Evga Voltage Tuner

[Diverge]

Install instructions for us software challenged people?

It's all in the readme.

[Protocol_48]

little nice tool from my german collegues of Awardfabrik.de
Voltage Factory V1.0

http://rapidshare.com/files/19383349...dfabrik.de.rar

no more needed EVGA Voltage Tuner tool

Supports all VT11XX IC Cards (GX2,260-65nm,280,295,4850,4870)

Great utility! thanks for posting.


For people using the gtx 295, you will need to use device /sd0 for the first and /sd2 for the second gpu, for this tool that means choosing the sli/cf mode option rather than multi-gpu (as multi gpu assumes /sd0 and /sd1)

This can ofcourse be confirmed using rivatuner's monitoring graph.

[Luka_Aveiro]

Well, we went from the EVGA Voltage Tuner which was the subject of this thread onto Riva Tuner, then onto RT Advanced Command Line stuff, then into all the different chip versions, then into coding and recoding, then into the issues with RT not working for some and why, then into C programming and GUI stuff, then into Sliders, and then into many branches of different people making different programs.

I should've dug up that trainwreck pic I had on another 'puter where a track switch went wrong and one trian went in two different directions at the same time.

It's all cool though. I was just being a smart butt trying to be funny.

that was what I first thought about that joke, but we are in XtremeSystems, some crazy minds flow around here

[fgw]

ok, here is probably my last post on this:

just uploaded a rivatuner plugin with allows changing voltages from rivatuner:

see my thread here

[Leeghoofd]

Good job mate , hope the 285 and 260 (55) will get support too soon

[BuBBle.D]

I feel like crap now that i sold my GTX280 and put in the new GTX285 that is not supported.

[SAE]

It's not supported yet, but it possibly will be later. I doubt EVGA would tell their users there's gonna be a version with support for 55nm card later, if they knew they could never program the Intersil!

[mikeyakame]

I just finished reading and re-reading the Intersil ISL6327CRZ multiphase closed-loop feedback and compensation PWM Voltage Reg datasheet, the VR used by NV in the die shrink G200 reference cards, GTX 260 55nm and GTX 285 55nm.
Seems NV screwed you guys over with VR cost cutting this time around, mind you the IC itself only costs $5usd or so.

It's a VR designed with cost savings and affordability in mind so it lacks any kind of external low-pin count bus interface accessibility and relies on the pcb designer to set up all the necessary operational values in design and testing phase. There are 8 VID pins which when pulled up the base VID is read by which pins are 0 or 1 and converted to digital signal by DAC. '

Voltage adjustment is done through current offset via the OFS pin, which is polarity dependant for negative or positive offset, switch to Vss (gnd) for neg, switch to Vcc (vin) for pos. Offset is configured through external resistors connected between the DAC and REF pins, the impedance dynamically controls the current which in turn determines the Vout (voltage out).

Long story short, unless NV have some kind of low pin count programmable interface to adjust a pull up or pull down resistor connected between DAC digital VID signal pin and REF voltage reference pin there is absolutely NO CHANCE AT ALL to control Voltage Adjustments via software on either the GTX 260 55nm or GTX 285 55nm. The only way would be to change the resistor between DAC and REF pins, to give a greater current offset to drive Voltage out higher.

If EVGA is saying it is possible on their cards, short of physical changes to the NV reference cards to add this basic functionality I can't see how it is possible at all. I read over the datasheet a good 4 or 5 times just to make sure I completely understood the electrical design and don't see any way to do it.

I posted this earlier in the RT plugin thread.

[W1zzard]

on the gtx 285 (ISL6327CRZ) nvidia most probably connected some GPIO pins to the VID input pins but this will give you like 3 or 4 selectable voltages in a rather small range. it will definitely not be what you can get with the i2c based chips.

[G4h4o8s6T]

on the gtx 285 (ISL6327CRZ) nvidia most probably connected some GPIO pins to the VID input pins but this will give you like 3 or 4 selectable voltages in a rather small range. it will definitely not be what you can get with the i2c based chips.

Well as long as we're atleast able to push it to the point OVP kicks in, Ill be happy

[jaredpace]

I just bought a 65nm GTX 260. With this voltmod, it has a slight advantage, IMO.

1. larger process, handles more volts than 55nm

2. higher quality stock HSF and shroud

3. voltera VRM that is programmable for the softmod

4. $190 new

this thing, even with 192 shaders, might be able to perform somewhere between a gtx280 and 285 once i get the core > 800mhz.

[mikeyakame]

on the gtx 285 (ISL6327CRZ) nvidia most probably connected some GPIO pins to the VID input pins but this will give you like 3 or 4 selectable voltages in a rather small range. it will definitely not be what you can get with the i2c based chips.

That makes a lot of sense Wiz!

Something along those lines would be executed by controlling say 3 or 4 dedicated GPIO pins which would be used to assert/deassert the VID pins based on the required voltage table bit mask to obtain 3 or 4 preset Performance Level VIDs? Ie. use different open GPIO pin combination to get 4 bit mask range, 00 01 10 11 of which each has a preset VID entry to connect VID0-VID7 pins based on what Voltage Table mask is needed to output each performance level voltage.

I'm not all that knowledgable on the GPIO functionality so I've just taken a stab at how i think it may be done, feel free to correct me if i've missed the target heh.

[Sailindawg]

I just bought a 65nm GTX 260. With this voltmod, it has a slight advantage, IMO.

1. larger process, handles more volts than 55nm

2. higher quality stock HSF and shroud

3. voltera VRM that is programmable for the softmod

4. $190 new

this thing, even with 192 shaders, might be able to perform somewhere between a gtx280 and 285 once i get the core > 800mhz.

I've been using Riva Tuner to adjust voltage on my GTX 260. Running clocks at 706/1501/1306, on 1.15V (up from 1.13V stock),in COD:WAW I'm getting a solid 60 fps with V-sync enabled, all eye candy enabled, run at 19x12 resolution. Other games behave similarly. I also force triple buffering with Riva Tuner. How much more could I get with a GTX 280 or 285?

Just being able to adjust voltage to stabilize clocks helps these 65 nm cards render at the same level of the higher cards. I really don't think that you'll see all that much of an improvement on the higher level cards as compared to the 260's. With this voltage tuner, EVGA just extended the life of the 65nm 260's.

[clo007]

Finally released:

http://www.evga.com/articles/00462/

[Unwinder]

That makes a lot of sense Wiz!

Something along those lines would be executed by controlling say 3 or 4 dedicated GPIO pins which would be used to assert/deassert the VID pins based on the required voltage table bit mask to obtain 3 or 4 preset Performance Level VIDs? Ie. use different open GPIO pin combination to get 4 bit mask range, 00 01 10 11 of which each has a preset VID entry to connect VID0-VID7 pins based on what Voltage Table mask is needed to output each performance level voltage.

I'm not all that knowledgable on the GPIO functionality so I've just taken a stab at how i think it may be done, feel free to correct me if i've missed the target heh.

Correct. But normally amount of such GPIO controlable pins is reduced to required minimum during the PCB design (the rest VRM's VID pins are hardwired), so normally there is just 1 GPIO pin connected to VRM if 2 different voltages are needed, 2 pins if up to 4 voltages are needed etc. So maximum you can expect is having limited set of fixed voltages. And the maximum one is normally aready in use.

[Sailindawg]

Finally released:

http://www.evga.com/articles/00462/

That just made a lot of people happy!

[DMH]

I've been using Riva Tuner to adjust voltage on my GTX 260. Running clocks at 706/1501/1306, on 1.15V (up from 1.13V stock),in COD:WAW I'm getting a solid 60 fps with V-sync enabled, all eye candy enabled, run at 19x12 resolution. Other games behave similarly. I also force triple buffering with Riva Tuner. How much more could I get with a GTX 280 or 285?

Just being able to adjust voltage to stabilize clocks helps these 65 nm cards render at the same level of the higher cards. I really don't think that you'll see all that much of an improvement on the higher level cards as compared to the 260's. With this voltage tuner, EVGA just extended the life of the 65nm 260's.

Damn your GTX 260 has a high stock voltage mine was 1,06V,also I doubt most people will see 800 core on their GTX260

[newls1]

For some reason or another, I cant log into EVGA? It seems that there server or some sh*t is down. If I click "Member login" I get a "PAGE CAN NOT BE DISPLAYED" error everytime. Damn it, i've been waiting for so long for this app. EVGA FIX YOUR WEBSITE.

[clo007]

I'm on the site now and it is fine

[xtant25]

Just downloaded it will be testing it on my GTX260 216 (65nm) hopefully some higher clocks.

[G4h4o8s6T]

Correct. But normally amount of such GPIO controlable pins is reduced to required minimum during the PCB design (the rest VRM's VID pins are hardwired), so normally there is just 1 GPIO pin connected to VRM if 2 different voltages are needed, 2 pins if up to 4 voltages are needed etc. So maximum you can expect is having limited set of fixed voltages. And the maximum one is normally aready in use.

Hope this isnt the case with the 285, but thanks for sharing this info Unwinder, much appreciated!

[Sailindawg]

Damn your GTX 260 has a high stock voltage mine was 1,06V,also I doubt most people will see 800 core on their GTX260

Stock voltage at 2D settings is 1.11V by RT. I also have an EVGA Superclock card. That accounts for the higher 2D stock voltage. The 3D stock voltage is 1.13V. Don't know how high the gpu core will go for max clocks, but I have noticed that with a slight adjustment, with <<max clocks, the performance increase is very noticeable.

Edit: EVGA Voltage Tuner works like a charm. RT is better because it makes you think about what the hell you're actually doing!

[jaredpace]

Damn your GTX 260 has a high stock voltage mine was 1,06V,also I doubt most people will see 800 core on their GTX260

here is 820mhz core stock cooled from August 08 @ techpowerup

http://img.techpowerup.org/080801/bios.jpg




@Sailindawg: how high does your 260's core go?

[newls1]

Finally have been able to get 720c/1570shader/1170 stable using only 1.15v on my GTX295 Awesome job evga

[Sailindawg]

@Sailindawg: how high does your 260's core go?

Wow, that post was from Tech Power-Up was pretty sweet. And it's stock cooled like mine. Don't know what max clocks are, I ought to figure it out!

Update: Been playing around all afternoon. The most stable clocks I was able to get were 774/1548/1306 @ 1.165V, by EVGA Voltage Tuner. Shaders did not like getting much beyond 1548 or so. This was Vantage stable. Higher volts had no impact on stability. It only helped things crash faster. Max gpu temps were 58C with fan at 96% duty cycle. I ran this using windows 7 64 bit beta and 185.20 beta drivers. MMMV with XP 32 bit and a different driver. Still, this 260's got a lot of unlocked power.