HD 3800: First DX10.1, 55nm and Four-Way GPU
When the R600 graphics processor and the Radeon HD 2900 series launched, I stated that AMD had hardware that was more forward-looking than Nvidia's G80 technology. I still feel that way after looking at the latest information we obtained from AMD about RV670. On the same day Nvidia is launching its
GeForce 8800 GT, Rick Bergman, Vice President of the Graphics Product Group at AMD disclosed some details about the Radeon HD 3800 series and beyond. However, he kept most of the juicy bits to himself pending product launch on November 15th. We do know that this launch will focus DX 10.1 hardware. Microsoft updated its software developers kit (SDK) in August and revealed some of the changes that would be taking place.
Due to the changes to 10.1, the RV670 graphics processor is not just a die shrink. Primarily this die shrink will be a 55 nm process. RV670 should take less silicon per wafer to produce than Nvidia's 8800GT meaning higher margins per part. AMD hinted but did not disclose that it should be able to beat Nvidia's thermal envelope especially at idle as it chose to implement some of its mobile technology into the desktop parts.
This is the sweet spot that was missing for almost a year. Only high, low and entry level cards have had a presence in the marketplace. PC Gamers were forced to spend above the traditional midrange price point for hardware that is clearly high end or purchase inferior performance DX10 hardware. The only card that came close was Nvidia's 320 MB model of the GeForce 8800GTS. Looking forward there will be at least three models (2 from AMD and 1 from Nvidia) that will service the "real" midrange. Traditionally midrange parts offered 75% of the performance of high end models at 50% or less of their price. The GeForce 8800GT and Radeon 3800 models should service this segment well with the new PCIe 2.0 interface.
Beyond DX 10.1 and a 55 nm process, users will be able to use more cards. Two, three and four-way CrossFire will be supported on Vista. Bergman also hinted at an asymmetric version of CrossFire. This means that cards of the same core but different memory and clock frequencies could be configured in CrossFire, stretching a consumer's dollar further. The Radeon HD 3800 series will also have an updated Universal Video Decoder (UVD) for the hardware acceleration of HD DVD and BluRay movies.
So, if the launch goes as planned, AMD will be able to claim three firsts: first to DX 10.1, first to 55nm and first to four way GPU performance on Vista.
There will be two versions of the Radeon HD 3800, with pricing (yet unconfirmed and subject to change) between $150-250 depending on model, clock frequency and memory configurations. These will be competitive with cards based on the technology Nvidia announced today. We wanted mid-range cards and now it appears we have them. The question that remains is "what does the change to the graphics component of DirectX in D3D 10.1 mean to consumers?" That is the real key to both launches.
PC Gaming Is Alive and More Than Well
Several developers at the Nvidia 2007 Editor's conference last week told us that they were too far into the production cycle to consider the upcoming D3D standard. According to NPD research 254 new PC game titles are expected this fall. Roy Taylor, Nvidia Vice President of Developer Relations, stated that one undisclosed publisher reported online purchase and downloads of new PC Game titles were four times that of retail purchases. Microsoft General Manager of Games for Windows Kevin Unangst told us that there will be over "15 million Direct X 10 GPUs installed by the end of 2007 and over 102 million by the end of 2008."
Unangst continued noting that "gaming is expected to grow 80% over the next 5 years" and that Microsoft has shipped "over 60 million units" of its Vista operating system. All of the numbers point to something phenomenal: PC Gaming is alive and well. Recent DFC data projects PC gaming will bring in over $13 billion through 2012. In the short term and long, DX10 is here and will have plenty to fuel the cards that already service the market. Additionally, the future of the market will shift to the new standard and existing hardware will not be able to service indefinitely.
Vista is almost a year old and the Direct X graphics component Direct 3D is slated to get an update with the first service pack for the OS. Microsoft's Unangst stated that Direct3D 10.1 update "is not something we think any developer will target exclusively" and concluded by stating that "the hardware you see today (GeForce 8800GT) is the hardware people will be targeting for years to come." While that should make those who bought AMD and Nvidia DX10.0 hardware feel better, you should take those comments with a grain of salt. He made that statement to the press during an Nvidia sponsored event with high profile Nvidia personnel in the room.
In the short run there is no need to worry as the code for most games coming out over the next quarter will not be changed. With the holiday sales period already starting and demands for maximum seasonal earnings on their minds, publishers are pressuring developers to finalize their work sooner rather than later. That being said, those who are looking to purchase cards will need to think about what they really want past this season. Nvidia's GeForce 8800GT is a nice card for the price but it's only 10.0 compliant where AMD's upcoming releases of RV670, RV680, and R700 will all be DX 10.1 ready.
What's New in DX 10.1
So what's new in DX 10.1 that should lead you to rethink your purchasing decisions?
There's a new Shader Model (SM). Direct3D 10.1 introduced SM 4.1 to replace SM 4.0. Seeing that DX 10 hardware unified the shader core in terms of what operations can be handled, the changes can be can be divided into their three components: tighter specifications, shading and texturing capabilities, and anti-aliasing improvements. The debate over which cards to buy would have been simplified if some of these changes had been implemented in DX 10.0 instead of delaying them until DX 10.1. But, that didn't happen.
Additionally, Shader performance has been improved. Shaders will be able to output to any MRT (multiple render target) or buffer with their own blending mode per target. Vertex shaders now can have 32 128-bit values instead of 16. Level of detail (LOD) instructions were added which allow custom filtering techniques to be used. Gather4 enables a block of 2x2 unfiltered texture values instead of a single bilinear filtered lookup. The last of the major improvements to shader model 4.1 is the use of cube map arrays. In the world of gamers, this is one of the most significant changes, as it allows the use of efficient global illumination techniques which can deliver ray trace quality effects including indirect lighting, color bleeding, soft shadows, refraction, and high quality glossy reflections to a scene. Below is a screenshot of new demo that AMD has been working on that shows the use of cube maps for global illumination. On the right it has been turned on and on the left it is off. Notice the shadows and color blending per sphere with the technique enabled.
DX 10.1 takes some of the optional precision and formatting items from Direct3D 10 and forces their compliance in DX 10.1. Two data formats which are required are floating point 32 (FP32) and integer 16 (Int16). 0.5 ULP (Unit in the Last Place) is also enforced in 10.1. Both FP32 for 128-bit texture format filtering and Int16 64-bit integer pixel blending ensures all 10.1 compliant hardware vendors will support these high precision data formats in their products. A tighter ULP makes rounding floating point numbers even more precise. This means that if a developer wants to use these specific formats or requires tighter rounding precision, they don't have to code around the hardware. The same is true for the blending, 10.1 will support all unorm and snorm formats and 10.1 hardware will support all of it regardless of manufacturer.
Antialiasing (AA) will get cleaned up a lot as well. Multisampling antialiasing (MSAA) under D3D 10.1 now requires a minimum of four samples per pixel for 32-bit and 64-bit (MSAA 4x). AMD/ATI and Nvidia have kept their AA sampling patterns under wraps for competitive reasons. 10.1 will force many improvements regarding AA sample patterns as well. The first being a standardized sample pattern for AA for 2x, 4x, 8x and 16x sampling. 10.1 will also enable pixel coverage masking and multi-sample read and writes to the buffer. Now color and depth can be accessed directly by the shader and then can be manipulated via a predetermined pixel shader AA program. Long gone will be the days of HDR without AA. Developers can determine what will work best with their applications, performing custom AA, edge detecting AA, and even programming their own custom AA filter for adaptive and temporal AA.
Computer is a fast developing thing, but PC BIOS isn’t. From ancient 386 to modern quad core PC, the Mainstream PC BIOS looks the same, the only difference is more adjustments can be done with today’s BIOS. That’s not enough, we knew BIOS with beautiful UI have been developed for a long time but because of the production cost, no company will bring this kind of “next gen” BIOS to US.ASUS’P5E3 ASUS Express Gate show us some features of the next gen BIOS, But now MSI announces that they will use EFI BIOS on their upcoming mobos, the first one is a new P35-Neo3.
