Well, I posted these in my livejournal, but I figured a few people here might be interested. Here's the quick and dirty cut and paste with my wonderful commentary. (naughty)
<TABLE cellSpacing=0 cellPadding=5 width="100%" summary="" border=0><TBODY><TR align=left><TD class=caption>G33K T1M3</TD><TD class=index align=right>[05 Aug 2004|05:02pm]</TD></TR><TR align=left><TD bgColor=#ffffff colSpan=2>Well, I took a half day, going to upgrade my nephews computer (add more ram from 256 to 768) so that he can play doom3 with better framerates.
This got me thinking. I had several hours to kill, so I decided to benchmark both my machines using the "timedemo demo1" command in doom3. ( CTRL+ALT+~ to pull up the console)
ATI released new drivers today which were specifically designed to raise performance in doom3. Not only do they work, but they are actually amazing the hell out of me.
Machine 1 - AMD 2400XP, 1.5GB DDR266, Soyo Dragon 333 Ultra Platinum, ATI 9800pro 128MB Game is installed on 1250GB Raid5 array (might decrease load times on the demo - I dunno - but once it's been run once, it's already cached).
Machine 2 - Intel P4 2.8E, 1GB DDR400 Dual Channel, Intel 865GBF, Chaintech Geforce FX5950 256MB Game installed on normal IDE Drive ( 80GB WD )
Machine 1's CPU holds an average running temp of 154 degrees F during the test.
Machine 2's CPU holds an average running temp of 124 degrees F during the test.
Benchmarks were made 3 times at each setting. The first is always a bit slower because of caching, but - it is indicative of performance when entering a new area. Granted, the timedemo might not be the best benchmark, but it's good enough, and was easy to use.
First I benchmarked Machine 1 using the most current Certified Driver
Medium Detail 800x600
NO AA - 27 / 32.8 /32.8
2x AA - 28 / 30.9 /31
4x AA - 24.7 / 26.9 / 26.8
Then with the New improved Driver released today
No AA - 27.9 / 33.9 / 34.4
2x AA - 28.9 / 32.2 / 32.1
4x AA - 25.2 / 27.5 / 27.4
Yeah, so - it bumped performance up about a frame or two straight across the board. So I started thinking... surely my p4 would fare better. After all, it's faster, the cpu has more cache, the video card has double the ram, and is faster. Granted the P4 has 1GB of ram as compared to the AMD's 1.5GB. But it's Dual Channel in the p4 and also DDR400 compared to the AMD's DDR266. Hmm... I should mention that DDR266 vs. DDR400 isn't going to make a big difference. I've seen benchmarks, and built lots of machines. While you would expect it to make a huge difference, it really doesn't. Also, the AMD is my daily use machine, got tons of stuff loaded, Inksaver, Livejournal, Norton AV, Winamp Agent, Creative Volume Control and of course Spybot. The P4 on the other hand, has loaded, norton AV, Norton Ghost, and Intel's Active Monitor. Last of all, the FSB on the AMD is 333mhz, while on the Intel it's 800mhz. The Intel machine also cost a hell of a lot more to build ( not counting the RAID array on the AMD ) as compared to the AMD.
So, I Benched the p4, at the same settings.
Medium Detail 800x600
No AA - 33.2 / 35.6 / 35.6
2x AA - 31.4 / 33 / 33
4x AA - 29.1 / 30.4 / 30.4
Hmm... Not bad, the intel pulls ahead a bit. but only marginally. Maybe I'm just not pushing the machines hard enough. Lets push things a little harder and kick up those settings.
AMD Tests
800x600 Very High Quality
No AA - 27.9 / 33.4 / 33.3
2x AA - 27.2 / 31.1 / 31
4x AA - 23.3 / 25.7 / 25.8
1024x768 Very High Quality
No AA - 25.9 / 29.3 / 29.3
2x AA - 23.9 / 26.5 / 26.6
4x AA - 19 / 20.4 / 20.5
Intel Tests
800x600 Very High Quality
No AA 28.7 / 31 / 30.9
2x AA - 26.8 / 28.8 / 28.8
4x AA 24.8 / 26.4 / 26.4
1024x768 Very High Quality
No AA - 22.3 / 23.6 / 23.6
2x AA - 20.7 / 21.6 / 21.7
4x AA - 18.8 / 19.7 / 19.7
So - what the **** happenned??? My best guess is that the extra 512MB of RAM really helps alot. I've had both the ATI and the nVidia card in the P4, and in 3dmark the scores were significantly higher with the nVidia. That's why the ATI card is in my AMD machine, I thought the best performing video card should be in the stronger machine. Also, Running 3dmark 03 on both machines, the P4 scored higher. Since I've gone through all this trouble though - might as well run 3dmark too on both machines. That will be the next post.
Also the driver on the nVidia card is 5672. A new version is available 6177. I'll upgrade tonight and see how it changes. </TD></TR></TBODY></TABLE>
<TABLE cellSpacing=0 cellPadding=5 width="100%" summary="" border=0><TBODY><TR align=left><TD class=caption>G33K T1M3</TD><TD class=index align=right>[05 Aug 2004|05:02pm]</TD></TR><TR align=left><TD bgColor=#ffffff colSpan=2>Well, I took a half day, going to upgrade my nephews computer (add more ram from 256 to 768) so that he can play doom3 with better framerates.
This got me thinking. I had several hours to kill, so I decided to benchmark both my machines using the "timedemo demo1" command in doom3. ( CTRL+ALT+~ to pull up the console)
ATI released new drivers today which were specifically designed to raise performance in doom3. Not only do they work, but they are actually amazing the hell out of me.
Machine 1 - AMD 2400XP, 1.5GB DDR266, Soyo Dragon 333 Ultra Platinum, ATI 9800pro 128MB Game is installed on 1250GB Raid5 array (might decrease load times on the demo - I dunno - but once it's been run once, it's already cached).
Machine 2 - Intel P4 2.8E, 1GB DDR400 Dual Channel, Intel 865GBF, Chaintech Geforce FX5950 256MB Game installed on normal IDE Drive ( 80GB WD )
Machine 1's CPU holds an average running temp of 154 degrees F during the test.
Machine 2's CPU holds an average running temp of 124 degrees F during the test.
Benchmarks were made 3 times at each setting. The first is always a bit slower because of caching, but - it is indicative of performance when entering a new area. Granted, the timedemo might not be the best benchmark, but it's good enough, and was easy to use.
First I benchmarked Machine 1 using the most current Certified Driver
Medium Detail 800x600
NO AA - 27 / 32.8 /32.8
2x AA - 28 / 30.9 /31
4x AA - 24.7 / 26.9 / 26.8
Then with the New improved Driver released today
No AA - 27.9 / 33.9 / 34.4
2x AA - 28.9 / 32.2 / 32.1
4x AA - 25.2 / 27.5 / 27.4
Yeah, so - it bumped performance up about a frame or two straight across the board. So I started thinking... surely my p4 would fare better. After all, it's faster, the cpu has more cache, the video card has double the ram, and is faster. Granted the P4 has 1GB of ram as compared to the AMD's 1.5GB. But it's Dual Channel in the p4 and also DDR400 compared to the AMD's DDR266. Hmm... I should mention that DDR266 vs. DDR400 isn't going to make a big difference. I've seen benchmarks, and built lots of machines. While you would expect it to make a huge difference, it really doesn't. Also, the AMD is my daily use machine, got tons of stuff loaded, Inksaver, Livejournal, Norton AV, Winamp Agent, Creative Volume Control and of course Spybot. The P4 on the other hand, has loaded, norton AV, Norton Ghost, and Intel's Active Monitor. Last of all, the FSB on the AMD is 333mhz, while on the Intel it's 800mhz. The Intel machine also cost a hell of a lot more to build ( not counting the RAID array on the AMD ) as compared to the AMD.
So, I Benched the p4, at the same settings.
Medium Detail 800x600
No AA - 33.2 / 35.6 / 35.6
2x AA - 31.4 / 33 / 33
4x AA - 29.1 / 30.4 / 30.4
Hmm... Not bad, the intel pulls ahead a bit. but only marginally. Maybe I'm just not pushing the machines hard enough. Lets push things a little harder and kick up those settings.
AMD Tests
800x600 Very High Quality
No AA - 27.9 / 33.4 / 33.3
2x AA - 27.2 / 31.1 / 31
4x AA - 23.3 / 25.7 / 25.8
1024x768 Very High Quality
No AA - 25.9 / 29.3 / 29.3
2x AA - 23.9 / 26.5 / 26.6
4x AA - 19 / 20.4 / 20.5
Intel Tests
800x600 Very High Quality
No AA 28.7 / 31 / 30.9
2x AA - 26.8 / 28.8 / 28.8
4x AA 24.8 / 26.4 / 26.4
1024x768 Very High Quality
No AA - 22.3 / 23.6 / 23.6
2x AA - 20.7 / 21.6 / 21.7
4x AA - 18.8 / 19.7 / 19.7
So - what the **** happenned??? My best guess is that the extra 512MB of RAM really helps alot. I've had both the ATI and the nVidia card in the P4, and in 3dmark the scores were significantly higher with the nVidia. That's why the ATI card is in my AMD machine, I thought the best performing video card should be in the stronger machine. Also, Running 3dmark 03 on both machines, the P4 scored higher. Since I've gone through all this trouble though - might as well run 3dmark too on both machines. That will be the next post.
Also the driver on the nVidia card is 5672. A new version is available 6177. I'll upgrade tonight and see how it changes. </TD></TR></TBODY></TABLE>
I'll post any info I have from my sad sad GeForce2 and my soon to arrive FX5200 just for comparison sake..
