PDA

View Full Version : 5200 Ultra owners thread


ragejg
07-29-03, 10:08 AM
I know, not many have this card, so I may bump this thread once in a while...

To further confirm validity of bjorn3d's figures on the ASTRONOMICAL increase in AF/AA performance when overclocking, I'd like some 5200U owners to please run 3dmk2k1 @ 1600x1200 with 4XAA/AF @ stock speeds... then OC it as far as possible and record the results (with 4xAA/AF)...

Same for 1280x1024...

Same for 1024x768...

Same for 800x600...

Not everyone has time for this, but MAYBE we can tabulate enough figures to make some sense out of this... 16x12 is the most important for studying this fact with 3dmk2k1...

ALSO: This thread can also be used as a "general user experience" thread FOR 5200U owners... no 5200, no 5600/U.... you know what I mean... :rolleyes: :)

I DON'T have this posted everywhere else... I think we might be able to get enough info from nvnews users...

:)

Malfunction
07-29-03, 12:04 PM
I'm game, although I would like to ask you for a bit more detail in what settings you wish to use:

Performance and Quality Settings

Would you like it set to: High Performance/Performance/Quality

Direct 3D Settings

What would you like Mip Map Detail set to?

Just trying to be as accurate for you as you wish. This would also help me out as well knowing we are all on the same track. I have no problem doing this, so let me know. :)

Peace,

:)

ragejg
07-29-03, 12:35 PM
Performance and Quality Settings:
~middle setting (corrcet perf/IQ balance IMO)

Mipmap detail:
~middle setting (correct perf/IQ balanc IMO)

...I am also interested in the Single Texturing/MultiTexturing capability increase with OC'ing on this card...

:)

Malfunction
07-30-03, 11:58 AM
Hey ragejg,

don't wanna be a pain...lol, but did 4xAA/AF mean 4xAA/4xAF? I was looking over the Bjorn3D review and noticed that it was benched at 0xAA/0xAF and 4xAA/8xAF. If 4xAA/AF meant 4xAA/4xAF then that is cool, because I have already started... hehe. Thing that confused me is you made reference to the Bjorn3D review yet it seemed like you suggested we bench differently than was done in the review is all. :thinker:

I stopped to clarify though and because I am getting myself some new goodies that will arrive next week (Woohoo!:D ). So, if you don't mind you are gonna have to wait a bit. :(

Thanks again for the clarification. :)

Peace,

:)

ragejg
07-30-03, 02:10 PM
here's the review I was looking @...

http://www.************/_preview.php?articleID=241&pageID=243

and here's the thread with the pics I pruned...

http://www.nvnews.net/vbulletin/showthread.php?s=&threadid=15432

I see nothing making reference to 8XAF...

And by 4XAA/AF, yes I mean 4x for both...

oh, and i may have forgotten... I think it'd be good if tex. sharpening was turned off...

sorry if I muddled the requirements... ;)

Malfunction
08-06-03, 07:04 AM
I am a man of my word. :D I ran 3DMark2001SE Build 330 with the following settings:

Asus P4C800-E Deluxe 875P
Pentium 4 2.8GHz,800FSB@3.0GHz,860FSB
Micron [Model#: MT16VDDT6464AG-40BC4] 1024MB DDR @ 344Mhz 2-3-3-7, Dual Channel
8xAGP
Windows XP Pro/SP1-DX9b-Detonator 44.67

FX5200U @ Default 325/650
FX5200U @ O/C 370/740 *Forgot I can go to 750Mhz on the memory... sorry. :( I will update later
Vsync Off-Perfrmance Mode-4xAA/AF-Mip Map Detail-Blend

1600x1200 4xAA/AF

Default: 2015points (http://service.futuremark.com/compare?2k1=6849587)

Overclocked: 2324points (http://service.futuremark.com/compare?2k1=6849630)

1280x1024 4xAA/AF:

Default: 2846points (http://service.futuremark.com/compare?2k1=6849805)

Overclocked: 3277points (http://service.futuremark.com/compare?2k1=6849726)

1024x768 4xAA/AF:

Default: 4520points (http://service.futuremark.com/compare?2k1=6849777)

Overclocked: 5121points (http://service.futuremark.com/compare?2k1=6849756)

800x600 4xAA/AF:

Default: 6574points (http://service.futuremark.com/compare?2k1=6849818)

Overclocked: 7453points (http://service.futuremark.com/compare?2k1=6849654)

**************************************************

Well, that's it. :) I am very pleased with my EVGA FX5200U regardless of what people feel about Nvidia or the NV3x line of videocards. I still feel confident I made the right purchase for myself. I was able to overclock my CPU to 3.47Ghz thus far, although the retail HSF doesn't cool well enough for my needs and that is why it was benched at 3.0Ghz instead. Run's stable @ 3.47Ghz and can go higher... just wanna get my Swiftech MCX-4000 to be on the safe side. :) I like to be sub 50c not @ 53c... hehe. ;)

If ya need anything else, just hollar.

Peace,

:)

ragejg
08-06-03, 08:34 AM
I'm really wondering what that extra memory mhz is gonna do...

Nice numbers, btw, thanks a lot for doin that...

Bjorn3d's ran @ 362/775, so your numbers should be close...

I'm not gonna say "bjorn3d had a typo in that review" till we see what scores look like with the memory a teeny bit higher... maybe the card scales great in that respect... maybe it doesn't... hmmm...

:)

Malfunction
08-06-03, 02:53 PM
Originally posted by ragejg
I'm really wondering what that extra memory mhz is gonna do...

Nice numbers, btw, thanks a lot for doin that...

Bjorn3d's ran @ 362/775, so your numbers should be close...

I'm not gonna say "bjorn3d had a typo in that review" till we see what scores look like with the memory a teeny bit higher... maybe the card scales great in that respect... maybe it doesn't... hmmm...

:)

Thanks for the compliment :D , and you are welcome. I decided I was going to increase my AGP voltage from 1.5v to 1.8v. After making sure everything was ok, I hit Auto-Detect too see what it would say I could overclock to. Found that it suggested 385/720. This was pretty low for the memory, however the core was just amazing from the .3v increase. After finding that out I decided to put the core back to default clock speed and see how far I could push the memory since I already found I could hit 750Mhz. Well, it was good for a 10Mhz increase for a total of 760Mhz. I tried 385/760 and it locked up, so I lowered the core to 382Mhz and left the memory unchanged. Success!!! :D The following are updated 3DMark2001 SE Build 330 results.

1600x1200 4xAA/AF:

2375points (http://service.futuremark.com/compare?2k1=6850886)

1280x1024 4xAA/AF:

3296points (http://service.futuremark.com/compare?2k1=6850933)

1024x768 4xAA/AF:

5151points (http://service.futuremark.com/compare?2k1=6850968)

800x600 4xAA/AF:

7433points (http://service.futuremark.com/compare?2k1=6850987)

I have no idea how bjorn3d managed 775Mhz on the memory. Although they are probly wondering how I managed 382Mhz on my core...lol.:D My case/ambient temp is 35c. Kinda more than happy enough with this card and think I will just avoid getting ram sinks or a new HSF for it. Not bad for all retail components and no volt mode, just an AGP voltage increase. :D

Hope everyone enjoyed the results, I have... hehe. :D

Peace,

:)

ragejg
08-06-03, 05:10 PM
hmmm...

how in the world would they have gotten such a high score, then?

Malfunction
08-06-03, 05:19 PM
I have no idea... seems a bit rediculous to me. I did my best and they were only using a AMD XP2000+ lol.

I have a huge core Mhz increase compared to them. I didn't see what drivers they were using but I think it is a type-o more than anything. :(


Peace,

:)

ragejg
08-06-03, 05:22 PM
...and, um... what's your flat out regular 10x7 score, no AF/AA, all options on highest performance?

My guess in 10.5k...

:)

Malfunction
08-06-03, 05:39 PM
Originally posted by ragejg
...and, um... what's your flat out regular 10x7 score, no AF/AA, all options on highest performance?

My guess in 10.5k...

:)

Actually...

High Performance Mode/No Vsync/Mip Map Detail: Best Performance/382/760Mhz

1024x768 NoxAA/AF:

11349points (http://service.futuremark.com/compare?2k1=6851560)

Just ran it... lol

Peace,

:)

ragejg
08-06-03, 07:14 PM
sorry folks, but I have to at least say that THAT score is NICE! THough the bjorn3d figure appears to mislead, IMO when OC'ed, the card gives nice results...

Malfunction
08-09-03, 11:56 PM
Hey ragejg,

thought you might enjoy this. I reran the bench tonight with improved results. I decided to run the memory @ 1:1 ratio this time. I bumped the voltage to 2.85v for it and the timmings remained the same: 2.5-3-3-5. It is running at 430FSB now stable. CPU is still running @ 215FSB or 3012.8Mhz. It is a little warm in here so videocard wouldn't allow memory to go back to 760Mhz, so I settled for 758Mhz.

1024x768 32bit/NoxAA/AF/Vsync: Always off/MipMap Detail: Best Performance/High Performance Mode

Here are the results:

11570points (http://service.futuremark.com/compare?2k1=6864114)

*I can't wait to get that Swiftech MCX-4000 (http://www.swiftnets.com/products/mcx4000.asp) and run CPU at 250FSB with DDR @ 5:4 ratio! :D

Peace,

:D

bkswaney
08-10-03, 12:10 AM
Originally posted by Malfunction
Hey ragejg,

thought you might enjoy this. I reran the bench tonight with improved results. I decided to run the memory @ 1:1 ratio this time. I bumped the voltage to 2.85v for it and the timmings remained the same: 2.5-3-3-5. It is running at 430FSB now stable. CPU is still running @ 215FSB or 3012.8Mhz. It is a little warm in here so videocard wouldn't allow memory to go back to 760Mhz, so I settled for 758Mhz.

1024x768 32bit/NoxAA/AF/Vsync: Always off/MipMap Detail: Best Performance/High Performance Mode

Here are the results:

11570points (http://service.futuremark.com/compare?2k1=6864114)

*I can't wait to get that Swiftech MCX-4000 (http://www.swiftnets.com/products/mcx4000.asp) and run CPU at 250FSB with DDR @ 5:4 ratio! :D

Peace,

:D


Why on earth r u running a 5200 with your rig? :confused:
A high end system should have a high end card. :)

Malfunction
08-10-03, 12:14 AM
Didn't want to spend a whole lot reintroducing myself to Nvidia products. Now that I see features that I like more than ATi's control panel I will be awaiting the arrival of that NV38 :D

Then... my PC should be complete. ;) I figured someone would have asked sooner or later...lol. Thanks for asking.

Peace,

:D

yoladude
08-10-03, 01:12 AM
Originally posted by ragejg
sorry folks, but I have to at least say that THAT score is NICE! THough the bjorn3d figure appears to mislead, IMO when OC'ed, the card gives nice results...

i'll have to call their results b.s. sorry, i'm not an owner, and i know u probably wont like me getttin off topic, but i think they must have made the mistake of resetting the settings to 1024x768 while oc'ed. because if you look into an above graph, it gets 4259 marks when stock clocked with 4xAA/AF at 10x7. at 1600x1200 stock, it gets 1905, which makes sense. but when oc'ed, it gets 5337. so, this leads me to believe that they ran that test at 1024x768 oc'ed.
just my .02

Malfunction
08-10-03, 03:31 AM
Originally posted by yoladude
i'll have to call their results b.s. sorry, i'm not an owner, and i know u probably wont like me getttin off topic, but i think they must have made the mistake of resetting the settings to 1024x768 while oc'ed. because if you look into an above graph, it gets 4259 marks when stock clocked with 4xAA/AF at 10x7. at 1600x1200 stock, it gets 1905, which makes sense. but when oc'ed, it gets 5337. so, this leads me to believe that they ran that test at 1024x768 oc'ed.
just my .02

Did you miss what mine does at default clock speeds?

FX5200U @ Default 325/650
FX5200U @ O/C 382/760
Vsync Off-Perfrmance Mode-4xAA/AF-Mip Map Detail-Blend

1600x1200 4xAA/AF

Default: 2015points (http://service.futuremark.com/compare?2k1=6849587)

Overclocked 2375points (http://service.futuremark.com/compare?2k1=6850886)

1280x1024 4xAA/AF:

Default: 2846points (http://service.futuremark.com/compare?2k1=6849805)

Overclocked 3296points (http://service.futuremark.com/compare?2k1=6850933)

1024x768 4xAA/AF:

Default: 4520points (http://service.futuremark.com/compare?2k1=6849777)

Overclocked 5151points (http://service.futuremark.com/compare?2k1=6850968)

800x600 4xAA/AF:

Default: 6574points (http://service.futuremark.com/compare?2k1=6849818)

Overclocked 7433points (http://service.futuremark.com/compare?2k1=6850987)

Peace,

:confused:

ragejg
08-11-03, 07:18 AM
Ok, cool, we've made things a little learer on the "performance hit of AF/AA" side of things...

Now what I'd like to do is get some more definite words on IQ for the 5200 series... I know that it's missing some features... Intellisample being one of them... something else, too... I can't seem to remember yet (I have not had any coffee yet :rolleyes: )...

I'm operating under the assumption that the 5200 series uses the same algorythyms and hardware as the GF4 series for AF & AA...

This is true, correct?

... I'd just like to narrow this down in THIS thread...

btw, Malfunction: thatnks a lot for all the effort you've put into this thread. You've made it a good read.

:)

Malfunction
08-11-03, 12:52 PM
Originally posted by ragejg
btw, Malfunction: thatnks a lot for all the effort you've put into this thread. You've made it a good read.

:)

Hey it's been my pleasure. I kinda know what the guys with the 5800's have gone through now I feel. Although there was some advantages to that card, it was mainly the focus of attention for slamming... that was just wrong. :( I know the FX5200U wasn't suppose to compete with any mid-range or high end graphics cards, although it performs surprisingly well for a sub $150 DX9 card. I am a very happy owner to say the least and don't mind at all showing others what I see when playing games on it. The card has been one of the best purchasing decisions I have made. It has totally allowed me to see both sides of the spectrum, and I never thought I would be saying this but I do like the features in Nvidia's control panel better than the ATi control panel. I can also see how many former Nvidia based users get confused when using ATi cards/control panel as well becuase it is some what different. I wish ATi adopted a hidden overclocking utility like Nvidia's coolbits as well as the onboard GPU Temp/Ambient temp monitoring features that are very accurate. I notice that the temps matched that of The Asus Probe software.

Till then, I will be looking foward to the NV38 for my first Nvidia Enthusiast card in ages. :D I know there will be alot of people performing test with it, although if you would like me to run test/benching I will be more than happy to oblige. :)

Peace,

:)