Towering Inferno: A Caselabs STH10 Production

[jrees]

Well, that used 680 FTW+ still sits on kijiji. The guy hasn't responded to any of my messages. Memory Express has them on sale for $39.99 more than he is asking, so I went ahead and ordered one. I figured it was well worth it to not deal with somebody else's potential problems, and to have the 30-day return/exchange window with ME in case this card also suffers from the voltage issue. I hope to have it before Friday so I can put it through the paces this weekend.

With that being said, I'll have to put in an order with Daz for a third block, and a new CSQ connector. Sadly, for tri-sli on my board in the preferred config there is no acrylic option from EKWB for a tri-sli (maching problems according to them???). Will have to go with a black connector, which will look good as well..........just not as nice unfortunately

[stephen]

hey, i just read about your problem. to me this sounds excatly like a psu problem in some cases when a gpu isnt getting enough power it will underclock.

have your tried them on another psu?

i recently had similar problem with my bitcoin mining rig the psu is multi rail and i had to split the 6 and 8 pin on two different rails.

worth a try if you can

edit: just looked into your psu and its also a multi rail, try each pci-e connector on a seperate rail.

edit edit: your psu rails are 20 amps each and from evga page for your gpus (Minimum recommended power supply with +12 Volt current rating of 38 Amps.) youll need two rails for sure.

so your kinda pushing it to on your rails

[jrees]

hey, i just read about your problem. to me this sounds excatly like a psu problem in some cases when a gpu isnt getting enough power it will underclock.

have your tried them on another psu?

i recently had similar problem with my bitcoin mining rig the psu is multi rail and i had to split the 6 and 8 pin on two different rails.

worth a try if you can

edit: just looked into your psu and its also a multi rail, try each pci-e connector on a seperate rail.

edit edit: your psu rails are 20 amps each and from evga page for your gpus (Minimum recommended power supply with +12 Volt current rating of 38 Amps.) youll need two rails for sure.

so your kinda pushing it to on your rails

Appreciate your feedback, but I am absolutely certain it's not my PSU as it works fine with 1 of the 2 card I received from EVGA. The bum card performs the exact same in a different system with a Corsair HX850. See post 168 for the basic troubleshooting that I did.

I don't think my 12v rails amperage is an issue. I'm not well-versed in all of that, but this PSU is rated for 124A across the 12v rails. If I'm not mistaken, when EVGA says 38 amps on the cards page, that is for the whole system, is it not? It's not 38 per card the way I understand it. If anybody has a link to a guide for understanding all of this I'd appreciate it as it'll give me a better understanding of my rig.

I'd hate to think EVGA's flagship PSU couldn't support a measly SLI system!

[Izerous]

Edit: When I look up the 680FTW I found the following TDP.

TDP: 195 W
(195 watts) / (12 volts) = 16.25 amperes

Which means on a multi rail PSU with 20A rails you would want each card on it's own individual rail.

If each card was consuming 38A @ 12V...
(38 amperes) * 12 volts = 456 watts

which would bring it into the area of max power consumption of a 6990 in the boosted BIOS

[bungwirez]

I'd hate to think EVGA's flagship PSU couldn't support a measly SLI system!

Sorry to disappoint Jrees, but this is not a favorable review of the EVGA SuperNOVA NEX1500 PSU IMO...

Among the tragic missteps, we find a multi-12v rail unit with a numbered and color coded modular interface that DOES NOT match up to the 12v rail assignment. What was EVGA thinking?

www.hardocp.com

[stephen]

sorry early morning post, should have specified that the rails are usualy populated by other connectors aswell, as the chart that bung posted.

nice enough that they specified whats on what rail so you could probably divide and find out.

thats mostly why i stick to single rail psu's now. as overcurrent protection per rail just isnt worth being underpowered.

but like you said it doesnt work on a single rail psu aswell. other then trying a down clock to stock to see if chip just cant handle it.

ive also read i think it was from evga that stress test programs like 3dmark and crap stress certin parts of the card that no game would touch which can kill a card or make it preform weird.

[gdesmo]

Well that`s one opinion, but I don`t think they touched on the fact that the software package allows you to change rail configurations and voltages. One of the OCN club members is the psu editor and worked for EVGA in the development of that psu. He says it is an excellent psu and that is good enough for me, I also take Hardcore`s results with a grain of salt because they are not saying what kind of measurement hardware they are using. Who knows how accurate it is ! Jrees I would first look into the software settings you have setup because it is possible that there is a default setting that may not be optimal for your setup. Your cards do not use as much power as people think they do, main card will use full power and others will only use 50 -65 % of TDP because they share the memory. You may want to switch the cables around as well, you never know when you get a wire that has not been crimped incorrectly or not pushed in far enough. Have read a few other reports on the psu and they were favorable, Hardcore was probably peed-off that they did not get a donor. Best of luck and stay positive !

[stephen]

gpu's dont share memory, if a card has 2 gb's and you have 3 cards you still only have 2 gb's of vram

[jrees]

Sorry to disappoint Jrees, but this is not a favorable review of the EVGA SuperNOVA NEX1500 PSU IMO...

Yeah, read the review a while back. I have been following the H for years, and it's one of my go-to sites. They really do seem to have a hate on for EVGA, and I am certain they set out with the intent to give a bad review of this unit. They got a launch unit, and apparently some things have been ironed out with future shipments.

Everything else out seems to suggest it's fine and I'm comfortable with my purchase even though it was ripped at the H. Time will tell I guess.

[bungwirez]

...I'm comfortable with my purchase even though it was ripped at the H. Time will tell I guess.

Really, this is what matters the most, I'm just devils advocating for the sake of education is all.

[jrees]

sorry early morning post, should have specified that the rails are usualy populated by other connectors aswell, as the chart that bung posted.

nice enough that they specified whats on what rail so you could probably divide and find out.

thats mostly why i stick to single rail psu's now. as overcurrent protection per rail just isnt worth being underpowered.

but like you said it doesnt work on a single rail psu aswell. other then trying a down clock to stock to see if chip just cant handle it.

ive also read i think it was from evga that stress test programs like 3dmark and crap stress certin parts of the card that no game would touch which can kill a card or make it preform weird.

The unit comes with a layout suggesting 2 different configs. I received an email from a higher-up at EVGA support and I've asked for his suggestions on how to configure my PSU cabling for what I want done. He also said they have arranged to have my card "burned-in prior to shipping out to ensure there are no problems with the new card".

That being the case, my card arrives there tomorrow, so hopefully it's just a matter of shipping my new one out upon receipt of the RMA. The one I order from Memory Express will be here on Wednesday according to Canada Post.

[stephen]

to me thats some supreme support right there, just sucks to be in the situtation.

[jrees]

sorry early morning post, should have specified that the rails are usualy populated by other connectors aswell, as the chart that bung posted.

nice enough that they specified whats on what rail so you could probably divide and find out.

thats mostly why i stick to single rail psu's now. as overcurrent protection per rail just isnt worth being underpowered.

but like you said it doesnt work on a single rail psu aswell. other then trying a down clock to stock to see if chip just cant handle it.

ive also read i think it was from evga that stress test programs like 3dmark and crap stress certin parts of the card that no game would touch which can kill a card or make it preform weird.

Just grabbed all the documentation for the PSU and have been looking it over again. The PSU has 5 manual dip switches and I have switches 3 and 5 enabled which allows for Single Rail operation which is recommended by EVGA for multi-GPU systems. Didn't even bother with the multi rail config now that I think about it.

[jrees]

to me thats some supreme support right there, just sucks to be in the situtation.

That's why I have liked EVGA for so long. Never really needed much support in the past, but their reputation is unmatched when it comes to support. It's nice to have quick answers, and communication is good.

[jrees]

RMA'd card arrived today. Disabled my current card, and ran this one solo without a problem. It doesn't OC as far as my other card, but it seems that one is pretty golden. I expect the 3rd card to be here Wednesday, but the latest shipping update says it's in transit from Mississauga, so perhaps it'll be here Tuesday.

Here's hoping it's problem-free. Will put the new ones through more strenuous testing tomorrow to be sure I can go ahead and put the waterblocks on'em.