View Full Version : 5900 against 9800- The Showdown
In reference to This (http://forum.oc-forums.com/vb/showthread.php?threadid=214847) thread, I am going to benchmark an ATi Radeon 9800 Pro and An eVga GeForce FX 5900 Ultra (450/850) against each other, and am then going to overclock each card to its max and benchmark them again. I will be using 3dMark2001 (Default)
3dMark2003 (Default) Unreal Tournament 2003 (1280x960 8x antialiasing 8x aniso) Quake 3 Arena (8xaa 8x aniso) and one test selected by popular demand. (read- suggest another test).
I will be doing the testing later today on this machine, aiming to remove all possible bottlenecks:
P4 2.4c @3.45GHz
512 MB 384MHz 2-2-2-5 RAM
Abit IC7 Motherboard
WD 1200JB HDD
Radeon 9800 Pro (Stock/Overclocked)
Geforce FX 5900 Ultra (Stock/Overclocked)
Antec Tru Power 480, Tested with volt meter to supply correct voltages.
I will be testing as soon as it is decided what the final test will be.
Why dont you use some games that arnt used everytime. I mean Ut2k3... 2k1, 2k3, quake3, they have all been done. Use some where cheats/optimizations arnt present. I mean you can still use those... but maybe throw in a couple oddballs.
Maybe throw in Nolf2, or WC3. Just a thought.
funnyperson1
07-17-03, 12:08 PM
Yes, but whatever you do don't use 3dmark03. Use a couple popular games.
I will still use 3dmark03, because 3.3.0 doesnt have any *known* cheats, but I would not trust it. UT2003 has been removed. Quake 3 has been removed. Final Lineup:
3dmark2001
3dmark2003
Warcraft 3
Soldier of Fortune 2
MOHAA
GTA Vice City
RTCW Enemy Territory
America's Army
Ghost Recon
Tests will be completed by late today/Friday
Preliminary results (have to wait to get other disks off of friend)- 1024x768 all in game detail set to highest:
SOF2 was bugged so the score will not be included.
9800 Pro Scores are listed first, Geforce FX 5900 Ultra Second
Warcraft 3 ( 59) (61)
Mohaa ( 93 ) ( 90)
GTA Vice City (209) ( 212)
RTCW Enemy Terr. ( 67) ( 63)
3dmark03 3.3.0 (6109) ( 6894)
3dmark2001 ( 20,101) (20,021)
Currently, the cards are TIED! Neither card so far, at standard settings, is fast enough to be declared a champion. 3dMark03 is purely for refrence, some drivers for that benchmark are "bugged"
Not to sound rude but why bother? I mean Are you going to install this 9800pro on a clean system?? Then format again and then install the 5900? Because thats what your going to have to do to get some accurate results. Also, are you going to be using games that run faster on Pentiums vs. AMD's? Sorry I mean if yah wanna make your own review thats cool, I just dont see the point when there are a million of them out there comparing these cards. We ALL know that the 9800 pro is a better card for games. It might change as the 5900's have just came out this month so maybe in 3 months driver updates might show otherwise. They are both fantastic cards no matter how you look at them.
Atari its about the cards, hes trying to take the processor out of the equasion. I dont see why you gotta be so criticle.
AMD_Me can you do some AA and AF tests? And maybe different resolutions plz. Thanks.
I find it... interesting how they are both very close on all the game benchmarks but when it comes to 2k3 the FX pulls ahead big time. Very interesting....
Originally posted by IR1
Atari its about the cards, hes trying to take the processor out of the equasion. I dont see why you gotta be so criticle.
Ummm when it comes to benchmarking video cards and all the controversy that surrounds the tests being conducted, isn't being "critical" the only way to get true numbers???
I doubt AMD_me is fudging anything of course, hes is doing this as a courtesey. your deal about it being a P4 and not an AMD... is just odd to me, Pentiums are the fastest right now, hes got his overclocked well... which takes the CPu out of the equasion... this is exactly what we want.
The criticle part is when you complain about it not being an Athlon system. Sorry this isnt a review site.. he may not have an Athlon system handy right now.
Did I say he was fudging anything?? Where did you get this?
All i'm stating is that every computer is different and when your benchmarking cards without a clean install then your going to run into problems. We all know ATI and Nvidia drivers dont mix. By the way some games do run faster on Pentium systems I.E Quake 3. Thats why I make the statement of software choices.
ATI and Nvidia drivers can be cleanly removed. I didnt say you said he was fudging. I was stating that outright. And im glad you agree that he isnt.
Originally posted by Atari
By the way some games do run faster on Pentium systems I.E Quake 3. Thats why I make the statement of software choices.
That has nothing to do with benchmarking the video cards at all, since any optimisations for CPU will take effect on BOTH video cards, and the end result will still be the same (one being faster than the other)
Evnas is exactly right. This isnt an AMD vs P4 benchmark. As long as everything stays constant except for the video cards the results should be fine.
The conrol must be the video cards.. not the processor.
Evnas... one thing. You need to update the Your System statement in your profile, I think its a little dated. ;)
.
rad 9800 doesn't have 8 sample AA only 6 sample but it's quality is far better than the 5900's 8 sample
Yeah so you have to compare based on the quality.. not the number. You gotta sorta eye ball it I guess.
Originally posted by IR1
Evnas... one thing. You need to update the Your System statement in your profile, I think its a little dated. ;)
.
Yea, i notice that everytime i go in there to update something else, but im to lazy to change that too, heh
Could you run the tests with AA/AF? If you own one of those cards you will probably be using AA/AF all the time.
-Rav
I'd still like to see a test where it's equal IQ's and each tested like that, but I guess it'd be pretty hard to do...
Originally posted by Hurk
I'd still like to see a test where it's equal IQ's and each tested like that, but I guess it'd be pretty hard to do...
I forget who exactly it was on the HardOCP team, but he actually got hte IQ equal in the 9700 vs 5800 Ultra tests...probably just need to do that same thing
Originally posted by Evnas
I forget who exactly it was on the HardOCP team, but he actually got hte IQ equal in the 9700 vs 5800 Ultra tests...probably just need to do that same thing
Yeah right. Got a link?
How do you benchmark for Americas Army?
Originally posted by Hurk
Yeah right. Got a link?
http://www.hardocp.com/article.html?art=NDQ0
It was in the review of the 5600 and 5200 (not hte 5800 like i thought). Its on the second page
Sean Pelletier was the writer
Originally posted by Hurk
How do you benchmark for Americas Army?
I don't know if you can record demos in it or not but you could open the console and type stat fps, then type start bridge_se (so you will start in the same spot for both cards, maybe try a few different maps too) then type playerlock 0 (so you can move) and then just move straight forward without touching the mouse to make sure both cards would be rendering the exact same thing.
It wouldn't be the most accurate way to do it though as you couldn't record every spike/drop in framerate but it should give you a pretty good idea, maybe using FRAPS would help.
Originally posted by Atari
Not to sound rude but why bother? I mean Are you going to install this 9800pro on a clean system?? Then format again and then install the 5900? Because thats what your going to have to do to get some accurate results. Also, are you going to be using games that run faster on Pentiums vs. AMD's? Sorry I mean if yah wanna make your own review thats cool, I just dont see the point when there are a million of them out there comparing these cards. We ALL know that the 9800 pro is a better card for games. It might change as the 5900's have just came out this month so maybe in 3 months driver updates might show otherwise. They are both fantastic cards no matter how you look at them.
These take place on 2 seperate systems running at the same time with a clean install of windows.
I have a liscence to FRAPS. I will run AA/AF tests, and eyeball to see image quality, and run a basic level of similar IQ tests.
this is awesome, finnally we know we're going to get a pretty equal benchmark
Thanks alot AMD_Me, you're doing us all a big favor.
vBulletin® v3.8.7, Copyright ©2000-2012, vBulletin Solutions, Inc.