Gigabyte GTX 1070 ITX - Problems?

[philipma1957]

So I did have msi drop setting and go full speed ahead on my gigabyte aorus  but with big fans/heat sink  it survived

once I did this to the clock settings they stayed put and no runaway tdp

Looks like sometimes the software looses control on the cards, this happened sometimes here too. Thanks for the tip.

For mining purposes, I always set TDP bellow 80%, no voltage added, fans at 100%, Temperatures limited to 75ºC maximum, also all my cards are on an open case with strong ventilation and they're all custom cards with beef coolers and at least/minimum 2 fans each.

This thread scares me a little, since my rig stays online 24/7, I was thinking in these Gigabyte 1070GTX Mini cards, but they're too expensive here in my country, I think the dual fan Galax 1070 GTX Mini OC is a better choice.

yeah  i have mined with cards since 2011  and sometimes the cards drop the software  mostly msi  afterburner and just go full blast.

So any card with a really good cooling system stands to do okay  but the mini cards with 1 fans and a small heat sink  can suffer.

so far I have used  these nvidia

aorus 1080 ti = dropped msi setting
gigabyte 1070 itx = no issues since I set to third party.
msi aero 1070 itx = no issues as it is a msi
msi aero 1080 ti   = no issues as it is a msi
zotac 1080 ti founders = no issues set at   third party.
asus  1080 ti blower turbo = no issues set at third party.
zotac 1080 mini    = no issues set at third party.
asus 1070 strix  = no issues set at third party
msi 1070 gamer = no issues  a msi card
zotac 1060 mini = no issues set at third party
zotac 1050 ti      = no issues set at third party
pny 1080 ti founders = no issues set at third party.

Someone  else has complained about his 1080 ti an aorus doing a jump from 60 tdp to 85 tdp  I told him try the third party setting.
that was in my main alt coin thread.

https://bitcointalk.org/index.php?topic=1877588.0

some where past page 70

[1714324295]

1714324295

[1714324295]

1714324295

[1714324295]

1714324295

[jpouza]

So I did have msi drop setting and go full speed ahead on my gigabyte aorus  but with big fans/heat sink  it survived

once I did this to the clock settings they stayed put and no runaway tdp

Looks like sometimes the software looses control on the cards, this happened sometimes here too. Thanks for the tip.

For mining purposes, I always set TDP bellow 80%, no voltage added, fans at 100%, Temperatures limited to 75ºC maximum, also all my cards are on an open case with strong ventilation and they're all custom cards with beef coolers and at least/minimum 2 fans each.

This thread scares me a little, since my rig stays online 24/7, I was thinking in these Gigabyte 1070GTX Mini cards, but they're too expensive here in my country, I think the dual fan Galax 1070 GTX Mini OC is a better choice.

yeah  i have mined with cards since 2011  and sometimes the cards drop the software  mostly msi  afterburner and just go full blast.

So any card with a really good cooling system stands to do okay  but the mini cards with 1 fans and a small heat sink  can suffer.

so far I have used  these nvidia

aorus 1080 ti = dropped msi setting
gigabyte 1070 itx = no issues since I set to third party.
msi aero 1070 itx = no issues as it is a msi
msi aero 1080 ti   = no issues as it is a msi
zotac 1080 ti founders = no issues set at   third party.
asus  1080 ti blower turbo = no issues set at third party.
zotac 1080 mini    = no issues set at third party.
asus 1070 strix  = no issues set at third party
msi 1070 gamer = no issues  a msi card
zotac 1060 mini = no issues set at third party
zotac 1050 ti      = no issues set at third party
pny 1080 ti founders = no issues set at third party.

Someone  else has complained about his 1080 ti an aorus doing a jump from 60 tdp to 85 tdp  I told him try the third party setting.
that was in my main alt coin thread.

https://bitcointalk.org/index.php?topic=1877588.0

some where past page 70

I also have an Aorus non Extreme 1080Ti card, btw wonderful card, temps never go above 64ºC, just tried this third part settings and it's totally stable until now.

I had a Titan X Pascal, and even with reference card settings enabled and 75ºC max temp, sometimes temps just went to the moon (about 83-87ºC).

[QuintLeo]

Today I soldered two AWG20 wires across the dead fuse and hooked up a cheap ass multimeter capable of measuring 10A DC current. The damn card works flawlessly and pulls 6.6A over that fuse (WTF!!) under full overclocked load. That's 80W of power via PCIe.

 Slightly out of spec as PCI-E allows for a 75 watt draw from the bus.
 Seems odd that they would prioritise draw out of the bus and not the plug-in connector though.


 "Target Temp" settings in both Afterburner and PrecisionX have never worked well for me, it's WHY I insist on being able to set fan curves (they work MUCH more reliably).


 I suspect that the reason Afterburner will sometimes "lose control of the card" is due to the driver resetting - that DEFINITELY can confuse NVIDIA-SMI and I don't see why it would not confuse 3'd party solutions as well.

[philipma1957]

Today I soldered two AWG20 wires across the dead fuse and hooked up a cheap ass multimeter capable of measuring 10A DC current. The damn card works flawlessly and pulls 6.6A over that fuse (WTF!!) under full overclocked load. That's 80W of power via PCIe.

 Slightly out of spec as PCI-E allows for a 75 watt draw from the bus.
 Seems odd that they would prioritise draw out of the bus and not the plug-in connector though.


 "Target Temp" settings in both Afterburner and PrecisionX have never worked well for me, it's WHY I insist on being able to set fan curves (they work MUCH more reliably).


 I suspect that the reason Afterburner will sometimes "lose control of the card" is due to the driver resetting - that DEFINITELY can confuse NVIDIA-SMI and I don't see why it would not confuse 3'd party solutions as well.

so far  all my 10 series nvidia  it only happened on my gigabyte aorus.  So  it is hard to say what to do for other cards.  but I do set all gigabyte cards to 3rd party and nothing goes wrong.

mind you I mostly moved all rigs to nvoc0017 my preferred os for nvidia

[un4given]

i can also confirm strange behavior on some of my ITX OC cards. I have been using them since January with 70% TDP and +200 core for ZEC Mining. Temps were pretty ok with 65 degrees but it is a no long run solution. No in Summer Temps go up to 80 degree and i told miner to switch off the card if it reaches 80 degrees celsius.

One Cards Fan dies 2 days ago and i just plugged the whole fan thingie off and using an adapter plugged a normal 120 MM Fan to it using cable binders

Another card in 6x 1070 rig not showing anymore in Windows, need to investigate on weekend...hope it didnt burn down...maybe riser is dead

[SCSI2]

Oh yeah! I had one out of eight die on me on Sunday for no apparent reason. Didn't want to wait for newegg to RMA it and took a closer look at it.

Out of all cards I have, this one is the only 1070 that has a regular 10A FUSE (not polyfuse) marked F4 mounted on the back side right next to the PCIe fingers. It's a small white ceramic part (1206 size I think) with 10 written on it and it was blown open! No visible damage to it, just blown inside.

Today I soldered two AWG20 wires across the dead fuse and hooked up a cheap ass multimeter capable of measuring 10A DC current. The damn card works flawlessly and pulls 6.6A over that fuse (WTF!!) under full overclocked load. That's 80W of power via PCIe. I would never expect that much and now wonder how much does it pull from the 8-pin connector on the top... probably 2A?

So... to confirm, just get a multimeter and measure continuity of the F4 fuse. I think it is simply undersized for this design.

P.S. The risers I use are all V006C with 20A poly fuse onboard, but that obviously didn't help and the poor card blew itself up.

Let me know if any of you have this same fuse blown...

-scsi

nice find  let us know if it keeps working.

The card keeps working like nothing happened for two days non-stop. Multimeter acts as a "fuse" for now and still shows 6.6A under full load. Tomorrow will get clamp-on current meter and measure other wires.

[philipma1957]

Oh yeah! I had one out of eight die on me on Sunday for no apparent reason. Didn't want to wait for newegg to RMA it and took a closer look at it.

Out of all cards I have, this one is the only 1070 that has a regular 10A FUSE (not polyfuse) marked F4 mounted on the back side right next to the PCIe fingers. It's a small white ceramic part (1206 size I think) with 10 written on it and it was blown open! No visible damage to it, just blown inside.

Today I soldered two AWG20 wires across the dead fuse and hooked up a cheap ass multimeter capable of measuring 10A DC current. The damn card works flawlessly and pulls 6.6A over that fuse (WTF!!) under full overclocked load. That's 80W of power via PCIe. I would never expect that much and now wonder how much does it pull from the 8-pin connector on the top... probably 2A?

So... to confirm, just get a multimeter and measure continuity of the F4 fuse. I think it is simply undersized for this design.

P.S. The risers I use are all V006C with 20A poly fuse onboard, but that obviously didn't help and the poor card blew itself up.

Let me know if any of you have this same fuse blown...

-scsi

nice find  let us know if it keeps working.

The card keeps working like nothing happened for two days non-stop. Multimeter acts as a "fuse" for now and still shows 6.6A under full load. Tomorrow will get clamp-on current meter and measure other wires.

6.6amp is really high.  I suspect that the firmware is not ideal on gigabyte 1070's and on 1080ti aorus.

I have a few msi itx and the part/fuse in question is different.
The pcb board looks very similar .  I will photo the msi late today.

[SCSI2]

...

6.6amp is really high.  I suspect that the firmware is not ideal on gigabyte 1070's and on 1080ti aorus.

I have a few msi itx and the part/fuse in question is different.
The pcb board looks very similar .  I will photo the msi late today.

Most likely you don't have a fuse there but rather a low ohm current-sensing resistor. It's probably somewhat large and has an R005 or some similar value. Very easy to monitor current consumption by simply measuring the voltage drop across that part and using ohms law.

On all cards in my possession which don't have the backplate covering the parts, I see they use current-sensing resistors in that circuit and only Gigabyte 1070 ITX has plain vanilla fuse in there.

[nzalog]

Here we go...

http://oi67.tinypic.com/2rq2j35.jpg

Any idea on the full spec of that fuse?
I'm thinking of just soldering an external fuse holder.

This is the exact issue 2 cards have had already and shipping it back for RMA just to replace the fuse seems ridiculous. Rather not waste my money on shipping since fuses are much cheaper.

[SCSI2]

Here we go...

Any idea on the full spec of that fuse?
I'm thinking of just soldering an external fuse holder.

This is the exact issue 2 cards have had already and shipping it back for RMA just to replace the fuse seems ridiculous. Rather not waste my money on shipping since fuses are much cheaper.

No exact spec sorry. My educated guess would be a fast blow 10A ceramic fuse in a 1206 SMT package. Something like this should work just fine: https://www.digikey.com/product-detail/en/bel-fuse-inc/0685H9100-01/507-1942-1-ND/5466714

I'll probably order ten or so since they're cheap...

My only 1070 ITX with this issue (out of 8 total) is still going strong and i still have that cheap multimeter acting as a fuse on it because I have other things to do. It's also fun to watch current readings fluctuate around 6.0A

-scsi

[szafa]

Sell this card if the problem.

[nzalog]

Here we go...

http://oi67.tinypic.com/2rq2j35.jpg

Any idea on the full spec of that fuse?
I'm thinking of just soldering an external fuse holder.

This is the exact issue 2 cards have had already and shipping it back for RMA just to replace the fuse seems ridiculous. Rather not waste my money on shipping since fuses are much cheaper.

No exact spec sorry. My educated guess would be a fast blow 10A ceramic fuse in a 1206 SMT package. Something like this should work just fine: https://www.digikey.com/product-detail/en/bel-fuse-inc/0685H9100-01/507-1942-1-ND/5466714

I'll probably order ten or so since they're cheap...

My only 1070 ITX with this issue (out of 8 total) is still going strong and i still have that cheap multimeter acting as a fuse on it because I have other things to do. It's also fun to watch current readings fluctuate around 6.0A

-scsi

So I guess this was not the same issue on mine.
The fuse is blown, but there seems to be a more serious underlying cause. The PSU instantly went into a protect mode if I shorted across the fuse.
Wouldn't doubt if I made things worse.

[SCSI2]

Here we go...

Any idea on the full spec of that fuse?
I'm thinking of just soldering an external fuse holder.

This is the exact issue 2 cards have had already and shipping it back for RMA just to replace the fuse seems ridiculous. Rather not waste my money on shipping since fuses are much cheaper.

No exact spec sorry. My educated guess would be a fast blow 10A ceramic fuse in a 1206 SMT package. Something like this should work just fine: https://www.digikey.com/product-detail/en/bel-fuse-inc/0685H9100-01/507-1942-1-ND/5466714

I'll probably order ten or so since they're cheap...

My only 1070 ITX with this issue (out of 8 total) is still going strong and i still have that cheap multimeter acting as a fuse on it because I have other things to do. It's also fun to watch current readings fluctuate around 6.0A

-scsi

So I guess this was not the same issue on mine.
The fuse is blown, but there seems to be a more serious underlying cause. The PSU instantly went into a protect mode if I shorted across the fuse.
Wouldn't doubt if I made things worse.

Well, looks like they had a reason to put the fuse in there after all. And I will repeat myself by saying that I've never seen a regular fuse on any of the other 1070's where I can actually see the parts, i.e. open back. Sounds like an afterthought on Gigabyte part and a cheap choice of fuse to boot.

[jojo01]

Here we go...

http://oi67.tinypic.com/2rq2j35.jpg

can you give me the pic again?

[Bohr256]

Last week (around the same time you posted), the one gygabyte gpu I have (sans itx) stopped working too. Mine had to do w/ the drivers, nothing an update couldn't fix. Just kinda weird.

[BreadPit]

Here we go...

http://oi67.tinypic.com/2rq2j35.jpg

Any idea on the full spec of that fuse?
I'm thinking of just soldering an external fuse holder.

This is the exact issue 2 cards have had already and shipping it back for RMA just to replace the fuse seems ridiculous. Rather not waste my money on shipping since fuses are much cheaper.

No exact spec sorry. My educated guess would be a fast blow 10A ceramic fuse in a 1206 SMT package. Something like this should work just fine: https://www.digikey.com/product-detail/en/bel-fuse-inc/0685H9100-01/507-1942-1-ND/5466714

I'll probably order ten or so since they're cheap...

My only 1070 ITX with this issue (out of 8 total) is still going strong and i still have that cheap multimeter acting as a fuse on it because I have other things to do. It's also fun to watch current readings fluctuate around 6.0A

-scsi

So I guess this was not the same issue on mine.
The fuse is blown, but there seems to be a more serious underlying cause. The PSU instantly went into a protect mode if I shorted across the fuse.
Wouldn't doubt if I made things worse.

Well, looks like they had a reason to put the fuse in there after all. And I will repeat myself by saying that I've never seen a regular fuse on any of the other 1070's where I can actually see the parts, i.e. open back. Sounds like an afterthought on Gigabyte part and a cheap choice of fuse to boot.

I bought a card that turns out to be dead (no signal, no fan). so pretty likely it is this exact issue. See this Ghetto repair, looks promising: https://www.reddit.com/r/pcmasterrace/comments/9wlcse/ghetto_fuse_repair_on_gigabyte_gtx_1070_mini_itx/

I am looking for a more refined fix though I am not an electrics expert, but why not replace this with the same parts that you mention for other 1070s?

- Polyfuse like your other (Gigabyte Mini?) 1070s seem to have?
- Low ohm current sensing resistor? 

Or does this not fit into the rest of the Gigabytes 1070 Mini circuitry?