Thanks
Topic: Crate rig FIRE!! (Read 3055 times)
Ok.
Thanks
Thanks
I'm runnin 3 7970s on unpowered risers through a 1200w FSP Group on a Gigabyte B75 board + Celeron g1610 combo with no problems mining ltc for roughly a month now...
OP, you never mentioned your PSU specs? Or maybe i'm blind...
OP, you never mentioned your PSU specs? Or maybe i'm blind...
With only 3 cards you should be reasonably safe. With more than 3 use powered risers.
Better safe than sorry, OP was running 4.
I'm runnin 3 7970s on unpowered risers through a 1200w FSP Group on a Gigabyte B75 board + Celeron g1610 combo with no problems mining ltc for roughly a month now...
OP, you never mentioned your PSU specs? Or maybe i'm blind...
OP, you never mentioned your PSU specs? Or maybe i'm blind...
quad GPU needs LOOOOT of cooling while in closed - case. I would personally build/buy open case for such rig
No luck on a case, but I did find a rack that will probably suit my needs. In the crate, video cards are staggered vertically so they touch each other at the corners, but fans have enough open space between them for good ventilation. The rack I want to use would not give me as much vertical room. So that brings me to another stupid question!
If I got my 2 powered risers, and had 2 cards on those with the other 2 cards directly in the 16x slots on the motherboard and it runs stable in that configuration.... Could I use one of my non-powered risers just to spread the cards out for better airflow? Meaning I would have 2 cards on powered risers, one card in the motherboard 16x slot, and one non-powered 1x->16x riser in the other 16x slot for the last card. I don't see how this would be any different for mining than having 2 cards directly on the motherboard, unless the 1x (non-powered) riser is not capable of transferring as much power as a regular 16x slot? Would this work?
I could always just get another riser for the rack that I want to use, but I hate to waste the ones I already have.
If I got my 2 powered risers, and had 2 cards on those with the other 2 cards directly in the 16x slots on the motherboard and it runs stable in that configuration.... Could I use one of my non-powered risers just to spread the cards out for better airflow? Meaning I would have 2 cards on powered risers, one card in the motherboard 16x slot, and one non-powered 1x->16x riser in the other 16x slot for the last card. I don't see how this would be any different for mining than having 2 cards directly on the motherboard, unless the 1x (non-powered) riser is not capable of transferring as much power as a regular 16x slot? Would this work?
I could always just get another riser for the rack that I want to use, but I hate to waste the ones I already have.
LOL that might defeat the purpose 
I know that back in the ancient times of about a decade ago when I was in college, a buddy of mine had built his pc into a large server tower that had TONS of extra slots / bays.... but I can't seem to find anything like that out there right now
I know that back in the ancient times of about a decade ago when I was in college, a buddy of mine had built his pc into a large server tower that had TONS of extra slots / bays.... but I can't seem to find anything like that out there right now
Thanks, I watched it for a couple hours today and the mosfets got warm, not too hot to touch just warm, and it was stable with no issues. So I am pretty sure it was the non-powered risers that caused the issue. I will probably be ordering some more so all 4 cards will be on powered riser cables and that will keep the motherboard from struggling to provide power.
I was going to leave it as a crate rig but I may want to put it in a real case now that this incident has me paranoid about it.... anyone know of an atx case with 12+ back panel slots to properly space these cards?? lol
I was going to leave it as a crate rig but I may want to put it in a real case now that this incident has me paranoid about it.... anyone know of an atx case with 12+ back panel slots to properly space these cards?? lol
You can buy one of these, no problem leaving the house.
Thanks, I watched it for a couple hours today and the mosfets got warm, not too hot to touch just warm, and it was stable with no issues. So I am pretty sure it was the non-powered risers that caused the issue. I will probably be ordering some more so all 4 cards will be on powered riser cables and that will keep the motherboard from struggling to provide power.
I was going to leave it as a crate rig but I may want to put it in a real case now that this incident has me paranoid about it.... anyone know of an atx case with 12+ back panel slots to properly space these cards?? lol
I was going to leave it as a crate rig but I may want to put it in a real case now that this incident has me paranoid about it.... anyone know of an atx case with 12+ back panel slots to properly space these cards?? lol
That stinks. Imagine if you hadn't been there:(. If I build a rig, I better not let my parents see this or that's a straight out no for me:). Hope you figure out the problem and it works out fine next time.
LOL right
I got the new board today, am running it now with only the 2 cards in the provided 16x slots / no risers
The mosfets are still getting pretty hot to touch but no fires yet
Not sure how hot they are supposed to get / are rated for, I will probably go ahead and put all the cards on powered risers over the next week or 2 just to reduce the load on the motherboard
I got the new board today, am running it now with only the 2 cards in the provided 16x slots / no risers
The mosfets are still getting pretty hot to touch but no fires yet
Not sure how hot they are supposed to get / are rated for, I will probably go ahead and put all the cards on powered risers over the next week or 2 just to reduce the load on the motherboard 
and knowing is half the battle!
thanks for sharing
thanks for sharing
Interesting.... I was under the impression that the cpu would not be used when mining, so a rig just mining and doing nothing else wouldn't really put considerable use / stress on a cpu.
It is not.
These chips power the CPU that is the real power drawer from motherboard. Some AMD FX processors consume 140W and more when pushed really hard.
The cards take most of the power from additional power connectors from PSU. If the motherboard is unable to supply all PCI-E slots with nominal power it most likely will result in instability not fire. So I think the problem was low quality board components that gave up when put to use not configuration of the devices.
The cards take most of the power from additional power connectors from PSU. If the motherboard is unable to supply all PCI-E slots with nominal power it most likely will result in instability not fire. So I think the problem was low quality board components that gave up when put to use not configuration of the devices.
Interesting.... I was under the impression that the cpu would not be used when mining, so a rig just mining and doing nothing else wouldn't really put considerable use / stress on a cpu.
Anyway this thing was sitting not 3 feet from a large stone slab in front of my fireplace, which will be the new home for it when it gets rebuilt. It will have several days of mining under a watchful eye before I even think about leaving it. I'm now wondering whether that bamboo board is a good idea or not, but maybe I will just get a nice (real) case with insane ventilation for it when the funds allow.
The GPUs are indeed drawing power from additionally cables, but the memory is still drawing power from PCI slots. (Issue amplified when mining LTC)
This, combined with low quality components will lead to melting cables or even fire.
A problem that powered risers can solve.
This, combined with low quality components will lead to melting cables or even fire.
A problem that powered risers can solve.
I did not see the pictures but according to description the PWM area / mosfets caught fire. Nothing to do with PCI-E lanes. ASSrock is making much better boards than they were few years ago but this happens from time to time. AM2+ boards combined with 6core black edition phenoms were prone to PWM failures with fire back then. And MSI and Gigabyte were best burning boards back then.
That is correct, those tiny chips near the 24 pin connector is what caught fire. But how does that have nothing to do with the PCIe lanes? I am no expert, just thinking that if the PCIe lanes were asking the board for more power than it could supply, it would make sense for the PWM area of the board to become overloaded / unstable / or in this case on fire, wouldn't it?
The cards take most of the power from additional power connectors from PSU. If the motherboard is unable to supply all PCI-E slots with nominal power it most likely will result in instability not fire. So I think the problem was low quality board components that gave up when put to use not configuration of the devices.
Something like this: https://bitcointalksearch.org/topic/m.1703391 ?
Yep lol, that is my board and that is exactly where my fire started. Notice he also was doing LTC without powered risers.... at least I know now, but I am still gonna be real scared to leave the rig running unattended after this. Maybe I can set it up to run in my fireplace, just incase
Something like this: https://bitcointalksearch.org/topic/m.1703391 ?
Yep lol, that is my board and that is exactly where my fire started. Notice he also was doing LTC without powered risers.... at least I know now, but I am still gonna be real scared to leave the rig running unattended after this. Maybe I can set it up to run in my fireplace, just incase
I did not see the pictures but according to description the PWM area / mosfets caught fire. Nothing to do with PCI-E lanes. ASSrock is making much better boards than they were few years ago but this happens from time to time. AM2+ boards combined with 6core black edition phenoms were prone to PWM failures with fire back then. And MSI and Gigabyte were best burning boards back then.
That is correct, those tiny chips near the 24 pin connector is what caught fire. But how does that have nothing to do with the PCIe lanes? I am no expert, just thinking that if the PCIe lanes were asking the board for more power than it could supply, it would make sense for the PWM area of the board to become overloaded / unstable / or in this case on fire, wouldn't it?
Something like this: https://bitcointalksearch.org/topic/m.1703391 ?
I mine BTC on one rig with 5 7970 cards without powered risers, just simple risers.
For BTC mining we usually down-clock the memory as low as we can go.
LTC mining OTOH is using high memory clocks and pulling a lot of power from the motherboard.
Thanks, so I can at least use the risers I have, though I probably won't unless I build a btc specific rig.
Jump to: