@cflush: you have to follow up previous posts in this thread. The first thing after you fresh install Ubuntu is update the system and the dependencies. That step helps your system update what it needs for many new drivers and apps.
## Update the system sudo apt-get update && sudo apt-get -y dist-upgrade
# All the dependencies for Cuda & ccminer (I think) sudo apt-get -y install gcc g++ build-essential automake linux-headers-$(uname -r) git gawk libcurl4-openssl-dev libjansson-dev xorg libc++-dev libgmp-dev python-dev
@Phantom Success!!! Getting 19+ mH/s with my two 750 Ti s in ethminer!
A big thanks to you and Genoil. I have two more 750Ti s in another rig and plan to set that up tomorrow, and then I plan to do some mining for Genoil. I would appreciate it if you would reconsider and give me your ethpool.org address so I could mine for you. It is the least I can do.
p.s. Is there a preferred way to stop the miner, or do I just close out the terminal?
I got my ethminer compiled to 100% and then I had to run off to work. I only had enough time to start the miner only to get an error. Didn't have time to post. So I run ethminer and it tells me that -U is an invalid argument.
@TheFiendishOne : I saw that error when I tried to copy and run Ethminer on another rig. Base on my very little knowledge about Ubuntu, I guess that Ethminer doesn't run as root. So, it cannot access CUDA library. Make sure that you change to root before run ethminer.
@Genoil: could you please remove "-Werror" by default as cudaminer branch. It will make a setup guide a lot simple. I did update my guide, but leaved it in my PC. I will upload it when I wake up. OMG, 5am already, I have to get some sleep before my wife wakes up.
@cflush Thanks that worked! I actually tried that after it failed as I remembered in some windows tutorials it showed the -U at the end, and it still failed. After some observing I noticed the (-) was slightly different from when I copied it from word and it started telling me that -F was an invalid argument. I backspaced the dash and typed it in after reading your post and it worked so there must have been a double dash the first time.
Thanks @Phantom for all the help, there is some good info on this thread now.
Most importantly thanks to the one who made it all possible @Genoil
Running at 9 MH/s which is 3x what I was getting on Windows (3 MH/s)
What's a good tip?
*EDIT*
Ran the overclock and added -t 1 --cuda-turbo to the end and am now at 9.3 MH/s. Let the eth poor in. Now I want a 2nd card.
Hi guys, is it possible to run miner on GeForce GTX650? I have this old server sitting in the corner doing nothing, I'd like to run mining on it just for fun. So far managed to assemble cuda miner on ubuntu 14.04, but when I run `./ethminer -U -M` it says:
CUDA device GeForce GTX 650 has insufficient GPU memory.1070268416 bytes of memory found < 1423739904 bytes of memory required No CUDA device with sufficient memory was found. Can't CUDA mine. Remove the -U argument
As far as i can see, there's no workaround, am I correct?
@alexey_petrenko Does that 650 have 1 or 2 gb of RAM? From what I have read, cards need at least 2 gbs RAM for ethminer. Maybe consider a used 750ti for around $100. Good bang forbuck and low power consumption. Cflush
Wouldn't @alexey_petrenko be able to GPU mine using opencl with his 650 instead of using CUDA? I assume the hashrate may not be worth it, but I was just wondering.
Wouldn't @alexey_petrenko be able to GPU mine using opencl with his 650 instead of using CUDA? I assume the hashrate may not be worth it, but I was just wondering.
I assume OpenCL miner has the same memory requirements?
@kenshirothefist I believe you might be using the wrong version which is why you'd get 2 MH/s instead of 3 MH/s using windows. The other problem is that you're using windows. it's best to use Ubuntu Linux (I just made the switch) and it was well worth it.
@kenshirothefist: Nothing's wrong here. This issue has mentioned before. ATM, GTX750ti does not run well in Windows. GTX 9xx series will run good in Windows, but hashrate will be slower than in Ubuntu.
Whew- just finished installing linux / cuda ethminer on my second rig.. and I can personally vouch that it would not compile with 4 gb of RAM. Gave me errors and then locked up my system tighter than Dick's hatband. I thought since I had swap enabled it would chug though it (not). Had to grab 8gb kit out of my server and recompile..and it worked.
Thanks to contributions from the very helpful @RoBiK, we now have solid CUDA stream synchronization, flexible scheduling options and the removal of a bug that could potentially lead to a lower submitted share rate on pools. This means the --cuda-turbo flag is gone and replaced by --cuda-schedule:
--cuda-schedule Set the schedule mode for CUDA threads waiting for CUDA devices to finish work. Default is auto. Possible values are: auto - Uses a heuristic based on the number of active CUDA contexts in the process C and the number of logical processors in the system P. If C > P, then yield else spin. spin - Instruct CUDA to actively spin when waiting for results from the device. yield - Instruct CUDA to yield its thread when waiting for results from the device. sync - Instruct CUDA to block the CPU thread on a synchronization primitive when waiting for the results from the device.
The mode that is working best for you depends on the number of CPU cores and number of installed GPUs, so please try it out and see if it produces better results for you.
all in cudaminer-frontier (default) branch, Win64 binaries attached.
--edit--
binary removed. rolled back to version with --cuda-turbo. This commit was not very nice to CPU.
Comments
sudo apt-get install libglew-dev libcheese7 libcheese-gtk23 libclutter-gst-2.0-0 libcogl15 libclutter-gtk-1.0-0 libclutter-1.0-0
I got the same error before. Then, I took note this command in my note, but forgot to add in the guide.
## Update the system
sudo apt-get update && sudo apt-get -y dist-upgrade
# All the dependencies for Cuda & ccminer (I think)
sudo apt-get -y install gcc g++ build-essential automake linux-headers-$(uname -r) git gawk libcurl4-openssl-dev libjansson-dev xorg libc++-dev libgmp-dev python-dev
Source: https://gist.github.com/zcshiner/4b32980792d367222304
Thank you! You guys are awesome.
can anyone explain this? at this point I can't install cuda because libgbm1 wont upgrade.
If so, please advise. Thank u.
Cflush
## Update the system
sudo apt-get update && sudo apt-get -y dist-upgrade
# All the dependencies for Cuda & ccminer (I think)
sudo apt-get -y install gcc g++ build-essential automake linux-headers-$(uname -r) git gawk libcurl4-openssl-dev libjansson-dev xorg libc++-dev libgmp-dev python-dev
Source: https://gist.github.com/zcshiner/4b32980792d367222304
@TheFiendishOne Did you get your miners up and running yet??
Getting 19+ mH/s with my two 750 Ti s in ethminer!
A big thanks to you and Genoil. I have two more 750Ti s in another rig and plan to set that up tomorrow, and then I plan to do some mining for Genoil. I would appreciate it if you would reconsider and give me your ethpool.org address so I could mine for you. It is the least I can do.
p.s. Is there a preferred way to stop the miner, or do I just close out the terminal?
Regards, Cflush
I got my ethminer compiled to 100% and then I had to run off to work. I only had enough time to start the miner only to get an error. Didn't have time to post. So I run ethminer and it tells me that -U is an invalid argument.
[email protected]dioXPS-Ubuntu:~/cpp-ethereum/ethminer# ./ethminer –U –F http://xxx
Invalid argument: –U
If you have 4 GTX750ti, why don't you put them all in one rig. It saves you about 80-100W/hour.
I'm kind of lost on which parts of the setup guide I should now copy and paste together..
./ethminer -F http://xxx -U --cuda devices 0 1...
Hope that helps
@Genoil Those 750Tis are PNY overclocked version, factory set at 1202 mhz (1281 boost clock), mem 6008 mhz effective
-cflush
Thanks @Phantom for all the help, there is some good info on this thread now.
Most importantly thanks to the one who made it all possible @Genoil
Running at 9 MH/s which is 3x what I was getting on Windows (3 MH/s)
What's a good tip?
*EDIT*
Ran the overclock and added -t 1 --cuda-turbo to the end and am now at 9.3 MH/s. Let the eth poor in. Now I want a 2nd card.
I have this old server sitting in the corner doing nothing, I'd like to run mining on it just for fun.
So far managed to assemble cuda miner on ubuntu 14.04, but when I run `./ethminer -U -M` it says:
CUDA device GeForce GTX 650 has insufficient GPU memory.1070268416 bytes of memory found < 1423739904 bytes of memory required
No CUDA device with sufficient memory was found. Can't CUDA mine. Remove the -U argument
As far as i can see, there's no workaround, am I correct?
Cflush
@alexey_petrenko You're probably right, I over looked that.
p.s.: is there any way to check the hashrate of ethminer via API or any similar way? It's annoying to watch ethminer output to see the hashrate ...
Anyway, again a big thanks to Genoil and Phantom.
mining for Genoil in 3..2..1
./ethminer -F http://us1.ethpool.org/miner/0xeb9310b185455f863f526dab3d245809f6854b4d.worker10/18 -U --cuda-devices 0 1
deleteddeletedJust to report back, I am getting 49MH/sec as a pretty stable average on 6 750Ti's.
Testing system specs:
GIGABYTE GA-990FXA-UD3
AMD FX 6300
4Gb Memory
32Gb SSD
Corsair CX750M
Windows 7 64 bit Ultimate
Ethminer-Opencl-0.9.41
1. Sapphire HD 7950 Dual-X Boots
- Stock clock: 22.75Mh/s ~ 190W
- Overclock: Core 1060/ Mem 1450/ Power Limit 120%
+ .bat file: ethminer -G -F http://xxx
24.2Mh/s ~ 209W
+ .bat file: ethminer -G -F http://xxx --cl-local-work 128
25.2Mh/s ~ 219W
+ .bat file: ethminer -G -F http://xxx --cl-global-work 16384
26.4Mh/s ~ 224W
I got driver crash when I added both --cl-global-work 16384 and --cl-local-work 128.
2. Sapphire R9 290 TriX OC
- Stock clock: 23.42Mh/s ~ 241W
- Overclock: Core 1180/ Mem 1425/ Power Limit 120%
+ .bat file: ethminer -G -F http://xxx
27.81Mh/s ~ 273W
+ .bat file: ethminer -G -F http://xxx --cl-local-work 128
29.43Mh/s ~ 286W
+ .bat file: ethminer -G -F http://xxx --cl-global-work 8192 --cl-local-work 128
31.5Mh/s ~ 291W
I got driver crash when I added --cl-global-work 16384 on R9-290. So used 8192 instead, and add --cl-local-work 128.
Finally, I run both cards with --cl-global-work 8192 --cl-local-work 128. Stable at 58 Mh/s ~ 640W
Thanks to contributions from the very helpful @RoBiK, we now have solid CUDA stream synchronization, flexible scheduling options and the removal of a bug that could potentially lead to a lower submitted share rate on pools. This means the --cuda-turbo flag is gone and replaced by --cuda-schedule:
--cuda-schedule Set the schedule mode for CUDA threads waiting for CUDA devices to finish work. Default is auto. Possible values are:
auto - Uses a heuristic based on the number of active CUDA contexts in the process C and the number of logical processors in the system P. If C > P, then yield else spin.
spin - Instruct CUDA to actively spin when waiting for results from the device.
yield - Instruct CUDA to yield its thread when waiting for results from the device.
sync - Instruct CUDA to block the CPU thread on a synchronization primitive when waiting for the results from the device.
The mode that is working best for you depends on the number of CPU cores and number of installed GPUs, so please try it out and see if it produces better results for you.
all in cudaminer-frontier (default) branch, Win64 binaries attached.
--edit--
binary removed. rolled back to version with --cuda-turbo. This commit was not very nice to CPU.