PCI-E x16 - x16 riser trouble on Gigabyte motherboard

[waichet]

Hi,

I am in desperate need of help.

I have 4 x Gigabyte GV-R797OC-3GD on Gigabyte GA-Z77x-UP7 motherboard but found that they are running way too hot with all 4 cards sitting on the motherboard since there is not much gap between the graphics card. Since its impossible for me to run all cards in the motherboard without overheating them. I bought 4 PCI-E x16 -x16 extender cable with molex connector from cablesaurus in hope that I could place the graphics card further away from the motherboard and have more gaps between the graphics card.

Problem is with the extender cable connected between the motherboard and the graphics card, the motherboard could only detect 2 of the 4 graphic cards and of 2 detected card only one is getting information of its state (temperature, core MHz, memory MHz, etc).

I found some possible solution to my problem with the motherboard not finding some of the graphics card when connected through a cable riser / extender. It is possible to force the motherboard to detect the graphics card on some motherboard by shorting short pins on pci-e x16 slot of the motherboard. Although for most motherboard when connecting PCI-E x16 to x16 slot through extender / riser cable, it is not required to short the pins for the motherboard to detect the graphics card, but I did find some cases where some people only got their PCI-E x16 slot working on their motherboard (possibly only for Gigabyte's motherboard) after shorting them when using PCI-E x16 riser / extender.

So what I've tried so far is:
1. Tried it on Xubuntu and Windows 7. Both on Windows 7 and Xubuntu are only seeing 2 of the 4 graphics card. While Xubuntu is able to get information (temp, core Ghz, memory Ghz, etc.) from one of 2 detected the graphics cards, windows just BSOD when i try to run GPU-Z.
2. Tried shorting pin A1 - B17 (doesn't work) in PCI-E x16 slot on motherboard using wire from LAN cable. (Could be because shorting of pin A1 - B17 only work for connecting motherboard PCI-E x1 - x16 graphics card) - not working
3. Tried Shorting ping A1 - B81 (for PCI-E x16 card presence detection) - not working
4. Tried plugging monitor on DVI port of the UN-detectable / not working graphic card connecting to motherboard through PCI-E x16 -x16 riser / extender. - not working

I'm running out of idea on what I should do next to have the graphics card working on the motherboard when connecting through riser / extender. Am hoping if you guys could share your experience when getting problematic PCI-E x16 - x16 extender to work on gigabyte motherboard.

[drlukacs]

Go step by step:

First, connect only one card using the riser, and see what happens. (No other cards or risers.)

If i works fine, try the other riser cable, again, with a single card setup.

If either of these steps fail, try to plug the riser at different slots, and see if that makes any difference.

If that works too, then try one card plugged into the mother board, and one card with a riser. See what happens.

Then try only two cards into the motherboard, and no risers, and see again what happens.

Then try two cards into the motherboard, and one card with a riser.

Basically, you should try to isolate the cause of the problem.

[waichet]

Go step by step:

First, connect only one card using the riser, and see what happens. (No other cards or risers.)

If i works fine, try the other riser cable, again, with a single card setup.

If either of these steps fail, try to plug the riser at different slots, and see if that makes any difference.

If that works too, then try one card plugged into the mother board, and one card with a riser. See what happens.

Then try only two cards into the motherboard, and no risers, and see again what happens.

Then try two cards into the motherboard, and one card with a riser.

Basically, you should try to isolate the cause of the problem.

Hi Thanks for the input. You're are right, I'll definitely do that first thing I got home from work. I was so focused on the problem in whole that I forgot to try and isolate the tests to find if the problem only happens on certain slot / if one of the riser cable is bad.

[waichet]

Go step by step:

First, connect only one card using the riser, and see what happens. (No other cards or risers.)

If i works fine, try the other riser cable, again, with a single card setup.

If either of these steps fail, try to plug the riser at different slots, and see if that makes any difference.

If that works too, then try one card plugged into the mother board, and one card with a riser. See what happens.

Then try only two cards into the motherboard, and no risers, and see again what happens.

Then try two cards into the motherboard, and one card with a riser.

Basically, you should try to isolate the cause of the problem.

So this is what i've done for testing, I isolate the test to only use the 1st PCIe x16 slot on the motherboard while trying on different different riser. From my observation, one of the cable seem to work better for the motherboard to detect the graphics card. The way I carried out the test was by having my boot-able HDD removed (so that I don't mess it up from constant powering on and off without proper shutdown), have a monitor connected to the graphics card and riser cable is used to connect PCIe x16 slot 1 on the motherboard to the graphics card (just 1 graphics card on the motherboard). I then power on the computer to see if it can display the boot-up screen on the monitor, with some of the riser the graphic card manage to display the bios boot up screen on the monitor while other just doesn't.

Here is the puzzling part, those risers that used to display the bios boot up screen on first try might stop showing the boot up screen after switching from other riser cable I test on or if I disconnect and reconnecting it (I'm certain everything is plugged in correctly between the motherboard and the graphics card and power supply cables are connected to the graphics card as well). There are times power cycling the motherboard could get it to show the boot up screen on the monitor while other times it just refuse to show anything on the monitor (the monitor is not showing black screen but rather it goes to standby mode thinking that the computer is turned off). Is it possible that the signal between the motherboard and graphics card is corrupted or the signal that the motherboard / graphics card is sending out is too weak for either components to pick up correctly?

To clear out the suspicion if the graphic card is not getting enough power from PSU, I'm using a CoolerMaster Silent Master 1500W so that is well enough for all 4 graphics card plugged in.

[drlukacs]

So this is what i've done for testing, I isolate the test to only use the 1st PCIe x16 slot on the motherboard while trying on different different riser. From my observation, one of the cable seem to work better for the motherboard to detect the graphics card. The way I carried out the test was by having my boot-able HDD removed (so that I don't mess it up from constant powering on and off without proper shutdown), have a monitor connected to the graphics card and riser cable is used to connect PCIe x16 slot 1 on the motherboard to the graphics card (just 1 graphics card on the motherboard). I then power on the computer to see if it can display the boot-up screen on the monitor, with some of the riser the graphic card manage to display the bios boot up screen on the monitor while other just doesn't.

Here is the puzzling part, those risers that used to display the bios boot up screen on first try might stop showing the boot up screen after switching from other riser cable I test on or if I disconnect and reconnecting it (I'm certain everything is plugged in correctly between the motherboard and the graphics card and power supply cables are connected to the graphics card as well). There are times power cycling the motherboard could get it to show the boot up screen on the monitor while other times it just refuse to show anything on the monitor (the monitor is not showing black screen but rather it goes to standby mode thinking that the computer is turned off). Is it possible that the signal between the motherboard and graphics card is corrupted or the signal that the motherboard / graphics card is sending out is too weak for either components to pick up correctly?

To clear out the suspicion if the graphic card is not getting enough power from PSU, I'm using a CoolerMaster Silent Master 1500W so that is well enough for all 4 graphics card plugged in.

Did you try this experiment with all four graphical cards and all possible combinations of graphical card and riser?

In general, it sounds like something is wrong with the risers, perhaps the contact is not perfect. I would try then the same experiment with a different motherboard (single card + riser), to see if the instability is related to the motherboard or the riser.

[waichet]

So this is what i've done for testing, I isolate the test to only use the 1st PCIe x16 slot on the motherboard while trying on different different riser. From my observation, one of the cable seem to work better for the motherboard to detect the graphics card. The way I carried out the test was by having my boot-able HDD removed (so that I don't mess it up from constant powering on and off without proper shutdown), have a monitor connected to the graphics card and riser cable is used to connect PCIe x16 slot 1 on the motherboard to the graphics card (just 1 graphics card on the motherboard). I then power on the computer to see if it can display the boot-up screen on the monitor, with some of the riser the graphic card manage to display the bios boot up screen on the monitor while other just doesn't.

Here is the puzzling part, those risers that used to display the bios boot up screen on first try might stop showing the boot up screen after switching from other riser cable I test on or if I disconnect and reconnecting it (I'm certain everything is plugged in correctly between the motherboard and the graphics card and power supply cables are connected to the graphics card as well). There are times power cycling the motherboard could get it to show the boot up screen on the monitor while other times it just refuse to show anything on the monitor (the monitor is not showing black screen but rather it goes to standby mode thinking that the computer is turned off). Is it possible that the signal between the motherboard and graphics card is corrupted or the signal that the motherboard / graphics card is sending out is too weak for either components to pick up correctly?

To clear out the suspicion if the graphic card is not getting enough power from PSU, I'm using a CoolerMaster Silent Master 1500W so that is well enough for all 4 graphics card plugged in.

Did you try this experiment with all four graphical cards and all possible combinations of graphical card and riser?

In general, it sounds like something is wrong with the risers, perhaps the contact is not perfect. I would try then the same experiment with a different motherboard (single card + riser), to see if the instability is related to the motherboard or the riser.

Thank you so much!!!! You just lifted the weight off me, almost a week of late nights just trying to figure out the problem. You are right that the riser cables is bad. I tried it on my ASUS motherboard with ASUS graphics card and out of 4 riser cables only 1 is working without problem the other 3 is bad.

Apparently the problem was from the riser cable not from the Gigabyte motherboard or graphics card. The guy from cablesaurus offer to refund me if i send the bad cables back (not sure what use getting the bad cables back).

[drlukacs]

Thank you so much!!!! You just lifted the weight off me, almost a week of late nights just trying to figure out the problem. You are right that the riser cables is bad. I tried it on my ASUS motherboard with ASUS graphics card and out of 4 riser cables only 1 is working without problem the other 3 is bad.

Apparently the problem was from the riser cable not from the Gigabyte motherboard or graphics card. The guy from cablesaurus offer to refund me if i send the bad cables back (not sure what use getting the bad cables back).

You are very welcome. I am glad I was able to help.

I would insist that the seller pay for the return shipping of the defective item. In most states, you are probably entitled to it also under the law.

[waichet]

Thank you so much!!!! You just lifted the weight off me, almost a week of late nights just trying to figure out the problem. You are right that the riser cables is bad. I tried it on my ASUS motherboard with ASUS graphics card and out of 4 riser cables only 1 is working without problem the other 3 is bad.

Apparently the problem was from the riser cable not from the Gigabyte motherboard or graphics card. The guy from cablesaurus offer to refund me if i send the bad cables back (not sure what use getting the bad cables back).

You are very welcome. I am glad I was able to help.

I would insist that the seller pay for the return shipping of the defective item. In most states, you are probably entitled to it also under the law.

As per their policy the guy from cablesaurus offer to send me replacements as it is believe the risers was damaged during shipment, he has offer to throw in an extra but from my recent bad experience of such high failure rate from riser cable I'm hoping he could do extra of 2 instead of 1 especially when they are sending me 3 for replacement (from my previous order I received 3 out of 4 bad risers).

[drlukacs]

As per their policy the guy from cablesaurus offer to send me replacements as it is believe the risers was damaged during shipment, he has offer to throw in an extra but from my recent bad experience of such high failure rate from riser cable I'm hoping he could do extra of 2 instead of 1 especially when they are sending me 3 for replacement (from my previous order I received 3 out of 4 bad risers).

That is very nice of them!