Highend CPU's vs High End Graphics cards

http://www.guru3d.com/article/article/136/

Hello and welcome, now weeks ago both ATi and NVIDIA suddenly released a new family of graphics cards on the market, obviously we reacted immediately,you’ve seen our previews. So you think a Pentium 4 2.8 Gigahertz with 1 Gigabyte DDR400 memory would be sufficient for the hottest graphics cards ? Think again, as it is not fast enough at all. So we worked tirelessly through the night thinking up excuses … for not bothering to test them :wink:

So here’s the thing, our recent reviews where showing a lot of CPU limitation on cards like Radeon x800 Pro and XT, GeForce 6800 GT and Ultra. What I noticed was simple, a lot of scores where close together or even equal. That’s usually a strong indication that the GPU can’t go any faster as it’s awaiting data from the CPU, the processor. Our recent GeForce 6800 GT review for example showed that the GT and the Ultra are extremely close together when it came to the end-result which made me believe that the Ultra was not working at it’s maximum capability.

Was a 2.8 Ghz system actually holding back performance of the new generation graphics cards ? Oh yeah, definitely. Make no mistake with such a system you’ll get your 90% maximum performance, yet the new graphics cards are limited by such a CPU. In today’s article we are going to look at the behavior of the new generation graphics cards by placing them in one of the fastest rigs available.

After I explained my little worry towards AMD they agreed to shop out an Athlon 64 3800+ which I could use to experiment a little with for a few weeks. The results are quite shocking …

We took the Radeon x800 Pro and Radeon x800 XT, GeForce 6800 GT and GeForce 6800 Ultra and started measuring performance on that Pentium 4 2.8 GHz system and then did the exact same thing on the Athlon 64 3800+. Now before we beging I need to make clear that you need to see things in relation to the processor. A faster CPU obviously means a newer platform thus a newer mainboard and memory utilisation also. Primary the advantage will come from the CPU though.

If you like to learn more about the graphics cards used in this article then you can find the reference reviews with all technological explanations here: GeForce 6800 GT preview, GeForce 6800 Ultra Preview, Radeon x800 Pro & XT (PE) preview.

This article is not so much a shootout of both brands graphics cards, no contrary it has been made with one thing in mind. To observe the effect of the fastest processor around on the new graphics cards. We’ll dive immediately into the benchmarks … first up the test systems and then we’ll start off with Splinter Cell.

Hello and welcome, now weeks ago both ATi and NVIDIA suddenly released a new family of graphics cards on the market, obviously we reacted immediately,you’ve seen our previews. So you think a Pentium 4 2.8 Gigahertz with 1 Gigabyte DDR400 memory would be sufficient for the hottest graphics cards ? Think again, as it is not fast enough at all. So we worked tirelessly through the night thinking up excuses … for not bothering to test them :wink:

So here’s the thing, our recent reviews where showing a lot of CPU limitation on cards like Radeon x800 Pro and XT, GeForce 6800 GT and Ultra. What I noticed was simple, a lot of scores where close together or even equal. That’s usually a strong indication that the GPU can’t go any faster as it’s awaiting data from the CPU, the processor. Our recent GeForce 6800 GT review for example showed that the GT and the Ultra are extremely close together when it came to the end-result which made me believe that the Ultra was not working at it’s maximum capability.

Was a 2.8 Ghz system actually holding back performance of the new generation graphics cards ? Oh yeah, definitely. Make no mistake with such a system you’ll get your 90% maximum performance, yet the new graphics cards are limited by such a CPU. In today’s article we are going to look at the behavior of the new generation graphics cards by placing them in one of the fastest rigs available.

After I explained my little worry towards AMD they agreed to shop out an Athlon 64 3800+ which I could use to experiment a little with for a few weeks. The results are quite shocking …

We took the Radeon x800 Pro and Radeon x800 XT, GeForce 6800 GT and GeForce 6800 Ultra and started measuring performance on that Pentium 4 2.8 GHz system and then did the exact same thing on the Athlon 64 3800+. Now before we beging I need to make clear that you need to see things in relation to the processor. A faster CPU obviously means a newer platform thus a newer mainboard and memory utilisation also. Primary the advantage will come from the CPU though.

If you like to learn more about the graphics cards used in this article then you can find the reference reviews with all technological explanations here: GeForce 6800 GT preview, GeForce 6800 Ultra Preview, Radeon x800 Pro & XT (PE) preview.

This article is not so much a shootout of both brands graphics cards, no contrary it has been made with one thing in mind. To observe the effect of the fastest processor around on the new graphics cards. We’ll dive immediately into the benchmarks … first up the test systems and then we’ll start off with Splinter Cell.

Hello and welcome, now weeks ago both ATi and NVIDIA suddenly released a new family of graphics cards on the market, obviously we reacted immediately,you’ve seen our previews. So you think a Pentium 4 2.8 Gigahertz with 1 Gigabyte DDR400 memory would be sufficient for the hottest graphics cards ? Think again, as it is not fast enough at all. So we worked tirelessly through the night thinking up excuses … for not bothering to test them :wink:

So here’s the thing, our recent reviews where showing a lot of CPU limitation on cards like Radeon x800 Pro and XT, GeForce 6800 GT and Ultra. What I noticed was simple, a lot of scores where close together or even equal. That’s usually a strong indication that the GPU can’t go any faster as it’s awaiting data from the CPU, the processor. Our recent GeForce 6800 GT review for example showed that the GT and the Ultra are extremely close together when it came to the end-result which made me believe that the Ultra was not working at it’s maximum capability.

Was a 2.8 Ghz system actually holding back performance of the new generation graphics cards ? Oh yeah, definitely. Make no mistake with such a system you’ll get your 90% maximum performance, yet the new graphics cards are limited by such a CPU. In today’s article we are going to look at the behavior of the new generation graphics cards by placing them in one of the fastest rigs available.

After I explained my little worry towards AMD they agreed to shop out an Athlon 64 3800+ which I could use to experiment a little with for a few weeks. The results are quite shocking …

We took the Radeon x800 Pro and Radeon x800 XT, GeForce 6800 GT and GeForce 6800 Ultra and started measuring performance on that Pentium 4 2.8 GHz system and then did the exact same thing on the Athlon 64 3800+. Now before we beging I need to make clear that you need to see things in relation to the processor. A faster CPU obviously means a newer platform thus a newer mainboard and memory utilisation also. Primary the advantage will come from the CPU though.

If you like to learn more about the graphics cards used in this article then you can find the reference reviews with all technological explanations here: GeForce 6800 GT preview, GeForce 6800 Ultra Preview, Radeon x800 Pro & XT (PE) preview.

This article is not so much a shootout of both brands graphics cards, no contrary it has been made with one thing in mind. To observe the effect of the fastest processor around on the new graphics cards. We’ll dive immediately into the benchmarks … first up the test systems and then we’ll start off with Splinter Cell.

Splinter Cell
New in our Benchmark suite is the very popular game Splinter Cell. Making a believable world for a spy to play in is quite a daunting task, but the levels are varied, filled with appropriate objects, and designed so that you usually dont have to choose between too many paths. It wouldve been great if you couldve had several points of entrance and that way get a lot more replay-value. Sam and the rest of the characters do look terrific, with high polygon models and both crisp and appropriate looking textures. What really separates Splinter Cell from most recent action games is the use of shadows. Splinter Cell uses the Unreal engine, which weve seen in several great looking games the past months, but UbiSoft also added improved lighting. By using real-time cast shadows, lightmaps, etc, this title gives you some of the best looking shadows to date.
In response to the growing use of sophisticated digital encryption to conceal potential threats to the national security of the United States, the NSA (National Security Agency) has ushered forth a new dawn of intelligence-gathering techniques. This top-secret initiative is dubbed Third Echelon. Denied to exist by the U.S. government, Third Echelon deploys elite intelligence-gathering units consisting of a lone field operative supported by a remote team. Like a sliver of glass, a Splinter Cell is small, sharp, and nearly invisible.

You have the right to spy, steal, destroy and assassinate, to ensure that American freedoms are protected. If captured, the U.S. government will disavow any knowledge of your existence.

You are Sam Fisher.

You are a Splinter Cell.

Splinter Cell is a DirectX 8/9 title and can handle Pixel Shaders if your card supports it. The downside of this nice piece of software is that it has different modes for different classes of hardware. We designed a configuration that is nearly the same for all graphics cards, however any low-end graphics card that does not support Pixel Shaders will reproduce a slightly different score. Secondly Splinter Cell has two shadowing techniques, Projector and Buffer mode. We force Projector mode in high detail on all graphics cards. Again, graphics cards without shader capabilities will run into a problem as they do not support it. We are talking about GeForce4 MX and earlier models (excluding the GeForce 3 series) only. With that in mind this software really is an excellent benchmark. Small sidenote, we are not using the standard timedemo’s. We made one ourselves that stresses the fillrate of a graphics card and will utilize a CPU very little.

All cards clearly could use a faster CPU other then a 2.8 GHz Pentium 4. As you can see the x800 XT from ATI is the card that is most CPU dependant as even in 1600x1200 it showed a noticeable difference in overall FPS.

Conclusion
Well, this was certainly an interesting experience. This article clearly shows the importance of a faster CPU versus the fastest graphics cards around.
The faster test-system proved that performance wise things will shift in advantage for the x800 XT series over the Ultra. It’s really a tie though, both cards are just so close to each other. But when we look purely at performance rankings in the highest resolution this would be the end-result : on the 4th place the x800 pro, 3rd place the 6800 GT, 2nd place the GeForce 6800 Ultra and at the first place the x800 XT, although a shared first place would probably be better wording. Remember I’m only talking about 1600x1200x32 here.

On a somewhat slower system (I refuse to call a 1 gig, 2.8 Ghz system mid-end) the results however are way closer to each other and definitely in favor of the GeForce 6800 series. The x800 XT is hungry for a faster processor that’s for sure.

If you are to buy a new system or graphics card then bare in mind that you really need to match everything to get the best bang for your bucks.

Right now I look at it this way, up-to a 3000 MHz you simply are better suited with a Radeon x800 Pro or GeForce 6800 GT. If you have a faster than 3 GHz (3000+) system and want the best of the best, then and only then I can recommend you the GeForce 6800 Ultra or the Radeon x800 XT. That system would be ridiculously expensive though and you need to consider whether that’s worth all that hard earned dough.