Talk:GeForce 3 series

Fair use rationale for Image:Geforce3 logo.png

Image:Geforce3 logo.png is being used on this article. I notice the image page specifies that the image is being used under fair use but there is no explanation or rationale as to why its use in this Wikipedia article constitutes fair use. In addition to the boilerplate fair use template, you must also write out on the image description page a specific explanation or rationale for why using this image in each article is consistent with fair use.

Please go to the image description page and edit it to include a fair use rationale. Using one of the templates at Wikipedia:Fair use rationale guideline is an easy way to insure that your image is in compliance with Wikipedia policy, but remember that you must complete the template. Do not simply insert a blank template on an image page.

If there is other fair use media, consider checking that you have specified the fair use rationale on the other images used on this page. Note that any fair use images lacking such an explanation can be deleted one week after being tagged, as described on criteria for speedy deletion. If you have any questions please ask them at the Media copyright questions page. Thank you.

BetacommandBot (talk) 05:30, 19 December 2007 (UTC)

FLOPs calculation?

Hi, I was reading about this GPU architecture at both Tom's Hardware and Khronos group's wiki about GPU programmability. Seems the Vertex Shader was fully programmable while the Pixel Shader not so much (but, it was capable of register combiners, so not fully fixed-hardware either).

Vertex Shader Details - High-Tech And Vertex Juggling - NVIDIA's New GeForce3 GPU | Tom's Hardware (tomshardware.com) History of Programmability - OpenGL Wiki (khronos.org)

What caught my attention on the reading was this:

"The above diagram shows that the vertex shader is able to compute vertices with up to 16 data entries. Each entry consists of 4 32 bit floating-point numbers. 16 entries are quite a lot. It easily fits an average vertex with its position coordinates, weight, normal, diffuse and specular color, fog coordinate and point size information, leaving plenty of space for the coordinates of several textures.

Inside the vertex shader, the data is computed in form of entries. We just learned that each entry is a set of four 32 bit numbers. This makes the vertex shader a SIMD (single instruction multiple data) processor, as you are applying one instruction and affect a set of four variables. This makes perfect sense, because most transform and lighting operations are using 4x4 or 3x3 matrix operations. Each data is treated as floating point value, which shows that all computations executed by the vertex shader are actual floating-point calculations. Basically, the vertex shader is a very powerful SIMD FPU, barely touched by Pentium 4's SSE2 unit."

If I'm correct, it says that in a vector it's components (x,y,z,w) are treated each as a float, which means that the Vertex Shader is capable of doing 4 FLOPS, right? This is also backed up by the comparison against Pentium 4's SSE2 unit, which is also capable of 4 FLOPS according to the own Wikipedia's FLOPS article.

Floating point operations per second - Wikipedia

On modern GPUs we measure the floating-point capabilities of a GPU by looking at it's programmable Shaders, the pixel shader is still not fully programmable yet on this architecture so we are not counting that, would it be accurate to say that a GeForce 3 at 175MHz it's around 700MFLOPs? (175x4). I heard that this arquitecture has support for Vec4 MADD making it 8 FLOPS instead of 4, that would make the GPU 1.4GFLOPs, but I can't find reliable sources pointing to that.

Thanks. JidoBarbeiro (talk) 19:17, 24 October 2024 (UTC)