Schleicher
|
|
January 08, 2014, 06:35:52 AM |
|
Downloded the latest nVidia drivers that were released today (332.21), and Cudaminer wont work now ("Unable to query CUDA driver version! Is an nVidia driver installed?"). Never had this problem before when updating drivers, so sorry if I've missed over the answer somewhere in here!
No problem on my computer. I've just updated the driver too.
|
|
|
|
Valnurat
|
|
January 08, 2014, 08:04:55 AM |
|
I see alot of different setup with for an exampel Lx15 ect.
How do I find a setup for FX 4600?
I'm not sure if it is related, but when I start cudaminer the applications stops as you can see. The error is: that Windows will shutdown the app and will inform me if there is a solution for it. I hope someone can help me? Really!! No one can help me?
|
|
|
|
bigjme
|
|
January 08, 2014, 08:09:13 AM |
|
It is setting config L0x0 which isnt a valid config. Run the benchmark, I find it easier to print screen the results then look through them for a valid response.
Failing that, have a google and try a random config from online
|
Owner of: cudamining.co.uk
|
|
|
Sarg338
Newbie
Offline
Activity: 8
Merit: 0
|
|
January 08, 2014, 08:31:31 AM |
|
Downloded the latest nVidia drivers that were released today (332.21), and Cudaminer wont work now ("Unable to query CUDA driver version! Is an nVidia driver installed?"). Never had this problem before when updating drivers, so sorry if I've missed over the answer somewhere in here!
No problem on my computer. I've just updated the driver too. I should have known this, and shouldn't have even posted without doing this first, but a computer restart fixed it, and it's mining away now.
|
|
|
|
CaptainBeck
|
|
January 08, 2014, 08:46:26 AM |
|
Downloded the latest nVidia drivers that were released today (332.21), and Cudaminer wont work now ("Unable to query CUDA driver version! Is an nVidia driver installed?"). Never had this problem before when updating drivers, so sorry if I've missed over the answer somewhere in here!
No problem on my computer. I've just updated the driver too. I should have known this, and shouldn't have even posted without doing this first, but a computer restart fixed it, and it's mining away now. Is it mining better or worse??? Because my machines are only for mining so if its not seeing an improvement i'm not updating them.
|
|
|
|
patoberli
Member
Offline
Activity: 106
Merit: 10
|
|
January 08, 2014, 10:42:46 AM Last edit: January 08, 2014, 11:45:56 AM by patoberli |
|
I was able to compile it in VS2012 Professional and actually run it on my puny Quadro 4000 (Fermi based) and tested it for scrypt-jane. I Could run it with -l F4x3 and received a fairly low 0.41 khash/s. Sadly interactive mode didn't seem to work, so I can't use it on this computer (I use it for normal work too).
Edit: For comparison, normal scrypt is running with around 83 khash/s.
|
YAC: YA86YiWSvWEGSSSerPTMy4kwndabRUNftf BTC: 16NqvkYbKMnonVEf7jHbuWURFsLeuTRidX LTC: LTKCoiDwqEjaRCoNXfFhDm9EeWbGWouZjE
|
|
|
cbuchner1 (OP)
|
|
January 08, 2014, 11:35:23 AM |
|
I was able to compile it in VS2012 Professional and actually run it on my puny Quadro 4000 (Fermi based) and tested it for scrypt-jane. I Could run it with -l F4x3 and received a fairly low 0.41 khash/s. Sadly interactive mode didn't seem to work, so I can't use it on this computer (I use it for normal work too).
The order in which I will enable interactive mode is Titan, Fermi. I also noticed terrible hash rates running Fermi scrypt-jane on Windows. Even on GPUs that otherwise work really well for scrypt hashing. Generally I get much better results on Linux. Likely Windows (the WDDM driver model) has a performance problem when allocating a lot of GPU memory... I haven't been able to try Fermi devices under Linux yet. Yesterday I tried to escape the texture cache size limit (2GB) by splitting up the data into two textures. Didn't work at reasonable speed unfortunately, so I dropped the idea. Christian
|
|
|
|
manofcolombia
Member
Offline
Activity: 84
Merit: 10
SizzleBits
|
|
January 08, 2014, 03:09:12 PM |
|
[/quote] Is it mining better or worse??? Because my machines are only for mining so if its not seeing an improvement i'm not updating them. [/quote] Same. Just waiting for a guinea pig to say whether or not todays drivers do anything for mining
|
|
|
|
bigjme
|
|
January 08, 2014, 03:11:35 PM |
|
My 780 is mining scrypt-jane the same with the new drivers. I havent tried normal scrypt though
|
Owner of: cudamining.co.uk
|
|
|
bathrobehero
Legendary
Offline
Activity: 2002
Merit: 1051
ICO? Not even once.
|
|
January 08, 2014, 03:27:43 PM |
|
What does having a scrypt-jane cryptocurrency having an Nfactor of 6 and N of 128 potentionally mean? I mean could cudaMiner be optimized for that for higher hashrates?
|
Not your keys, not your coins!
|
|
|
bigjme
|
|
January 08, 2014, 03:34:00 PM |
|
higher N factor means lower rewards for mining blocks and a higher memory requirement every time it increases. eventually it will get so high that gpus wont have enough memory to even hash once so the hashrate will drop very quickly. this is why cpu mining will carry on as desktop memory is cheap and can be replaced
personally I think scrypt-jane when the N factor goes up is going to force up memory prices when the cpu miners start to be the only ones able to mine. unless we can find a way to use our system memory for gpu crunching on scrypt-jane it will eventually be impossible to mine with a gpu
|
Owner of: cudamining.co.uk
|
|
|
manofcolombia
Member
Offline
Activity: 84
Merit: 10
SizzleBits
|
|
January 08, 2014, 03:53:52 PM |
|
Just patiently waiting for the update with scrypt-jane Wanna jump on ya coin quick. Its actually a bit more profitable than ltc right now if I mine ya and convert to btc= $$
|
|
|
|
bigjme
|
|
January 08, 2014, 03:59:47 PM |
|
It is a lot more for me. Although solo mining is un predictable. I could maybe mine 0.16 ltc in a day. I can get abiut 150 yacoins in a day which is just over 0.35 litecoins.
Then again I still have a system on multipool, mining doge the dieing dog still as tue most profitable one. And if thats the most profitable with its price over halving in a week. It shows how bad the rest of the alt coins are doing
edit: saying that. I got 150 coins in 7 hours solo mining its been 29 hours since then and I have added an extra 1khash/s and not got anything.
I just hope the damn config im using that causes the driver crash is actually doing something and not just wasting my time saying it is mining. As I have no way to tell if its even sending off anything
|
Owner of: cudamining.co.uk
|
|
|
cbuchner1 (OP)
|
|
January 08, 2014, 04:07:05 PM Last edit: January 08, 2014, 04:17:22 PM by cbuchner1 |
|
I have to strongly recommend to mine yacoin using Linux on Kepler devices (Compute 3.0 or 3.5). For Linux, github has always been the preferred way of getting the source code.
The Windows version (for some bizarre reason) does not achieve comparable performance on the same kernel launch configurations.
I will attempt to optimize scrypt-jane mining for Fermi later, but that may take a bit longer.
Solo-Mining is a lottery. Only when you average several days or weeks you get a reliable estimates of your expected income. And even that estimate may change, as people start jumping on yacoin with CUDA, driving up the difficulty. Today for example the dominant pool had a failure of its stratum port, losing more than half of its hash rate. That's more profit for us solo'ers, right there.
Christian
|
|
|
|
bigjme
|
|
January 08, 2014, 04:09:01 PM |
|
Can you explain to me how to get cuda 5.5 running? I mean I have ubuntu 13.10 but setting up cuda was so long winded I couldnt do it.
If you can explain that I will give it a go and let you know how things run. Because I just have a feeling my windows one is doing nothing right now
|
Owner of: cudamining.co.uk
|
|
|
cbuchner1 (OP)
|
|
January 08, 2014, 04:13:58 PM |
|
Can you explain to me how to get cuda 5.5 running? I mean I have ubuntu 13.10 but setting up cuda was so long winded I couldnt do it.
If you can explain that I will give it a go and let you know how things run. Because I just have a feeling my windows one is doing nothing right now
Typically you download a RPM or .deb package for the CUDA toolkit from nvidia, and if there isn't one for Ubuntu 13.10, there may be one for 13.04 or 12.10 or 12.04 at least. In my experience versions for older Ubuntu releases still run on more recent Ubuntu installations. Installation requires root privileges. If the driver shipping with the toolkit doesn't install easily (it requires installation from a text console, without the X server running and it needs linux kernel headers etc...), maybe try getting nvidia-331 with apt-get. There is a ppa repository called xorg-edgers (and another one called X-Swat) which may have even more recent nvidia driver versions. Good luck! Christian
|
|
|
|
manofcolombia
Member
Offline
Activity: 84
Merit: 10
SizzleBits
|
|
January 08, 2014, 04:17:38 PM |
|
I have to strongly recommend to mine yacoin using Linux on Kepler devices (Compute 3.0 or 3.5). For Linux, github has always been the preferred way of getting the source code.
The Windows version (for some bizarre reason) does not achieve comparable performance on the same kernel launch configurations.
I will attempt optimizing for Fermi once more, but that may take a bit longer.
Solo-Mining is a lottery. Only when you average several days or weeks you get a good, reliable estimates of your expected income. And even that may change, as people start jumping on yacoin with CUDA, driving up the difficulty. Today for example the dominant pool had a failure of its stratum port, losing more than half of its hash rate. That's more profit for us solo'ers, right there.
Christian
*sigh* thats annoying...my linux box (Home server) is an old Pentium 4 with PCI...
|
|
|
|
cbuchner1 (OP)
|
|
January 08, 2014, 04:18:39 PM |
|
higher N factor means lower rewards for mining blocks and a higher memory requirement every time it increases. eventually it will get so high that gpus wont have enough memory to even hash once so the hashrate will drop very quickly. this is why cpu mining will carry on as desktop memory is cheap and can be replaced
personally I think scrypt-jane when the N factor goes up is going to force up memory prices when the cpu miners start to be the only ones able to mine. unless we can find a way to use our system memory for gpu crunching on scrypt-jane it will eventually be impossible to mine with a gpu
I think a lookup-gap will allow me to GPU mine even with N factors of 15,16,17,18 but the requirements on the number of CUDA cores will increase. You can always trade memory for more computation power (and extra electricity waste) - but at some point that power will become the bottleneck.
|
|
|
|
bigjme
|
|
January 08, 2014, 04:18:55 PM |
|
I have edgers for the driver and that works. It was only cuda that I hadnt got. I foubd instructions online and they were sooo long. It wasnt a case of just running the cuda installer. But does this work if I just run the cuda installer?
|
Owner of: cudamining.co.uk
|
|
|
cbuchner1 (OP)
|
|
January 08, 2014, 04:20:20 PM |
|
I have edgers for the driver and that works. It was only cuda that I hadnt got. I foubd instructions online and they were sooo long. It wasnt a case of just running the cuda installer. But does this work if I just run the cuda installer?
after running the cuda installer you should have a /usr/local/cuda-5.5 folder and the ./configure script of cudaminer will hopefully run through. There may be other required dependencies, such as libssl-dev, libcurl-dev, ... Christian
|
|
|
|
|