Just a little later this week, NVIDIA will unveil the GeForce GTX 550 Ti mainstream graphics processor (GPU), successor to GeForce GTS 450. NVIDIA partners seem to be ready with their products based on the new GPU. One of NVIDIA's largest partners, Galaxy, is even ready with a highly overclocked model with its core running at 1 GHz (900 MHz reference), and 1150 MHz (1025 MHz reference) memory speed. Galaxy's card will use the HOF (Hall of Fame) identifier, which is supposed to indicate the company's highest level of factory-overclock speeds.

Based on the 40 nm GF116 silicon, the GeForce GTX 550 Ti packs 192 CUDA cores, and a 192-bit wide GDDR5 memory interface, holding 1 GB of memory, that's right, 1 GB, not 768 MB or 1536 MB. It looks like NVIDIA will be doing a mixed density memory chip setup. We can hypothesize that four of the six chips are 1 Gbit (128 MB), amounting for 512 MB; and two 2 Gbit (256 MB), amounting for the other 512 MB. Galaxy's HOF card will use a 3+1 phase VRM to power the card, and feature a dual-slot cooler. Its display connectivity includes one each of DVI, HDMI, and DisplayPort.

AMD today announced the launch of the fastest graphics card in the world, the AMD Radeon HD 6990, packing so much raw performance it delivered a new single graphics card world record score of P11865 in the industry standard 3DMark11 benchmark. AMD Radeon HD 6990 graphics cards are available immediately from retailers worldwide, with select models starting at US $699 MSRP.

Designed for the power gamer, AMD Radeon HD 6990 graphics cards include AMD's second-generation, Microsoft DirectX 11-capable architecture, advanced image quality features for the best looking games, and AMD PowerTune intelligent technology to enable higher clock speeds and faster gaming. The AMD Radeon HD 6990 offers the ultimate in immersive HD gaming and computing with native support for up to five displays using AMD Eyefinity multi-display technology, with six display support possible via DisplayPort 1.2 components available for purchase later this year. The AMD Radeon HD 6990 graphics cards also feature a dual-BIOS toggle switch, allowing enthusiast gamers to unlock the card and rev up the clocks and voltages for even more performance. Finally, AMD HD3D technology provides a cutting-edge stereoscopic 3D gaming and movie experience, thanks to HDMI 1.4a support.

Sparkle computer's CeBIT bit booth featured an eye-catching new graphics card, the Sparkle One GTX 570. This is the first air-cooled single-slot GeForce GTX 570 series graphics card. While it is thin, the card is extremely long. We estimate around 13-inches long, if not more. Sparkle seems to be making use of the extra length to squeeze in as much metal (heatsink) as possible. A fan pushes air backwards. The card features clock speeds of 752/1504/4000 MHz (core/shader/memory).

Like every other GTX 570, it's based on the 40 nm GF110 GPU with 480 CUDA cores enabled, and 1280 MB of GDDR5 memory across a 320-bit wide memory interface. Display outputs include one each of DVI, HDMI, and DisplayPort. Sparkle expects the card to be released by next month.

While it is slated for release on the 8th of this month, AMD's Radeon HD 6990 was detailed to sections of the industry. It wasn't long before someone leaked the press deck, revealing all the specifications and AMD performance estimates. The Radeon HD 6990 is a dual-GPU graphics card making use of two AMD Cayman GPUs in internal CrossFire. Cayman is the same GPU that is at the center of Radeon HD 6950 and HD 6970.

In the HD 6990, each Cayman is configured to use all 1536 of its VLIW4 stream processors. The GPU core is clocked at 830 MHz, and memory at 1250 MHz (5.00 GHz GDDR5 effective). The 256-bit wide memory interface of each GPU is populated with 2 GB of memory, setting the total board memory to 4 GB. Display IO includes five connectors, including one DVI and four mini-DisplayPort 1.2. While the HD 6990 is said to outperform GeForce GTX 580, AMD is also bracing itself for NVIDIA's dual-GPU GTX 590, with HD 6990 OC variants following an official specification of 880 MHz core, and the same 5.00 GHz memory.

One of the first partner-branded NVIDIA GeForce GTX 550 Ti graphics cards surfaced, listed on a European online store, priced at a little over €159. It is carrying PNY's part number GMGX55TN2H1IZPB, and doesn't carry the XLR8 branding. Based on the 40 nm GF116 silicon, the PNY GeForce GTX 550 Ti packs 192 CUDA cores, and clock speeds of 900/1800/1025 MHz (core/CUDA cores/memory actual). It packs 1 GB of GDDR5 memory across a 128-bit wide memory interface. €159 does look like a steep price, but there's a lot of uncertainity surrounding it.

MSI is ready with a new Radeon HD 6950 2 GB graphics card that makes use of the company's new Twin Frozr III cooling assembly. The R6950 Twin Frozr III Power Edition from MSI makes use of a custom-design PCB with strong, overclocker-friendly VRM circuitry, out of the box overclock speeds of 850 MHz core and 1300 MHz (5200 MHz effective) GDDR5 memory speeds, against reference speeds of 800/1250 MHz. The Twin Frozr III cooler makes use of a dense aluminum fin array to which heat is conveyed by a number of nickel-plated copper heat pipes, and is ventilated by two 80 mm fans that use "propeller-design" fan blades that deliver higher air-flow at lower noise outputs.

Based on the 40 nm Cayman silicon, Radeon HD 6950 packs 1408 VLIW4 stream processors, is DirectX 11 compliant, and makes use of a high-speed GDDR5 memory controller with 256-bit bus width. Display output connectors include two DVI, two mini-DisplayPort 1.2, and one HDMI 1.4a. AMD-exclusive technologies such as Eyefinity and HydraVision are supported. The card can pair with up to three more Radeon HD 6900 series GPUs for 4-way CrossFireX multi-GPU operation. MSI's new Radeon HD 6950 2 GB Twin Frozr III Power Edition graphics card is priced at around 300€.

Gigabyte released a new GeForce GTX 560 Ti Super Overclock graphics card model. The GV-N560SO-1GI-950 from Gigabyte features an out of the box core clock speed of 950 MHz, 1900 MHz CUDA core speed, and 1145 MHz (4580 MHz effective) GDDR5 memory speed. The new model is slightly slower than the GV-N560SO-1GI the company released last month, which has clock speeds of 1000/2000/1145 MHz. The GTX 560 Ti SOC from Gigabyte is based on a custom design PCB that uses Ultra Durable VGA+ high grade components, and WindForce 2X cooler.

The WindForce 2X cooler makes use of a large aluminum fin array to which heat is conveyed by a number of copper heat pipes that make direct contact with the GPU. The heatsink is ventilated by two 80 fans. Like every other GeForce GTX 560 Ti, the DirectX 11 compliant GPU packs 384 CUDA cores, and makes use of 1 GB of GDDR5 memory over a 256-bit wide memory interface. Display outputs include two DVI and one mini-HDMI. Gigabyte did not give out pricing, but one can expect it to be lower than that of the 1000 MHz SOC variant.

GIGABYTE, a leading manufacturer of motherboards and graphics cards, introduces a powerful new addition to its Super Overclock Series: GTX 580 SOC (GV-N580SO-15I). To qualify for Super Overclock Series, a GPU must go through GIGABYTE's innovative GPU Gauntlet Sorting process and it must be equipped with Ultra Durable VGA or VGA+ boards. Designed specifically for overclocking competition and gaming battle, GIGABYTE's secret weapon GTX 580 Super Overclock Edition is built with NVIDIA GeForce GTX 580 GPU.

Based on 40nm processor and advanced GDDR5 memory technology, the powerful new GPU consists of ground-breaking features including Microsoft DirectX 11 and NVIDIA PhysX which lead to incredibly immersive HD gaming experience and unrivaled performance. To satisfy hardcore gamers and extreme overclockers, GV-N580SO-15I is especially designed with GIGABYTE Extreme Dual BIOS technology. GV-N580SO-15I also features WINDFORCE 3X, 14 phase PWM design, OC Guru graphic card smart tuner and voltage read point.

NVIDIA is reportedly working on a new mainstream GPU for a mid-March launch. The graphics giant is planning to name it GeForce GTX 550 Ti, that's right, "GTX" prefix and "Ti" suffix monikers being extended to the core mainstream, in place of the typical "GTS". The new GPU will be based on the 40 nanometer GF116 silicon. Its exact specifications are not known, except that the GPU will use a 128-bit wide GDDR5 memory interface, and will be pin-compatible with GF106, on which is based the GeForce GTS 450. Its performance is estimated to be up to 35% higher than ATI Radeon HD 5770. Due to an electrical redesign over GF106, the GTX 550 Ti is expected to have a TDP of just 110W, nearly the same as that of the GTS 450, except having higher performance. The NVIDIA GeForce GTX 550 Ti is slated for release on 15 March, 2011.

Club3D announced its Radeon HD 6950 1 GB graphics card (model no.: CGAX-69524F). Making use of a completely in-house design by Club3D, the card uses a red PCB that is 10.5 inches long, and a custom-design cooling assembly that uses a large aluminum fin heatsink to cool the GPU, ventilated by two 80 mm spinners. The card uses 1 GB of memory over a 256-bit wide GDDR5 memory interface. It uses AMD-reference clock speeds of 800 MHz core and 1250 MHz (5.00 GHz effective) memory. Like every other HD 6950, it features 1408 stream processors. Display outputs are standard, two each of DVI and mini-DisplayPort 1.2, one HDMI 1.4a. Club3D did not give out pricing information.

Leading graphics card supplier, SAPPHIRE Technology has just launched a new TV tuner card with a difference - the SAPPHIRE HD 5570 XtendTV card. The SAPPHIRE HD 5570 XtendTV card is a single slot PCI-Express graphics card with 1GB of GDDR5 memory and HDMI and DVI outputs - and a built in programmable TV tuner. Connected to a standard TV antenna signal, it will receive and decode all of the local free to air TV and radio stations*. Just as with a regular TV tuner, the user can watch TV on the host PC using Media Center - record, save and playback programs at will.

But XtendTV does more. Installing the FlexiStream software on the PC in which the card is installed enables the PC as a streaming server and users can watch all of their home TV programmes anywhere in the world using the XtendTV client on a remote PC or laptop.

After the launch of NVIDIA's GeForce GTX 560 Ti, AMD made it easier for partners to release overclocked and custom-design Radeon HD 6870 graphics cards. Partners seem to be responding to this. HIS released two new custom-design Radeon HD 6870 graphics card models, based on the Barts IceQ X board design the company implemented on its HD 6850 IceQ X series. Like with the HD 6850, there's the base model that sticks to AMD reference clock speeds of 900/1050 MHz (core/memory), and a Turbo-X model that uses overclocked speeds of 975/1150 MHz. The "X" in Turbo-X denotes that it's faster than the Turbo model based on AMD's reference design board, by HIS. The cooler makes use of a slightly large aluminum fin array to which heat is conveyed by 8 mm thick nickel-plated copper heat pipes, which is centrally ventilated by a 92 mm fan. The cooler is covered by a frosty-looking shroud. Like with every other HD 6870, the card features 1120 stream processors, and connects to 1 GB of GDDR5 memory over a 256-bit wide memory interface. HIS did not give out pricing details.

Galaxy's first wave of graphics cards featuring NVIDIA GeForce GTX 560 Ti graphics processor includes at least two non-reference designs. The first one, the GC Version, makes use of Galaxy's publicized "Gemini" GPU cooler, which uses two large 90 mm fans to cool a large aluminum heatsink which is fed by four copper heat pipes. The GC Gemini features slightly upped clock speeds of 835/1670/1000 MHz (core/CUDA core/memory actual).

The second card is the Galaxy GTX 560 Ti White Edition, which features a white PCB, possible digital PWM circuitry, a full-coverage single fan cooling assembly, and high overclock speeds of 950 MHz core, and 1100 MHz memory. Both cards feature Galaxy's typical spacecraft-styled cooler shrouds, connectivity that includes two DVI and a mini-HDMI. While the GC Version draws power from two 6-pin connectors, the White Edition requires 6-pin plus 8-pin. Based on the 40 nm GF114 GPU, GeForce GTX 560 Ti features second-generation Fermi architecture, and is powered by 384 CUDA cores, and aided by 1 GB of GDDR5 memory over a 256-bit wide memory interface. The GTX 560 Ti is slated for release on the 25th of this month.

What happens in Gainward, stays in Palit. Gainward's parent company, Palit, is also readying a GeForce GTX 580 graphics card with 3 GB memory, like its subsidiary. Featuring a non-reference PCB and cooling assembly, Palit's latest card uses twelve 2 Gbit GDDR5 memory chips over a 384-bit wide memory interface to achieve 3072 MB (3 GB) in total. It features a more traditional-looking cooler compared to the Gainward GTX 580 Phantom; which makes use of two fans to cool a large aluminum fin array. Palit's card is clocked at 783 MHz (core), 1566 MHz (CUDA cores), and 1005/4020 MHz (memory actual/effective). Based on the 40 nm GF110 core, the GTX 580 features 512 CUDA cores, and is compliant with the latest in consumer graphics technologies.

Gigabyte is ready with a Super Overclock (SOC) variant of its upcoming GeForce GTX 560Ti graphics card, already. The GV-N560SO-1GI clocks at an impressive 1000 MHz core (vs. 822 MHz reference), and 1145 MHz (actual) or 4580 MHz (effective GDDR5) memory speed, with a CUDA core clock speed of 2000 MHz. Apart from these, the card features 384 CUDA cores, and 1 GB of memory over a 256-bit wide GDDR5 memory interface.

The GV-N560SO-1GI uses Gigabyte's Ultra Durable VGA+ PCB, which is rich in copper and uses top-grade components. It is cooled by a larger version of WindForce 2X, which the company earlier used on older high-end GPUs. Gigabyte also run its own performance tests on the card. The card yielded X11515 points in 3DMark Vantage Extreme preset, and P24590 in Performance preset. The card is said to be at least made public, if not released to market, on the 25th of this month.

One of MSI's first graphics cards based on NVIDIA's upcoming GeForce GTX 560 Ti GPU got listed on a European store. The N560GTX-Ti Twin Frozr II OC from MSI features the company's iconic Twin Frozr II GPU cooler, paired with the GTX 560 Ti GPU, overclocked out of the box at 880 MHz core and 1050 MHz (4.20 GHz GDDR5 effective) memory, against reference speeds of 820 MHz core and 1000 MHz (4.00 GHz) reference.

The Twin Frozr II, used on several MSI graphics card models till date, uses a dense aluminum fin array to which heat is conveyed by 8 mm thick heatpipes, ventilated by two 80 mm fans. MSI claims this cooler to keep the GPU up to 20°C cooler compared to NVIDIA reference cooler. It is currently listed at €280.33, but one can take this pricing with a pinch of salt, as pre-release prices tend to be a little higher.

Priced point of US $279, AMD's new 1 GB Radeon HD 6950 stands testament to the competitiveness AMD left for itself in the performance segment, a crucial market segment. AMD's Radeon HD 6950 with 1 GB of memory will be launched by partners soon, along with 1 GB Radeon HD 6970. The two will compete with NVIDIA's GeForce GTX 560Ti which is slated for 25th of this month.

The new Radeon HD 6950 was able to shed $21 of its price by halving its memory amount. The original 2 GB version sells for an average $300. The 1 GB cards use eight 1 Gbit memory chips over a 256-bit wide GDDR5 memory interface. Memory clock speeds and bandwidth remain unchanged. Apart from memory amount, partners will most likely use cost-effective board and cooler designs. Apart from the HD 6950, AMD will also introduce "Turbo" variants of the Radeon HD 6870, which will feature very high GPU core clock speeds.

HIS unveiled one of its first Radeon HD 6950 1 GB graphics cards. The HIS HD 6950 1 GB Fan Edition features AMD's high-end GPU with 1 GB of memory, and a cost-effective design that makes it competitive with upcoming GeForce GTX 560 Ti. HIS used its own design PCB that's shorter in length, and uses cost-effective VRM since the card isn't dealing with high-density 2 Gbit GDDR5 memory chips anymore. It uses a GPU fan-heatsink that uses a large fan to ventilate an aluminum fin array that cools the GPU and other components. the cooler is covered by a well-ventilated shroud.

That and the memory aside, other specifications are the same: the 40 nm Cayman GPU features 1408 VLIW4 stream processors, and connects to 1 GB of GDDR5 memory over a 256-bit wide memory interface. The core is clocked at 800 MHz, and the memory at 1250 MHz (5.00 GHz GDDR5 effective). HIS did not release pricing yet, but it will be well below the $300 mark.

Gigabyte released a new overclocked non-reference design GeForce GTX 570 graphics that makes use of its WindForce3X custom cooler. The GV-N570OC-13I from Gigabyte uses clock speeds of 780 MHz core and 1560 MHz CUDA cores, against reference speeds of 732 and 1464 MHz, respectively. The WindForce3X cooler keeps the beast cool. It makes use of a dense aluminum fin array to which heat is conveyed by a number of copper heat pipes, and ventilated by three 80 mm fans. The blue Ultra Durable VGA PCB by gigabyte makes use of top-quality components and 2 oz copper layer board. Other specifications are standard: 480 CUDA cores, and 1280 MB of memory across a 320-bit GDDR5 memory interface. Gigabyte did not give out pricing information.

SPARKLE Computer Co., Ltd., the professional VGA card manufacturer and supplier, today unveils SPARKLE GeForce GTS 450 Low Profile Graphics Card, an ideal product for home theater enthusiasts and gamers. Exclusively on the SPARKLE GeForce GTS 450 Low Profile Graphics Card, users get the best cooling performance with Axial Fan design.

This design features dual axial fans which bring greater amount of wind volume. The unique airflow guide technology can bring the heat generated by GPU to the outside smoothly. Thanks to the axial fan design, now users can get lower working noise from the SPARKLE GeForce GTS 450 Low Profile Graphics Card, making it a ideal products for home theater enthusiasts who seek quiet operation environment.

ASUS displayed a huge new graphics card based on the AMD Radeon HD 6970 GPU. The EAH6970 DirectCu is perhaps the largest among the company's graphics cards that use a "direct-copper" GPU cooler in which heat pipes make direct contact with the GPU die. Such is the size of the cooler, that the card requires three expansion slots in your system. To make optimal use of the expansion brackets, ASUS wired out all TMDS links from the GPU, yielding a display connectivity consisting of two DVI and four full-sized DisplayPort 1.2 connectors.

The GPU cooler makes use of a large aluminum fin array which uses at least three 8 mm thick copper heat pipes. The aluminum fins are cooled by the airflow of two 120 mm fans. The card uses 2 GB of high-speed GDDR5 memory over a 256-bit wide memory interface, it packs 1536 VLIW4 stream processors. The card also seems to be using a 9-phase VRM that allows software voltage control. Power is drawn in from two 8-pin PCI-E power connectors.

After the successful launch of the World's First Independent Produced GeForce GTX 580 V-Go Graphics Card, SPARKLE Computer Co., Ltd., the professional VGA card manufacturer and supplier, today announced the GeForce GTX 570 V-Go Graphics Card, becoming the world's first manufacturer who can independently produce high-end graphics cards using Fermi 2.0 architecture. "V-Go, the series is represent the independent produced graphics cards, inherit the expertise spirit on graphics cards manufacturing for 16 years" said Kevin Wang the General Manager of SPARKLE ."We are pleased with the new high level of SPARKLE R&D and production capabilities. As another masterpiece of V-Go series, GeForce GTX 570 V-Go Graphics Cards inherit the blood of Fermi 2.0 Architecture also has a better price advantage."

First reported in October, Galaxy's GeForce GTX 460 HOF (Hall of Fame) edition graphics card seems to have finally taken shape, with the company giving out pictures of the card with its cooling solution installed. The card sports a milky-white colored PCB that makes use of 1 Gigabit memory chips and digital PWM power circuitry that makes use of proadlizers to condition power.

Its cooling solution makes use of an aluminum fin array to which heat is conveyed using four copper heat pipes, cooled by a single 80~90 mm fan. The card makes use of a 6+2 phase digital PWM power circuit that allows software voltage control. Out of the box, the card is clocked at 850/1700/1000 MHz (4 GHz GDDR5 effective), core/CUDA cores/memory. More pictures can be found at the source.

TUL Corporation, a leading manufacturer of AMD graphics cards, today launches the latest enthusiast graphics solution: the PowerColor HD6900 series. The PowerColor HD6900 series is the most advanced architecture to date. Consuming only 190W under the most demanding game titles and increasing performance at any given power by taking advantage of AMD PowerTune technology, the latest series enhances anti-aliasing ability to deliver the best image quality in minimal performance change.

PowerColor's HD6900 series includes the HD6970 and HD6950. The HD6970 clocks in at 880MHz core speed and 1375MHz of memory speed, while the HD6950 clocks in at 800MHz and 1250MHz GPU and memory clock speeds. Both are equipped with the latest 2GB GDDR5 memory buffer, offering more than 2.25 teraFLOPs of computing power; perfect for massive multi-monitor high-res gaming.

ASUS rolled out its second AMD Radeon HD 6870 graphics card, and its first non-reference design one: the HD 6870 DirectCu. The card features two features that make it supposedly better than the reference design, the Direct Cu GPU cooler, and a refined PCB design that make use of high-grade components. To begin with, the DirectCu GPU cooler combines a GPU base in which copper heatpipes make direct contact with the GPU die, conveing heat to a dense aluminum fin array which is cooled by a quiet fan.

The PCB makes use of "Super Alloy" voltage regulation components, such as Super Alloy chokes, Super Alloy capacitors, and Super Alloy MOSFETs, that not only increase durability, but also result in a cooler operation. The card is also factory-overclocked slightly, with the GPU running at 915 MHz but memory staying at 4200 MHz GDDR5 effective. Display connectivity includes two DVI, and one each of DisplayPort and HDMI. Other features include Super Hybrid Engine (which manages clock speeds based on load), and Voltage Tweak (which allows adjustment of voltages). ASUS' new card is yet to be priced.