Who did that ? i'm quite surprised :confused:Quote:
Originally Posted by flytek
Printable View
Who did that ? i'm quite surprised :confused:Quote:
Originally Posted by flytek
this poor guy: http://www.xtremesystems.org/forums/...9&postcount=21Quote:
Originally Posted by Ebola
i have also seen at least two revisions of the vmem vreg ic on 7900gt's.
so maybe they figured there was a problem with the earlier cards and changed it.
i'm not convinced iether way yet but it seems reasonable to treat the vmem mod with caution for the moment.
does this card drive anyone else crazy? with conductive ink mods i was getting bad flashing screen problems, so i did the 'resistor k' mod as well as replacing 'j' for 1.5v. still having a lot of problems with this card, although it's clocking very well when it works. i am not looking forward to replacing resistors 'i' or 'h' so was wondering if anyone had any suggestions?
Step 1 & 2 done, can someone pls tell me if the line i've drawn in Aqua (attatched pic) is the closest ground point for 1.6v mod?
Will be using ink pen to draw line.
Thanks
C-BuZz
The ink will also come off really easy if you use a fine point sewing needle and magnify glass.Quote:
Originally Posted by quijonsith
Ok, 1.6vmod done. 800/500 artifact free after a few 06 runs. 1.7v seems unstable & keeps crashing my system in full 3D. I can run ATITool 830 core ok but crashing in 06.
Any ideas?
C-BuZz
somewhere in one of these threads i read you should use the Vmem mod to keep your core clock stable. So if you havent done the Vmem mod, you may want to try that avenue to see if it gets 1.7V stable. It could also be a heat issue as well...
I've done the vmem mod.Quote:
Originally Posted by Haltech
The card is now loading @ 1c so it's definately not a heat issue :)
Somthing interesting, If I lower mem I can pass 06 @ 825/450 1.7v.
Anything higher it just straight blackscreen & have to reboot. Whereas I use 1.6v on core & I can do 800/500 runs no prob :confused:
C-BuZz
Sounds like a power regulation problem, these cards have their limits using the very basic power reg they have :)Quote:
Originally Posted by C-BuZz
Did you already try to increase the switching frequency? (resistor K)Quote:
Originally Posted by C-BuZz
Your problem seems to be a current limit, either in the Vgpu or in the Vmem supply.
My guess is that it is the Vgpu supply.
To change the current limit in the Vgpu supply, change resistor H from 3k5 to 6k8.
Step 2 in the drawing that you showed does not remove the current protection as mentioned already at the beginning of this thread.
BTW the 825Mhz that you can get with your board is unbelievable. What manufacturer and type of board do you have ?
If this does not help you, we can change the current limit of your Vmem supply.
sorry not to sure what you mean there, 3k5 to 6k8? I removed & replaced the resistor with a 51K resistor as described in step 1, is that suitable? & I removed both current protection resistors. Then I used the cunductive ink to ground for 1.7vcore. Was crashing in 3D after about 3secs.Quote:
Originally Posted by t024484
Are you saying that Step 2 pictured above does not remove Over current protection??
1.6vcore seems quite stable.
I going to have to try a new power supply as I beleive it could possibly be my crap psu giving me issues.
Here is my latest run with 1.6vcore
http://forums.overclockers.com.au/sh...&postcount=272
Card is Forsa.
C-BuZz
Step 1 with 51K is O.K.Quote:
Originally Posted by C-BuZz
Step 2 in the picture does not remove OCP.
Look at the first posting in this thread, and you will understand what I said.
Great, so i've removed 2 load balancing resistors for no reason :slapass: nice 1 vrzone..Quote:
Originally Posted by t024484
ok power supply has now been ruled out so its definately a problem with the card.
I've done some further testing & 1.6v works fine up to around 800 core very stable & no artifacts. Any volotage over 1.6v on the core & it dies in 3D after 3seconds. I can boot into Windows no prob @ 1.7vcore, but run 3D & it dies.
SO, it looks like the OCP Resistor "H" is kicking in after all. Could I use a pencil on this to increase the resistance?
Thx for your help :)
C-BuZz
Resistor "K" was the key to my problems.
I penciled it & got this. 1.7v on the core, Totally artifact free.
C-BuZz
Well I didnt come this far not to try 1.75v on the core :) Finally got the card loading in the positives.. around +3/5c. Think my modified Vapo is beginning to struggle a little, as evap temps are around -13c when fully loaded.
This run artifact free. I can forsee 850/860mhz core benchable no probs :) I was hoping for 100% clock but I guess I cant have everything..
I think the card is becomming quite bottlenecked by cpu.
C-BuZz
C-BuZz, how long are you running the artifact scanner? I've had plenty of times that I got no artifacts for ten minutes only to let it go a little longer and have it burst into artifact central. I run it for atleast 25min personally to be sure it's good. Also, what cooling are you using and what are your temps?
I've found artifacts behave differently when temps are in the negatives & the core doesnt get the chance to fully heat up. I have run the scanner for 1hr before & they where no different to a 10 second scan.Quote:
Originally Posted by quijonsith
With a normal heatsink/fan setup they usually wont show straight away/or they get worse as the card becomes hotter & the core heats up.
With the gpu running @ these speeds, the core temps are +5c MAX so the artifacts wont get any worse over time as the core can not get any hotter. Idle core I think it drops to around -10/-15c.
I have completely rebuilt a vapochill PE for the task :) Although i'm beginning to think it's not quite up to it heh..
C-BuZz
Ah, you got extreme cooling and then some :) I am using a vf900cu myself. Even when my temps were below 60c I have had artifacts show up after 10min on some speeds until I lowered the clock slightly. Then it makes it to 25min and I call it good. Nice to see someone pushing the limits past 1.7v.
Dunno about extreme, but it's a lot better than what I had heh...Quote:
Originally Posted by quijonsith
Cards behave a lot differently @ very low temps :)
I could barely squeeze 700mhz out of the core with my ghetto heatsink mod & 1.55vmod. But when I strapped my Vapo to it I was doing 780mhz runs with 1.55vcore easy :)
Considering pumping 1.8v through it, but i'd crack it if she died :D
C-BuZz
Had to try it lol, but she didnt like 1.8vcore to much. Was exhibiting similar problems as before, however 3D mark would just close to desktop. Wouldnt actually crash the whole system like before.
That was @ 860 core heh..
1.75vcore seems like a winner.
C-BuZz
oh yeah definitely. lower temps result in higher stable clock speeds in general. There's atleast a 15mhz difference between my a/c working so that the core stays at or below 60c, and when my a/c doesn't work and the card can get to 65 or in some cases 68 (this is using atitool to heat er up of course; in game temps are much better).Quote:
Originally Posted by C-BuZz
since i'm on air cooling I can use my temps as a guideline for how much voltage I can push. With vapochill (and IMO it is extreme cooling) you really only have your own guts as judgement.
On the upside, had your card died you would have provided the entire 7900gt vmod community with the kill zone voltage for the core :D....a very expensive lesson it would've been. Glad your card survived.
hehe me to. I redid the 1.8v mod & did some further testing @ 860core. This time crashing to desktop after about 30sec of running 3D. Seems similar to last problem. I'll try pencil the resistor a bit more & see how we go.Quote:
Originally Posted by quijonsith
C-BuZz
Congrats C-BuZz :clap:Quote:
Originally Posted by C-BuZz
I never thought I would ever see such a high OC :toast: :toast:
wow.. (on 1.8v w/ 860 core) impressive!!
Please read the first posting of this thread.Quote:
Originally Posted by xit