I've had this problem for a couple years now, and given the situation it's more an issue of pride than necessity, still I feel like an ass to be unable to get to the bottom of it. Originally the system was
i5 2500k
Asrock H61M-S
4 gb g.skill ram
hd 6870
ocz 550w modular
Never overclocked cpu (because h61), fast forward to two years later and after the regularly scheduled hardware cleaning (and your friendly neighbourhood power mass outage), psu starts acting up and the only pcie x16 slot dies. The rig doesn't even post with the card in it, or it boots as if there's nothing in that slot, black screen with hdd led flashing up as if it's loading the OS. After spending days trying other cards and confirming they all work somewhere else, I just decide to upgrade the system, which becomes
i5 2500k
Asus z77-a
8gb kingston hyper x
hd 6870 (later became gtx 960)
thermaltake berlin 630w
Mobo is crossfire-ready, 1 slot 3.0 (sandy bridge = 2.0 max) and a 2.0 x4 slot below.
Top slot doesn't work from the get-go, literally the same as the older faulty (?) mobo. Unfortunately I had to wait two weeks for the psu to build the rig, so in the meantime I lose right of withdrawal and they tell me "tough *, you gotta deal with warranty", which means buying another motherboard, shipping the broken card is on me, wait 80 days for asus all with the risk of getting a refurbished motherboard for nothing. Meanwhile I checked everything (other OS, no peripherals, breadboarding, heatsink, psu, bent pins, bios updates, disabling igpu, pcie gen settings etc.) short of trying the cpu somewhere else (found NOBODY with 1155). Now, since the bottom slot works flawlessly (southbridge-controlled, not the cpu), and I read that as unlikely as that sounds it can happen for one part of the cpu to malfunction while the rest lives on, I started suspecting it could be the pcie part of the cpu that died. The cpu doesn't overheat, never a bluescreen, it can even take overclock up to 4.5 ghz 24/7 without fail, supports any stress test with good scores including unigine 4.0 at max detail alongside my new 960 without crashing or breaking down (except somewhat lower score I guess due to x4 2.0). Before you ask: no, the problem appeared long before I even started overclocking the cpu.
Two final things:
1.I might have forgotten "take out the ram and let the speaker beep, if it doesn't then it doesn't post and it's the motherboard". I had an old speaker but it's broken now
2. When I put two gpu in the two slots (video hooked to secondary pcie or even igpu) the system boots fine, and this is what I have with card in and out (albeit in doesn't show in any software)
(for those who can't see pictures: x0 with card in, not present with card out)
i5 2500k
Asrock H61M-S
4 gb g.skill ram
hd 6870
ocz 550w modular
Never overclocked cpu (because h61), fast forward to two years later and after the regularly scheduled hardware cleaning (and your friendly neighbourhood power mass outage), psu starts acting up and the only pcie x16 slot dies. The rig doesn't even post with the card in it, or it boots as if there's nothing in that slot, black screen with hdd led flashing up as if it's loading the OS. After spending days trying other cards and confirming they all work somewhere else, I just decide to upgrade the system, which becomes
i5 2500k
Asus z77-a
8gb kingston hyper x
hd 6870 (later became gtx 960)
thermaltake berlin 630w
Mobo is crossfire-ready, 1 slot 3.0 (sandy bridge = 2.0 max) and a 2.0 x4 slot below.
Top slot doesn't work from the get-go, literally the same as the older faulty (?) mobo. Unfortunately I had to wait two weeks for the psu to build the rig, so in the meantime I lose right of withdrawal and they tell me "tough *, you gotta deal with warranty", which means buying another motherboard, shipping the broken card is on me, wait 80 days for asus all with the risk of getting a refurbished motherboard for nothing. Meanwhile I checked everything (other OS, no peripherals, breadboarding, heatsink, psu, bent pins, bios updates, disabling igpu, pcie gen settings etc.) short of trying the cpu somewhere else (found NOBODY with 1155). Now, since the bottom slot works flawlessly (southbridge-controlled, not the cpu), and I read that as unlikely as that sounds it can happen for one part of the cpu to malfunction while the rest lives on, I started suspecting it could be the pcie part of the cpu that died. The cpu doesn't overheat, never a bluescreen, it can even take overclock up to 4.5 ghz 24/7 without fail, supports any stress test with good scores including unigine 4.0 at max detail alongside my new 960 without crashing or breaking down (except somewhat lower score I guess due to x4 2.0). Before you ask: no, the problem appeared long before I even started overclocking the cpu.
Two final things:
1.I might have forgotten "take out the ram and let the speaker beep, if it doesn't then it doesn't post and it's the motherboard". I had an old speaker but it's broken now
2. When I put two gpu in the two slots (video hooked to secondary pcie or even igpu) the system boots fine, and this is what I have with card in and out (albeit in doesn't show in any software)


(for those who can't see pictures: x0 with card in, not present with card out)