GTX 1080 volt issues past 1.25v

[Vellinious]

8 minutes ago, Lays said:

The PCB itself is fine, they just have a bit lackluster coolers on em it seems.  

I wasn't a fan because all the 970 STRIX guys went in blind, thinking they were going to be getting something awesome, then stuck with a single 8 pin connector and limited to 1.212v.  Granted...they should have done their homework before they bought them, but.....that's a pretty shitty thing for ASUS to do.  That'd be like EVGA slapping a Classy sticker on something, and then giving it some criminal restrictions.

I'm not a fan of ASUS for anything but motherboards anyway, and even that's starting to slip, because their customer service is just horrible.

[done12many2]

3 hours ago, Vellinious said:

I'm not a fan of ASUS for anything but motherboards anyway, and even that's starting to slip, because their customer service is just horrible.

 

OT, but you're dead on.  I love the boards, but each time I buy one it's with the understanding that I'm rolling the dice if something goes wrong.

[Hunched]

Lets not forget EVGA fucked all their earliest 970 customers by using last gen coolers that weren't designed for the 970.

Resulted in having a heatpipe that didn't even contact the GPU, and a heatpipe (think the same one) that wasn't enclosed on the end, making it essentially useless.

 

They simply did it because they needed to sell old stock, and did so at the expense of their customers.

It's not just Asus who are assholes/idiots.

[Lays]

7 minutes ago, Vellinious said:

I wasn't a fan because all the 970 STRIX guys went in blind, thinking they were going to be getting something awesome, then stuck with a single 8 pin connector and limited to 1.212v.  Granted...they should have done their homework before they bought them, but.....that's a pretty shitty thing for ASUS to do.  That'd be like EVGA slapping a Classy sticker on something, and then giving it some criminal restrictions.

I'm not a fan of ASUS for anything but motherboards anyway, and even that's starting to slip, because their customer service is just horrible.

I went after my Matrix instead of a KP after seeing some reviews of it, I was just hoping they delivered the same quality the 980 Matrix had before it, but it triggers me when people throw ALL of the GPU's ASUS made for the 900 series under the bus, despite the Matrix being hands down one of the best 980 ti's on the market.  Triggers me so hard

[Hunched]

2 minutes ago, Lays said:

it triggers me when people throw ALL of the GPU's ASUS made for the 900 series under the bus

That's what happens when by far your most popular 900 Series GPU is a piece of shit, you damage your brand name.

[Vellinious]

Yeah..the Matrix was good.  Connect that with ASUS customer support though, and.....meh.

Seems as if EVGA and EK are having a bit of a tiff, so, EK isn't going to be making blocks for the FTW and Classy.....may have to move away from my go to.

 

[Vellinious]

8 minutes ago, Hunched said:

Lets not forget EVGA fucked all their earliest 970 customers by using last gen coolers that weren't designed for the 970.

Resulted in having a heatpipe that didn't even contact the GPU, and a heatpipe (think the same one) that wasn't enclosed on the end, making it essentially useless.

 

They simply did it because they needed to sell old stock, and did so at the expense of their customers.

It's not just Asus who are assholes/idiots.

The biggest problem with the early ACX 2.0 design was the criminally low power limits....at least, in my estimation.  I don't air cool, so....don't much care about the coolers.

[Hunched]

I'm curious, when was the last time Gigabyte and MSI completely botched a GPU design?

Cause in my eyes they seem to be doing a consistently better job than the rest of the competition.

[Vode]

5 hours ago, Lays said:

Does the same not apply to haswell and skylake then?  Skylake has higher voltage tolerance, and higher VID than Haswell chips did.

 

Generally what the guy said should be true but Haswell had an intregrated voltage regulagor (FIVR) that generates a lot of extra heat and Skylake does not. 

[boboman342]

6 hours ago, othertomperson said:

Overclock a 980 to 2GHz and see how well that goes. Dramatic clock speed increases are part of the improvements of Pascal over Maxwell. You can't just artificially limit them for Pascal while allowing Maxwell to run at stock and then claim that they are therefore the same.

Tom Petersen from Nvidia said that, not me. Plus he and Ryan talked about how there are like 25% more shaders running at a 40% faster clock speed. Transistor count doesn't change from 28nm to 16 either... You can just fit more gates and stuff. Of course the 1080 spanks the 980. I would assume if both the 980 and 1080 had the same number of cuda cores running at the same clock speed, as per Tom, theoretically they should have the same performance, minus the GDDR5X part. 

 

It seems (to me at least) that pascal so far has been shrunken maxwell with the new space filled with more refined shaders and a few new pipelines to do stuff like SMP and async-compute. 

[Big-G]

1 hour ago, Lays said:

The PCB itself is fine, they just have a bit lackluster coolers on em it seems.  

And below par VBIOS, stupid keeping the stock VBIOS' power limit...

[App4that]

43 minutes ago, boboman342 said:

Tom Petersen from Nvidia said that, not me. Plus he and Ryan talked about how there are like 25% more shaders running at a 40% faster clock speed. Transistor count doesn't change from 28nm to 16 either... You can just fit more gates and stuff. Of course the 1080 spanks the 980. I would assume if both the 980 and 1080 had the same number of cuda cores running at the same clock speed, as per Tom, theoretically they should have the same performance, minus the GDDR5X part. 

 

It seems (to me at least) that pascal so far has been shrunken maxwell with the new space filled with more refined shaders and a few new pipelines to do stuff like SMP and async-compute. 

The benchmarks don't show async. I think the async in Pascal is like the driver update to enable it in Maxwell.

[Lays]

12 minutes ago, GidonsClaw said:

And below par VBIOS, stupid keeping the stock VBIOS' power limit...

It's not like it's hard to fix that though, it only takes a few mins of reading, and like 30 seconds in the bios tweaker.  These days bricking a GPU is damn near impossible in my experience, I've flashed probably 30-40 times on my card, and tons of times on the past like 20 gpu's I've owned

[Vellinious]

51 minutes ago, Lays said:

 

Out of curiosity, what kind of voltage are you seeing stock with the GPUs under load?  And, what kind of voltage are you seeing as a limit without hardware mods (just bios modding)?

[Lays]

11 minutes ago, Vellinious said:

Out of curiosity, what kind of voltage are you seeing stock with the GPUs under load?  And, what kind of voltage are you seeing as a limit without hardware mods (just bios modding)?

On the 1080s? I haven't tested one yet, that'd be a question for those guys in that thread.  I was sort of wanting to buy a 1080, but unless that 1.25v thing gets fixed it's sort of useless for me

[Vellinious]

I won't do any hardware mods, so I'm curious as to what the voltage is after bios mods, and how well they're reacting to voltage increases.  I was going to get one right away, but, I may wait to see what kinds of overclocks people are seeing on the custom boards.

[Lays]

13 minutes ago, Vellinious said:

I won't do any hardware mods, so I'm curious as to what the voltage is after bios mods, and how well they're reacting to voltage increases.  I was going to get one right away, but, I may wait to see what kinds of overclocks people are seeing on the custom boards.

no idea on scaling and stock volts etc, I just know it's capped somehow at 1.25, and nobody can figure it out so far. 

 

I don't think we're going to see much higher than ~2100, not with this voltage handicap.

 But I do think these aftermarket designs will actually hold the boost, not boost up for 5 mins then throttle down 200 mhz.

[stconquest]

16 minutes ago, Lays said:

no idea on scaling and stock volts etc, I just know it's capped somehow at 1.25, and nobody can figure it out so far. 

 

I don't think we're going to see much higher than ~2100, not with this voltage handicap.

 But I do think these aftermarket designs will actually hold the boost, not boost up for 5 mins then throttle down 200 mhz.

" The answer is a firm “maybe.” MSI is still finalizing designs for the card, but it's currently slated to host a custom VBIOS and should have “slightly” more overvoltage headroom than the FE VBIOS. Even an extra few millivolts will grant additional overclocking potential, as that became the primary throttle point (after thermals) for the GP104-400 GPU. We are unsure of how much more mV of power can be provided through VBIOS at this time, but it will still be locked in part by nVidia recommendations for manufacturers. This is standard for both AMD and nVidia. "

 

http://www.gamersnexus.net/news-pc/2454-exclusive-with-msi-new-gtx-1080-twin-frozr-vi-overclocking

 

It should be interesting to see, as better cards start flooding the market.

[Hunched]

10 minutes ago, stconquest said:

" The answer is a firm “maybe.” MSI is still finalizing designs for the card, but it's currently slated to host a custom VBIOS and should have “slightly” more overvoltage headroom than the FE VBIOS. Even an extra few millivolts will grant additional overclocking potential, as that became the primary throttle point (after thermals) for the GP104-400 GPU. We are unsure of how much more mV of power can be provided through VBIOS at this time, but it will still be locked in part by nVidia recommendations for manufacturers. This is standard for both AMD and nVidia. "

 

http://www.gamersnexus.net/news-pc/2454-exclusive-with-msi-new-gtx-1080-twin-frozr-vi-overclocking

 

It should be interesting to see, as better cards start flooding the market.

Hopefully MSI can deliver, 8+6pin is good.

I believe it was revealed today, but the Gigabyte GTX 1080 G1 has a single 8-pin, so it's shit nobody buy it.

At least there's the Gigabyte GTX 1080 Xtreme, has dual 8-pins... I wish it looked better though.

It's going to be silly if they can't solve the bios voltage limit but are slapping down dual 8-pin connectors on their card.

"Wow I'm so glad I can do literally nothing with this extra power!"

 

[stconquest]

3 minutes ago, Hunched said:

Hopefully MSI can deliver, 8+6pin is good.

I believe it was revealed today, but the Gigabyte GTX 1080 G1 has a single 8-pin, so it's shit nobody buy it.

At least there's the Gigabyte GTX 1080 Xtreme, has dual 8-pins... I wish it looked better though.

It's going to be silly if they can't solve the bios voltage limit but are slapping down dual 8-pin connectors on their card.

"Wow I'm so glad I can do literally nothing with this extra power!"

 

The question, I believe, is:  Is Nvidia locking down the voltage for all their board partners?

 

We may get the potential for better power delivery only to be voltage throttled by Nvidia's magic hardware/software.

[Big-G]

5 minutes ago, Hunched said:

Hopefully MSI can deliver, 8+6pin is good.

I believe it was revealed today, but the Gigabyte GTX 1080 G1 has a single 8-pin, so it's shit nobody buy it.

At least there's the Gigabyte GTX 1080 Xtreme, has dual 8-pins... I wish it looked better though.

It's going to be silly if they can't solve the bios voltage limit but are slapping down dual 8-pin connectors on their card.

"Wow I'm so glad I can do literally nothing with this extra power!"

 

I'm actually keeping my eyes on the Water Cooled version of the Xtreme

[DELTAprime]

6 minutes ago, Hunched said:

It's going to be silly if they can't solve the bios voltage limit but are slapping down dual 8-pin connectors on their card.

"Wow I'm so glad I can do literally nothing with this extra power!"

So it take it as things currently are there is no reason to have anything but a single 8 pin connector on a 1080 because of the voltage limitation?

[Hunched]

1 minute ago, stconquest said:

The question, I believe, is:  Is Nvidia locking down the voltage for all their board partners?

 

We may get the potential for better power delivery only to be voltage throttled by Nvidia's magic hardware/software.

I know, which would to a degree make the better power delivery pointless until it can be unlocked.

Which makes me believe MSI and Gigabyte will fix this if they're allowed. Asus is just retarded and likely rushed their card to market without testing anything, as usual.

If none of the more competent board partners can fix it, well hopefully it can be fixed by users somehow.

[stconquest]

1 minute ago, Hunched said:

I know, which would to a degree make the better power delivery pointless until it can be unlocked.

Which makes me believe MSI and Gigabyte will fix this if they're allowed. Asus is just retarded and likely rushed their card to market without testing anything, as usual.

If none of the more competent board partners can fix it, well hopefully it can be fixed by users somehow.

I am curious to see how they are doing it.  Probably just code hidden somewhere.

[Hunched]

7 minutes ago, DELTAprime said:

So it take it as things currently are there is no reason to have anything but a single 8 pin connector on a 1080 because of the voltage limitation?

I wouldn't go as far as saying a single 8-pin is acceptable because of a voltage lock, no.

Especially if the issue is ever resolved, you would regret the decision of getting a single 8-pin even more. You'd miss out on some unlocked performance.