Failed Upgrade

Message boards : Number crunching : Failed Upgrade
Message board moderation

To post messages, you must log in.

AuthorMessage
Profile Shaggie76
Avatar

Send message
Joined: 9 Oct 09
Posts: 282
Credit: 271,858,118
RAC: 196
Canada
Message 1807666 - Posted: 7 Aug 2016, 0:58:01 UTC

I was excited to bring home pair of MSI GTX 1070's last night; after putting them in my rig and letting it cook over night I couldn't have been more pleased. Then, when I had time this morning, I tried installing an SLI-bridge -- and that's when my day went pear-shaped.

I have a relatively fancy rig -- Win7, X99 chipset, enthusiast motherboard, i7-5960X -- no other cards in it and a 860W PSU with lots of headroom. There was no reason for trouble -- it should have been painless.

The first problem is that whenever I had the 2 cards installed, bridge or otherwise, the "Starting Windows" screen would freeze and my monitor would go to sleep. At first I thought it was locking up -- I thought it could be my relative modest overclock so I put everything down to stock (even underclocking the ram a bit for good measure). It took me hours of farting around to realize that it wasn't always freezing but would sometimes wake up after about 3 minutes of stalling. IT was bizarre: "Starting Windows" then frozen for a bit, then monitor off, waiting a while, then monitors ticking back on and there was the login screen. When I first installed the cards I must have gone off to dinner during the first reboot and missed this.

I tried all kinds of stuff: switching the cards around, swapping one of the 1070's for my old 980ti, unplugging all but one monitor, switching the cables to DVI, did a DDU clean driver install, checked for firmware updates on each card, and nothing seemed to make this startup stall go away -- something was definitely wrong. I checked the event log and it was uncanny: you could clearly see a 3 minute gap in the startup event log where nothing happened. I don't know the current state of Linux support but maybe if I had more time I'd try booting something off a thumb-drive to see if I can get more diagnostics.

What's worse is that half the time it wouldn't ever recover -- so I'd be stuck waiting and wondering if I should punch the reset switch. I was nowhere near my PSU limit either -- the two 1070's actually used less power over all than my 980 Ti did even though on paper I'd have expected a bit more pull.

Eventually I got fed up and just yanked out one of the cards thinking I'd use the second in another cruncher. With just one of the 1070's rebooting was fine, things seemed ok, until I let it start cooking and the monitors turned off: now when I go wiggle the mouse only one out of three monitors turns on but doesn't show anything. I know the OS is still ticking away under it all because I can remote in and control it -- the card is just not cooperating anymore.

Anyway, I'm feeling super bummed about the whole experience and will be taking them back to the shop tomorrow and suffering the restocking fee. My 980 Ti might waste power but at least everything works. I guess being an early adopter has its perils -- maybe I'll try again in a year or two.
ID: 1807666 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 13720
Credit: 208,696,464
RAC: 304
Australia
Message 1807694 - Posted: 7 Aug 2016, 2:23:40 UTC - in response to Message 1807666.  
Last modified: 7 Aug 2016, 2:24:00 UTC

I was excited to bring home pair of MSI GTX 1070's last night; after putting them in my rig and letting it cook over night I couldn't have been more pleased. Then, when I had time this morning, I tried installing an SLI-bridge -- and that's when my day went pear-shaped.

Unless you're a gamer, there's no point in using SLI.

Have to say I've happy with my single GT1070 chugging away, in a rather old system.
Probably too late for you to check for motherboard BIOS updates & see if there's anything there relating to GPUs?

I know there are some issues with systems that support Optimus switching between different graphics- some extremely long boot times in Windows. But haven't heard of any issues once it's up.
Grant
Darwin NT
ID: 1807694 · Report as offensive
rob smith Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer moderator
Volunteer tester

Send message
Joined: 7 Mar 03
Posts: 22160
Credit: 416,307,556
RAC: 380
United Kingdom
Message 1807740 - Posted: 7 Aug 2016, 8:48:04 UTC

When I first got the pair of GTX970 I tried SLI, just to see what happened. To put it mildly "disappointing". Similar issues to those you have observed. I rapidly ditched SLI and the results were beyond what I expected. That pair of cards are now sitting alongside a GTX1080 and are performing very well indeed, with SLI disabled, and no SLI connector in place.
The SETI@Home applications are not configured to use SLI (or, I believe, Crossfire on AMD cards), and as we have both seen using it can cause some rather unpleasant effects.
Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?
ID: 1807740 · Report as offensive
Profile Shaggie76
Avatar

Send message
Joined: 9 Oct 09
Posts: 282
Credit: 271,858,118
RAC: 196
Canada
Message 1807752 - Posted: 7 Aug 2016, 11:09:50 UTC - in response to Message 1807694.  

[Unless you're a gamer, there's no point in using SLI.

I used my "game programmer" day-job as an excuse to rationalize getting the setup :)

Probably too late for you to check for motherboard BIOS updates & see if there's anything there relating to GPUs?

Well whaddya know -- I'd updated the BIOS last month but there was another update just over a week ago. Sadly the release notes say nothing other than "Improve system compatibility."

Maybe I'll give that a spin today.

The only other long-shot is I've heard that DisplayPort cables can cause problems; something about power feeding back through one of the pins if not wired correctly; it'll work most of the time but can cause serious problems. Maybe my old card was more forgiving or something.
ID: 1807752 · Report as offensive
Profile Shaggie76
Avatar

Send message
Joined: 9 Oct 09
Posts: 282
Credit: 271,858,118
RAC: 196
Canada
Message 1807769 - Posted: 7 Aug 2016, 14:09:14 UTC

Latest BIOS, still hating life. Put the 980 Ti back in this morning, latest drivers, boots are fast, no TDRs. I could probably find the right flavor of driver to make just one card happy but without the pair it's a side-grade from the 980Ti so there's no point.

I'll have to find something else to satisfy my mid-life crisis.
ID: 1807769 · Report as offensive
Profile M_M
Avatar

Send message
Joined: 20 May 04
Posts: 76
Credit: 45,752,966
RAC: 8
Serbia
Message 1807795 - Posted: 7 Aug 2016, 15:57:25 UTC - in response to Message 1807769.  

I would suspect that that nVidia drivers are still long way to be "bulletproof" and optimized when it comes to Pascal, i.e. 10x0 family, and specially when it comes to SLI setups, so some patience is needed...
ID: 1807795 · Report as offensive

Message boards : Number crunching : Failed Upgrade


 
©2024 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.