The Last Bout of ‘03 – NVIDIA’s GeForce FX 5700 Ultra
by Derek Wilson on October 23, 2003 9:30 AM EST- Posted in
- GPUs
Architecture
There was a great deal of talk about why architectural decisions were made, but we will concern ourselves more with what exists rather than why this path was chosen. Every architecture will have its advantages and disadvantages, but understanding what lies beneath is a necessary part of the equation for developers to create efficient code for any architecture.
The first thing of note is NVIDIA's confirmation that 3dcenter.de did a very good job of wading through the patents that cover the NV3x architecture. We will be going into the block diagram of the shader/texture core in this description, but we won't be able to take quite as technical a look at the architecture as 3dcenter. Right now, we are more interested in bringing you the scoop on how the NV36 gets its speed.
For our architecture coverage, we will jump right into the block diagram of the Shader/Texture core on NV35:
As we can see from this diagram, the architecture is very complex. The shader/texture core works by operating on "quads" at a time (in a SIMD manner). These quads enter the pipeline via the gatekeeper which handles managing which ones need to go through the pipe next. This includes quads that have come back for a second pass through the shader.
What happens in the center of this pipeline is dependent upon the shader code running or the texturing operations being done on the current set of quads. There are a certain few restrictions on what can be going on in here that go beyond simply the precision of the data. For instance, NV35 has a max of 32 registers (less if higher precision is used), the core texture unit is able to put (at most) two textures on a quad every clock cycle, the shader and combiners cannot all read the same register at the same time, along with limits on the number of triangles and quads that can be in flight at a time. These things have made it necessary for developers to pay more attention to what they are doing with their code than just writing code that produces the desired mathematic result. Of course, NVIDIA is going to try to make this less of a task through their compiler technology (which we will get to in a second).
Let us examine why the 5700 Ultra is able to pull out the performance increases we will be exploring shortly. Looking in the combiner stage of the block diagram, we can see that we are able to either have two combiners per clock or complete two math operations per clock. This was the same as NV31, with a very important exception: pre-NV35 architectures implement the combiner in fx12 (12 bit integer), NV35, NV36, and NV38 all have combiners that operate in full fp32 precision mode. This allows two more floating point operations to be done per clock cycle and is a very large factor in the increase in performance we have seen when we step up from NV30 to NV35 and from NV31 to NV36. In the end, the 5700 Ultra is a reflection of the performance delta between NV30 and NV38 for the midrange cards.
If you want to take a deeper look at this technology, the previously mentioned 3dcenter article is a good place to start. From here, we will touch on NVIDIA's Unified Compiler technology and explain how NVIDIA plans on making code run as efficiently as possible on their hardware with less hand optimization.
Facebook
Twitter
RSS
114 Comments
View All Comments
Anonymous User - Thursday, October 23, 2003 - link
#7, way to go! that's some hard numbers you got there. now that's what I call objective analysis! i agree with #12 and #14 as well.this review IMHO has been very subjective. even if they mentioned they would follow up with image quality reviews, it may be too late because "simple minded" individuals looking only for frame rate numbers may have already been influenced in their graphics card future buying decisions.
..and people, listen up! if you disagree with the site reviews then don't visit them and don't recommend them to friends. that way they get less hits and vendors leave them, and they die sooner or later. if we keep on visiting their sites even with good or bad intentions, they get hit counts and people see their rotating ads. that's how web site businesses operates now and a way for them to generate income. money talks in the hardware and technology business.
let's not keep debating or bashing each other for this. haven't you noticed they are playing us all like fools. I pity the people who will believe these reviews without really evaluating them.
so again, vote with your wallet! i'm no fanboy and i will always evaluate and buy the best product i see out there.
tech reviewers please be responsible! you got big in the first place because the community supports you. you owe it to them. you start without ads at first, people come and now you see "hits" coming you see an oppurtunity to generate income then you confuse us with your AD BLOATED web sites and BIASED reviews.
Anonymous User - Thursday, October 23, 2003 - link
I agree with everyone about the conclusion not making much sense. And what is the price before rebate? Ya know...the money I ACTUALLY HAVE TO HAND OVER? Rebates are sometimes(not all the time) a risky business. Who is honoring these rebates? Why wasnt this mentioed? Thats rather odd.If there is a rebate involved im assuming that the card must be $250. If this is the case then the 9700pro is the same price. So actually....If your going to shell out $250, just get a 9700pro and forget about the stupid rebate =)
Anonymous User - Thursday, October 23, 2003 - link
This is a FACT. Its not an excuse...its just a fact that im putting out.To be 100% unbiased is NOT human.
So the boys at anandtech tend to lean towards Nvidia alittle more then ATI...maybe 55-45, dont get made about it though...just take what they say with a grain of salt =)
Anonymous User - Thursday, October 23, 2003 - link
i tought that tom pointed ati as the cheaterAnonymous User - Thursday, October 23, 2003 - link
#24 You should try taking English 101...it might help you a bit :)Anonymous User - Thursday, October 23, 2003 - link
This IQ comparison better not be like the last one. If every other site gives me full uncompressed screenshots and shows me Nvidia filtering issues, anandtech better show me too if they dont want to lose respectAnonymous User - Thursday, October 23, 2003 - link
I like AnandTech, and respect their reviews. I agree the conclusion doesn't match the test results, but the conclusion is also subjective to actually using the device reviewed.Besides, HardOCP is way more biased than Anand or Tom's, it's just that their bias changes based on Kyle's latest whim.
Anonymous User - Thursday, October 23, 2003 - link
If the IQ testing for this card is the same as the last IQ article, no thanks. Let's hope they put more effort into actually showing the obvious differences that were passed up last time.Anonymous User - Thursday, October 23, 2003 - link
ej guys 9or better ati fans) if derek made this review besides posting the numbers he probably saw the image quality. all you knows perfectly whose the best driver writer. and it's been so seen i got my first tnt. but anyways it is obvious that some games work better on ati's card somework better on nvidia. in xbitlabs forum yesterday one guy had posted after he upgrated his GF3 to radeon 9800 pro he started experiencing quality promblems. so don't bitch about the quality it all dipends on the test suit.it was the same situation with the reviews of fx-51 and pentium EE. every single review gontradict to another.
btw those of you trade better start buying nvidia'a stocks.
Anonymous User - Thursday, October 23, 2003 - link
#10, The ads rotate. I just saw two different ATI ads on the same page. Keep trying!