Two 7970s on one motherboard; one fine, the other hashing at exactly 1/8 speed
-
Good people of the forum, I need your help. I have three Sapphire 7970s [url=http://www.sapphiretech.com/presentation/product/?cid=1&gid=3&sgid=1157&pid=1481&psn=&lid=1&leg=0]http://www.sapphiretech.com/presentation/product/?cid=1&gid=3&sgid=1157&pid=1481&psn=&lid=1&leg=0[/url], which had all been fine mining at between 660-670kh/s for the last couple of months. One is on an
Asus P6T Deluxe Version 2
Core i7 920
12GB RAM
450W bequiet! PSUrunning with
[code]cgminer --scrypt -o http://127.0.0.1:1111 -u 1 -p 1 -o http://p2pool.org:9377 -u XXXXXXXXXXXXXXXXXXXXXXXXXXXXXX -p password --auto-fan --gpu-engine 1008 --gpu-memclock 1675 --thread-concurrency 8192 -I 13 -g 2 -w 256
[/code]the other two are on an
Asus P8B WS
Xeon E3-1230
8GB RAM
1000W Zalman PSUusing
[code]cgminer --scrypt -o http://127.0.0.1:1111 -u 1 -p 1 -o http://p2pool.org:9377 -u XXXXXXXXXXXXXXXXXXXXXXXXXXXXXX -p password --auto-fan --gpu-engine 1008,1008 --gpu-memclock 1675,1675 --thread-concurrency 8192,8192 -I 13,13 -g 2 -w 256,256[/code]
Recently, though, GPU1 on the Asus P8B WS started mining at exactly one-eighth speed, which I initially thought could have been some kind of hardware problem, but I then swapped the cards out, tried every combination, and found that any one of my GPU cards on the P8B WS will run a 1/8th speed, unless there are two cards installed and it is the second card; also, all three are fine on their own in the other machine. I tried the first cgminer configuration above with the machine with two GPUs, but had the same results. I am using Windows 7 x64 and have had the same results with both beta and stable ATi drivers. Since the issue started, I tried to configure the machine with Xubuntu, then Ubuntu, then Debian, then gave up and reinstalled Windows. I have also tried each of the GPUs in three of the four accessible PCI-e 16x slots (including with a ribbon cable) on the P8B WS, but this didn’t make and difference. I tested with a SHA-256 currency as well, and found that the card in this case mined at exactly 1/2 the speed it should have been (in case that helps).
Please let me know your thoughts - I’m not aware of having changed anything just before the issue started :(
-
So you are saying if you have one card in it hashes at 1/8th the speed and if you have two then the first runs full speed and the other does 1/8th?
-
Sounds like you may have narrowed it down to an issue w/ the MB.
-
[quote name=“sl1982” post=“9455” timestamp=“1370383558”]
So you are saying if you have one card in it hashes at 1/8th the speed and if you have two then the first runs full speed and the other does 1/8th?
[/quote]That’s right - and it’s always this way round:
[code]GPU 0: | 670.7K/670.3Kh/s
GPU 1: | 84.21K/83.72Kh/s[/code][quote author=zythen link=topic=1208.msg9457#msg9457 date=1370383630]
Sounds like you may have narrowed it down to an issue w/ the MB.
[/quote]Yeah, it’s looking that way. I suppose this screenshot from GPU-Z isn’t looking too good, either, come to think of it.
[IMG]http://gpuz.techpowerup.com/13/06/04/2c9.png[/img] [IMG]http://gpuz.techpowerup.com/13/06/04/rb.png[/img]
You can see that the current is much higher in the second screenshot; I suppose now it’s just a case of finding a smoking gun (outside of just mining) before I can get ASUS to RMA the motherboard.
-
can you try erasing this -g 2.
btw, don’t forget to put this --failover-only, to enable failover-only.
like these
-o http://127.0.0.1:1111 -u 1 -p 1 --failover-only -o http://p2pool.org:9377 -u XXXXXXXXXXXXXXXXXXXXXXXXXXXXXX -p password -
How many monitors are running on the machine with this problem?
-
[quote name=“RIPPEDDRAGON” post=“9756” timestamp=“1370453772”]
How many monitors are running on the machine with this problem?
[/quote]That is a good question, could it perhaps be a card without a monitor or dummy plug?
-
[quote name=“silentgaps” post=“9479” timestamp=“1370391676”]
can you try erasing this -g 2.btw, don’t forget to put this --failover-only, to enable failover-only.
like these
-o http://127.0.0.1:1111 -u 1 -p 1 --failover-only -o http://p2pool.org:9377 -u XXXXXXXXXXXXXXXXXXXXXXXXXXXXXX -p password
[/quote]I removed “-g 2”, and this decreased the speed of GPU0 below the normal speed, but increased the speed of GPU1:
[code]GPU 0: | 508.5K/507.9Kh/s
GPU 1: | 91.87K/93.08Kh/s[/code][quote author=RIPPEDDRAGON link=topic=1208.msg9756#msg9756 date=1370453772]
How many monitors are running on the machine with this problem?
[/quote]I have been running this machine headless via RDP; perhaps now is the time for me to try some dummy plugs, in that case. I gather one per card is what is generally recommended. I will report back.
-
Time for a visit to radio shack. :) Have you tried plugging a spare monitor into that card for giggles?
-
[quote name=“F8ECBB6F02F7C” post=“9882” timestamp=“1370477565”]
[quote author=silentgaps link=topic=1208.msg9479#msg9479 date=1370391676]
can you try erasing this -g 2.btw, don’t forget to put this --failover-only, to enable failover-only.
like these
-o http://127.0.0.1:1111 -u 1 -p 1 --failover-only -o http://p2pool.org:9377 -u XXXXXXXXXXXXXXXXXXXXXXXXXXXXXX -p password
[/quote]I removed “-g 2”, and this decreased the speed of GPU0 below the normal speed, but increased the speed of GPU1:
[code]GPU 0: | 508.5K/507.9Kh/s
GPU 1: | 91.87K/93.08Kh/s[/code][quote author=RIPPEDDRAGON link=topic=1208.msg9756#msg9756 date=1370453772]
How many monitors are running on the machine with this problem?
[/quote]I have been running this machine headless via RDP; perhaps now is the time for me to try some dummy plugs, in that case. I gather one per card is what is generally recommended. I will report back.
[/quote]
Eeeep! RDP is nasty for mining. Grab tightVNC instead, should save the dummy plugs and you’ll get all the proper stats in cgMiner. -
im sure that is doing great