If this miner is running on a site and I open the site in 5 tabs or in tabs in 2 different browsers, how is this handled? Will only one instance of the miner run? Even if people opt in, I think by default it should still limit the rate on their box so as to not hurt performance.
That's another thing that is worth investigating. Of course at present every webpage loads its own miner, and that can be a problem if you have more than one page open. But... is it possible for an applet to detect the presence of other applets running? Well, it turns out it is possible after all.  I have updated the jar file; now if you open a second tab the second applet should refuse to start mining. I have not had the time to test it extensively; as always, if you encounter any issue please let me know. |
|
|
|
Any chance PoolServerJ will support difficulties less than 1? I have tried setting it up for Litecoin (current difficulty is about 0.66), but with version 0.4.0rc2 all I get is a bunch of com.google.bitcoin.core.VerificationException: Difficulty target is bad: 40474064220290374717375238241067511033434108403022755503819899011072
at com.google.bitcoin.core.Block.getDifficultyTargetAsInteger(Block.java:582)
|
|
|
|
The problem is that, unless I'm mistaken, it is not possible to make native calls from an unsigned applet. Another suggestion/question.
If this miner is running on a site and I open the site in 5 tabs or in tabs in 2 different browsers, how is this handled? Will only one instance of the miner run? Even if people opt in, I think by default it should still limit the rate on their box so as to not hurt performance.
That's another thing that is worth investigating. Of course at present every webpage loads its own miner, and that can be a problem if you have more than one page open. But... is it possible for an applet to detect the presence of other applets running? |
|
|
|
|
I have done some research, and it looks like at present no mobile phone can run ordinary Java SE applets.
That would mean one less problem to deal with.
Can anyone confirm or refute this?
Of course the problem of laptops running on battery power still remains. I think the only satisfactory solution implies using cookies to remember users that have opted-out. My Javascript is a bit rusty, but if I remember correctly it shouldn't be too difficult to set these cookies client-side.
|
|
|
|
Um - hang on a tick ... that should never be an issue. You do ask them before taking their CPU cycles right? You just simply say that "if they are on a mobile phone that they shouldn't" do it. Edit: and the same issue regarding a laptop on a battery ... if you ask and explain the consequences there should be no issue .... unless you hid the consequences ... but you wouldn't do that coz that would be scamming them Unfortunately it's not that simple. I think opt-out is more practical than opt-in for most sites, otherwise I'm pretty sure that many visitors who don't mind simply wouldn't bother turning the miner on. Even if you start the applet on an opt-in basis, you certainly don't want the user to manually start the miner every time he goes from one page to another. You would probably use a cookie to remember the user's decision. So it would still be nice if the miner could automatically detect when a laptop is running on battery power and when it is not, as this could change at any time. |
|
|
|
Another question-- is this, or will it be, open-sourced?
Didn't read the thread, eh? Yes, the source will be released soon. That would be nice, but sadly a standard way of getting battery level information in Java doesn't seem to exist. You may be able to get it on some phones, but I don't think an unsigned applet can get that kind of information when running on a laptop. You also wouldn't want to mine if someone visits on a smartphone or some other low-powered device, but I guess the 2-core threshold will take care of that.
Not completely: dual-core mobile phones already exist. I will investigate, thank you for the suggestion! |
|
|
|
Also, if you set threads to -1, will it not run on systems with 1 cpu?
Yes, that's what I proposed a couple days ago. The current default is: 1 cpu = leave it alone (0% of all resources) 2 cpus = use 80% of one (40% of all resources) 3 cpus = use 1.6 (80% of two) (53% of all resources) 4 cpus = use 2.4 (80% of three) (60% of all resources) This looks very similar to what you proposed. The only noticeable difference is in the single-core case. But again, I'm not sure we should put any stress on single-core machines. Correct me if I'm wrong, but these would mainly be relatively old 32-bit cpus, and at 25% you wouldn't get much out of them. We should also consider that every connection to the work provider puts a little more load on the server, so I wonder if it would be worth it. |
|
|
|
I'm solo mining, and my reward frequency has dropped a lot in the last couple of weeks.
Last blocks were found on November 1st, 6th, and 15.
Given I have 8 threads mining at about 1.1 khash/sec ~ 9 khash, I should find a block every
2^32 * .64 / 9,000 seconds =
305,000 seconds =
3.5 days
Given the above calculation, how likely is 14 days without a block?
I would say (1-1/(0.64*2^32))^(9000*86400*14), or about 2%. Not very improbable. |
|
|
|
On a side note, it would also be nice to be able to throttle back the miner so it doesn't use 100% of the cpu(s) it's running on.
Nice idea, I will definitely look into that. |
|
|
|
Do you mind if I poll your API every five-or-so minutes?
No problem. |
|
|
|
I would like to know what everybody thinks about embeddable miners, such as the one by BitcoinPlus ( http://www.bitcoinplus.com/miner/embeddable). First off, I want to make it clear that if Litecoin is successful somebody will eventually come up with a (potentially evil) embedded miner, no matter what. (Especially since I am going to release the source code of the Java miner very soon - stay tuned. ) As terrytibbs pointed out, for some sites such miners may represent a nice alternative to ads. On the other hand, they might be even more annoying to some visitors. Personally I think they can be fine as long as the visitor is informed of what's going on (that would also bring some publicity to Litecoin), and can choose to opt-out at any moment. The problem is, of course, that this cannot be enforced by the pool to which the miner submits its work. On the technical side, I think it would be wise for such miners to leave single-core systems alone, and only start n-1 threads for an n-core system. I still haven't decided if I will release an embeddable miner, I would like to hear other people's opinions first. |
|
|
|
Doesn't auto-populate on the latest version of Chrome.
Thank you, that should be fixed now. i3, running 1 thread, 64 bit W7 and 64 bit Java JRE 7, latest Chrome:
Minerd -- 1.44Khash
Online Miner -- .60Khash
So like I said, little less than half. Still pretty good though. I don't have a 64 bit browser -- I'll DL one of the firefox ones if I can find a stable-ish one.
I have revived an old single-core laptop (32-bit Pentium-M @ 1.7 GHz) and I get 0.37 kH/s, versus 0.75 kH/s of minerd. So far it looks like a 67% ratio can be achieved on 64-bit systems, while 32-bit systems cannot get past 50%. Given these results, my theory is that having a fully 64-bit environment (OS+browser+JVM) is the only way to achieve full performance. If you think about it, it's a long way from the Java browser plugin to basic CPU instructions. And, as someone said, a (software) chain is only as strong as its weakest link. |
|
|
|
It does correctly detect my quad core -- I'd just like to be able to limit it to at least 3 ( I like have 1 core free for other stuff) but I'd also like to simultaneously run 1 core comparisons.
Done! You should now be able to select the number of threads you want to start. The "Threads" field should be auto-populated with the number of cores your CPU has when the page is loaded. Please let me know if that doesn't work in your browser. |
|
|
|
All browsers with Java installed for the browser, allow Java code to be run on them - that's the point of installing it.
I feel obliged to mention that you can enable/disable Java, Flash etc. on a per-site basis. I do. Browser plugins like Noscript and Flashblock have become very popular lately. Personally, I find them very valuable, especially when I have to work on older computers. that also means people can put this miner on their web site and hide it in the page and have it run on site visitor's computers without asking them for permission.
Come on, people can do much worse things than that, if they are malicious enough. Without the aid of this miner. It's the same old story: the same technology can be both "good" and "bad", depending on what use you make of it. Unfortunately we can't stop the march of technology. If it can be done, someone will eventually do it. I did not start this project with the intention of stealing the CPU time of unaware surfers. I started it because I think it can help the diffusion of Litecoin. (And because I love programming ) Of course this miner will always be somewhat slower and less profitable than compiled miners, but it has other advantages. First and foremost, it will run on almost any platform, and with little to no configuration. I am sure that, as kjlimo said, "simple minded" people will enjoy that. Also think about this: the average user (who might not be familiar with Bitcoin and its software) will be (rightly) reluctant to download and run an executable file from the Internet. And for sure he/she won't have the time or knowledge to check out the source code. With a Java applet, you don't have to trust the author of the code that much, because the guys at Sun Microsystems have worked hard to make Java as secure as possible. That's why, by default, browsers just run Java applets without asking too many questions. |
|
|
|
I've just checked the online miner on a Phenom II X6 @ 3.5 GHz. Using all six cores, I can get 13.8 kH/s (vs 21.3 kH/s of the optimized minerd). This was tested under Linux, this time with IcedTea 6. I'm still not sure why other people are not getting similar figures. The only thing I can think of is that you are probably using a 32-bit version of the Java virtual machine. Correct me if I'm wrong, but I think that to get the most out of Java you should be running a 64-bit version of the JVM in a 64-bit browser... all, of course, under a 64-bit OS. EDIT: according to rTech's edit above, it looks like that was really the problem. Everybody, please update your browser and JRE/JDK to a 64-bit version if you want to double your hash rates. |
|
|
|
I am kinda confused. I run the Minerd. I don't even remember where I downloaded it from. I run a AMD Phenom 8400 Triple-Core 2.10 GHz and I run 3 threads. I get about 3.5 Khashes. Is there a software way to increase my hashes?
If that 3.5 is on each core then you are doing good my X3s get about 3.05 per core 3.10 GHz speed, if total for the box then that sucks you will want to compile your own minerd with ./configure CFLAGS="-march=amdfam10 -Wall -O3" as your configure option to get the most out of your processor. Or try this binary if you are running Windows (replace minerd.exe if you use the ScryptMiner GUI): http://content.wuala.com/contents/jbw9/pub/Bitcoin/Tenebrix/miner/win/minerd-amdfam10-sse4a.exe/?dl=1 |
|
|
|
I seem to be getting around 5khash/sec on my i5 2400 (@ 3.8ghz), which I usually get about 12 with minerd (the optimized sse4 one).
I thought it could do much better on such a machine. Consider I'm getting 2.2 kH/s on a dual-core laptop at 1.6 GHz! I wonder if this could depend on your Java setup... in particular, are you using the latest and, perhaps most importantly, 64-bit version of the Java VM? |
|
|
|
|
Show Posts
)
