NVIDIA's GeForce 8800 (G80): GPUs Re-architected for DirectX 10
by Anand Lal Shimpi & Derek Wilson on November 8, 2006 6:01 PM EST- Posted in
- GPUs
Branching
In order to talk generally about SPs and their capabilities, all the vertices, primitives, pixel components, etc. to be processed are referred to as threads. This way we can look at each SP as handling its own thread no matter what type of data is being processed. G80 is able to sustain "thousands" of threads at a time, but the actual number of threads that can be active at any given time is not disclosed. While all SPs can handle any type of thread, SPs that share resources must be running the same type of thread at any given time. In this way, each block of 16 SPs can be running one type of shader program on 16 threads. This indicates something about branch granularity as well. For vertex shaders, branch granularity is 16 vertices. For pixel shaders, branch granularity is 32 pixels (arranged in pairs of blocks of 4x4 pixels).
Branch granularity defines how many threads must follow the same path through data. When a group of 32 pixel threads all take the same branch, we don't have a problem. If even one thread must take a path that is different from the others, all 32 threads must be evaluated with both paths following the branch. The branch then defines what result each individual thread will keep and which it will discard. It's easy to see that optimum granularity is 1 thread, as no unnecessary work would be done. The way resources are allocated and the way instructions are run on SPs grouped together currently doesn't allow any more fine-grained branching. Here's a chart that address branch granularity:
| GPU | Branch Granularity |
| NVIDIA NV4x | ~1K pixels |
| NVIDIA G70 | ~256 pixels |
| ATI R580 | 48 pixels |
| NVIDIA G80 | 16 vertex 32 pixels |
Clearly G80 has the advantage here, as it's less likely that smaller groups of pixels will take different directions through a branch. This gives programmers the ability to more easily integrate branching into their code without getting a massive performance hit. If programmers are able to incorporate more branches, shader code can become more general purpose and we will see many more effects make their way into games. Now that G80 has caught up to ATI in terms of potential branch performance, we hope developers will take the reality of more complex code seriously.
Early-Z, Memory Interface
NVIDIA has added hardware for Early-Z to G80, after their current Z-Cull hardware which removes regions of pixels completely occluded by other geometry. Early-Z is a more fine-grained occlusion culling method that looks at a calculated Z value of a fragment before it hits the pixel pipeline. Z-Cull doesn't look at per fragment Z values, but uses a Z value based on geometry. While Z-Cull can get rid of large blocks of data it has issues handling surfaces that are only partially occluded or intersecting surfaces. Looking at individual depth values per pixel can help remove unnecessary fragments from heading down the pipeline only to be thrown out when the ROPs get to them.
The memory interface has been dramatically redesigned to support the access patterns of all of G80's independent stream processors. Given the theme of increasing granularity within G80 it's no surprise that we are now seeing 5 and 6 channels of GDDR rather than the 2 or 4 channels we have been used to for the past few years. 8800 GTX will have a 384 bit bus (6 x 64-bit channels), while the 8800 GTS will have a 320 bit wide connection to DRAM (5 x 64-bit channels). We would love to delve further into the details of G80's new memory interface, but NVIDIA isn't discussing the details of this aspect of their hardware.
Facebook
Twitter
RSS
111 Comments
View All Comments
Nightmare225 - Sunday, November 26, 2006 - link
Are the FPS posted in this article, Minimum FPS, Average FPS, or Maximum? Thanks!multiblitz - Monday, November 20, 2006 - link
I enjoyed your reviews always a lot as they inclueded the video-capbilities for a HTPC on previous cards. Unfortunately this was this time not the case. Hopefully there will be a 2. Part covering this as well ? If so, it would be nice to make a compariosn on picture quality as well against the filters of ffdshow, as nvidia is now as well supporting postprocessing filters...DerekWilson - Tuesday, November 21, 2006 - link
What we know right now is that 8800 gets a 128 out of 130 on HQV tests.We haven't quite put together an HTPC look at 8800, but this is a possibility for the future.
epsil0n - Sunday, November 19, 2006 - link
I am not agree with this:"It isn't surprising to see that NVIDIA's implementation of a unified shader is based on taking a pixel shader quad pipeline, and breaking up the vector units into 4 scalar units. Now, rather than 4 pixel quads, we see 16 SPs per "quad" or block of stream processors. Each block of 16 SPs shares 4 texture address units, 8 texture filter units, and an L1 cache."
If i understood well this sentence tells that given 4 pixels the numbers of SPs involved in the computation are 16. Then, this assumes that each component of the pixel shader is computed horizontally over 16 SP (4pixel x 4rgba = 16SP). But, are you sure??
I didn't found others articles over the web that speculate about this. Reading others articles the main idea that i realized is that a shader is computed by one and only one SP. Each vector instruction (inside the shader) is "mapped" as a sequence of scalar operations (a dot product beetwen two vectors is mapped as 4 MUD/ADD operations). As a consequence, in this scenario 4 pixels are computed only by 4 SPs.
DerekWilson - Tuesday, November 21, 2006 - link
Honestly, NVIDIA wouldn't give us this level of detail. We certainly pressed them about how vertices and pixels map to SPs, but the answer we got was always something about how dynamic the hardware is able to dynamically schedule the SPs optimally according to what needs to be done.They can get away with being obscure about how they actually process the data because it could happen either way and provide the same effect to the developer and gamer alike.
Scheduling the simultaneous processing one vec4 MAD operation on 4 quads (16 pixels) over 4 groups of 4 SPs will take 4 clock cycles (in terms of throughput). Processing the same 16 pixels on 16 SPs will also take 4 clock cycles.
But there are reasons to believe that things happen the way we described. Loading components of 16 different "threads" (verts, pixels or whatever) would likely be harder on the cache than loading all 4 components of 4 different threads. We could see them schedule multiple ops from 4 threads to fill up each block of shaders -- like computing 4 consecutive scalar operations for 4 threads on 16 SPs.
At the same time, it might be easier to maximize SP utilization if 16 threads were processed on one block of SPs every clock.
I think the answer to this question is that NVIDIA knows, they didn't tell us, and all we can do is give it our best guess.
xtknight - Thursday, November 16, 2006 - link
This has been AT's best article in awhile. Tons of great, concise info.I have a question about the gamma corrected AA. This would be detrimental if you've already calibrated your display, correct (assuming the game heeds to the calibration)? Do you know what gamma correction factor the cards use for 'gamma corrected AA'?
DerekWilson - Monday, November 20, 2006 - link
I don't know if they dynamically adjust gamma correction based on monitor (that would be nice though) ...if they don't they likely adjusted for a gamma of either (or between) 2.2 or 2.5.
Also, thanks :-) There was a lot more we wanted to pack in, but I'm glad to see that we did a good job with what we were able to include.
Thanks,
Derek Wilson
bjacobson - Sunday, November 12, 2006 - link
This comment is unrelated, but could you implement some system where after rating a comment, on reload the page goes back to the comment I was just at? Otherwise I rate something halfway down and then have to spend several seconds finding where I just was. Just a little nuissance.Thanks for the great article, fun read.
neo229 - Friday, November 10, 2006 - link
This is a very suspect quote. A card that requires two PCIe power connectors is going to dissipate a lot of heat. More heat means there must be a faster, louder fan or more substantial and costly heat sink. The extra costs associated with providing a truly quiet card mean that the bulk of manufacturers go with the loud fan option.
DerekWilson - Friday, November 10, 2006 - link
If manufacturers go with the NVIDIA reference design, then we will see a nice large heatsink with a huge quiet fan.Really, it does move a lot of air without making a lot of noise ... Are there any devices we can get to measure the airflow of a cooling solution?
We are also seeing some designs using water cooling and theres even one with a thermo-electric (peltier) cooler on it. Manufacturers are going to great lengths to keep this thing running cool without generating much noise.
None of the 8 retail cards we are testing right now generate nearly the noise of the X1950 XTX ... We are working on a retail roundup right now, and we'll absolutely have noise numbers for all of these cards at load.