Anyway, I dont seem to have lost any work, perhaps thats the advantage of using slower cards
. 
Topic: Bitminter client (Windows/Linux/Mac) - page 45. (Read 654946 times)
disclaimer: I dont know what Im talking about, but if the server was up and pinging showed no problem, could it be bitcoind having (had) trouble connecting to the bitcoin network? Is bitcoind using a different connection? If so, could this possibly be related to a DDoS attack?
Anyway, I dont seem to have lost any work, perhaps thats the advantage of using slower cards.
Anyway, I dont seem to have lost any work, perhaps thats the advantage of using slower cards
. 
Did something change on the back-end tonight?
Nope, nothing changed.
Not sure what that was, but I also see "idling" messages intermittently in the log for 15 minutes. I wasn't watching as it happened, so I didn't get a chance to investigate. But if it was accepting proofs of work but was very slow with giving out new work, then perhaps it was a problem with bitcoind. Strangely, everything seems to have cleared up as suddenly as the problem appeared.
Did something change on the back-end tonight?
Im constantly running out of work to do.. server is taking all the proofs of work no problem.
spare work units on hand would just keep dropping to zero.. now the funny thing is exactly at 3:30 -6GMT everything cleared up..
the instant I post a proof of work I get a new work unit and the buffer stays full.
no packet loss or high ping during the problem time either.
Im constantly running out of work to do.. server is taking all the proofs of work no problem.
spare work units on hand would just keep dropping to zero.. now the funny thing is exactly at 3:30 -6GMT everything cleared up..
the instant I post a proof of work I get a new work unit and the buffer stays full.
no packet loss or high ping during the problem time either.
btw...i cant find the donation box??
Voluntary donations most likely coming after merged mining, signature banners and automated mining. So hold that thought!
I just want to be able to start it up from the command line 
For regular: javaws https://bitminter.com/client/bitminter.jnlp
For beta: javaws https://bitminter.com/beta/beta.jnlp
Yes, beta is fine. I just increased the work intervals to 20 (I used to keep it at 10 with previous version) to obtain the same performance I got with the previous version. But in addition my puter is much more snappy and my cards a bit cooler with the beta.
Beta is working great for me..
K9A2 Platinum MB / dual core CPU forget what it is.
4GB Ram / no HD
2 x 6970's
Ubuntu
getting 428 a card 52121 with 68 rejects
----
890FXA-GD70 / 1090T CPU
8GB ram / HD
3 x 6970's
Windows 7 64 bit
Getting 420 a card. 19898 with 12 rejects
I just want to be able to start it up from the command line
K9A2 Platinum MB / dual core CPU forget what it is.
4GB Ram / no HD
2 x 6970's
Ubuntu
getting 428 a card 52121 with 68 rejects
----
890FXA-GD70 / 1090T CPU
8GB ram / HD
3 x 6970's
Windows 7 64 bit
Getting 420 a card. 19898 with 12 rejects
I just want to be able to start it up from the command line
btw...i cant find the donation box??
im still running last vers, 16days 2 hours 363290 and 418 rejects, im staying with this one till you get the bugs worked out.
Damn, performance wasn't supposed to drop. Well, this is what beta testing is for. Thanks for the reports, guys! I will look into this.
How much higher intervals help depends on your hardware and what other programs you got running. Basically you just have to try it out.
Well, this is what beta testing is for. Thanks for the reports, guys! I will look into this.How much higher intervals help depends on your hardware and what other programs you got running. Basically you just have to try it out.
How much can you bring it up with higher work intervals ?
Even my 5750 on another pc is slower: from 175 Mhs with the previous version to 163 Mhs now. Same settings, of course.
Even my puters seems more responsive though, so I am happy anyway.
Even my puters seems more responsive though, so I am happy anyway.
I changed my work intervals from 100 to 500 ms. Perhaps i see no big difference because the card is so slow. At the same intervals, the computer is more responsive than before. Nice for background mining.
When you get lower performance, are you sure you are running with the same settings? Remember default intervals changed from 50ms to 20ms.
Yep, it is. Probably going to do merged mining (big job) and signature banners (small job) first, then automation.
Automation should be pretty high up on list of priorities, imho.
Yep, it is. Probably going to do merged mining (big job) and signature banners (small job) first, then automation.
Also seems a tad slower on my 5850, think I lost around 10 Mh/s
I take that back. its a fresh ubuntu beta install and I seem to have lost performance on the 1.0.1 version too
I take that back. its a fresh ubuntu beta install and I seem to have lost performance on the 1.0.1 version too
It seems a bit slower on my 5870: now i get 395-405 Mhs against 415-420 Mhs with the previous version using the same settings.
Even my card seems a couple of degrees C cooler though.
the new version runs my card 2 degrees C cooler than before at the same hashrate. 1600 hashes and zero stales ! Very nice
Beta version 1.1.0-M1 (milestone 1) now available.
Changes:
What I would still like to add for v1.1 is saving GPU settings, automated mining, and an options window with at least settings for automation and audio. I'm releasing this now though to get it tested - it may take a while to finish the rest.
Changes:
- Reduce stales through better handling of stale data and block changes.
- Improved buffer handling. Should fix the performance degradation
that some users observed with v1.0.1 - Added sound effects
- Log and play sound (if enabled) when submitting a block-generating
proof of work. Display number of generated blocks on status bar. - Use larger log buffer - fit more log messages before old ones disappear
- Base "BTC/day" display on up-to-date difficulty. Log changes in difficulty.
- Bugfix: no longer automatically tries wrong password multiple times
- Adjusted power meters for AMD Caicos and Intel i7 2600. Your GPU meter
not in the middle of the green area at stock clock? Let me know. - Lowered minimum work interval (formerly break interval) from 10 to 1.
Lowered default setting from 50 to 20.
What I would still like to add for v1.1 is saving GPU settings, automated mining, and an options window with at least settings for automation and audio. I'm releasing this now though to get it tested - it may take a while to finish the rest.
The only way I found to fix this problem is to revert back to cataCLYSM 11.5. With them javaw takes only 2-3% of cpu while mining.
So I propose to launch a petition to ATI to fix this bug (maybe they see it as a "feature").
On another miner I am still keeping cataCLYSM 11.9 keeping the cpu underclocked by CrystalCPUID on XP, since I think to add a second card and I know that old drivers need a dummy plug that I don't have and don't want.
Setting it to low priority isn't going to help the fact that you're burning tons of CPU heat with a runaway CPU process performing no work... great way to screw over your power bill is letting your CPU run full speed when the GPU is doing all the work. At the very least, set a power profile with a maximum CPU state of 50% in Windows 7 power options just for mining.
But the proper fix is to untangle the cause of the CPU usage. ...
[/quote]
So I propose to launch a petition to ATI to fix this bug (maybe they see it as a "feature").
On another miner I am still keeping cataCLYSM 11.9 keeping the cpu underclocked by CrystalCPUID on XP, since I think to add a second card and I know that old drivers need a dummy plug that I don't have and don't want.
But the proper fix is to untangle the cause of the CPU usage. ...
[/quote]
AFAIK, its the exact same thing. Its just that "aggression" is defined so that higher aggression means higher GPU kernel priority. Break interval is the opposite, as I understand it, its the time between opportunities for the OS to do something else, so higher interval means lower GPU priority.
Yeah, just two ways of deciding how big each job for the GPU should be. If you give it so much work in one go that it takes more than 5 seconds to process, Windows will think something locked up and it will restart the graphics driver. And even before that the desktop will of course get extremely laggy. On the other hand, there is a little overhead every time you give the GPU new work. So if you give it tiny jobs things get slowed down from the overhead.
But the proper fix is to untangle the cause of the CPU usage. BitMinter doesn't have aggression options (just a vague "delay period" option that doesn't conceptually tie in well with "Aggression" IMO), but most others do.
AFAIK, its the exact same thing. Its just that "aggression" is defined so that higher aggression means higher GPU kernel priority. Break interval is the opposite, as I understand it, its the time between opportunities for the OS to do something else, so higher interval means lower GPU priority.
Jump to: