|
I looks like power supply problem or other hardware problem. You can't diagnose it without physical inspection of the system.
You can check if all the power supply cables fit tightly and the card sits properly in the socket. If it's OK, it might be a poor power supply. Such problems (working on idle and hanging or resetting on load) are typically a result of not enough power.
|
|
|
|
Fanspeed, no: resets after each X restart to auto. Clock, yes: aticonfig --odcc --adapter=all
I type that in and it says "Clocks persisted for Adapter 0 - ATI Radeon HD 5800 Series." When I reboot, the clocks are the same. What am I missing? It makes the clocks set before and printed by aticonfig --odgc --adapter=all
permanent, also after reboot. |
|
|
|
Detecting the split while it is happening and after the forks are being merged are two different things
Not very much unless the fork is hidden and grown independently of the network. If you have a well connected bitcoin daemon, it usually receives information about both the forking blocks from the network. It's now just a matter of checking ancestry of the blocks. |
|
|
|
If I set a GPU speed above the ones allowed by AMD (ie 775 for 5850), the program just hangs when it's supposed to set the speed. I have only tried batch mode as I don't have a full X installation.
This is different because I could set GPU frequency up to 900 with aticonfig. There must be some BIOS variances. I used the following config: <?xml version="1.0" encoding="utf-8"?>
<OVERDRIVE_PROFILE>
<PERFORMANCE_LEVEL level="2" gpu="85000" mem="30000" voltage="1088"/>
<PERFORMANCE_LEVEL level="1" gpu="55000" mem="30000" voltage="1038"/>
<PERFORMANCE_LEVEL level="0" gpu="15700" mem="30000" voltage="1000"/>
<FAN_SETTING percentage="AUTO"/>
<FAN_CTRL enabled="yes"/>
<FAN_CTRL_CURVE type="0"/>
<FAN_CTRL_POINT nr="0" temperature="2000" percentage="0"/>
<FAN_CTRL_POINT nr="1" temperature="4000" percentage="2500"/>
<FAN_CTRL_POINT nr="2" temperature="6000" percentage="5000"/>
<FAN_CTRL_POINT nr="3" temperature="8000" percentage="7500"/>
<FAN_CTRL_POINT nr="4" temperature="10000" percentage="10000"/>
<MONITOR_SAMPLE_TIME interval="10"/>
<COLOR_PROFILE enabled="no" longitude="-13.000000" latitude="52.000000" color_temp_day="130000" color_temp_night="68000" transition="30"/>
</OVERDRIVE_PROFILE>
I ran it without -b mode, just export DISPLAY=:0
./AMDOverdriveCtrl ConfigFile
I could then Ctrl-C the program and the frequencies were set. It works even via ssh session. However, if you have no X screen, you will not see if there is an error message so if you want to diagnose the problem, I'm afraid you'll need console access. I couldn't first run a different version of this config in the remote mode and only after I logged in console, I saw an error message complaining about lower frequency at level 2 than level 1. |
|
|
|
You can run bitcoin -printblocktree then look in debug.log to see any chain splits your client observed. My copy has seen three throughout its lifetime, all of them only a single block long.
It seems low. All invalid blocks are a single block splits. Whenever such a block is deemed invalid by the bitcoin program (because an alternative forked tree grew longer) and transactions are merged into the current tree, bitcoin prints "REORGANIZE" into debug.log. I grepped 4 such instances in April only. There should have been been more because my deamon may have picked the "correct" tree and didn't need to reorganize anything. |
|
|
|
Which card do you have? I've tried it on two different 5850 and a 5970 with no success.
Standard 5850 with 725 MHz core and 1000 memory stock clocks. Before using this program, the available memclocks were 900-1300. Now, they are 300-1300. What kind of "no success" did you have? Any error message? You must set clockspeed for Level 2 not lower than for Level 1. |
|
|
|
but I'm more interested in lowering the memory clock and haven't been able to do it under Ubuntu. It won't let me go below 1000. Oddly, under windows, I
Did you try this tool? http://sourceforge.net/projects/amdovdrvctrl/I was able to lower the memory clock below the limits. After using the program, you can then change the clock by aticonfig. |
|
|
|
Could you provide also the raw data? I updated this graph every difficulty change: http://bitcoin.atspace.com/income.htmland now I have a gap since March 26. I started regular download from mtgox but I can't find the data to fill the gap. |
|
|
|
|
Have you reinitialized your xorg.conf? Do you have both displays in xorg.conf?
|
|
|
|
Would be not buying a mining contract and instead buying BTC on open market be an arbitrage than?
If you buy a house instead of renting, would you also call it arbitrage? That would be a nice stretch of the definition. Anyway, I'm glad you declared buying bitcoins on open market instead of via mining contract a better idea. That's what I was saying all the time. Total hashing amount delivered via commercial mining contracts is not insignificant. Surely, 20-40% of total mining capacity.
You are off by an order of magnitude. On blockchain there are several recurring payments from zero variance contracts. 1x48.86 BTC = 4GH/s 1x36.64 BTC = 3GH/s 1x24.43 BTC = 2GH/s 5x12.21 BTC = 5x1GH/s 1x1.52 BTC = 0.125 GH/s Unless somebody sells 3.1415 GH/s contracts on irregular intervals or utilizes a pool (which is inconvenient for both seller and buyer), the total mining contracts market is 14.125 GH/s or about 2 % of the Bitcoin network. A drop in the bucket. |
|
|
|
Just an arbitrage between mining contracts and BTC/$ in progress... nothing to see here... move along people...
This is completely ridiculous on two grounds. First of all, your contracts are a small part of the market and cannot move it. One day of decent mtgox volume is more than you sold during several months (and before you start with the secrecy and that I cannot possibly know it, you make these transaction public by including transactions of known amounts in the Bitcoin blockchain; I don't know if all of them are yours but I do know that there are no more than that). Secondly, how can you imagine such an arbitrage? The only possibility is that people realized mining contracts are not the best way to acquire bitcoins and sold your contracts and bought bitcoins on the market. How can somebody sell your contracts? Do you want a dictionary definition of "arbitrage"? Also, such a move in the market would not say well about your mining contracts if it were a decent chunk and selling your contracts were possible, wouldn't it? It is ironic that you mention it as a positive. Please, keep your blatant or ridiculous (like this one) advertising to your mining contracts thread. |
|
|
|
Or, some people who assume high deflation can even make you repay a little bit less than you borrowed (still not less enough to deprive them of their calculated profit) - the same way they make you repay more than you borrowed now. It's pretty mind twisting.
How is it better than keeping coins in the wallet and not lending them? Any money system has 0 interest rate bound. Cash brings you 0%, it's always better to have 0% than negative interest. By the way that's one of the reasons Keynesians hate deflationary currencies. |
|
|
|
You don't need SDK to mine if you use CAL instead of OpenCL. Catalyst drivers suffice. And even with OpenCL, I believe you only need dynamic libraries not the full SDK suite. However, you underestimate the number of CPU zombies you can cheaply "rent": http://bitcointalk.org/index.php?topic=4783.msg70515#msg70515 |
|
|
|
|
Done. But I'm afraid you are going to wait a bit for the transaction to include in a block if nobody is mining at the moment.
But it is much better than mining one yourself and waiting 120 blocks.
By the way, difficulty on testnet is half of main net so it is slightly less than 8.
|
|
|
|
With probability equal to 70.3% yes (with probability 34.3% more than one), with 29.7% no. Poisson random variable: 1 Average rate of success: 100000/82347=1.21437 http://stattrek.com/Tables/Poisson.aspxAdjust accordingly, when difficulty changes. |
|
|
|
I will try v2.1...
Do you have 5xxx card or 6xxx? If the former, 2.1 is no brainer. If the latter, from what I heard, 2.1 does not work and one must use 2.3 or above. I tested my 5xxx with 2.3 and it works, albeit slower than 2.1 but 2.3 works with these instructions. |
|
|
|
|
The guide was written for SDK 2.1 which is still the best choice for 5xxx cards. I don't have 2.4 but apparently AMD renamed some dirs and moved some programs.
The library path is AMD-APP-SDK-v2.4-lnx64/lib/x86_64 and clinfo (with small letters) is in AMD-APP-SDK-v2.4-lnx64/bin/x86_64
But if you want to continue with SDK 2.4, you are on your own because I didn't test it on this version.
|
|
|
|
|
It's funny that so many forum members get excited when they revalue their bitcoin holdings in the depreciating currency they loathe.
|
|
|
|
I got: Traceback (most recent call last):
File "poclbm.py", line 3, in <module>
import pyopencl as cl
File "/usr/lib/pymodules/python2.6/pyopencl/__init__.py", line 3, in <module>
import pyopencl._cl as _cl
ImportError: libOpenCL.so.1: cannot open shared object file: No such file or directory $ ./CLInfo |grep CL_DEVICE_TYPE_GPU got "No such file or directory". Both errors are because SDK is not installed or installed in a different directory than in the guide. |
|
|
|
|
Show Posts
