During a press conference at Computex, AMD gave the world's first official DirectX 11 GPU demonstration, saying the new API will debut before the end of 2009. When it does, AMD promised it would beat the competition to the punch and "deliver DirectX 11 first."
"Games and other applications are about to get a lot better as a result of AMD's new graphics hardware and DirectX 11," AMD stated in a press release. "DirectX 11 features such as tessellation will bring consumers higher quality, superior performing games making use of 6th generation AMD technology."
AMD also said its DX11-based videocards will improve Windows 7 performance in a wide number of applications and in a way that's "completely transparent to users," such as accelerating the conversion of video playback on portable media players.
Nivida and Super Micro have worked together in order to create a 1U server that ties together the power of massively parallel Tesla GPUs with multi-core CPUs. The system is said to deliver 12 times the performance of a traditional quad-core CPU-based 1U server.
The SuperServer 6016T-GF-TM2 is on display at Computex this week. “Our new Tesla GPU-based SuperServer 6016T-GF Series delivers a much higher performance-per-watt and per-rack than any other 1U solution in the market today," said Don Clegg, Super Micro‘s Vice President of Marketing. "This 2-Teraflop SuperServer meets the most demanding enterprise data center requirements for reliability and manageability."
Reportedly, Brazilian energy company Petrobras has already installed a cluster of 190 Tesla GPUs and is seeing a 5x to 20x improvement over their previous, multi-core CPU-based clusters.
You'll have to look to Mars to spy the fastest desktop consumer graphics card in the galaxy, which is the name of the new GPU Asus is showing off at Computex. Instead of two semi-custom GPUs that sit "between" a GTX 260 and 280, the Asus Mars 295 Limited Edition stuffs two higher end GTX 285 chips -- the fastest single GPU in Nvidia's lineup -- into a single package.
All told, the new card boasts all 240 shader processors on each GPU, a full 512-bit GDDR3 memory interface, 32 memory chips for 4GB total (2GB accessible per GPU), and the same core/shader/memory clockspeeds as the GTX 285 (648/1476/2400 MHz). By comparison, a traditional GTX 295 sports 896MB of GDDR3 per GPU on a 448-bit memory bus with core/shader/memory clockspeeds checking in at 576/1242/2000 MHz.
According to TechPowerUp, the funky looking cooler uses the same basic internal construction as the reference design for GTX 295 cards, albeit extending "slightly higher."
No word on price or availability, but Asus did say it would limit the run to 1,000 individually numbered cards.
It took a bit of waiting, but those Catalyst 9.5 downloads that started appearing on the web days ago are finally available direct from AMD. For those of you who grew impatient and snagged the download from an alternate source, they should be the same ones now appearing on AMD's site.
Catalyst 9.5 promises a handful of performance improvements, including:
Performance gains of up to 11 percent for single GPU and ATI CrossFireX configurations with AA enabled in Unigine Tropics DX9
Up to a 10 percent boost in Company of Heroes
Up to a 10 percent boost in BattleForce for CrossFireX configurations
Several bugs have been squashed in the new release, a handful of which apply to Windows 7. Win 7 owners should no longer see dropped frames while playing certain Blu-ray interlaced content, and moving a video clip across an extended desktop should no longer cause the system to flicker or for the media player to stop working.
Intel's Atom platform has been such a resounding success, one has to wonder what the No. 1 chip maker has planned for a follow-up. You don't have to wonder anymore, as Intel this week officially unveiled 'Pine Trail', the codename for Atom's successor.
The CPU used in Pine Trail, called 'Pineview,' moves the memory controller and GPU onto the same die as the CPU. This means Pine Trail will be a two-chip solution, one less than Intel's current netbook platform. In theory, this should result in cost savings and lower power consumption.
Pineview is being built on a 45nm manufacturing process. Intel hasn't said what type of memory controller it will use, though previous speculation pointed to single-channel DDR2. But what's most interesting is how the war between Intel and Nvidia is shaping up. Like Pine Trail, Nvidia's Ion platform is also a two-chip solution and will have had time to mature by the time Pine Trail debuts later this year. Performance looks to be better on the 9400M-based Ion as well, but Intel's price structure for selling standalone Atoms could put Nvidia at a disadvantage. Moreover, what chips will Nvidia use once Intel makes the move to a CPU+GPU solution?
Try to imagine where 3D gaming would be today if not for the graphics processing unit, or GPU. Without it, you wouldn't be tredging through the jungles of Crysis in all its visual splendor, nor would you be fending off endless hordes of fast-moving zombies at high resolutions. For that to happen, it takes a highly specialized chip designed for parallel processing to pull off the kinds of games you see today, the same ones that wouldn't be possible on a CPU alone. Going forward, GPU makers will try to extend the reliance on videocards to also include physics processing, video encoding/decoding, and other tasks that where once handled by the CPU.
It's pretty amazing when you think about how far graphics technology has come. To help you do that, we're going to take a look back at every major GPU release since the infancy of 3D graphics. Join us as we travel back in time and relive releases like 3dfx's Voodoo3 and S3's ViRGE lineup. This is one nostalgiac ride you don't want to miss!
Just in case you were worried that Intel wasn’t committed to it’s heavily delayed Larrabee platform, a 12 million dollar investment in a new Visual Computing Institute should help convince you otherwise. Located at Saarland University in Saarbrücken, Germany, this is the largest joint project ever formed between Intel and a European university. The institute will help Intel explore advanced graphical computing technologies, which includes everything from more realistic gaming, to advanced 3D user interfaces.
The primary focus of the research will be applied to Intel’s terascalling program. This will help them better understand how they can apply Larrabees unique multi x86 core architecture to achieve sustainable performance increases over modern day GPU’s. Larrabee has been delayed until some unknown date in 2010, presumably because it hasn’t yet achieved the type of performance gains they were hoping for against Nvidia & AMD.
In addition to terascalling research, Intel will also work with other hardware design labs in Barcelona, Spain, and Braunschweig, Germany to help optimize the Larrabee design. Z-buffering, clipping, and even ray tracing are all promises made by the Larrabee team, but clearly the software needed to make all this happen still requires some work.
Want more details? Click here to watch the press video.
So is Larrabee really the future? Or does this only prove Nvidia’s case that its promise is overhyped?
Intel's Larrabee project might rank as one of the most anticipated technology releases in a long while, and it looks like we'll have to wait just a bit longer than originally thought. It was expected that Intel would launch its many-cored cGPU sometime in late 2009, however the chip maker is now saying it plans to launch Larrabee in 2010.
Not a whole lot of details are known about Larrabee, only that it's a x86-based discrete graphics solution built around the second generation Pentium processor technology with the P54C core. When Larrabee launches, it will come in several iterations, the lowest of which will comprise no less than 8 cores. On the higher end, look for at least 32 cores and a 2GHz or faster clockspeed.
While it all sounds impressive, Intel's Jospeh Schultz did say that it would be a "big challenge" to compete with products from Nvidia and AMD.
Much was made over the race to 1GHz on the CPU front, a race AMD won with its Athlon processor. Markedly less exciting (but still an impressive feat) has been the sprint to churn out the first factory-clocked 1GHz GPU, with AMD again claiming victory, this time over Nvidia instead of Intel.
"Throughout the 40-year history of AMD, we have continually focused on technology firsts that deliver superior value to the customer," said Rick Bergman, senior VP, Products Group, AMD. "The 1GHz ATI Radeon HD 4890 continues that tradition by increasing the performance and compute power of our flagship singleGPU solution, ensuring a great experience whether our customers are playing the latest DirectX 10.1 game or running GPU accelerated applications built with OpenCL."
At 1GHz, the HD 4890 is able to deliver 1.6 TeraFLOPs of computing power, or "50 percent more than that of the competition's best single-GPU solution." In terms of real-world performance, however, the HD 4890 trails slightly behind Nvidia's GTX 285 in most benchmarks, or at least it does at 900MHz (see review of Asus Radeon EAH4890 Top in the June 2009 issue of Maximum PC on page 74).
FPS jockeying aside, it's good to see AMD aggressively going after the top spot in the graphics market rather than concede the high-end sector to Nvidia like it had done with its last generation of GPUs.