PDA

View Full Version : AMD64 w/ 6800GT vs. Intel P4 w/ X800XT PE


tret
08-08-04, 05:06 PM
I have two systems which I will detail below. I ran the timedemo and decided to see the performance differences and post them here for others to compare.

Rig 1:
AMD 64 FX53 939 Pin (Stock Clocks)
Gigabyte GA-K8NSNXP-939
1GB TwinX Corsair 3200XL 2-2-2-10-1T
2x 36GB Raptor Raid 0 Array
Sound Blaster Audigy 2 ZS
BFG6800GTOC @ 400/1100 (No Volt Mods)

Rig 2:
Intel P4 3.2Ghz Northwood (Stock Clocks)
Asus P4C800-E Deluxe
1GB TwinX Corsair 3200LL 2-3-2-6-1T
1x 74GB Raptor
ATI X800XT Platinum Edition w/ Overdrive Enabled



Results TimeDemo1 2nd Run Score, No VSync

Rig 1: Forceware 61.77 Drivers & Dx9.0c

800x600 0xAA High Quality = 102.1
800x600 4xAA High Quality = 90.7

1280x1024 0xAA High Quality = 67.0
1280x1024 4xAA High Quality = 53.6

1600x1200 0xAA High Quality = 67.3
1600x1200 4xAA High Quality = 40.7



Rig 2: Catalyst 4.7 Drivers & Dx9.0c

800x600 0xAA High Quality = 69.1
800x600 4xAA High Quality = 63.2

1280x1024 0xAA High Quality = 59.8
1280x1024 4xAA High Quality = 39.8

1600x1200 0xAA High Quality = 49.5
1600x1200 4xAA High Quality = 29.8



Rig 2: Catalyst 4.7 Drivers w/ 4.9b OpenGL Hotfix & Dx9.0c

800x600 0xAA High Quality = 78.5
800x600 4xAA High Quality = 72.6

1280x1024 0xAA High Quality = 65.1
1280x1024 4xAA High Quality = 46.0

1600x1200 0xAA High Quality = 51.6
1600x1200 4xAA High Quality = 34.2



Well, I have nothing to complain about with either system, the 4.9b OpenGL drivers from ATI boosted performance but it still couldn't even come close to the 6800GT. Nothing you all haven't heard about a billion times before by now.

tret

cam9786
08-08-04, 05:13 PM
Just shows how much nVidia rules. Great Report, man! :thumbsup:

Sazar
08-08-04, 05:27 PM
the a64 kicks ass in games... comparing a64 - a64 would have made things more balanced...

that said nothing new here, I think most woulda expected the nv40 to come out on top anyways in this compare... :)

i think unless ati's alleged complete OGL driver re-write comes about they are gonna be playing second fiddle in d3...

nice work regardless :thumbsup:

PaiN
08-08-04, 05:43 PM
Actually...the ATI card didn't look to bad considering the CPU you had feeding it.
I'd like to see an apples to apple comparison.

mustrum
08-08-04, 05:57 PM
Comparing a FX-53 to a crappy P4 3.2.
You must have a sense of humor for sure!
(newb)

tret
08-08-04, 06:08 PM
Hmm, I see alot of places like anandtech comparing varying platforms and graphics cards, Guess they are Newbies too huh? I haven't gotten around to putting the X800XT in my FX53 system, But I will and do some more timedemo Benchmarks so people won't have to call me a "Newbie"

tret

Xplode
08-08-04, 06:22 PM
Isnt it suppose to be AMD/Ati vs Pentium/Nvidia??

Stoneyguy
08-08-04, 06:34 PM
Isnt it suppose to be AMD/Ati vs Pentium/Nvidia??

errrr... why is that? Because the First letters are closer together? (crazy)

Nvidia makes chipsets for AMD based motherboards not Intel. So it only seems about right to pair those two together.

UDawg
08-08-04, 06:59 PM
The only thing you can do it test each card in each system there-by getting 4 scores.

tret
08-08-04, 07:01 PM
Ok, I just Uninstalled the 6800GT, ran DriverCleaner in SafeMode, Installed the X800XT PE w/ 4.7 Cats and the 4.9b OpenGL Driver. Here are the results of the X800XT PE w/ Overdrive Enabled in Rig 1: (FX53 System)

Results TimeDemo1 2nd Run Score, No VSync

Rig 1: X800XT PE | ATI Catalyst 4.7 w/ 4.9b OpenGL Driver

800x600 0xAA High Quality = 108.4
800x600 4xAA High Quality = 86.2

1280x1024 0xAA High Quality = 72.1
1280x1024 4xAA High Quality = 46.5

1600x1200 0xAA High Quality = 53.8
1600x1200 4xAA High Quality = 34.2

4.7 Catalyst OpenGL
1600x1200 4xAA High Quality = 29.8


Pretty Interesting Results, Clearly the P4 System was a huge Bottleneck at 800x600 as the X800XT Gained about 30FPS at that resolution beating out the 6800GT @ Ultra Speeds. With 4xAA @ 800x600 there was another nice improvement but still just a little slower than the Nvidia Card. At 1280x1024 the ATI Card again showed a nice jump beating the Nvidia Card but not able to do so once 4xAA was turned on. At 1600x1200 we can see that the CPU is no longer the bottlneck as there was less than 2FPS increase with 0xAA and Exactly the same FPS with 4xAA enabled.

Seems pretty clear that the X800XT was being bogged down at the lower resolutions by the slower system but showed it could not keep up with the 6800 when AA was turned on. At 1600x1200 it could not perform any better on the FX53 Platform than it could in the P4 System.

tret

burningrave101
08-08-04, 07:04 PM
Comparing a FX-53 to a crappy P4 3.2.
You must have a sense of humor for sure!
(newb)

At 1600x1200 w/ AF enabled, the CPU isn't going to matter much.

CPU's aren't nearly as important for gaming as people think they are. The majority of games are GPU intensive so having a blindingly fast CPU wont mean crap unless you have a video card thats CPU limited. The GPU does most all the work at high resolutions and especially with high AA and AF enabled.

At 1600x1200 in DOOM 3 at High Quality you are not CPU limited so it wont matter if you have an A64 3500+ or a 3.0c Northwood. In fact i've seen the 6800's on Pentium 4 setups slightly outperforming the A64's with 6800's quite often. I would imagine it has to do with the fact Intel has better chipsets.

http://www2.hardocp.com/article.html?art=NjQ0

You can check out the performance differences there between Pentium 4's and A64's in DOOM 3 at High Quality.

Unless you sit around playing games at 800x600 then an A64 is not going to give you better gaming performance. The A64 will only be a benefit in CPU intensive games like Morrowind.

I can run my 3.0c at stock and get nearly the same performance numbers or a little better then someone with an A64 is nearly every game because of the fact i play at 1600x1200.

Vagrant Zero
08-08-04, 07:05 PM
Heh, ATI owners should put that in their pipes and smoke it.

UDawg
08-08-04, 07:08 PM
Oh knock the fanboy stuff off.

tret
08-08-04, 07:19 PM
I can run my 3.0c at stock and get nearly the same performance numbers or a little better then someone with an A64 is nearly every game because of the fact i play at 1600x1200.

I would like to see your timedemo results compared to mine, we have the same GPU at the same clocks. It will be a good way to see the difference between the P4 3.0 and the FX53 939

tret

shinrai
08-08-04, 07:20 PM
At 1600x1200 w/ AF enabled, the CPU isn't going to matter much.

CPU's aren't nearly as important for gaming as people think they are. The majority of games are GPU intensive so having a blindingly fast CPU wont mean crap unless you have a video card thats CPU limited. The GPU does most all the work at high resolutions and especially with high AA and AF enabled.

At 1600x1200 in DOOM 3 at High Quality you are not CPU limited so it wont matter if you have an A64 3500+ or a 3.0c Northwood. In fact i've seen the 6800's on Pentium 4 setups slightly outperforming the A64's with 6800's quite often. I would imagine it has to do with the fact Intel has better chipsets.

http://www2.hardocp.com/article.html?art=NjQ0

You can check out the performance differences there between Pentium 4's and A64's in DOOM 3 at High Quality.

Unless you sit around playing games at 800x600 then an A64 is not going to give you better gaming performance. The A64 will only be a benefit in CPU intensive games like Morrowind.

I can run my 3.0c at stock and get nearly the same performance numbers or a little better then someone with an A64 is nearly every game because of the fact i play at 1600x1200.



The days when Intel had far superior chipsets to those available to AMD cpu's are all but gone. With the memory controller on the cpu performance figures across all AMD platforms are very close and performance is already tweaked to max (due to this onboard memory controller).

And besides morrowind there are other games, such as Far Cry, that also require a fast cpu. There are plenty of levels (areas) in Far Cry that donot respond much to GPU tweaking or overclocking. But overclock your cpu and you'll get a nice little increase.

I'd guess that there may be areas like this in Doom3 so i'd take an AMD system anyday.

burningrave101
08-08-04, 07:32 PM
The days when Intel had far superior chipsets to those available to AMD cpu's are all but gone. With the memory controller on the cpu performance figures across all AMD platforms are very close and performance is already tweaked to max (due to this onboard memory controller).

And besides morrowind there are other games, such as Far Cry, that also require a fast cpu. There are plenty of levels (areas) in Far Cry that donot respond much to GPU tweaking or overclocking. But overclock your cpu and you'll get a nice little increase.

I'd guess that there may be areas like this in Doom3 so i'd take an AMD system anyday.

Well VIA which is one of AMD's major chipset manufacturers also manufactures chipsets for Intel platforms and they aren't as good as Intel chipsets. Intel chipsets have better memory compatibility and are usually more stable. There isn't a board out there more stable then the reference Intel motherboards. The only bad part is you can't overclock them.

When the A64 boards first started coming out they had horrible memory compatbility. They are getting better now though.

I like the nForce chipsets and i'm probably going to upgrade to an nForce board and A64 next year unless Intel comes out with something nice like they did with the Pentium 4 Northwoods vs the Athlon XP's.

Intel has already sent out their first 64-bit Prescott series chips and Intel was planning to add an on-die memory controller to their chips which would eliminate any advantage on the A64's.

HardOCP ran all their tests in-game using FRAPS so it will be a little better representation of actual performance over a recorded demo probably.

http://www2.hardocp.com/image.html?image=MTA5MDc4NzE0M1RPNjJBTU9FV1hfOV8yX 2wuZ2lm

On the 3.0c at 1600x1200 w/ 8xAF the 6800GT scored 56.5 fps.

http://www2.hardocp.com/image.html?image=MTA5MDc4NzE0M1RPNjJBTU9FV1hfMTdfM 19sLmdpZg==

On the FX-53 it scored 57.2 fps.

Its definitely not worth the cost to upgrade to an FX-53 over a lower end Pentium 4 or A64 for gaming. Nearly every game out there is GPU intensive and there are only a handful that aren't.

There are a few exterior parts of Far Cry that can benefit from a faster CPU i suppose but i've seen Far Cry benches running Pentium 4's and A64's head to head and there isn't much difference.

Gaming performance shouldn't be the first thing you look at when picking a CPU. There are alot more CPU intensive tasks like Multitasking, Rendering, Encodeing, Compression, Productivity Software, and so on that is far more reliant upon CPU performance. Thats part of the reason why i chose a 3.0c of an A64 3000+. The Pentium 4 is faster in nearly all those areas.

tret
08-08-04, 07:38 PM
Do Benchmarks of your own, and post the results. Alot of the time the hardware reviews and personal tests have little in common.

tret

burningrave101
08-08-04, 07:46 PM
Do Benchmarks of your own, and post the results. Alot of the time the hardware reviews and personal tests have little in common.

tret

Well like i said, i dont have a copy yet. I will probably grab a copy in the next few weeks though.

And the reason personal tests are sometimes different is because the person is testing a different portion of the level and may have done windows tweaks to their machine and not everyone can benchmark as accurately as the veteran reviewers out there. In-game testing with FRAPS is very tedious because of the AI influence.

tret
08-08-04, 07:48 PM
Well like i said, i dont have a copy yet. I will probably grab a copy in the next few weeks though.

I see, missed that part hehe, get it ASAP, your missing out.

And the reason personal tests are sometimes different is because the person is testing a different portion of the level and may have done windows tweaks to their machine and not everyone can benchmark as accurately as the veteran reviewers out there. In-game testing with FRAPS is very tedious because of the AI influence.

Just Keep it simple then and do the TimeDemo Benchmarks, I haven't done any windows tweaks personally so i don't know about all that.

tret