posted
I was interested in startin' a topic for computer benchmark comparisions on our machines at work, home, or otherwize.
What is the most widely accepted benchmark program lately?
I just downloaded PerformanceTest V4.0 from passmark.com. It seems simple enough, it tests a bunch of things, like cpu, ram, HD, Video, CD-rom etc. and gives you one combined score for the machine.
Anyone have a beter program?
I thought we could all post our machines scores. up here once we decide on the best benchmark program....
Posts: 2331 | From: Rosemount | Registered: Mar 2001
| IP: Logged |
Sandra lets you pick and choose what tests to run - CPU/File System/Memory/Cache/Lan. Or do a burn in and keep doing them over and over.
Memt86 is the standard memory module tester if you think that might be an issue.
Posts: 5484 | From: St. Paul, Mn | Registered: Apr 2002
| IP: Logged |
posted
Here are some real world benchmarks for 1942. These are all 1024x768 60hz. There is no official demo - you just have to turn on the FPS display and keep your eye on it. This can be done by typing in the console "console.showfps 1" (no quotes). Change the 1 to a 0 to turn off.
FX5900 128m XP2800 - 99 FPS 99% of the time. GF4 4200 128m XP2200 - 90FPS 80% of the time. GF3 500ti 64m XP1800 - 35FPS 80% of the time.
Posts: 5484 | From: St. Paul, Mn | Registered: Apr 2002
| IP: Logged |
posted
FX5200 128m XP1600 512m RAM runs 1942 20-30 FPS at 800x600 medium quality. IMO this card is a piece of shit. It's about equal to a GF4 MX400 (junk) card. I do not recommend this card even for a value system.
I haven't tried to overclock it yet - will post results when I get around to it.
Anyone have a ATI card? I was going to go that direction with HL2 coming out (free copy when you buy certain models) but with more delays for that game I decided against it.
Posts: 5484 | From: St. Paul, Mn | Registered: Apr 2002
| IP: Logged |
posted
I thought you were buying that FX5900 non-ultra?
I ran the 3dmark2001 last weekend on my machine, but really couldn't make any sense of the results. Overall score was in the 8800 level, but I didn't see results about FPS or anything, plus most of this is still another language to me when you start talking about how to optimize a system.
I added another 256 RAM after I ran that bench test, seems to run smoother. I also attempted messing with the built in overclocking setup on my vid card, but I must be running different drivers or something, because the setup/help they have does not match the displays/settings my computer is showing me (ie. I'm afraid to mess with it for fear of f'n up somehthing that works fine)
Posts: 2469 | From: Anchorage, AK | Registered: Oct 2001
| IP: Logged |
posted
Yes, I bought the FX5900 listed below. But I have a FX5200 that I purchased last summer to throw in that computer I was building for the shop. On Saturday night when we had 4 people playing 1942 (remember that Joe ) I had to set one of the systems up with the 5200 and it ran like shit.
If you run that 3dmark 2001 after the test there is an online results browser that will show you similar systems and their results. I would compare those scores. I think it gives you a summary of each test it ran and what the average fps was also. 3dmark 2003 saves a text file with all that info in it. Also run the in game console.showfps 1 in 1942 for your fps.
If you follow the directions for enabling coolbits in the Nvidia drivers you should see the clock freq area in the display properties. That's what you want to mess with. Test your system before and after. 3dmark is also a good way to test the stability of your overclocked settings. Just run the demo in a loop.... If it freezes up you have overclocked to high...
Posts: 5484 | From: St. Paul, Mn | Registered: Apr 2002
| IP: Logged |
posted
I've downloaded 3dmark 2003 and hope to try it this weekend: I have 1GB of new PC2700 RAM coming today and I'd like to compare before/after with my current PC133 @512MB, then I'll try the Coolbits and see where I end up. Right now everything runs smooth at 1024x768 with 2xQ AA and 2xAF forced. I can actually lightly recommend the FX5200 if you are upgrading from a GeForce 3 or less (it's what Jack's PC uses now) and things run smooth @ 800x600, Medium settings, NO forced AA or AF (and his system only has 128MB of PC100 RAM!).
Posts: 559 | From: Blaine, MN | Registered: Oct 2000
| IP: Logged |
posted
Make sure to do 3dmark 2001 also. The 2003 is only really accurate with dx9 cards. You can get a full serial number if you google for it. I have one what works with 2003 - will post it later.
Posts: 5484 | From: St. Paul, Mn | Registered: Apr 2002
| IP: Logged |
posted
Well I ran 3dMark'03 over the weekend and got some pretty shitty scores. Before the Memory upgrade to 1Gig I was around 2300 to 2500, after I only went up to 3500 (all perfomance settings). Where I score lowest is the CPU tests - I even drop to 1 FPS at points during the troll scene. This seems pretty jaded as I have an Athlon XP 2000 (and noe 1 GIG of PC2700).
I try to get 3dMark'01 and see where that gets me.
Scott, if more people try this, maybe you can set up a folder on the FTP site where we can upload those *.3dr files?
Posts: 559 | From: Blaine, MN | Registered: Oct 2000
| IP: Logged |
posted
Good idea about uploading the results. I will see what I can come up with.
I would try the same tests with 3dmark 2001. The 2003 tests expect a Directx 9 video card. If you don't have one your CPU will be the bottle neck and your results will be lower. Even my 5900 (pretty much the lowest Dx9 card) had trouble with the troll scene. 2003 is mostly for the super high end cards like the 5950 ultra and the Radieon 9800 Pro- everything else scores like shit. There has even been rumors that Mad Onion designed the tests to do alot better on ATI cards.
I like real-world tests the best. I remember Quake 3 had that built in demo that would run and return a FPS at the end. I however have lost my CD. And Unreal Turny 2003 has a built in one also (I can't find my key). I think there is a small program called FRAPS that you can run to give a FPS in any game, but I haven't tried it.
Posts: 5484 | From: St. Paul, Mn | Registered: Apr 2002
| IP: Logged |
posted
Call of Duty is supposed to have a timedemo, I'm going to try that as well as AquaMark3. Still haven't tried Coolbits to see where I can push it, maybe this weekend...
Posts: 559 | From: Blaine, MN | Registered: Oct 2000
| IP: Logged |
posted
So I ran my benchmarks and compared them to the online results. Seems I can probably get another 400 out of my computer. I got 6766, but some online results with my exact setup post in the 7100's. What are things I can do to up the score?
-------------------- "Rarely has it been so clear how much we, the ordinary people of this country, are better than our rulers. I hope that lesson is not lost on anyone, of any political persuasion." Posts: 520 | From: Shakopee, MN USA | Registered: May 2002
| IP: Logged |
posted
Overclocking is usually the difference: you can OC your RAM, CPU and Video card. OCWorkbench Tweaktown Otherwise look for bottlenecks: AGP slot speed, harddrive access time. I have successfully OC'd my RAM to a CAS Latency of 2 (from 2.5; Crucial is the king of memory) and so far SW:KOTOR has been stable (that's all I've been playing). It was a simple BIOS setting. Have not tried coolbits yet, but that's next...
HEADS UP: If you smell something like wood burning and you don't have a fireplace, your CPU fan went out! 121 Celcius baby!
posted
There are a bunch of BIOS setting you can mess with to squeek out a bit more. Everything by default is set low. I know I increased my AGP size and turned on the fast cacheing and got an performance increase. Consult a discussion forum about your video card or motherboard and see what others have done - that's how I figured out what to mess with.
Figure out where your bottle neck is.... Is it the CPU or the video card? Once you find that out you will know where to upgrade ect.
Posts: 5484 | From: St. Paul, Mn | Registered: Apr 2002
| IP: Logged |
My FX5700U started at 475/900 and I now have it at 480/909 and it's stable. I changed my CAS by .5 in the BIOS with no problems. My BIOS doesn't allow for CPU OCing, and I don't want to put in a hacked BIOS to do it, so that stands at XP2000. My new favorite benchmark is Aquamark3, its quick to test and you can see the immediate results of overclocking.
Current AM3 score: 28,694
I started under 26k, so my tweaking is definitely improving performance. I'll keep you posted where my OCing tops off.
Posts: 559 | From: Blaine, MN | Registered: Oct 2000
| IP: Logged |
posted
Here are my latest results. First, I thought I'd push that button that said "Automatic". Well, after some clicks and screen blanks, it came back with 522/1.0GHz! I just about pooped my pants. So I ran Aquamark 3 and got this: 29,885 But it got me thinking, so I looked at my Temp Gauge and for the first time ever, I had a couple of green temp bars showing. This made me think I might start having heat issues during long gaming sessions, so I trimmed back to 500/950 and AM3 was this: 29,086
Funny thing, though. The next time I boot up my PC, my memory clocking has been rolled back to 906. This has happened several times, so I'm going to leave it at that. I might try some real-game benchmarking next with TimeDemos to see how I fare there as AM3 by default sets AF to x4 and AA turned off (you can change these if you PAY for the software). I do like to run at x2Q for AA to get rid of most of the jagged lines; sure makes environments look much better.
C'mon bitches! Are you all afraid my rig will make your PCs look like tiny penises? You spoiled brats have broadband connections, download AM3 and see where you are.
My current system: Athlon XP 2000 1 Gig PC 2700 RAM (Which my MB cannot handle, so it only runs at 133) IBM Deathstar 40G @ 7200RPM (hope it lasts!) GeForce FX 5700 Ultra 475/900 OC'd to 500/906
Posts: 559 | From: Blaine, MN | Registered: Oct 2000
| IP: Logged |
It's a Nvidia Nforce2 chipset - supposed to be the fastest chipset out if run in dual channel mode (matching amounts of RAM in each of the two channels).
If the chips multiplier is unlock (mine is) you can change that in the BIOS. The memory speed can also be stepped up in 1pt increments. I tried these setups with my 2500 Barton:
2500 (1.83Mhz) 166*11=1826 (Default) 2800 (2.08Mhz) 166*12.5=2075 (no change in temp/voltage - running stable for 6 months) 3200 (2.20Mhz) 200*11=2200 (Temp up 2-4 deg. - unstable - might need to up voltage)
Would like to find something around 3000 speed like 12*180 - but haven't messed with it yet.
quote:Originally posted by Cremator: Here are my latest results. First, I thought I'd push that button that said "Automatic". Well, after some clicks and screen blanks, it came back with 522/1.0GHz!
B. What were you using that gave you the automatic choice for OC'ing. Coolbits that I dwnloaded didn't have that.
Posts: 2469 | From: Anchorage, AK | Registered: Oct 2001
| IP: Logged |
posted
Coolbits is activated by going into the registery and changeing one value. This makes Coolbits show up under display properties. It's built into the Nvidia drivers.
posted
I know, I did that. But the feature it added for OCing didn't include a "Automatic" function so I was curious where Cremator was able to do this.
quote:Originally posted by Cremator: First, I thought I'd push that button that said "Automatic". Well, after some clicks and screen blanks, it came back with 522/1.0GHz!
posted
Sorry Joe, I don't know what the exact name of the button was, but it's right next to the TEST button when you enable Coolbits. Also, I am using the 56.xx drivers (which meant I had to re-enable Coolbits in the registry after upgrading).
Posts: 559 | From: Blaine, MN | Registered: Oct 2000
| IP: Logged |
posted
Under the clock freq area there is a automatic button. It will run a couple tests and then set it to the highest safe speed. I also had to redo the coolbits deal in the registry after the latest driver.
Posts: 5484 | From: St. Paul, Mn | Registered: Apr 2002
| IP: Logged |
quote:Originally posted by Cremator: but it's right next to the TEST button when you enable Coolbits.
Thanks, I don't have it. I havent' updated drivers on mine in a while. And they didn't work on Lukes Ti4200 ?? so I'm not messing with them, my machine works fine.
Posts: 2469 | From: Anchorage, AK | Registered: Oct 2001
| IP: Logged |
posted
Wow, that card overclocks pretty good tho. What were the scores before and after overclock??
Posts: 5484 | From: St. Paul, Mn | Registered: Apr 2002
| IP: Logged |
posted
ok I tried to do the auto detect and it came back 0/0 wtf???? anyone have any imput, gf4 ti4200 128ddr 8x so help if you can
Posts: 208 | From: Chaska MN | Registered: Aug 2002
| IP: Logged |
posted
Never seen it come back with zeros, but my old PC might have done that if it had worked: instead when I hit the auto, it tried a few settings, then REBOOTED!
I think your going to have to do this manually, bumping your speeds 1,5,10, etc. MHz at a time (whatever you're comfortable with) and test each setting - play a graphic intense game for a while and make sure the system is stable, then bump it again until you either become unstable of the temperature looks like it will become a problem.
If coolbits doesn't work for youm there are two others I know of but never tried: Rivatuner and nVHardpage.
Good luck!
Posts: 559 | From: Blaine, MN | Registered: Oct 2000
| IP: Logged |
posted
I always run the AM3 or 3DM Demo in a loop overnight to see if it's stable. I think the best thing I ever did to improve my overclocking was buy the big case with 6 fans in it. The Volcano CPU cooler set at 2000 RPMS (goes to 6000) and 2 input fans and 2 output fans have created a very stable system. Previously I had heat issues with only one intake and a very loud CPU fan - the case was just too hot (so it didn't matter how great my CPU cooler was)
Posts: 5484 | From: St. Paul, Mn | Registered: Apr 2002
| IP: Logged |
posted
If you really want to have a cool case you need a 120mm variable speed intake or exh fan. I have 80mm exh fans and 120mm variable speed intake fan. Having a variable speed intake in effect also makes the exhaust fans variable speed as well so it works out. Another good reason is when the 120mm fans start to go YOU KNOW IT.
Case starts to sound like a jet.
Posts: 2331 | From: Rosemount | Registered: Mar 2001
| IP: Logged |
Halo Time Demo (1024*768,AA=off, all settings max) 2500 (12*166=1.84GHZ) 44.34FPS 2800 (12.5*166=2.10GHZ) 44.57FPS 2800 (12.5*175=2.20GHZ) 45.26FPS
Halo Time Demo - GFX OC to 5950Ultra speed (475/950) 2500 (12*166=1.84GHZ) 52.20FPS 2800 (12.5*166=2.10GHZ) 52.34FPS 2800 (12.5*175=2.20GHZ) 52.43FPS
Conclusion - Video card is the bottle neck on my system. Changing the CPU speed did not net any real FPS gains.
Posts: 5484 | From: St. Paul, Mn | Registered: Apr 2002
| IP: Logged |
posted
One thing I noticed while using Coolbits this weekend with the newer drivers - I had to select Manual Overclock, then change the drop downlist to 3D - after that the auto button was active. I hit that and it tested and came up with a higher setting. I bumped it a further still and hit test (and it passed I guess). Then I checked apply at startup so that I wouldn't have to re-do the overclock everytime I reboot.
Posts: 5484 | From: St. Paul, Mn | Registered: Apr 2002
| IP: Logged |
posted
For some reason, my RAM clock speed (higher #) often will reset itself to default when using Coolbits. Have you seen this at all Scott? I use the "Apply at startup" checkbox and always choose test, then Apply, then OK. Both my PCs do this. This happens when using the 56.56 and 53.03, cards are from different mfg, both using Win XP, SP1 DX9.0b...
Posts: 559 | From: Blaine, MN | Registered: Oct 2000
| IP: Logged |