|
Old news is old.
Where do you think the inspiration for some of these miner implementations came from?
|
|
|
|
since what point?
Since a Chinese botnet got a CAPTCHA upgrade....  |
|
|
|
|
A built in nice would be nice (badumtish!) for both CPU and GPU. nice as in scheduling - I want this thread at 40% and this thread at 60% sort of thing. For both multi-pool participation as well as for non-dedicated miners and/or power peaks. Another example: if I pay more power during the day, I would like to run at 25% during the day at 100% at night.
The time/date scheduling can be done manually by the user, but the process scheduling would have to be a core part of the application.
Summarized output is what bugs me on a lot of miners. There's a request somewhere for a poclbm mod. Almost every miner will just spam the screen with "accepted", "stale", "MHs", blah blah blah. It would be nice to silence all of that and get a -
[CPU Thread 0 - 2.3MHs, 25% priority, 300 accepted, 25 stale, user: supa, pool: us.eligius.st]
[GPU Thread 0 - 300MHs, 100% priority, 15000 accepted, 100 stale, user: supa, pool: eu.eligius.st]
Summary on keypress or at some update interval (3 minutes?).
Finally, and this is just my personal preference and easy to script if no one else wants it, I would like a simple HTTP interface of stats. You could quite literally just accept connections on some port (2890?) and send "<html><head>Miner</head><body>300 MHs</body></html>"
If you're using Java, maybe future Tomcat integration?
|
|
|
|
|
Kind of and sort-of-ish.
You can certainly install (for most miners - this is just copying the .exe) on computer A.
You can then run that program from computer B using Windows file sharing or Linux file sharing (CIFS/SMB/NFS).
Example:
From Computer B:
\\computerA\c$\bitcoin\poclbm.exe --<my args go here>
For all intents and purposes, poclbm.exe is "installed" on computer A but being run from computer B. poclbm.exe is not the best example here, though - it has quite a few supporting files that are required. This method will probably not work with poclbm, but will work with ufasoft's miner or something similar.
For poclbm, you would need to map a network drive on computer B like this:
net use Z: \\computerA\c$\bitcoin\
Z:
poclbm.exe <my args here>
|
|
|
|
1 Megahash / second == 1,000 Kilohash / second == 1,000,000 hash / second
So, what you're saying is... divide my Kilohash score by 24, then multiple by powers of 1,000?  Joking aside, it's an easy mistake I suppose. For all intents and purposes, when you're counting things that aren't from base 2, it should be a completely safe assumption that K (as in kilometer) is 1,000 x 1 meter and not 1,024. Kilometers, gigawatts, megahashes, cycles per second (Hertz), etc, etc are all 1,000. Please do not try to multiply your 900MHz GPU card clock by 1024 * 1024.... |
|
|
|
Are you sure? Any proof please?
There's a ton of references on the forum for this. I, for one, welcome our non-dummy plug requiring Linux overlords. Additionally, if you happen to NVIDIA hardware already (worse hashrate than ATI, but *if you already have it*) you don't need to go above run level 3 - a welcome power saving. |
|
|
|
The way things are going I'd start buying AMD stock....
This. A warehouse store manager told me AMD is having some huge announcement in the next few months about new CPU/GPU combos and some other stuff. So either way.... AMD could be a big winning... |
|
|
|
Check ldconfig -p | grep -o opencl In my experience, make sure you - export DISPLAY=0:0 (or whatever your display is with the ATI driver) before you try to use something that needs OpenCL. I wish NVIDIA hashed better.... I don't even have to init the driver to use CUDA.  |
|
|
|
50 systems, *if* they can do 2MHs (this is about an Athlon X2), nets you ~100MHs. If you have something like Intel i7s... Now you're talking 500MHs. From the sounds of it (onboard video, office, etc) I'm going to guess you have some older PCs with 50/60W PSUs. PXE booting would be pretty easy to set up and ideal. Throw together a simple Linux image (or just use one of the LiveCDs) and add ufasoft's cpuminer. See what you get! |
|
|
|
10%? (300/100)*10 != 1 nor 2...
Also why use that extra power when you know it's so inefficient that it costs more in money than it produce?
I was thinking ahead of myself there, eh? I'm not even going to edit the typo out. I use multiple systems in parallel and was thinking in several systems instead of a single desktop. Define inefficient. My PC is on. With the CPU underclocked and at 0 load with no miner, it produces 0MHs. The CPU underclocked and at 100% load on a single core with a miner consumes no more power than it did previously. It produces 2MHs. Windows boot up on the same system at 10-20 seconds of complete load with no scaling is more expensive than not running a miner. It's also more expensive than running a miner. Hmmm... Add a GUI. That consumes far more cycles and far more power than running a single thread on a multi-core processor in many systems. If your definition of "inefficient" is "not producing anything and costing the same as if it were" then I suppose that's correct.... |
|
|
|
|
There should never, ever be a difference in GPU mining between OSes. With the exception of driver versions.
A WM? Really?
Consider a single sempron, no GUI, no DVD, no other disks, no monitor, 256M of RAM, no monitor, no GUI, no keyboard, no browsers, no GUI, on board sound disabled, AMD Cool n' Quiet on, BIOS-only RAM underclocking, run level 3, no GUI, no automatic coffee maker, no silliness.
People will scoff at CPU mining, but go crazy over 3% gain in a power hungry OS?
A simple Linux desktop can do 3MHs on a core without scaling up to full power load. Some ATOM processors can do 1-2MHs and never consume more power. There's your 10% gain overclocking a 300MHs card and performing extra cooling.
|
|
|
|
USB Persistence is over engineering. Here's a "live" example I use: I have a webserver at 192.168.2.10 with an index file that contains my BitCoin address. That's all it contains. The following... you could put it in /etc/init.d to auto-startup. Make sure it's late in the chain to avoid starting before network is ready. You could also put in some sleeps and wait for network up, but I don't bother....  --- export mybitcoinaddress=$(wget -q -O - http://192.168.2.10) /home/user/bitcoin-miner.exe -o http://mining.eligius.st:80 -u $mybitcoinaddress -p anything --- Done. For other miners that require login and pass, you could do some sed/awk/grep/whatever magic and include them all in a single page. Or for simplicity's sake, just make separate files on the web server with each piece of information (for example, http://webserver/username, http://webserver/password, etc). |
|
|
|
I remote in because I have pre-existing assets in other states prior to Bitcoin's conception. Not for anonymity. Perhaps the posts you read were for the same reason? There is currently no reason to sweat anonymity. Unless you want to support Tor out of the kindness of your heart. |
|
|
|
DEA Agent Anthony Marotta said high electricity usage does not always mean the residence is an indoor pot farm and has surprised federal agents. "We thought it was a major grow operation ... but this guy had some kind of business involving computers. I don't know how many computer servers we found in his home." That's from the terribly written Network World article. Was the DEA Agent speculating or did they encounter some other individual doing some "business involving computers" in the same manner as perpetuated by rumor on IRC (which they have a screenshot of in the article) ? |
|
|
|
Linux is easy - grab almost any common distro and add "3" to the end of the kernel parameters in grub. From the Grub menu press a key to enter selection mode (where you can move the cursor), choose a boot option and press "e". Select kernel, press "e" again. Go to the end of the line and add a "3". Hit enter, press "b". That will place you in run level 3 (no GUI, multiuser with networking) and you're already better off than Windows with almost no work. |
|
|
|
What about the publicized BitCoin Mine Raids when the power consumption "looked like" dope houses? Tor is Tor - by providing yourself as a node, you help the network. So yes, I say use Tor, solely to provide another node to Tor. ..... but in relation to bitcoin, it's probably silly to incur the latency cost of Tor. |
|
|
|
Ya, I see what you mean. I looked, too, and I only saw mention of slush/deepbit fees. Maybe he'll pop in and let us know. I know he must be making something on rounding - you can only payout in 0.01 increments and it will take you a significant amount of time to reach 0.01. I'm sure there's abandoned millicents abundant. |
|
|
|
|
Similar to pool hopping, I was questioning a parallel request versus more serial sequences.
Using simple numbers, let's say that I have one box (be it FPGA, desktop, CPU or GPU) that mines 1MHs. If I magically increased the speed to 5MHs, I am doing work requests for individual blocks and processing that work in a serial manner (that is, on a single processing source with no threads/timeslicing/multicore/multiprocessor).
If I instead had 5 x 1MHs boxes, I would be doing 5 times the number of requests in parallel.
Regardless of power and other factors... if I hear someone try to proselytize KWh/cost one more time I'm seriously going to have a breakdown.... what do you think the variance would be in these methods?
In small experiments (100MHs) over two months, I seem to have received more payouts with parallel computation rather than a single source of equal computation. Given that it could be pure coincidence, I was wondering if there is a possible justifiable explanation as to why which would be better (other than KW/h nonsense).
There is the obvious benefit of health to the Bitcoin network if the parallel sources are co-located. That is, if I have 100MHs in my bedroom, it's much better for the health of the network if I have 100 x 1 nodes scattered across datacenters. Consider me more of a philanthropist than redbull-fueled-overclocking-profit-rabid-miner.
|
|
|
|
So I suppose you'd crack the local admin password on each of the machines (which is remarkably simple) and set up the miner as an "all users" startup program. The only problem with that is that it's time consuming.
Just speculating, by the way. I love speculative hacking. =P
EDIT: Wait, you have to literally use a key card to sign into the computer? That's a bit odd, isn't it?
I think he meant manually sign on... On a side note, there was a lab that I visited before that required all users authenticate via SmartCards. It's been built into Windows GINA for quite a while and of course PAM for 'nix, too. Biometrics are another one that are common in some places - not for security, but for ease of access. Most Biometrics systems (cough*lenovo*cough) just store the users password along with a biometric key. When a user swipes their finger - it just provides their normal password. The opposite of improving security. http://msdn.microsoft.com/en-us/library/aa375457(v=vs.85).aspxIn order to "simply" break an admin password, you need physical access and the ability to change the boot media. At that point, it would be much simpler to use the revisor image I keep rattling on about...... with a little work, you could easily set up a service (init.d script) to automatically start mining on boot with no GUI. A crond to reboot or shutdown and there's absolutely no trace you were there. I only mention "no trace" because the goal is to keep the machines production ready - not for the legal implication of evidence. |
|
|
|
My current "arrangement" doesn't allow me to get that complex, so I can't really make good recommendations. There's an open source pool project you can grab from the forums here to create a pool, I think. I'm not sure what features it has or doesn't have. I was using the simple graphs available on eligius.st (a pool) to get some metrics when I was setting the environment up. If you find a way to collect legitimate metrics that aren't just each machines estimated MHs as displayed by the miner.... please let me know. I know that ufasoft and poclbm have worked fine on all the XP/7/2008 machines I've used them on. They're simple to set up and work fine in a console. For my limited environment... that's "good enough." For Linux, I use the minerd available from the forums here. The 4way algo seems to work best on my AMD machines - the sse for 64 has a 20% loss(  ). The age old combination of ssh+screen is all you need in Linux - and you have the option of streamlining your services and no GUI. poclbm works fine on both the ATI and NVIDIA cards in the machines I have access to - since it's a mixed environment, I haven't bothered with testing anything else. 2008 has some great new ways to streamline and tailor down the OS - you'll have to let us know what tricks you use to get them running flawlessly. |
|
|
|
|
Show Posts
