Intel just announced a Sandy Bridge page, which collects information on our new chip, both the CPU side and the GPU side. SNB is getting good reviews at Anands, Toms, and other key hardware review sites.
The team I work on has contributed a set of graphics samples, helper code, and whitepapters to the SNB launch, and that content is hosted on this page.
Two key SNB samples are:
The Onloaded Shadows sample highlights a technique of using the software rasterizer, WARP, to run low-frequency tasks on the CPU since in many cases those cores are underutilized. The article runs through the technique and has good references. The code includes automated cascade generation and thus may be useful independent of the load balancing technique. And there is a video walkthrough.
The AVX Cloth sample shows how to leverage the SIMD units on the CPU and get great performance. The article elaborates on the deep AVX SIMD voodoo involved, and with a little effort real performance wins are possible. The code is linked here. And there is a video walkthrough.
In addition, the SNB graphics micro-site provides the following graphics helper code and developer whitepapers:
SNB Dev Guide highlights techniques to best take advantage of the processor graphics available in SNB.
GPUDetect, linked here, provides some handy-dandy code to help you detect SNB processor graphics and then make use of the great information in the Dev Guide.
Sample Tweaker extends that to show how to use the Dev Guide info to tweak a sample. This whitepaper builds off the previously released Ocean Fog sample. There is an article to accompany the original sample.
Just to remind developers, the Intel graphics page for articles and samples can be found here. That includes the SNB content as well as older, non-SNB specific content.
On that page, there are 3 more graphics samples produced by the team I work on:
Colony shows rendering a lot of units as part of a crowd using Intel Thread Building Blocks ( TBB ) to distribute the crowd simulation processing workload. The article is here. The code is linked here. And there is a video walkthrough.
Fireflies is a particle system sample that uses Intel Thread Building Blocks ( TBB ) to distribute the particle simulation workload. The article is here. The code is linked here. And there is a video walkthrough.
And on that page are the Intel SIGGRAPH samples:
Adaptive Volumetric Shadow Maps, with the SIGGRAPH material here. The slides and code are linked here. This is a way cool shadow technique that supports shadowing from particles and other participating media.
Sample Distribution Shadow Maps, with the SIGGRAPH material here. The slides and code are linked here. This is an interesting technique to add to the kit bag of auto-partitioning techniques for cascaded shadow maps.
Deferred Rendering for Current and Future Rendering Pipelines implements conventional deferred shading, deferred lighting and tile-based deferred shading, with the SIGGRAPH material here. The slides and code are linked here. This is an interesting technique that uses DX11 Compute Shader.
One thing you might note is the art quality has improved with these samples over time. That is something we hope our developers appreciate, as it helps to showcase the value of these techniques. While obviously we should avoid judging a book by its cover, a little visual wow never hurts showcase how cool something could look. With that in mind, we will continue to provide visually interesting samples that highlight relevant techniques.
Another change you might note is the last several samples include a video walkthrough. We are hoping our developer community finds them valuable, and hope to be able to have a video walkthrough for each sample in the future.
With all of these changes, one thing you could expect is that over time not only will you see more of these upgraded samples, but we will also update the Game Code Resource Disk and the Intel Graphics Sample Browser with these new, updated samples so for those of you who want the convenience of a single download it will also be available.