Microsoft’s DirectX11 came with big fanfare for tesselation, compute shaders, and several other features, however there are many more tiny features that have gone largely unnoticed by the larger community. Over at the Danger Zone they take a look at some of these like Conservative Depth Output and Programmable Interpolation that even the simplest of applications can benefit from.
1. Conservative depth output: this is something you use for pixel shaders that manually output a depth value. Basically rather than using SV_Depth, you use a variant that also specifiea an inequality. For instance SV_DepthGreater, or SV_DepthLessEqual. The depth you output from the shader must then satisfy the inequality relative to the interpolated depth of the rasterized triangle (if you don’t, the depth value is clamped for you). This allows the GPU to still use early-z cull, since it can still trivially reject pixels for cases where the depth test will always fail for the specified upper/lower bound. So for instance if you render a quad and output DepthGreaterEqual, the GPU can cull pixels where the quad’s depth is greater than the depth buffer value. Don’t bother looking for this one in the documentation…it’s not in there.
Guru3D has posted a 1080p high-definition preview of one of the scenes from the new 3DMark 11 test suite. This is just a representation of what 3DMark 11 can do, and is not necessarily representative of what will be in the final product. Still, it looks really cool.
X-Bit Labs has an interview with ATI’s Neal Robison about some of their recent news like their Open Stereo 3D initiative, their OpenGL support, and the Eyefinity system. I particularly like this comment on the differences between DirectX and OpenGL.
X-bit labs: Congratulations about the industry’s first OpenGL 3.2/4.0 WHQL-certified driver. But does OpenGL make sense for video games nowadays in general?
Neal Robison: Although DirectX is probably the best known collection of APIs for games, OpenGL still remains an important part of gaming technology. AMD has worked extensively on shaping the OpenGL standard as well as providing ongoing support for it. As you alluded to in your question, in March of 2010 we were able to announce our support for OpenGL 3.3 and OpenGL 4.0, an incredible feat on the part of our OpenGL software team, and an act that speaks volumes to the commitment and continued support that we have for the many developers utilizing OpenGL. We believe in and encourage open and industry standards so maintaining OpenGL as a strong and viable graphics API is important to AMD.
Personally, I’ve never found ATI/AMD’s OpenGL support all that stellar, but it has been several years since I payed much attention to it. I’m personally glad to know that they’re dedicating engineers to improving it.
Now that NVidia finally has DX11 hardware available, Microsoft has the chance to compare how well the ATI RadeonHD cards stand up to the new GTX480/470. Specifically, they were looking for oddities in the DirectX11 implementation, and found one particularly interesting one in NVidia’s use of the “Feature Level” concept.
The ATI Radeon HD 5000 Series only provides one quality level per sample count, while the NVIDIA GeForce GTX 470/480 exposes a number of fine-grain quality levels per sample count. This highlighted a few UI bugs in some of the samples as well as DXUT/DXUT11 that were corrected in the June 2010 release. Be sure to test the behavior of any MSAA settings and quality levels in your DX10.x and DX11 programs on both vendor's hardware.
The guys at Geeks3d took both cards for a spin to get the actual results from the ‘CheckMultisampleQualityLevels’ Microsoft mentions, and sees the obvious difference.
Here are the details for a GTX 480 (with R257.15 drivers – Win7 64-bit):
EVGA whipped up a nice DirectX 11 Tessellation video which shows off the new visual enhancements you can expect on the latest GTX 480/470 cards from EVGA!
Managed to find an interesting video showing, very briefly, the capabilities of the NVidia GF100 in DirectX11 via Unengine. While the site is in Italian, the translation reads:
We managed to track exclusively for you from a Chinese website on the first video of the new series of graphics cards Nvidia Without dealing with the benchmark Unigine DirectX 11.
To you after the jump to preview the video GF100 in action.
See the video after the break.
Update: Found another better, longer video with narration showing not only the Unengine demo, but video of the open computer case showing the Card & Connections. After the break.
AMD plans to really rub DirectX11 support into NVidia’s face at CES by launching a whopping 12 new DirectX11 GPU’s, six for desktops and six for mobile platforms.
The Radeon HD 5600/5500 (codenamed Redwood) series is expected to replace HD4600 series. The leaked information indicated that Radeon HD 5670 features 400 stream processors, GDDR5 memory, 128-bit of memory interface, DisplayPort/DVI/VGA ports, and support for Eyefinity technology.
Additionally, the Radeon HD 5400/5300 is reportedly designed to take the place of HD4550/4350. The above mentioned cards are expected to sell around or below US$90.
With NVidia waiting on Fermi for DirectX11 compliant and the recently announced delays, this could finally give ATI a chance to jump NVidia in marketshare.
Jan Vlietinck has published a simple 200x200x200 fluid simulation that simulates and renders the result using the new DirectX11 DirectCompute GPU acceleration systems.
The calculations make use of a well known scheme of velocity advection, Jaccobi pressure solving and making the velocity divergence free by subtracting the gradient of the pressure.This is the so called Semi-Lagrangian scheme. A more accurate solver makes use of the second order MacCormack technique. The simulation makes use of the latter. However it makes the simulation unstable and introduces artifacts. Limiting generated extremes can fix this, unfortunately I was not able to get this working, so the simulation runs without limiters, still the result is some visual interesting turbulent behavior.
The amplitude of the speed vectors are visualized. To make a 3d rendering, a simple ray maximum projection is used. This shoots rays through the volume searching the maximum speed along the ray. With a linear interpolation the speed is given some color.
ATI currently owns the high-end DX11 market, and is now moving into claiming the low-end market. TechConnect has pictures of what’s believed to be the first Redwood-based Radeon HD5670, a sub-$100 card that’s DX11 compliant.
The quite compact card has a 128-bit interface and 400 Stream Processors, it boasts active cooling, supports CrossFireX and provides three connectivity options – D-Sub, DVI and HDMI. This Radeon is supported to have a GPU clock of 775 MHz and 1GB of GDDR5 memory @ 4000 MHz, and it doesn’t require more power than that delivered via the PCI-Express slot.
AMD’s Developer Central has been updated with a pair of new demos showcasing some of what’s possible in DirectX11 with the new Radeon HD5800 cards. First is the “Mecha Demo”:
The Mecha demo shows the results of a new approach to rendering semi-transparent objects without pre-sorting, known as order-independent transparency (OIT). It is made possible by the ATI Radeon™ HD 5800 Series of graphics processors and the new features of Microsoft® DirectX® 11 technology. Blending is an order-dependent operation that requires sorting objects before rendering them. Atomic operations and append buffers make it possible to construct per-pixel fragment lists and sort them on the GPU. The results are a significant increase in speed and accuracy over those possible with traditional techniques.
Second is the “Ladybug Demo”:
The Ladybug demo shows the results of a new approach to simulating lens-accurate depth-of-field effects based on real-world parameters of focal length and focus distance. This technique is made possible by the ATI Radeon™ HD 5800 Series of graphics processors and the Microsoft® DirectX® 11 and Direct Compute 11 technologies. Depth-of-field is used in feature films by cinematographers to subtly guide a viewer’s attention through a shot or to heighten the emotion of a scene. This technique finally provides developers with a way to achieve the same effects and bring new levels of cinematic realism to their games. This approach is enabled by features such as atomic operations and shared memory.
If you’ve got the necessary hardware & software, hit their website for the EXE’s to see for yourself. Otherwise, hit their website to download the H264 HD movies showing what you’re missing.
Comments