Rivatuner Software Volt Mod (Confirmed)

Yup, it does work. Tried it out yesterday and maxed out my core by uping the vcore to 1.24V.

Doesn't really help you if you already maxed out the shader clocks, more volts isn't going to help unless you've got watercooling or really good aftermarket aircooling set up.
 
if you can modify voltages through an app like rivatuner, does that mean you can flash a higher voltage onto gt200 cards with nibitor/nvflash?

if anyone can confirm this, it would be great.
 
dont think fashing works, just this edit everytime you start up your pc
 
oMek said:
dont think fashing works, just this edit everytime you start up your pc
Click to expand...

Just set up a task that launch every time riva tuner starts up. And anther that starts up when TR does at boot.

Personally I just tried this to under volt my 4870X2 idle. Where as in the bios I can only go down to 1.00v I can go all the way down to 0.85 with the softmodding.

My idle current on each GPU went from 12.19A down to 1.35A!!! I'll see if this is stable in a few days.
 
I heard about this but my VRMs hit 90c+ furmark as is and Im afraid to have them pumping out mre juice without better cooling. Im thinking about an HR-03 with some enzotech mosfet and memory sinks.
 
Here's a batch file I made for my 4870 if anyone wants.

http://www.sendspace.com/file/teeqis
4870tunerta7.jpg

4870tunerocai1.jpg
 
Well I followed the guide, it was especially easy because the guy who wrote it had a 280 and all my registers and values were exactly the same, so virtually all I had to do was copy and paste his commands and everything worked. Made a batch file and put it in my startup folder. So far, on my STOCK evga gtx280, I've been able to hit 760 core, 1520 shader, 2600 memory. Ran atitool fine with no artifacts. Will do more testing later to find my max clocks, but so far so good. Need to do actual gaming testing to see if I experience crashes, as in the past unstressful games like css have given me OC crashes even when atitool passed fine. Still not sure if the voltmod gave me headroom as I never found my max oc clocks on old voltage, but the amount of extra heat is marginal so I'm going to leave on the mod for now. Will report back with more results.
 
Got my MCW60 and GTX200 heatsink installed on my GTX260 (Core 192) and fiddled with voltages a bit to see how much difference it makes. Core/shader clocks shown against each voltage - memory was at a constant 1260 MHz

1.12v (stock) 783/1566
1.15v 810/1620
1.20v 837/1674

864/1728 isn't happening even at 1.25v, and with this much juice vrm 1 and 2 are hitting over 100c (#3 is 20c cooler and the core has yet to reach 50c). Not sure if the heatsink is making good contact with them though - not impressed with it tbh especially since one of the screwholes is borked (didn't tighten the screw particularly hard), but can't source a unisink here in the uk (have a Fuzion sitting idle) :(
Might try and clamp the damn thing, just to see if it is poor contact or simply a case if the heatsink not being particularly good!
Anyhoo, I'm seeing a definite gain from feeding the card more juice - watercooling is probably mandatory, as I couldn't get past 783/1566 regardless of voltage with the fan at 100%.
Since my card is a bit 'dodgy' and only gets 1.06v by default, 756/1512 was my maximum OC prior to tweaking. So I'm pretty happy with my new clocks, all with no noise and far less heat!

EDIT: 1728 on the shaders is impossible. Tried 1.3v for giggles - the card squeals like a stuck pig then the screen goes haywire and it's time to reach for the reset switch (only to find I bought a Cosoms S and it doesn't have one :p). Dammit, so close to a 50% OC on the core.
 
Confirmed to be working on my 4870. Bus 01, register 17. I'll post overclocking results later.
 
tacos4me said:
Confirmed to be working on my 4870. Bus 01, register 17. I'll post overclocking results later.
Click to expand...

odd balls for me..

from monitor it did change the voltage, but somehow I cant get it stable .....

when I push up to 840-850 it crash, not right away...

it doesnt seem to be temp problem, temp is around 50-60c, and voltage is at 1.39....
 
shansoft said:
odd balls for me..

from monitor it did change the voltage, but somehow I cant get it stable .....

when I push up to 840-850 it crash, not right away...

it doesnt seem to be temp problem, temp is around 50-60c, and voltage is at 1.39....
Click to expand...

From what I remember reading over at XS when I ran a 4870, a LOT of people had trouble OC'ing with anything more than about 1.35v - maybe OCP/OVP kicks in?
Also, check your VRM (VDDC Slave) temps. Over 100c is bad, especially when overclocking (though they can safely operate up to around 125c, apparently). 4870's generally don't clock well - highest RV770 clocks I've seen are all 4850's with vmods. Mine was good for 860, but tbh it didn't make a noticeable difference.
 
Just thought I'd mention this is also extremely useful for reducing idle power. I can get as low as 0.75v without ATiTool crashing my 2D clocks (145/310/252) - idle current draw is a miniscule 4.35a (compared to 10.16a @ 1.06v). Idle gpu temp is now 28c :cool:
 
rincewind said:
From what I remember reading over at XS when I ran a 4870, a LOT of people had trouble OC'ing with anything more than about 1.35v - maybe OCP/OVP kicks in?
Also, check your VRM (VDDC Slave) temps. Over 100c is bad, especially when overclocking (though they can safely operate up to around 125c, apparently). 4870's generally don't clock well - highest RV770 clocks I've seen are all 4850's with vmods. Mine was good for 860, but tbh it didn't make a noticeable difference.
Click to expand...

I guess I'll stick with 1.35v then...

VRAM temp is around 50-60c, I put up the fan to 100%, which it sounds like a vacuum cleaner LOL

but oh well, just giving it a shot....
 
rincewind said:
Just thought I'd mention this is also extremely useful for reducing idle power. I can get as low as 0.75v without ATiTool crashing my 2D clocks (145/310/252) - idle current draw is a miniscule 4.35a (compared to 10.16a @ 1.06v). Idle gpu temp is now 28c :cool:
Click to expand...

After experimenting with the under volting for a day I found that the minimum stable was 0.85 for my 4870X2. However during video playback the primary doesn't seem to like it and you get a driver no responding error. Bumping the primary to 0.9v seems to avoid it.

There appears to be a limit on the VRM ou put as the max it will go is 1.45V. Found this out while I accidentally keyed in 250 for the primary.:eek:

Overall pretty happy so far. 5.41A total power compared to 24.38A idle. The card stays under the fan threshold at all time and is nice and quiet.

This is definitely something anyone who complains about 4870X2 idle heat and power should check out.
 
finally...

4870 with this soft mod is 1.35v max....

850Mhz Core max, up to 860 will start seeing a bit of performance drop...

overall, 850 is max...
 
shansoft said:
I guess I'll stick with 1.35v then...

VRAM temp is around 50-60c, I put up the fan to 100%, which it sounds like a vacuum cleaner LOL

but oh well, just giving it a shot....
Click to expand...

Yeah, the 4870 fan gets REALLY loud - my ears tell me that 50% is louder than the GTX260 fan running flat out.
 
Hypernova said:
Overall pretty happy so far. 5.41A total power compared to 24.38A idle. The card stays under the fan threshold at all time and is nice and quiet.

This is definitely something anyone who complains about 4870X2 idle heat and power should check out.
Click to expand...

Very impressive savings there, that is some low power draw for a dual-gpu monster! The G80 GTS in my second rig is a lot thirstier than that - might have to replace it with the 4870 since that box has a lot of uptime.

BTW that power draw is for the core only as far as I can tell. The memory doesn't use the same voltage regulator.
 
rincewind said:
Very impressive savings there, that is some low power draw for a dual-gpu monster! The G80 GTS in my second rig is a lot thirstier than that - might have to replace it with the 4870 since that box has a lot of uptime.

BTW that power draw is for the core only as far as I can tell. The memory doesn't use the same voltage regulator.
Click to expand...

Right now I got another idea - Change the idle clocks the same way, is any of the registers related to the clocks?

All the valid registers I got
Code: 
Device 0 : 256-bit RV770 (800sp) with 1024MB DDR5 memory

I2C	01 70 00		: 30	48
I2C	01 70 01		: 80	128
I2C	01 70 02		: f0	240
I2C	01 70 03		: 60	96
I2C	01 70 04		: e6	230
I2C	01 70 05		: 5e	94
I2C	01 70 06		: 00	0
I2C	01 70 07		: 81	129
I2C	01 70 08		: 00	0
I2C	01 70 09		: 00	0
I2C	01 70 0a		: 00	0
I2C	01 70 0b		: 00	0
I2C	01 70 0c		: 00	0
I2C	01 70 0d		: d8	216
I2C	01 70 0e		: 03	3
I2C	01 70 0f		: 12	18
I2C	01 70 10		: 1b	27
I2C	01 70 11		: 02	2
I2C	01 70 12		: 01	1
I2C	01 70 13		: 06	6
I2C	01 70 14		: 00	0
I2C	01 70 15		: 2d	45
I2C	01 70 16		: 21	(idle voltage)
I2C	01 70 17		: 39	57
I2C	01 70 18		: 41	(load voltage)
I2C	01 70 19		: 01	1
I2C	01 70 1a		: 0a	10
I2C	01 70 1b		: 00	0
I2C	01 70 1c		: d3	211
I2C	01 70 1d		: 55	85
I2C	01 70 1e		: 04	4
I2C	01 70 1f		: 07	7
I2C	01 70 20		: ff	255
 
Damnit, I can't get 756 core actually game stable, still artifacts/crashes even at 1.26Vcore after 20-30 minutes of Crysis assault level game play. Oddly enough 1.21V is alright for 738 core. Still 25-30min of game play is a lot more then what I got when I tried out 1.25V, that crashed barely 10 minutes in. Since my shaders are still at 1512 I'm wondering whether its just the temperature build up that crashes 756 core or I just need to supply more voltage.
 
rincewind said:
I would guess not, but it would take Unwinder or someone else with in-depth knowledge to confirm.
Click to expand...

After some testing it seems none of the registers are clock related. Register 1F seems to change between 7 and 6 when I fiddle around with overdrive a bit but that's about it.:confused:
 
DeadSkull said:
Damnit, I can't get 756 core actually game stable, still artifacts/crashes even at 1.26Vcore after 20-30 minutes of Crysis assault level game play. Oddly enough 1.21V is alright for 738 core. Still 25-30min of game play is a lot more then what I got when I tried out 1.25V, that crashed barely 10 minutes in. Since my shaders are still at 1512 I'm wondering whether its just the temperature build up that crashes 756 core or I just need to supply more voltage.
Click to expand...
If you haven't yet, try cranking the fan to 100% and see if it stabilizes it. I wouldn't be surprised if it's temperature limited though, the GTX series generally is.
 
Mr. K6 said:
If you haven't yet, try cranking the fan to 100% and see if it stabilizes it. I wouldn't be surprised if it's temperature limited though, the GTX series generally is.
Click to expand...

Already got the fan cranked to 100%.
 
Mr. K6 said:
Looks like you're cooling limited then :(.
Click to expand...

Yup, no matter the voltage anything about 738 core or 1512 shader fails for me, just varies in time according to voltage. Ideally core temps should be kept at 58C or lower to achieve stability at higher clocks.
 
will this work with the 4850x2 sapphire card? ATI's CCC only lets me go to 700 which is too easy.
 
Hrmmm, not showing up with my 4850 in sig. :mad: Does this work with Vista x64?
 
I know some companies allow OC'ing in their warranty, so is a soft vold mod also allowed since it's part of the oc?
 
I tried to do the mod on my 55nM GTX260 and it doesnt have the same voltage regulator so no go ...yet.
Hopefully EVGA will release their voltmod tool soon.
At stock volts it gets 720 core at 60% fan / 62C stable, so could be promising.
 
Nenu said:
I tried to do the mod on my 55nM GTX260 and it doesnt have the same voltage regulator so no go ...yet.
Hopefully EVGA will release their voltmod tool soon.
At stock volts it gets 720 core at 60% fan / 62C stable, so could be promising.
Click to expand...

Weird the mod works with GTX 295 though.
 
DeadSkull said:
Weird the mod works with GTX 295 though.
Click to expand...

Yes that is odd.
The guys that have written apps and rivatuner mods to do the voltmod, acknowledge that 55nM GTX260 isnt supported.
I havent heard a whisper about support yet either so it may be down to EVGA.
Hopefully its just cos they havent got a card to test with.
 
I flashed my 55nm GTX260 to change the voltage. I lowered it from 1.18v to 1.06v since it didn't help OC'ing anyways, and saved a ton of heat.
 
xorbe said:
I flashed my 55nm GTX260 to change the voltage. I lowered it from 1.18v to 1.06v since it didn't help OC'ing anyways, and saved a ton of heat.
Click to expand...

Heh, I want to go up in voltage so darent risk that.
Nice one though :)
 
You must log in or register to reply here.
Back
Top