R9 stability issues

[miztaziggy]

Can anyone help?

I cannot get my R9 290 cards stable at all. 
- I have tried running 1 card, 2 cards, 4 cards and 5 cards. All configurations the same.
- I have tried on different motherboards (all GA-Z77) and using different PCIE slots.
- I have tried different cards and updated BIOS on the cards and the mobo.
- I am using powered 1x risers and have tried different risers in different slots. I have also tried unpowered 16x risers.
- I am using Powercool X-Viper 950W PSU to power mobo plus 2 cards. When I run more cards I have a second 950W PSU powering too. I have the same PSUs powering 5 card 6950 and 6970 rigs without fault.

I run them at I 18, TC 21000 and standard 947/1250 clocks. The computer crashes within 10 minutes of starting CGminer.

If I run them at TC 8192 I can only go up to I 13 without HW errors but only get 500 KH. As soon as I increase the intensity I need to increase the TC to stop the HW errors. If I have the TC over 8192 I get the crashes.

If I reduce the clocks to 900 / 1200 I can reduce the crashes to approx 1 per hour or so. Hash rate approx 750 KH.

If I reduce the clocks to 890 / 1190 I can reduce the crashes to approx 1 ever few hours. Hash rate approx 724 KH.

So do I have a bad batch of cards? Hard to imagine all 10 cards are bad....but it seems so.

Is the motherboard causing issues? Can the motherboard affect how fast the GPU can be clocked?

Everything I have read seems to sound like it should be easy to run at 947/1250 and most people can run at 1000/1500 without problem. I haven't found anything yet after lots of searching that indicates anyone else has the same problem.

I have the Sapphire R9 290 cards btw.

[coin123123]

in tc 8192 i 13 you use g 2 ?

[KWH]

I would suspect your PSU.

[miztaziggy]

in tc 8192 i 13 you use g 2 ?

Yep, tried g 1, g 2, g 3 and g 4 but it doesn't make much difference to the hash rate or stability.

I would suspect your PSU.

So what? The PSU spiking causing crashes?

[freddyfarnsworth]

Since all the cards are known good issueless performers and P/S is above 750w SINGLE rail type.
You should have 30a 5v and much higher 12v available.

The only unknowns are MB's everyone has different.

Gonna say MB as I have found similar power issues using the old 790i triple sli boards (have 5 of them) evga, xfx and NV reference. all have the same random shutdown issue with three cards. (all 1gb GTX250's I have 12 of them)

My stable MB is a ASRock Extreme7 Gen3 bios v1.34 (sandybridge) 2550k @3.8ghz with three zotac GTX465 1g (early models) rebadged 470's. I use Corsair TX series 950w (hard to get) single rail and No Green in em... (have many)

These were game Dev machines for FO3 a very particular game only likes certain hardware, now being used for BitCoin till I get real hardware.

So focus on MB, and win7 has a update for kernal mode driver, helps with the poweroff thing a bit.
Turn off all "GREEN" settings, do not let anything turn off if idling.

[af_newbie]

Can anyone help?

I cannot get my R9 290 cards stable at all. 
- I have tried running 1 card, 2 cards, 4 cards and 5 cards. All configurations the same.
- I have tried on different motherboards (all GA-Z77) and using different PCIE slots.
- I have tried different cards and updated BIOS on the cards and the mobo.
- I am using powered 1x risers and have tried different risers in different slots. I have also tried unpowered 16x risers.
- I am using Powercool X-Viper 950W PSU to power mobo plus 2 cards. When I run more cards I have a second 950W PSU powering too. I have the same PSUs powering 5 card 6950 and 6970 rigs without fault.

I run them at I 18, TC 21000 and standard 947/1250 clocks. The computer crashes within 10 minutes of starting CGminer.

If I run them at TC 8192 I can only go up to I 13 without HW errors but only get 500 KH. As soon as I increase the intensity I need to increase the TC to stop the HW errors. If I have the TC over 8192 I get the crashes.

If I reduce the clocks to 900 / 1200 I can reduce the crashes to approx 1 per hour or so. Hash rate approx 750 KH.

If I reduce the clocks to 890 / 1190 I can reduce the crashes to approx 1 ever few hours. Hash rate approx 724 KH.

So do I have a bad batch of cards? Hard to imagine all 10 cards are bad....but it seems so.

Is the motherboard causing issues? Can the motherboard affect how fast the GPU can be clocked?

Everything I have read seems to sound like it should be easy to run at 947/1250 and most people can run at 1000/1500 without problem. I haven't found anything yet after lots of searching that indicates anyone else has the same problem.

I have the Sapphire R9 290 cards btw.

disable ADL in cgminer or use reaper.  cgminer fan/clocks control is broken.

Use one thread, run TC at bus width*64