980 Ti Owners Thread

First card is ASUS 980Ti STRIX
Second card is Inno3D 980Ti X3 Ultra iChill

The second (Inno3D) card used to be overclocked to 1470 on the stock bios when single but the ASUS card doesn't seem to want to go further than 1350ish when I try to clock these in SLI.

I'd like them to not throttle under load and remain at default voltages.

 

On phone so multitasking plus reading you can forget the details or miss them.

IDK why you are saying I'm trying to dump on your fun.

You can say 1.27v is a bug but it's not.

I already told that the bios's on OCN do not enforce a max clock speed/voltage and drivers can override them.

So it's not an odd discovery.
It's a normal function because those bioses do not create a hard lock on max voltage, you can verify that with others on that thread.

I already told you they lack required changes to do such a thing.

It may seem like a bug but like i said, drivers can change that(mine did the same back when i first started messing with modded bioses on my Ti)

So the fact that you have to use MSI AB to fix that 'issue' you should know the only fix is to use your own modified bios with the correct settings.
Having to apply your OC profile is a bandaid fix.

You can also have clock fluctuations removed under 3d loads btw.

As for cooling, It usually doesn't matter what it is unless its a real H20 setup.

Most AIB cards have good core cooling but memory/vrm cooling are lacking.

Put a thermocouple between on the VRMs, i would be interested in seeing the temperatures.
I've seen a user review of someone a long time ago that was using a G1 iirc, core temps were 60s but VRM temps were in the 90c range.

Most VRM are rated for at least 100c but i still think those are pretty poor results when H20 can result in max temps of high 30s/low 40s.

Anyways i told you about the voltage and clock scaling because i have thoroughly tested a few maxwell cards and they have all resulted in similar behavior.

In fact the 'pro' OCer der8auer said that upon testing many cards, most will not scale much past 1.23.

I'm interested in what you said earlier though.

You gained 50mhz from going 1.24v >1.25?

Impossible for maxwell unless you got some extremely odd chip.
Most likely due to the fact that you went from stock > modded bios.

You should test that because that'd be very interesting if such a small increase in voltage results in that kind of clock scaling.

I'd wager you'd get around 12~mhz gain from that at best.

I have a lot of experience from testing all of this.
I was OCing via I2C commands and hex editing before these modded bios started showing so I'm just giving my 2c.

BTW electrical engineer I did not call you a newb so no need to act like that.

Nothing we are talking about has anything to do with being an electrical engineer(we aren't physically modifying anything nor discussing it)

Is it a problem to test each card individually to at least get a rough estimate of their max stable clocks at stock voltage?

Some cards might max out at 1380 and others can do 1500, in which case it might be a good idea to test before we set clock speeds on them.

Since you already know the Inno3d will do 1470, it's not much work to test the asus alone.

As for default voltage, tell me the voltage under load.
It will vary as well card to card.

You gave me a link to bios which is good, I'll verify the latest version of those.

 

Not a problem at all but I'm going away for a couple of days, next time I'll be able to test the ASUS card is on Friday. Anyway, GPU-Z reports that the Inno3D card's max VDDC is apparently 1.1240 - this is the correct voltage, right? I always ran my custom bios on 1.23 and 1470MHz.

 

Yeah you need to be careful with the spacers and choose the ones with the correct thickness (iirc the slimmest ones for 980 Ti), the differences are small so it's easy to mix them up.

I once tried putting the cooler on without spacers but as the spacers distribute the force caused by tightening screws I let them be.
Also tried using one size thicker spacers (can't recall why anymore) and indeed there was a gap left between HS and GPU.

 

Yeah they were the thinnest.
Maybe they had a bad batch the wrong size.

 

Check with HWINFO64 and list the max voltage it shows.(with single card)

VDDC is a rating for power usage/current supplied to VRM.

Anyways 1.1240 is most likely the voltage but really low.
This is not the full load voltage as you've said the card boosts to low speeds.

Once you test it alone max voltage will be near 1.2 +/-

 

OK

So for the ASUS card, max voltage is 1.187 under load - Clocks that seem to be okay are 1400MHz on the core and 3900MHz on the memory.

For the Inno3D card, max voltage is 1.193 under load. Core 1470MHz, memory 3900MHz.

 

Concerning the Strix 980Ti - it hits max 91c when I run BF4 in 4K that's gotta be bad right? In any other game it usually hits max 84c. I repasted it with some MX4 a week ago but that didn't seem to change anything. Should I put a fan/fans to push air between the two cards in addition to all my other 6 case fans (3x 140mm, 3x 120mm)?

 

Here are the modded BIOS for both cards.

Link

I don't like those temps for the strix so i changed the fan curves.

Boost clock is 1392 and 3801 memory @ 1.187 for both cards.

The clocks will be the same for both cards.
Your inno3d is a better card and in fact may be stable at a much lower voltage.

You may change this later after testing these two bios.

I assume you know the risks and understand how to use NVFLASH.

I recommend you to flash one card at a time for aforementioned risks so you can flash backup bios in case something goes wrong.

Of course you shouldn't have any issues just giving a warning.

Make sure to use latest nvflash with certs disabled.

BTW you should see much improved temps with the fan curve but it certainly would help to aim a single large fan(140mm) at or near the cards.

Also i was conservative with core and memory for a reason, multiple cards can have an affect on stability.
Quite annoying trying to figure out which is unstable.

 

Great stuff, I'll try them out right away and report back! :thumbup:

 

Well it seems something went very, very wrong. The ASUS GPU didn't like the bios at all, if I try to boot the PC with it as the primary GPU - black screen. If I try to boot the PC as it as the secondary GPU - Mouse starts lagging and BSOD. Now I've slapped in my 670 and am trying to flash the cards back.

 

Do you have the original bios to upload it?
Maybe there is something wrong with the techpowerup one(or maybe not compatible)

What about the inno3d card?

Exactly how did you flash the cards?
I mean with what commands

 

As me always trusting a new BIOS - I didn't backup the bios on the ASUS card :bang: as I'm 100% sure it's either one of these LINK or LINK
I was able to flash the card back to the same bios you used but without your changes. It doesn't seem to help though. I once experienced something very similar to this. When I had my 670 SLI setup and I flashed those cards I forgot to enable the second gpu in device manager after the flash and restarted so whenever I tried to use it as primary I would get a black screen or when in secondary it couldn't be recognized. However I did disable both 980's in device manager but when I had flashed the ASUS card and I enabled it the whole screen got black that early on.

I just used "--protectoff" and "nvflash -6 --index=X BIOSfilename.rom" with different index (0 and 1) for the cards.

 

So maybe the bios you gave me a link to was not the correct one?

Can you give an exact part number?
I saw several versions of strix on tpu.

Have you tried different display ports on the gpu?

IIRC i recall having a similar issue with forgetting to re-enable the primary GPU and had to put my monitor on the 2nd GPU.

Are you using v5.287 nvflash?

BTW protect off isn't necessary, required only when flashing different vendors usually.

 

I'm going to try the card in another PC to see if this is happening because of the device manager.

This right?
STRIX-GTX980TI-DC3-GD5

I'm using nvflash 645.218

 

If you can get into bios with it then there's an issue with windows.

Btw I can't find 645.218, you sure that isn't a typo?

 

It's an unofficial build I guess. Those can be found on some of the 980Ti custom bios threads on oc(.)net. I've been using it since I had some trouble with some custom bios' downloaded from there and that version didn't have any issues with them but the default v5.287 did.
Yes I can enter BIOS without an issue - testing in another PC is not gonna happen as my other rig has just 6+8 PCI-E power and that card requires 8+8, it won't let me boot without "please insert all the power..."
So is there a way to somehow "reset" the device manager settings, other than reinstalling windows?

 

You'll just have to fiddle around with different display ports to see if you can get any signal.

Maybe if you have a restore point you can go to prior to disabling any GPUs in device manager

 

I'm trying the flash the card now but I get a bsod everytime I try. In device manager in 980's properties the device status is "Windows has stopped this device because it has reported problems. (Code 43)"

 

Huge relief, phew! :roll: I was able to fix the GPU by doing a Device Manager disk cleanup, putting the GPU in the second PCI-E slot alone, plugging in the monitor and flashing the .48 BIOS on it. Now I just installed drivers and it's running fine again. I think I'm gonna take a break from GPU flashing for a while now, I was already losing hope :uhoh: