2008 XS Cheapazchips 3D challenge

[GOESA]

Kaffebord, I think water should be okay in standard class I don't think they do much better with water

My card is getting here tommorow! wow Iam exited to get to play with it

/Grau

[Johnny Bravo]

haha. thanks for the interest stevil. I will do. card kept cutting out, so i have linked the mem as well. only two wires, and solved everything. The gpu core right now is around 1.35-1.4v, and the mem i got a link giving 2.0-2.05v, so about right areas. GPU is going to be increased though. Get pic in like 5-10 minutes!

- Paul

Edit - here ya go. That is what it looks like now!

NICE!

Though get ALOT THICKER wiring on the go as that is gonna great massive droop under load

[G H Z]

Shortening the wire could help also.

[SpikeP]

well i dont have thicker wire at the moment. using the molex wire. What i have done is shorten the wires by about half, and it has made quite a difference. hopfully get a 9K+ bench in on 03 soon on ORB. Dam thing won;t let me delete them!

- Paul

[Johnny Bravo]

well i dont have thicker wire at the moment. using the molex wire. What i have done is shorten the wires by about half, and it has made quite a difference. hopfully get a 9K+ bench in on 03 soon on ORB. Dam thing won;t let me delete them!

- Paul

Just do multiple runs of the same wire and you'll be flying

[ownage]

I use a thick short wire now, but that doesn't change anything.

[zakelwe]

Well I am a bit behind with updates as I installed my Sli 780i motherboard today with dual 8600GT it is rather more exciting than updating a page.

Check out this score

http://service.futuremark.com/result...6&resultType=6

and they say 780i is a poor chipset. Not for 3dmark it's not. It does a 10.1s SuperPi 1M as well air cooled after only 4 hours of setup.

Now, been talking to the guys at Futuremark, Risto says this about publishing scores.

"the fix went live today, it was a case of wrong component information passed to the method which was then used to override the correct component information. Fastest way to fix this is to unpublish and then re-publish the project so the change reflects to the search table in the DB. I will make a script today that remaps all display adapters on published 3DMark 2001 Projects so the searching table will contain the projects correctly."

So if you cannot publish a result try unpublish and publish again. I also asked them to change that when you do a compare you have to put the criteria in again rather than just go back to the list to see what happens.

I will update scores in a minute.

Regards

Andy

[zakelwe]

Why limit det "standard" class to air only? Water isn't that much better IMO. You can run air outside in -20C, but you can't run water inside in +20C. I find that kinda silly. Anyone agree?

(The reason I write this is because I found out that my old ati 9800pro waterblock fittet perfectly )

I knew I would create some rules and then halfway through the comp somone would try to bend them completely. I guess it wouldn't be XS if that didn't happen

Considering we have cards wired into Frankenstein at the moment I doubt your water block will create too much of a tidal wave. So give it a bash and good luck.

Regards

Andy

[ownage]

Andy, check my Edit2, i had a higher score 26k. Thats more editing for you hehe. But first play with that 780i and hope it doesn't turn in an typical Nforce hell

[zakelwe]

Ok, added, but what are your card clocks for that run?

Regards

Andy

[ownage]

http://xs224.xs.to/xs224/08073/3dm01_xs_26k885.jpg

clocks: 834/1935/463

[[XC] DragonOrta]

Finally realized that my PCI-E x16 speed was at x1 on my P35-DS3P when after spending some time pushing my card, I was still only scoring around 7400 in '03 no matter how much I pushed the clocks.

Changed to my MSI Neo2-FR and am getting close to 8500 at the same clocks.

[zakelwe]

Ok ownage, I just checked your nature and with your original score you are getting 140fps with 800core and only 450 memory which is far higher than anyone else

Nature is memory limited and I only get 135 with 835/565. Stevil does 864/475 and gets less also. We are obeying rule 4 on the front page for this competion unlike the dutch comp so you may wish to resubmit taking this into account. You may not have seen this being a late entrant.

Regards

Andy

[dinos22]

SpikeP

far out man
that's the coolest thing i've ever seen lol

[[XC] DragonOrta]

24779 648/1512/549

Just a real quick '03 after changing mobos.

[Rol-Co]

Spike ....youre my man !!! , did the same overhere with a 6600



other pics

Memory cooling




bad contact on the psu fet....solved now, working great.


icecream anyone


Where using extreme lod overhere, and beyond....its allowed because you can't check it if somebody using it so ifits allowed everybody can use it.
not allowed is cheating, wrong resolution .
Its a bit different overhere , when you can win a high end card people do strange things, if its just for fun only like overhere you can play different.

can run nature at 1053/2700/506 atm with the stock vapochill ls.

[STEvil]

im going to try and get my water cooling setup going tomorrow and strap it to the GPU..

edit - I found a mosfet on the Asus 8400GS thats cresting 90c as well... very edge of the board next to the transistor used to do the vmem mod...

[dinos22]

im going to try and get my water cooling setup going tomorrow and strap it to the GPU..

edit - I found a mosfet on the Asus 8400GS thats cresting 90c as well... very edge of the board next to the transistor used to do the vmem mod...

PICs or ban

[Micutzu]

People, you don't need much stronger GPU supplies, the onboard one is good enough and with mods it can provide what's needed.

An off-board supply is not that easy to use properly, you need to pay attention to wire voltage drop, you need remote sense, you need a way to clean the voltage of the EMI picked up by the connection wires and the hardest thing, you need to take care somehow of the poor transient response a remote supply has - especially if it's a linear design.

The onboard supply doesn't have these problems and with upgraded RC components, raised switching frequency and decent cooling it can handle exactly what the tiny 8400GS chip needs.

[zakelwe]

Rol-co's poor old XFX ... Excellent work.

Yes, the Dutch competion rules on software tweaks are a lot more flexible than ours, which are more geared to people to just doing the hardware side.

Regards

Andy

[Deanzo]

The ORB is taking my 03 scores again

3D03: 9940 http://service.futuremark.com/result...eResultType=10


Temps seem fine, vmod is good, but as soon as I go over 1.35v it reboots and wont screen post till I go back to or under 1.35v.
In 01, I can run everything other than CH @ 740 on that vcore, if I could just get more volts into her it may just fly, or blow up. But I'm happy either way

1.26 is stock on my card.

Any ideas would be great, if I get it working right I my just mount my pot on it with zip ties, just for fun.

[G.Foyle]

DragonOrta, try changing PCI-E frequency to 105 and PCI-E voltage to +0.1 on that Gigabyte board. That should stop it from falling down do PCI-E 1x.

[Rol-Co]

People, you don't need much stronger GPU supplies, the onboard one is good enough and with mods it can provide what's needed.

An off-board supply is not that easy to use properly, you need to pay attention to wire voltage drop, you need remote sense, you need a way to clean the voltage of the EMI picked up by the connection wires and the hardest thing, you need to take care somehow of the poor transient response a remote supply has - especially if it's a linear design.

The onboard supply doesn't have these problems and with upgraded RC components, raised switching frequency and decent cooling it can handle exactly what the tiny 8400GS chip needs.

Not realy, power management of the xfx cards dieing in a straight line overhere,....in fact , the fets are burning literly of the cards, and cooling was sufficient (-8 air) ....if you want to go realy high you need external supply.

I have 0.03 voltage drop at 2.70v , clocks where 1070/2700/501 , biggest problem is youre psu, mine (1kw version) dropt to much on the 5v line, had to get the 620w enermax psu who was a little better at the 5v.

[LyP0]

Hey rol-co you here ?? :p

tried to bench my card yesterday but it went a bit wrong. The thing is, i didnt pay atention to the vgpu. Had quite a good run, 1080/2862/472 wich gave me a score of 164.8 fps in nature.
But sudenly the card started to do weird things, couldent even finish a run on very stable settings anymore. Then i looked over at the multimeter and saw " 2.78"
ofcourse i set it back to 2.3 ish and tried again. No luck.
This morning i woke up and saw my card had some weird colors round the gpu.

[[XC] DragonOrta]

DragonOrta, try changing PCI-E frequency to 105 and PCI-E voltage to +0.1 on that Gigabyte board. That should stop it from falling down do PCI-E 1x.

I've had the board for a while and done so much juggling around of settings that I give up when it goes back to x1 speed. The only thing that gets it back to x16 for me anymore is a BIOS flash.

I've tried setting the freq. to 101, 105, 110, trying every sigle PCI-E volt. It doesn't do anything for me.