I can resolve this problem just by set the power limit around 10-20%
I tried that and it didn't work for me. The easiest thing is just to close or minimize monitoring software.
Topic: Consolidated Litecoin Mining Guide for 5xxx, 6xxx, and 7xxx GPUs - page 12. (Read 624202 times)
April 15, 2013, 11:46:19 AM
I tried that and it didn't work for me. The easiest thing is just to close or minimize monitoring software.
April 15, 2013, 09:31:17 AM
i can resolve this problem just by set the power limit around 10-20%
what power limit? can you be more specific?
April 14, 2013, 11:06:15 PM
I'm using GUIminer scrypt to mine with CGminer and I just got a 7950. On 5000 series cards GPU usage was always steady while mining but my 7950 jumps up and down all the time. Any idea what could be causing this? The big drop is me stopping mining for a few seconds. It jumps from around 64% up to 99% and back.
https://i.imgur.com/x6aNHJQ.jpg
https://i.imgur.com/x6aNHJQ.jpg
I am also getting this one one of my 7950s.. Used to get it on two.. dont know what I did to reduce it by 1 now..
i can resolve this problem just by set the power limit around 10-20%
April 14, 2013, 09:59:02 PM
I have free electricity atm, so the voltage etc dosent really matter for me.
Should i still go for the 7950? Kinda had my mind set on that 7970
Should i still go for the 7950? Kinda had my mind set on that 7970
your temps will drop like a rock if you get that voltage to the proper level.. typically i have it at 0.05v above my mhz level. its WAY better for your cards, you'll reduce how fast you heat up the room and it'll be quieter with the fans slowing down some
for example: 1.0ghz = 1050 (1.05) voltage in afterburner
also.. the 7970 is about 5-8% faster mining vs a good 7950.. but 33% more money (400 vs 300 usd in the states)
April 14, 2013, 09:44:33 PM
I'm using GUIminer scrypt to mine with CGminer and I just got a 7950. On 5000 series cards GPU usage was always steady while mining but my 7950 jumps up and down all the time. Any idea what could be causing this? The big drop is me stopping mining for a few seconds. It jumps from around 64% up to 99% and back.
https://i.imgur.com/x6aNHJQ.jpg
https://i.imgur.com/x6aNHJQ.jpg
I am also getting this one one of my 7950s.. Used to get it on two.. dont know what I did to reduce it by 1 now..
I used to have really bad fluctuation with all four of my 7950's and I noticed it went away after I closed GPU-Z and all of my other monitoring software. I don't know if it is a related issue to what you are having but maybe this will help.
My 4x7870 LE do that if I have the intensity set at 13, but not 20. Minimizing GPU-Z and Afterburner solve the issue for me as well.
April 14, 2013, 09:29:44 PM
I'm using GUIminer scrypt to mine with CGminer and I just got a 7950. On 5000 series cards GPU usage was always steady while mining but my 7950 jumps up and down all the time. Any idea what could be causing this? The big drop is me stopping mining for a few seconds. It jumps from around 64% up to 99% and back.
I am also getting this one one of my 7950s.. Used to get it on two.. dont know what I did to reduce it by 1 now..
I used to have really bad fluctuation with all four of my 7950's and I noticed it went away after I closed GPU-Z and all of my other monitoring software. I don't know if it is a related issue to what you are having but maybe this will help.
April 14, 2013, 08:25:37 PM
I'm using GUIminer scrypt to mine with CGminer and I just got a 7950. On 5000 series cards GPU usage was always steady while mining but my 7950 jumps up and down all the time. Any idea what could be causing this? The big drop is me stopping mining for a few seconds. It jumps from around 64% up to 99% and back.
I am also getting this one one of my 7950s.. Used to get it on two.. dont know what I did to reduce it by 1 now..
April 14, 2013, 08:00:22 PM
Having a weird issue with one of my 5970s. Was working fine for weeks, hashing both cores, now one of the cores just crashes the driver.
Ive tried pretty much every setting in afterburner, going up and down with voltage, core speed, mem speed. But when it was hashing normal values worked fine, so I feel like its not my card settings. I had to reset all my systems at one point, since then the core hasnt worked.
GPUz still sees the core and it reports everything as fine. What other types of diagnostics/solutions/troubleshooting can you recommend?
Ive tried pretty much every setting in afterburner, going up and down with voltage, core speed, mem speed. But when it was hashing normal values worked fine, so I feel like its not my card settings. I had to reset all my systems at one point, since then the core hasnt worked.
GPUz still sees the core and it reports everything as fine. What other types of diagnostics/solutions/troubleshooting can you recommend?
April 14, 2013, 06:57:38 AM
When you use the --auto-fan and --auto-gpu features does your fan spike in speed and then fall back or run steady? Mine will run low then spike to bring temps down then run low again. Although annoying with the spikes all the time, I was wondering what was better on the fans. Seems to me that constantly ramping up fan speed would burn out the motor faster than a constant speed.
Another question for Taco, how do you make guiminer-scypt run multiple cards, do you have to start multiple miner instances? (cgminer)
Great software bro, please keep the fork going, will send LTC once the rigs produce me some profits!!
W0lfm4n
Another question for Taco, how do you make guiminer-scypt run multiple cards, do you have to start multiple miner instances? (cgminer)
Great software bro, please keep the fork going, will send LTC once the rigs produce me some profits!!
W0lfm4n
April 14, 2013, 12:00:04 AM
I'm uploading my reaper kernel for my 7950s because I'm wondering if some of the troubles people are having are related to poor compiling. I noticed that if I recompile my kernel now I no longer get 1.8 MH/s but get more like 1.6 and I have no idea why that is.
http://multiupload.nl/EYQOQQVFXK
Settings are below
litecoin.conf
http://multiupload.nl/EYQOQQVFXK
Settings are below
litecoin.conf
Code:
host yourpool.com ### Do not add http://!!
port 8344
user username
pass password
protocol litecoin
worksize 256
vectors 1
aggression 20
threads_per_gpu 1
sharethreads 32
lookup_gap 2
gpu_thread_concurrency 21712
port 8344
user username
pass password
protocol litecoin
worksize 256
vectors 1
aggression 20
threads_per_gpu 1
sharethreads 32
lookup_gap 2
gpu_thread_concurrency 21712
April 13, 2013, 11:53:20 PM
What are your overclock settings and temps? Any other cards/miners on this machines, and do they exhibit this problem?
April 13, 2013, 09:02:20 PM
I'm using GUIminer scrypt to mine with CGminer and I just got a 7950. On 5000 series cards GPU usage was always steady while mining but my 7950 jumps up and down all the time. Any idea what could be causing this? The big drop is me stopping mining for a few seconds. It jumps from around 64% up to 99% and back.
April 13, 2013, 05:37:57 PM
Glad to read you got it sorted. 
.
. 
April 13, 2013, 03:54:25 PM
Yes, because you need different settings for each card. You can't use the same settings for both cards. Need to follow this guide, get the settings for the 7850, and the settings for the 5870.
Check this out, see what other people use as for the settings.
https://github.com/litecoin-project/litecoin/wiki/Mining-hardware-comparison#wiki-ATI_AMD_GPUs
I know, I've already tried all sorts of thread concurrencies and intensities. I've tried every combination of settings there is, and I either get 30-50khash or 350khash with 50% HW errors. I've also tried tweaking clock ratios to no avail.Check this out, see what other people use as for the settings.
https://github.com/litecoin-project/litecoin/wiki/Mining-hardware-comparison#wiki-ATI_AMD_GPUs
Edit: I... somehow fixed the issue. Didn't change any of my settings. Uninstalled all AMD drivers and programs, searched for any remaining AMD and OpenCL dll files (which there were a few of in system32) and deleted them too. Rebooted the computer, installed 12.8 drivers and the 2.5 APP SDK again. Rebooted again. CGMiner is reporting 265kh/s with no HW errors and a WU of 250, so I think things are okay now. Looks like it was a driver issue.
April 13, 2013, 03:27:23 PM
Yes, because you need different settings for each card. You can't use the same settings for both cards. Need to follow this guide, get the settings for the 7850, and the settings for the 5870.
Check this out, see what other people use as for the settings.
https://github.com/litecoin-project/litecoin/wiki/Mining-hardware-comparison#wiki-ATI_AMD_GPUs
Check this out, see what other people use as for the settings.
https://github.com/litecoin-project/litecoin/wiki/Mining-hardware-comparison#wiki-ATI_AMD_GPUs
April 13, 2013, 01:45:27 PM
It would help yourself a ton if you used a cgminer.conf file, this way you can separate you card settings, rather having your 5870 setting being used for your 7850 as well. cgminer comes with an example.conf, rename it to cgminer.conf, and set the settings inside.
Don't use these, but here is the setting for my two same cards, but for yours, you would have settings to the left of the apostrophe for card 0 and to the right of the apostrophe would be for card 1. Example: Card 0 = 5870 Card 1 = 7850.
Once you have cgminer.conf configured, just double click cgminer.exe, and it will take off using the cgminer.conf settings.
As an example:
cgminer.conf left of the apostrophe for card 0 and to the right of the apostrophe would be for card 1
Thanks. I copied my settings into the .conf file and started cgminer again. I haven't changed any settings and now I'm getting 50khash on the card with really slow desktop responsiveness. What's even worse is that if I go back to the original shortcut I was using, I get the same exact thing.Don't use these, but here is the setting for my two same cards, but for yours, you would have settings to the left of the apostrophe for card 0 and to the right of the apostrophe would be for card 1. Example: Card 0 = 5870 Card 1 = 7850.
Once you have cgminer.conf configured, just double click cgminer.exe, and it will take off using the cgminer.conf settings.
As an example:
cgminer.conf left of the apostrophe for card 0 and to the right of the apostrophe would be for card 1
This is so frustrating.
April 13, 2013, 07:03:48 AM
Placing a order for my first 7970 on monday:)
SAPPHIRE RADEON HD7970 DUAL-X 3GB, thats the one i should get, right? Been reeding in this thread, seems like khash/s is pretty poor on those factory overclocked Ghz editions.
SAPPHIRE RADEON HD7970 DUAL-X 3GB, thats the one i should get, right? Been reeding in this thread, seems like khash/s is pretty poor on those factory overclocked Ghz editions.
If you want a MUCH better value, get this:
http://www.newegg.com/Product/Product.aspx?Item=N82E16814125414
I have 2 of those.. and 1 of the 7970 sapphire dual-x.
the 7970 gets 690khash..
each 7950 ($100 cheaper) gets 670+khash
I agree, everything I have read suggests the Gigabytes (7950 and 7970) are great LTC miners. Only problem is they are voltage locked so if you live in a country with high electricity prices they might not be the most efficient.
I have free electricity atm, so the voltage etc dosent really matter for me.
Should i still go for the 7950? Kinda had my mind set on that 7970
April 13, 2013, 04:15:39 AM
I'm having a lot of trouble getting the 7850 I purchased (1GB version) to mine efficiently.
GPU0 is the 7850, GPU1 is a 5870.
As you can see the 5870 is running perfectly fine, returning good results and achieving right around 400kh/s.
The 7850, on the other hand, is having a lot of problems. There are more hardware errors than returned shares. The hashrate is low, but I haven't tried optimizing it yet, so I'm not too worried about that (I did have it at around ~350kh/s earlier).
I'm running Catalyst 12.8 (13.1 gave me ridiculous artifacts) with APP runtime 2.5.
The card is at stock clocks (860 core/1200 mem) at 1.210V.
These are the settings for this card:
--scrypt -u beremat.cypress -p xxx -o stratum+tcp://coinotron.com:3334 -I 14 --lookup-gap 2 --thread-concurrency 6400 --shaders 1600
It doesn't crash while gaming, FurMark, or even mining, so what's the deal with the hardware errors? :/
GPU0 is the 7850, GPU1 is a 5870.
As you can see the 5870 is running perfectly fine, returning good results and achieving right around 400kh/s.
The 7850, on the other hand, is having a lot of problems. There are more hardware errors than returned shares. The hashrate is low, but I haven't tried optimizing it yet, so I'm not too worried about that (I did have it at around ~350kh/s earlier).
I'm running Catalyst 12.8 (13.1 gave me ridiculous artifacts) with APP runtime 2.5.
The card is at stock clocks (860 core/1200 mem) at 1.210V.
These are the settings for this card:
--scrypt -u beremat.cypress -p xxx -o stratum+tcp://coinotron.com:3334 -I 14 --lookup-gap 2 --thread-concurrency 6400 --shaders 1600
It doesn't crash while gaming, FurMark, or even mining, so what's the deal with the hardware errors? :/
It would help yourself a ton if you used a cgminer.conf file, this way you can separate you card settings, rather having your 5870 setting being used for your 7850 as well. cgminer comes with an example.conf, rename it to cgminer.conf, and set the settings inside.
Don't use these, but here is the setting for my two same cards, but for yours, you would have settings to the left of the apostrophe for card 0 and to the right of the apostrophe would be for card 1. Example: Card 0 = 5870 Card 1 = 7850.
Once you have cgminer.conf configured, just double click cgminer.exe, and it will take off using the cgminer.conf settings.
As an example:
cgminer.conf left of the apostrophe for card 0 and to the right of the apostrophe would be for card 1
Code:
{
"pools" : [
{
"url" : "pool1:port",
"user" : "username.1",
"pass" : "password"
},
{
"url" : "pool2:port",
"user" : "username.1",
"pass" : "password"
}
]
,
"gpu-platform" : "1",
"intensity" : "19,19",
"thread-concurrency" : "5760,5760",
"vectors" : "1,1",
"worksize" : "256,256",
"kernel" : "scrypt,scrypt",
"gpu-fan" : "0-75,0-75",
"gpu-vddc" : "1.088,1.088",
"gpu-engine" : "0-944,0-944",
"gpu-memclock" : "1180,1180",
"gpu-memdiff" : "0,0",
"gpu-powertune" : "0,0",
"temp-cutoff" : "0,0",
"temp-overheat" : "102,102",
"temp-target" : "100,100",
"api-port" : "4028",
"expiry" : "120",
"gpu-dyninterval" : "7",
"gpu-threads" : "1",
"log" : "5",
"queue" : "1",
"retry-pause" : "5",
"scan-time" : "60",
"scrypt" : true,
"temp-hysteresis" : "3",
"shares" : "0",
"api-listen" : true,
"api-network" : true,
"api-port" : "4028",
"kernel-path" : "/usr/local/bin"
}
"pools" : [
{
"url" : "pool1:port",
"user" : "username.1",
"pass" : "password"
},
{
"url" : "pool2:port",
"user" : "username.1",
"pass" : "password"
}
]
,
"gpu-platform" : "1",
"intensity" : "19,19",
"thread-concurrency" : "5760,5760",
"vectors" : "1,1",
"worksize" : "256,256",
"kernel" : "scrypt,scrypt",
"gpu-fan" : "0-75,0-75",
"gpu-vddc" : "1.088,1.088",
"gpu-engine" : "0-944,0-944",
"gpu-memclock" : "1180,1180",
"gpu-memdiff" : "0,0",
"gpu-powertune" : "0,0",
"temp-cutoff" : "0,0",
"temp-overheat" : "102,102",
"temp-target" : "100,100",
"api-port" : "4028",
"expiry" : "120",
"gpu-dyninterval" : "7",
"gpu-threads" : "1",
"log" : "5",
"queue" : "1",
"retry-pause" : "5",
"scan-time" : "60",
"scrypt" : true,
"temp-hysteresis" : "3",
"shares" : "0",
"api-listen" : true,
"api-network" : true,
"api-port" : "4028",
"kernel-path" : "/usr/local/bin"
}
April 12, 2013, 11:44:50 PM
I'm having a lot of trouble getting the 7850 I purchased (1GB version) to mine efficiently.
GPU0 is the 7850, GPU1 is a 5870.
As you can see the 5870 is running perfectly fine, returning good results and achieving right around 400kh/s.
The 7850, on the other hand, is having a lot of problems. There are more hardware errors than returned shares. The hashrate is low, but I haven't tried optimizing it yet, so I'm not too worried about that (I did have it at around ~350kh/s earlier).
I'm running Catalyst 12.8 (13.1 gave me ridiculous artifacts) with APP runtime 2.5.
The card is at stock clocks (860 core/1200 mem) at 1.210V.
These are the settings for this card:
--scrypt -u beremat.cypress -p xxx -o stratum+tcp://coinotron.com:3334 -I 14 --lookup-gap 2 --thread-concurrency 6400 --shaders 1600
It doesn't crash while gaming, FurMark, or even mining, so what's the deal with the hardware errors? :/
GPU0 is the 7850, GPU1 is a 5870.
As you can see the 5870 is running perfectly fine, returning good results and achieving right around 400kh/s.
The 7850, on the other hand, is having a lot of problems. There are more hardware errors than returned shares. The hashrate is low, but I haven't tried optimizing it yet, so I'm not too worried about that (I did have it at around ~350kh/s earlier).
I'm running Catalyst 12.8 (13.1 gave me ridiculous artifacts) with APP runtime 2.5.
The card is at stock clocks (860 core/1200 mem) at 1.210V.
These are the settings for this card:
--scrypt -u beremat.cypress -p xxx -o stratum+tcp://coinotron.com:3334 -I 14 --lookup-gap 2 --thread-concurrency 6400 --shaders 1600
It doesn't crash while gaming, FurMark, or even mining, so what's the deal with the hardware errors? :/
April 12, 2013, 11:07:12 PM
Do the 1GB, 2GB, and 3GB models affect KH/s? I'm not sure if I should spend more money on the cards with more RAM unless it adds to performance.
for the most part no.
(based on my memory)
1gb = 7700 series
2gb = 7800 series
3gb = 7900 series
for each of those cards, you won't ever need more ram than is required for optimal settings.
Memory bandwidth matters though, however that seems to scale decently with # of cores
7770 = 200khash
7870 400 khash
7950/7970 = 650-700 khash
Jump to: