©ASUSTEK COMPUTER INC. ALL RIGHTS RESERVED.

berryboy · ‎06-09-2015

Hey guys,
I've come to this forum hoping you guys will help solve my problem as this is my last resort. Please help!

I purchased the following build:
ASUS Rampage V Extreme X99 Motherboard
Intel i7-5820K
16GB 2400 Crucial Sport DDR4
EVGA Supernova 850W B2 PSU
Samsung Evo 850 250GB SSD
2x EVGA GTX 980 Reference

Before I purchased my second card, my system was running beautifully. I was getting very good performance and amazing 3DMark scores. I happened to get a very good deal and picked up a brand new EVGA GTX 980 reference card. Upon putting them into SLI (yes I enabled SLI), the performance fell apart. I didn't notice much increase in game and my 3DMark scores actually suffered to the point where they were scoring below a normal laptop.
I knew something was wrong so I started troubleshooting. Below I've listed what I've tried so far:

Memtest
New, stronger, more efficient power supply
New GTX 980s
New set of 970s to confirm the cards are not the issue
Replacement Rampage V Extreme
Reinstalled Windows 7
Reinstalled to Windows 8.1 Pro
Reinstalled drivers
Tested each of the 3 980's in both x16 slot and x8 slot - result was very close benchmark scores in 3DMark and Heaven

So as you can see I've tried a lot of different things and to no avail. It seems like it's either software related, such as me having some settings wrong, or the CPU is busted.

Here's some other info related to my issue: All overclocks have been removed. This issue still persists with stock settings and GSYNC turned off. When 3DMark finishes, the display driver crashes while trying to show my scores in Google Chrome. I've forced my cards into Gen 3 speeds in the BIOS but maybe I've missed something. I've also connected the additional 4 pin power for the CPU. When I check GPU-Z, both cards hit the normal voltage regulator limit, so they're getting enough power. Maybe one of them is still bad? My PC used to lose power and just restart when running something intensive so I thought it might be PSU, etc. But it wasn't. Drivers crash randomly and after 3DMark which is the strangest thing. I've been seeing a lot of issues like this and was wondering if anyone could help me out.

Please let me know if you need additional info and I'll respond and add it to this post for other people trying to help. Thank you all so much for your time.
Kevin

Chino · ‎06-09-2015

Have you tested each GTX 980 individually to see if you get the same score? Also have you tried swapping SLI cables as well? Also please indicate which PCIe slots you're using for your SLI configuration.

berryboy · ‎06-09-2015

Chino wrote:
Have you tested each GTX 980 individually to see if you get the same score? Also have you tried swapping SLI cables as well? Also please indicate which PCIe slots you're using for your SLI configuration.

Yes each card has been tested individually. I'm using the x16 and x8 slot so the top slot and the 4th one down. Also, I've tested multiple SLI bridges.

Korth · ‎06-09-2015

The i7-5820K supports 28 PCIe 3.0 lanes, so your GPUs can run in x16/x8 SLI. x16/x16 SLI requires a processor with more PCIe 3.0 lanes.
They should be plugged into the red PCIE_X16_1 and PCIE_X16/X8_3 (1st and 4th) slots, the red PCIE_X8_2 and PCIE_X8_4 slots shouldn't be used.
Check that the BIOS enables the GPU slots in x16 and x8 operation.
Check that the electrical DIP switches on the mobo enable power to the GPU slots.
Use the optional 4-pin "EZ Plug" power connector to deliver more power to the PCIE bus (along with the main 24-pin EATX, 8-pin EATX12V, and 4-pin EATX12V power connectors, of course).
Save changes, Clear CMOS just to be sure, restart.

You should test each GPU card separately, one at a time, to confirm they're both capable of stable x16 operation.
If you're overclocking them, do it for each card separately as well. But use the settings of the slowest-clocked card in your SLI link when running them together.

If you have any system boot/stability issues (QLED errors, BSoDs, etc) related to processor or memory overclocks, you should fully resolve them before complicating things with GPU overclocking.

"All opinions are not equal. Some are a very great deal more robust, sophisticated and well supported in logic and argument than others." - Douglas Adams

[/Korth]

berryboy · ‎06-09-2015

Korth wrote:
The i7-5820K supports 28 PCIe 3.0 lanes, so your GPUs can run in x16/x8 SLI. x16/x16 SLI requires a processor with more PCIe 3.0 lanes.
They should be plugged into the red PCIE_X16_1 and PCIE_X16/X8_3 (1st and 4th) slots, the red PCIE_X8_2 and PCIE_X8_4 slots shouldn't be used.
Check that the BIOS enables the GPU slots in x16 and x8 operation.
Check that the electrical DIP switches on the mobo enable power to the GPU slots.
Use the optional 4-pin "EZ Plug" power connector to deliver more power to the PCIE bus (along with the main 24-pin EATX, 8-pin EATX12V, and 4-pin EATX12V power connectors, of course).
Save changes, Clear CMOS just to be sure, restart.

You should test each GPU card separately, one at a time, to confirm they're both capable of stable x16 operation.
If you're overclocking them, do it for each card separately as well. But use the settings of the slowest-clocked card in your SLI link when running them together.

If you have any system boot/stability issues (QLED errors, BSoDs, etc) related to processor or memory overclocks, you should fully resolve them before complicating things with GPU overclocking.

Right. I researched the PCIE lane layout and read the manual to make sure. The cards are in the 1st and 4th slot down. I'm not sure how to check if the BIOS has x16 and x8 enabled. I can't find any option like this.
All PCIE slots are powered with the DIP switches.
I haven't tried the EZ Plug - is this actually necessary? I'm running these cards at stock 😕 The rest of the connections are powered, including the optional 4-pin EATX12V.

Regarding CMOS, is the CMOS info stored on the motherboard? I'll reset it but I'm just curious.
All overclocks have been removed until I fix this issue. The only stability issues that exist are the driver crashes and power loses that I've mentioned in the initial post.
Thanks!

Nate152 · ‎06-09-2015

Hello berryboy

Clearing the cmos is resetting the bios to defaults. You only need to connect power to the ez plug if you're running more than two gpu's but it won't hurt if you want to connect it.

To check if your gpu's are running at x16 x8, in the bios on the extreme tweaker tab go to GPU Dimm post, your 980's will be listed there.

To see if sli is working, open up the NVidia control panel and at the very top click on 3d settings, put a checkmark beside show sli visual indicator. Then open up a game and you'll see a thick green bar on the left, if it is getting longer and shorter sli is working.

When you reinstalled the NVidia driver did you have both 980's in the motherboard with the sli connector connected? This is how you want to do it.

ROG Strix 1070 OC Guide - ROG Spatha X Rapid Fire - ROG Falchion Touch Panel - z270 BCLK Overclocking - Xsplit Gamecaster Guide - ROG RX 460 OC Guide

berryboy · ‎06-10-2015

Nate152 wrote:
Hello berryboy

Clearing the cmos is resetting the bios to defaults. You only need to connect power to the ez plug if you're running more than two gpu's but it won't hurt if you want to connect it.

To check if your gpu's are running at x16 x8, in the bios on the extreme tweaker tab go to GPU Dimm post, your 980's will be listed there.

To see if sli is working, open up the NVidia control panel and at the very top click on 3d settings, put a checkmark beside show sli visual indicator. Then open up a game and you'll see a thick green bar on the left, if it is getting longer and shorter sli is working.

When you reinstalled the NVidia driver did you have both 980's in the motherboard with the sli connector connected? This is how you want to do it.

I couldn't find any settings like that inside the Extreme Tweaker tab, but in the Advanced section, I looked in PCH config and found where I can set the speeds of each slot. I set the 1st and 4th slot to Gen 3 speed instead of auto. It lets me know that the first slot is running at x16 and the 4th slot is knocked down to x8. Is this similar to what you were talking about?

I've enabled the SLI indicator a long time ago - I used it during all of my benches. The SLI usage never went to 90% or 100% but I figured that would be normal. In some of the tests, it was very low. I'm assuming this is normal for certain tests that are mainly Physx or something.

The drivers were reinstalled with both cards installed with the SLI bridge.

Menthol · ‎06-10-2015

I have seen some people report issues with chrome and any driver newer than 347.88, if you uninstall drivers to install older ones use DDU to uninstall them as it will remove all the files then install your drivers, are you using a flexible sli connector or one of the hard ones that came with the motherboard. Ideally use the 3 way hard sli connector that came with the motherboard your cards would be the first and third with the middle connector not used. If your cards are not the same reference models and you can't use the hard connector we can trouble shoot your issue, I have seen one of the sli connector on a card not working so if using a flexible sli connector try on the left connectors the the right, then left on one card and right on the other to identify if one of the cards have a bad sli connector. are you using any software to overclock or monitor your cards such as GPU teak or afterburner, etc. only install one of these type programs at a time

berryboy · ‎06-10-2015

Menthol wrote:
I have seen some people report issues with chrome and any driver newer than 347.88, if you uninstall drivers to install older ones use DDU to uninstall them as it will remove all the files then install your drivers, are you using a flexible sli connector or one of the hard ones that came with the motherboard. Ideally use the 3 way hard sli connector that came with the motherboard your cards would be the first and third with the middle connector not used. If your cards are not the same reference models and you can't use the hard connector we can trouble shoot your issue, I have seen one of the sli connector on a card not working so if using a flexible sli connector try on the left connectors the the right, then left on one card and right on the other to identify if one of the cards have a bad sli connector. are you using any software to overclock or monitor your cards such as GPU teak or afterburner, etc. only install one of these type programs at a time

Interesting. I'll downgrade the drivers I guess.

Regarding the SLI bridges, I've used two different ones so far (both hard): ROG 3-way bridge, NVIDIA 2-way spaced SLI bridge. Both have the same result. I haven't tried using a flexible bridge nor have I tried using a different side of the connection. Only MSI Afterburner is installed and it's mostly just for monitoring purposes. I'll try the different bridge set up when I get home from work!

Techotic · ‎06-10-2015

berryboy wrote:
Regarding the SLI bridges, I've used two different ones so far (both hard): ROG 3-way bridge, NVIDIA 2-way spaced SLI bridge. Both have the same result. I haven't tried using a flexible bridge nor have I tried using a different side of the connection.

Use the SLI flexible (ribbon cable) bridge, and connect it to the second set of SLI fingers on the cards.

I was running 2 EVGA GTX 980 ACX 2.0 cards in SLI flawlessly. When I added a third card for tri-SLI, I would get red flickering lights in games when running in tri-SLI mode. Turns out, the first set SLI connection fingers on the third card was faulty, but the second set would work fine in regular SLI. Unfortunately, with Tri-SLI you need both set of fingers on ALL cards to be working. Apparently, EVGA has a QA issue with faulty SLI fingers on the 980 cards (usually the first set, closest to the outside bracket).

SLI Issues / Horrific Performance w/ X99 (Rampage V Extreme)