM
Author: Ben Grantham

If reports by Bright Side of News are correct, ATI's upcoming 5700 series cards could be a little easier on the wallet than has previously been the case for newly released mainstream parts. Rather than going for between $150 and $200 that was previously expected, we may see 5700 series cards starting at just over $100 and topping out at around $160. That should also mean three 5700 series boards, rather than two, with the following models being in the pipeline:

• HD 5750 512MB GDD5 - $109.99 (possibly as low as $99.99 in some cases?)
• HD 5750 1GB GDDR5 - $119.99-$129.99
• HD 5770 1GB GDDR5 - $159.99

All of these cards will carry the same range of connectors found on the 5800 series, which means the 2x dual-link DVI, single HDMI and single Displayport configuration. Apparently, the HD 5770 will carry a cooler similar to that found on 5800 series cards as well, essentially giving it the appearance of a reduced size HD 5870. We have seen some pictures and performance numbers for the HD 5750 already (which I believe was the 512MB model). At the prices suggested here it would certainly be a reasonably good deal, and if the HD 5770 can improve on those numbers by a good margin, it should be an attractive proposition too.

NVIDIA Releases new GeForce 191.07 WHQL Drivers

Category: Video Cards
Posted: October 5, 2009 07:58AM
Author: redtigerdragon

NVIDIA just released new drivers for their cards (here for Windows 7/Vista 32-bit, here for 7/Vista 64-bit, here for 32-bit XP, and here for 64-bit XP), and they come with some improvements listed below. For all of the improvements and bug fixes, see the release notes (here for Windows 7, here for Vista, and here for XP).

Adds support for OpenGL 3.2 for GeForce 8, 9, 100, 200-series and ION GPUs.

Adds SLI support for Aion, Darkfall, Dawn of Magic 2: Time of Shadows, Dreamkiller, Fuel, Need for Speed: Shift and more.

Includes the following key fix: for graphics cards supporting multiple clock states, 3D clocks correctly return to 2D clocks after exiting a 3D application.

Accelerates performance in several games, such as (comparison between 191.07 and 190.62 drivers, results may vary):

o Up to 50% performance increase in Call of Juarez: Blood in Bound with SLI enabled
o Up to 34% performance increase in Prototype with antialiasing enabled
o Up to 14% performance increase in Fallout 3 (indoor scenes) with antialiasing enabled
o Up to 12% performance increase in ARMA 2
o Up to 10% performance increase in Far Cry 2 (DX9 version) with antialiasing enabled
o Up to 8% performance increase in Batman: Arkham Asylum with GPU PhysX enabled

 

NVIDIA Confirms Fermi Card was a Mock-up, but Demos were Real

Category: Video Cards
Posted: October 3, 2009 02:55PM
Author: Ben Grantham

After NVIDIA announced its new "Fermi" architecture this week, with CEO Jen-Hsun Huang holding up an example card during his keynote presentation, some began to wonder about the authenticity of that display model. Charlie Demerjian at SemiAccurate posted an article on Thursday, which used pictures from PC Watch to highlight suspicions that the card show on stage couldn't possibly have been a working Fermi board. A number of things about it just didn't look right, with components that obviously didn't line up with the PCB traces and the PCB itself being apparently cut down. NVIDIA have now confirmed that the card shown on stage was indeed a mock up model, though it also stresses that the demos shown during the conference claimed to be running on Fermi hardware were in fact doing just that. Apparently the real boards weren't in the kind of state that it wanted to be showing off, so we will have to wait for product shots once the designs have been finalised.

The question a number of people have therefore been asking is whether showing a mock up on stage without making it explicitly clear was a deliberately misleading move. We all know that in the GPU market, neither of the big two companies are above taking pots shots at the other to try and gain an advantage (even in purely marketing terms). From one perspective, mock ups of products certainly aren't unusual in presentations and NVIDIA would obviously like to give everyone present something representative of a final product to take a look at. On the other hand, the presentation of the card may have given the impression (intended or not) that development of the product was further along than it actually is. With the latest AMD GPUs already out there in the market, giving consumers a reason to hang on before making a purchase is desirable. I'll let you decide for yourselves on what NIVIDIA's intentions were, but either way, the GPU industry continues to be as exciting (or infuriating, depending on your outlook) as ever.

Radeon HD 5750 Pictures, Benchmarks

Category: Video Cards
Posted: October 2, 2009 08:01PM
Author: Ben Grantham

Apparently, a member of Chinese forum mymypc.com has had the opportunity to take a look at a HD 5750 card, as well as run a few benchmarks. If this all turns out to be correct, the HD 5750 board will be considerably smaller  than the recently released 5800 series cards (possibly even a small as some current lower end models), with 1120 stream processors and 56 texture units (TMU). Core and memory clocks come in at 700MHz and 1150MHz respectively. What is nice to see is the same 4 display ouputs that feature on the 5800 series cards. Performance wise, we only have 3DMark 06 and Vantage scores to go by, which put the 5750 in somewhere slightly above GTS250 and 4770, at 12931 for 3DMark 06 (with an E8400) and Vantage GPU score of 6881. Naturally driver versions are going to be pretty early at this stage.

Pricing is expected to be around the $149 mark for the HD 5750, with the HD 5770 around $199. It is also rumored that release dates for the 5700 series may have been brought forward, which would see them possibly appearing as early as October 12th. You can check out all of the pictures on the mymypc.com forums. Credit goes to VR-Zone for digging this up.

Oak Ridge National Laboratory to Use Fermi For Supercomputer

Category: Video Cards
Posted: October 2, 2009 01:41PM
Author: Nick Harezga

The Oak Ridge National Laboratory will be putting the recently announced Fermi GPU architecture by Nvidia to good use, announcing its intent to build a new supercomputer. The new supercomputer is estimated to be 10 times faster than the current fastest supercomputer. The Fermi based machined will be used for research in energy and climate change. Nvidia has been breaking into the scientific computing market with its Tesla based CUDA GPUs, a market previously dominated by CPUs. The computing power that will be available with GPU supercomputers is impressive already, with performance undoubtedly growing in the future.

NVIDIA: GT300 Will Launch in 2009; Beats 5870

Category: Video Cards
Posted: October 2, 2009 01:46AM
Author: Daryn Govender

NVIDIA revealed the specifications of its new Fermi architecture a few days ago leaving some people wanting more information, such as when the first Fermi cards will be available and how they stack up with ATI's latest offerings. Today we get partial answers to both of those questions.

An unnamed senior employee at NVIDIA has confirmed that the long-awaited GT300 will ship by the end of this year, according to Fudzilla. This means we have a definite time-limit on when us, the consumer should be able to get our hands on NVIDIA's next-gen cards although there is no information available on an exact launch date.

It has also been confirmed that Drew Henry, the General Manager of the GeForce and Ion product lines has said that the Fermi range will be faster than any video card available including the ATI Radeon 5870. This is not surprising based on the Fermi's specifiations on paper. Aditionally, NVIDIA has stated that it has a Fermi GX2 variant in the works and that the single and dual card TDP should be similar to that of the single-core GTX285 and dual-core GTX295 respectively.

It seems that NVIDIA has the upper hand over its red competitors in light of these new scraps of information and hopefully the red faction is concocting a counter-attack in terms of performance.

More Lucid HYDRA Talk, Still no Performance Figures

Category: Video Cards
Posted: October 1, 2009 04:48PM
Author: Ben Grantham

If you pay attention to our daily hardware roundup (and I'm sure you do), then you have probably already read Legit Reviews look at Lucid's HYDRA 200 chip, as well as the discussions they had with AMD and NVIDIA about the technology. Looking at the demo systems Lucid had on display both at IDF and behind closed doors, it is clear that HYDRA works largely as claimed, at least from a basic functionality stand point. Using MSI Big Bang motherboards, it was shown running NVIDIA and AMD cards together without need for bridge connectors. You will need to run Windows 7 to use cards from the different GPU vendors together though, as it is the only OS that allows the necessary drivers to be installed alongside each other. Being able to select the card which outputs to the display from the desktop is handily straight forward however.

When it comes to performance when compared with SLI or Crossfire, Lucid are still being very cagey, stating that they are unwilling to release numbers at this time. The furthest they were prepared to go was to say they expected "competitive performance". The reaction from AMD and NVIDIA is also somewhat predictable, defending their own technologies and generally towing the company line. They did however both say that if HYDRA delivers a substantial improvement for gamers then the industry as a whole would win (high performance and more straight forward multi-GPU gaming would likely lead to increased GPU sales). Take a look at the full article if you haven't already. Unfortunately, it looks like we will have to wait for motherboards featuring HYRDA to be released before we get some real world performance figures.

AMD Teams up with Pixelux for Open Source Physics

Category: Video Cards
Posted: September 30, 2009 05:01PM
Author: Ben Grantham

AMD has announced that it will be teaming up with Pixelux Entertainment, with the intent of making much wider use of the open source Bullet Physics engine. As we know, NVIDIA has attempted to strongly promote its own PhysX physics implementation, but hasn't seen widespread acceptance (at least on PCs) due to the fact you need to use an NVIDIA GPU to enable it. AMD has generally supported Intel-owned Havok physics, though as with PhysX we haven't seen it make any real impact on the games we play. The Bullet Physics engine currently lies third in the popularity stakes, behind both PhysX and Havok. AMD is hoping that encouraging the development of middleware built around Bullet Physics, OpenCL and DX11 will lead towards physics that can be used on PCs, games consoles and other hardware. Let's hope this actually goes somewhere meaningful, as we are going to need an open physics standard for gamers to really benefit independent of their hardware choices.

Premier Fermi Video Card Pictured

Category: Video Cards
Posted: September 30, 2009 03:07PM
Author: Daryn Govender

NVIDIA CEO Jen-Hsun Huang has followed up the release of the Fermi architecture specifications with the premier (first) glimpse of a Fermi-based video card at his opening keynote at the GPU Technology Conference in San Jose, California. The card he held was a Tesla card, intended for scientific use but I'd expect the GeForce series (and Quadro series) to look similar. The only feature which can be seen on the card is what looks like an 8-pin power connector. It looks as though NVIDIA is planning to tease consumers up to a big reveal to try and take their attention away from the recently released ATI 5000 Series. Now we all have to endure an agonizing wait for real-world performance and benchmark figures!

Update: According to bit-tech, there is also a 6-pin power connector on the board itself.

A special thanks to bit-tech.net for the pictures of the Tesla card.

NVIDIA GT300 'Fermi' Specifications Detailed

Category: Video Cards
Posted: September 30, 2009 02:48PM
Author: Daryn Govender

As most of you already know, the GT300 was first unveiled months ago along with some preliminary specifications. Today we get the bigger picture, thanks to bsn. NVIDIA's current next-gen GPU is known as the GT300 or N70. It is now known that the GPU architecture is called "Fermi" by NVIDIA engineers. A quick history lesson: the man behind the name is an Italian physicist, Enrico Fermi who is known for helping develop the first nuclear reactor. This may be why the GT300 board is also known as "reactor". Now, to the important details.

GT300 "Fermi" Architecture GPU Specifications

• 3.0 billion transistors
• 40nm GPU by TSMC
• 384-bit memory interface (6x64-bit memory controllers)
• 512 shader cores (renamed to CUDA Cores)
• 32 CUDA cores per shader cluster
• 1MB L1 cache memory [divided into 16KB Cache - Shared Memory]
• 768KB L2 unified cache memory
• Up to 6GB GDDR5 memory (1.5GB for GeForce and up to 6GB for Quadro/Tesla)
• Half Speed IEEE 754 Double Precision
• 16 Streaming Multiprocessors (new name for the former Shader Cluster) containing 32 cores each

The Fermi architecture also supports C (CUDA), C++, DirectCompute, Fortran, OpenCL, OpenGL 3.1 and OpenGL 3.2. This means that for the first time in history a GPU can run C++ code! NVIDIA's "Fermi" architecture will be released in three tiers, GeForce for high-end consumers, Quadro for commersial use and Tesla for scientific purposes. There is still no word on pricing or availablility, but I'd expect the new Fermi cards to break the bank a bit.

These revelations are huge for NVIDIA and the enthusiast community, now we need to see one in action!

Lucid Discusses HYDRA Multi-GPU Advantages

Category: Video Cards
Posted: September 28, 2009 03:58PM
Author: Ben Grantham

A couple of days after Lucid officially announced its HYDRA 200 chip, bsn* talked to Offir Remez (President and VP of Business Development at Lucid) about the technology and the approach it takes to multi-GPU rendering. Offir describes how the development of the technology essentially came from frustration of the balance of power constantly shifting between GPU manufacturers and how this meant frequent system changes to keep up as a gamer. Having an option that could be used with any GPU and approached the problem of rendering from a computational perspective rather than on a frame-by-frame basis seemed the logical way to go. Unlike implementations of SLi and Crossfire technologies, Hydra uses a real-time distribution engine that resolves the problem of frame dependency, which means no more AFR modes. This is apparently what has held back the scaling of multi-GPU rendering performance up to now.

Lucid has been working with the "majority of Top 5 motherboard makers", so we should be seeing the HYDRA 200 chip shipping on boards from at least a few manufacturers in Q4 this year. The board from MSI that we have mentioned previously is due on October 29th.

AMD Demos Dual-GPU 'Hemlock' Graphics Card

Category: Video Cards
Posted: September 24, 2009 01:58PM
Author: Ben Grantham

It looks as if AMD is keen to show that everything is on track with regard to its GPU release schedule. After officially launching brand new 5800 series cards this week, the guys over at VR-Zone were treated to a demo of  system running a dual "Cypress" GPU card, which they suspect was the HD 5850 X2. Displaying a DX11 tech demo of Alien vs Predator (a title due for release next year) at reportedly "encouraging" frame rates, the card comes in at an understandably longer length than the HD 5870. VR-Zone estimated it at around 12", so it's unlikely you'll be shoehorning this into anything but spacious cases. It was the fact that the card came equipped with 1GB of GDDR5 memory that lead to the conclusion it was likely a HD 5850 X2 model (you would expect a HD 5870 X2 to feature double this), with the core being clocked at 725MHz and the memory at 1GHz. Hemlock graphics cards are expected late this year, and should be coming in at a price below $500.

Intel Demos Its Own GPU

Category: Video Cards
Posted: September 23, 2009 06:57AM
Author: Nick Harezga

Intel demoed a functional version of its upcoming GPU, code named Larrabee, on Monday at the Intel Developer Forum. The Larrabee chip was used in a Gulftown system to render ray tracing environments. It was also revealed that Intel plans to build its own GPU based on Larrabee to compete with Nvidia and ATI in the future. However, Intel feels that Larrabee will really shine when used in hand held devices such as cell phones. Such applications would seem to indicate that Larrabee won't be extremely power hungry, an interesting change from current high end video cards.

ATI Radeon HD 5800 Series Officially Released

Category: Video Cards, General News
Posted: September 23, 2009 04:28AM
Author: MrAlex

Finally the long awaited ATI Radeon™ HD 5800 series cards have been released, fully supporting Microsoft’s DirectX® 11. These new cards utilize the 40nm fabrication process and pack double the amount of transistors found in the 4800 series, coming in with 2.15 billion of them. Both the 5870 and 5850 come with a GDDR5 memory interface, a PCI Express 2.1 x16 bus interface, DirectX® 11 support, OpenGL 3.2 support and ATI Eyefinity multi-display technology as well as a few more extras. The 5870 comes in as ATI’s fastest graphics processor, with 2.72 TeraFLOPS of processing power, 1600 Stream Processing Units, 80 Texture Units and 32 ROPs with a peak bandwidth of 153.6 GB/sec while the 5850 which comes in slightly less powerful with 2.09 TeraFLOPS of processing power, 1440 Stream Processing Units, 72 Texture Units and 32 ROPs with a peak bandwidth of 128.0 GB/sec. The full specifications for the 5870 can be found here and for the 5850 here. You can expect cards to be coming from ASUS, Club 3D, Diamond Multimedia, Force3D, GIGABYTE, HIS (Hightech Information Systems), MSI, Multimedia, PowerColor, SAPPHIRE Technology, VisionTek and XFX.

Hardware Roundup: AMD Radeon HD 5870 Edition

Category: Video Cards
Posted: September 23, 2009 02:31AM
Author: Dale Shuck

It's been talked about in the OCC forums with a great deal of speculation and anticipation and finally the release date for the newest series of cards from ATI is here with the unveiling of the Radeon HD 5870. Based on the RV870 core and produced using a 40nm manufacturing process, the HD 5870 features a core clock speed of 850MHz and 1200MHz on the 1GB frame buffer. There's a massive 2.15 billion transistors packed into this GPU that features 1600 stream processors and 2.72TFlops of compute performance. Of course, this is also the first DirectX 11 capable card out of the gate as well. There's been a lot of rumors and leaks about the card's supposed performance numbers so let's quit fooling around and check out the reviews using the links below!

Sapphire HD 5870 @ OCC
AMD's Radeon HD 5870 graphics processor @ Tech Report
ATI Radeon HD 5870 1GB Graphics Card and AMD Eyefinity @ PC Perspective
AMD's ATI Radeon HD 5870 Video Card @ [H]ardOCP
ATI Radeon HD 5870 DX11 Video Card @ Legit Reviews
AMD ATI Radeon HD 5870 DirectX 11 GPU @ HotHardware
Sapphire Radeon HD 5870 1GB Graphics Card @ TweakTown
ATI Radeon HD 5870 @ TechSpot
The AMD HD5870 - the fastest single-GPU card available? @ Bjorn3D

PowerColor to Include Special Character With Battle Forge Edition HD4890 Cards

Category: Video Cards, Gaming
Posted: September 21, 2009 04:04PM
Author: Dale Shuck

PowerColor has developed a pair of HD 4890 videocards featuring a Battleforge theme that include a full version of the game along some something extra - a Fallen Skyelf rare card inside the box. Battleforge is a card-based real-time strategy game where gamers can assemble their own army with collectible trading cards.

PowerColor offers two models of the HD4890 Battle Forge Edition card - the HD4890 PCS+ overclocked version (950MHz core/1100MHz(4.4Gbps) memory) and the standard HD 4890 version (850MHz core/975MHz(3.9Gbps) memory). Both versions come with the game which includes a standard set of cards, but PowerColor has added the Fallen Skyelf special card that has the ability to make the target more susceptible to damage, making it a good way to detonate an unwary enemy’s Unstable Demon. With full DirectX 10.1, Battleforge offers superior image quality and cinematic gaming at HD resolutions which can be enjoyed fully with the HD 4890 cards.

NVIDIA CEO to Talk with TSMC and Partners next Month

Category: Video Cards
Posted: September 21, 2009 02:57AM
Author: Daryn Govender

NVIDIA CEO Jen-Hsun Huang is going on a trip to Taiwan next month to meet TSMC chairman Morris Chang and NVIDIA's AIB and motherboard makers. It is speculated that most of the talks will involve the highly-anticipated GT300 GPU which is expected to launch by the end of the year, in particular the full-scale production of 40nm GPUs. Jen-Hsun will also check up on the current yield of 40nm wafer which is up 30-60% in July according to TSMC. Most sources agree that the GT300 will launch in November or December with general availability to consumers by Q1 2010. I just hope that Jen-Hsun Huang would hurry up with his discussions so all of us can finally see what NVIDIA's offerings are against the upcoming ATI 5000 series.