...
There is no many coz there are no implementations of any of the crap that you seem to think is better in GBT.
...
So yes GBT sux and no matter how often you make these deceptive comments, GBT is still crap.
GBT just wants to be friends.
Topic: OFFICIAL CGMINER mining software thread for linux/win/osx/mips/arm/r-pi 4.11.0 - page 447. (Read 5806004 times)
November 15, 2012, 12:38:39 AM
...
There is no many coz there are no implementations of any of the crap that you seem to think is better in GBT.
...
So yes GBT sux and no matter how often you make these deceptive comments, GBT is still crap.
GBT just wants to be friends.
November 15, 2012, 12:32:05 AM
Also I note Luke actually makes less then 1Diff shares only send >=1Diff shares so those could have losses.
It's not possible to send less than 1diff shares (except for CPU/GPU).The issue at the moment is that no BTC-FPGA bitstreams have been written that support <1 difficulty.
All the GPU ocl code that I have seen also only supports 1 difficulty.
However, there is no actual need for < 1 difficulty with BTC.
If there was, you'd just need to write a new OCL for the GPUs or a new bitstream for the FPGAs
However, for all Scrypt mining, it is of course possible and mandatory to support < 1 difficulty.
i.e. it does already.
Luke used the laziest way to fix it. A Compliant answer "Should" actually only send what the difficulty is.
Perhaps it is a bit lazy, but it works. Unfortunately, since stratum is representing the target as bdiff, it is impossible for the miner to know exactly what the target is. (Admittedly, it would be possible to get a closer guess of course.)Go do that else where.
Fractional difficulty is simple.
The only issue is that maybe one in a few billion-trillion shares might be rejected that are valid.
... and they wouldn't ever be a block either.
So who gives a damn if it is that rare to ever lose a share?
i.e. stop writing crappy hack code and do it properly for once.
My second suggestion posted shortly after the one I am quoting lower and before Luke quoted me wondered why a client couldn't ask for or submit a higher difficulty work unit. Say the pool asked for 1.09 and your miner only uses ints why not send diff 2 shares but before you do negotiate with the server for pay based on diff 2. Still runs ints but pays more per unit and requires less network traffic.
Just one of the many things GBT supports but stratum doesn't (yet?).There is no many coz there are no implementations of any of the crap that you seem to think is better in GBT.
Can you name even ONE that is implemented and used anywhere?
And if you can find ONE, then maybe consider listing this delusional many you keep having wet dreams about.
You've been told what is wrong with GBT and ignored it.
So yes GBT sux and no matter how often you make these deceptive comments, GBT is still crap.
November 14, 2012, 11:21:42 PM
Updated with support for fractional diffs and diffs just below 1.
November 14, 2012, 09:39:11 PM
Also I note Luke actually makes less then 1Diff shares only send >=1Diff shares so those could have losses.
It's not possible to send less than 1diff shares (except for CPU/GPU).Luke used the laziest way to fix it. A Compliant answer "Should" actually only send what the difficulty is.
Perhaps it is a bit lazy, but it works. Unfortunately, since stratum is representing the target as bdiff, it is impossible for the miner to know exactly what the target is. (Admittedly, it would be possible to get a closer guess of course.)My second suggestion posted shortly after the one I am quoting lower and before Luke quoted me wondered why a client couldn't ask for or submit a higher difficulty work unit. Say the pool asked for 1.09 and your miner only uses ints why not send diff 2 shares but before you do negotiate with the server for pay based on diff 2. Still runs ints but pays more per unit and requires less network traffic.
Just one of the many things GBT supports but stratum doesn't (yet?). 
November 14, 2012, 09:25:00 PM
On my system it looks about the same:
This is with Windows 7
Code:
cgminer (1).exe caused an Access Violation at location 0040cc8e in module cgminer (1).exe Reading from location 00000000.
Registers:
eax=00000000 ebx=00000000 ecx=00000000 edx=00000000 esi=00000000 edi=01e4f908
eip=0040cc8e esp=02b6f3d0 ebp=01e73df8 iopl=0 nv up ei pl zr na po nc
cs=0023 ss=002b ds=002b es=002b fs=0053 gs=002b efl=00010246
Call stack:
0040CC8E cgminer (1).exe:0040CC8E
77D2FA19 ntdll.dll:77D2FA19 RtlAnsiCharToUnicodeChar
77D2FAF6 ntdll.dll:77D2FAF6 RtlAnsiCharToUnicodeChar
77D30093 ntdll.dll:77D30093 LdrGetDllHandleEx
77D2FD2F ntdll.dll:77D2FD2F LdrGetDllHandle
76D81A35 KERNELBASE.dll:76D81A35 GetModuleFileNameW
76DA734E KERNELBASE.dll:76DA734E IsNLSDefinedString
76D81CFB KERNELBASE.dll:76D81CFB GetModuleFileNameW
75733362 kernel32.dll:75733362 RegKrnInitialize
November 14, 2012, 09:20:13 PM
Thanks. That is quite different actually. I've uploaded another set of executables into that temporary directory which may get us even further. Can you try them please? There is no way on earth it will compile on VS.
This is more like what I was expecting. The debug version is much much larger than the production one. But the output looks about the same...
Code:
cgminerdebug.exe caused an Access Violation at location 0040cb33 in module cgminerdebug.exe Reading from location 00000000.
Registers:
eax=00000000 ebx=624836d0 ecx=74712e09 edx=07c51858 esi=00000000 edi=00770ea0
eip=0040cb33 esp=03fff3d0 ebp=00801480 iopl=0 nv up ei pl nz na pe nc
cs=0023 ss=002b ds=002b es=002b fs=0053 gs=002b efl=00010202
Call stack:
0040CB33 cgminerdebug.exe:0040CB33
7713FAF6 ntdll.dll:7713FAF6 RtlAnsiCharToUnicodeChar
77140093 ntdll.dll:77140093 LdrGetDllHandleEx
7713FD2F ntdll.dll:7713FD2F LdrGetDllHandle
76641A35 KERNELBASE.dll:76641A35 GetModuleFileNameW
7666734E KERNELBASE.dll:7666734E IsNLSDefinedString
76641CFB KERNELBASE.dll:76641CFB GetModuleFileNameW
7666734E KERNELBASE.dll:7666734E IsNLSDefinedString
76641CFB KERNELBASE.dll:76641CFB GetModuleFileNameW
74AF3362 kernel32.dll:74AF3362 RegKrnInitialize
How about Watcom? I can dust it off and see if I can get it to run there.
M
November 14, 2012, 08:44:48 PM
Those with easy to reproduce crashes on windows can you try one of these quick hack builds please:
http://ck.kolivas.org/apps/cgminer/temp/
http://ck.kolivas.org/apps/cgminer/temp/
Not much different. Visual Studio gives me nothing to work with. Dr. MingW gives me this:
Code:
cgminer.exe caused an Access Violation at location 0040cb33 in module cgminer.exe Reading from location 00000000.
Registers:
eax=00000000 ebx=624836d0 ecx=74712e09 edx=07dfa910 esi=00000000 edi=004b0e80
eip=0040cb33 esp=0408f3d0 ebp=00579198 iopl=0 nv up ei pl nz na pe nc
cs=0023 ss=002b ds=002b es=002b fs=0053 gs=002b efl=00010202
Call stack:
0040CB33 cgminer.exe:0040CB33
7713FAF6 ntdll.dll:7713FAF6 RtlAnsiCharToUnicodeChar
77140093 ntdll.dll:77140093 LdrGetDllHandleEx
7713FD2F ntdll.dll:7713FD2F LdrGetDllHandle
76641A35 KERNELBASE.dll:76641A35 GetModuleFileNameW
7666734E KERNELBASE.dll:7666734E IsNLSDefinedString
76641CFB KERNELBASE.dll:76641CFB GetModuleFileNameW
7666734E KERNELBASE.dll:7666734E IsNLSDefinedString
76641CFB KERNELBASE.dll:76641CFB GetModuleFileNameW
74AF3362 kernel32.dll:74AF3362 RegKrnInitialize
Maybe if I can get it to compile in Visual Studio, I'll run in it dev and see what happens when I unplug the NIC.
M
November 14, 2012, 08:41:37 PM
Since eclipse uses fractional difficulties and it seems they are less reliable on JSON could you accept whatever eclipse says to work on then if it has a decimal truncate it to an int and add 1?
Eclipse is fully stratum/JSON/JSON-RPC compliant. Truncating it works, but you'll lose shares if you add a 1.In any case, GBT is still the preferred protocol.
This was in the post luke so hackily quoted. I even noted that it would lose acceptable shares. My first thought had been to actually just truncate it but that would cause higher rejects. My second thought was to add to it making sure that a person wouldn't send a ton of rejects. Technically a person could just spam the pool with any shares they wanted and though the difficulty would likely make them rejects it wouldn't keep it from working. Would it be ugly sure. Hacky sure. Really bad for your stats absolutely. A divine solution it really shouldn't be.
Also I note Luke actually makes less then 1Diff shares only send >=1Diff shares so those could have losses. I would call it a quick fix but for the guy trolling Con about not just blindly altering the data as a "Compliant" answer I really wonder. The whole point is to reduce network traffic. By spamming with lower results you do anything but lower it. What Luke did do is keep traffic the same as if the server sent the nearest lower int. This tells me Con isn't so much wrong as Luke used the laziest way to fix it. A Compliant answer "Should" actually only send what the difficulty is.
My second suggestion posted shortly after the one I am quoting lower and before Luke quoted me wondered why a client couldn't ask for or submit a higher difficulty work unit. Say the pool asked for 1.09 and your miner only uses ints why not send diff 2 shares but before you do negotiate with the server for pay based on diff 2. Still runs ints but pays more per unit and requires less network traffic.
Quote
Actually I see a good reason not to now that I think on it more. It seemed easy before but if I got a 1.9 difficulty I doubt 2 would be much worse but 1.09 would obviously cost me shares in theory. Unless the client could specify its difficulty it is working on that will cause an issue with not submitting work.
November 14, 2012, 08:23:56 PM
Those with easy to reproduce crashes on windows can you try one of these quick hack builds please:
http://ck.kolivas.org/apps/cgminer/temp/
http://ck.kolivas.org/apps/cgminer/temp/
Not much different. Visual Studio gives me nothing to work with. Dr. MingW gives me this:
Code:
cgminer.exe caused an Access Violation at location 0040cb33 in module cgminer.exe Reading from location 00000000.
Registers:
eax=00000000 ebx=624836d0 ecx=74712e09 edx=07dfa910 esi=00000000 edi=004b0e80
eip=0040cb33 esp=0408f3d0 ebp=00579198 iopl=0 nv up ei pl nz na pe nc
cs=0023 ss=002b ds=002b es=002b fs=0053 gs=002b efl=00010202
Call stack:
0040CB33 cgminer.exe:0040CB33
7713FAF6 ntdll.dll:7713FAF6 RtlAnsiCharToUnicodeChar
77140093 ntdll.dll:77140093 LdrGetDllHandleEx
7713FD2F ntdll.dll:7713FD2F LdrGetDllHandle
76641A35 KERNELBASE.dll:76641A35 GetModuleFileNameW
7666734E KERNELBASE.dll:7666734E IsNLSDefinedString
76641CFB KERNELBASE.dll:76641CFB GetModuleFileNameW
7666734E KERNELBASE.dll:7666734E IsNLSDefinedString
76641CFB KERNELBASE.dll:76641CFB GetModuleFileNameW
74AF3362 kernel32.dll:74AF3362 RegKrnInitialize
Maybe if I can get it to compile in Visual Studio, I'll run in it dev and see what happens when I unplug the NIC.
M
November 14, 2012, 06:47:20 PM
Since eclipse uses fractional difficulties and it seems they are less reliable on JSON could you accept whatever eclipse says to work on then if it has a decimal truncate it to an int and add 1?
Eclipse is fully stratum/JSON/JSON-RPC compliant. Truncating it works, but you'll lose shares if you add a 1.In any case, GBT is still the preferred protocol.
November 14, 2012, 06:32:19 PM
Those with easy to reproduce crashes on windows can you try one of these quick hack builds please:
http://ck.kolivas.org/apps/cgminer/temp/
http://ck.kolivas.org/apps/cgminer/temp/
November 14, 2012, 05:25:52 PM
Cgminer is already working with vardiff on EMC, it's just a format thing, I think. I have a patch in cgminer that works fine with Eclipse and the changes are pretty minor.
EMC stratum support should be fixed in git now, with support for fractional diffs and diffs just below 1. I used some of Inaba's code ideas and expanded it to include support for the actual fractional diffs as well. 
November 14, 2012, 04:24:17 PM
Cgminer is already working with vardiff on EMC, it's just a format thing, I think. I have a patch in cgminer that works fine with Eclipse and the changes are pretty minor.
November 14, 2012, 04:17:58 PM
Since eclipse uses fractional difficulties and it seems they are less reliable on JSON could you accept whatever eclipse says to work on then if it has a decimal truncate it to an int and add 1?
Eclipse is fully stratum/JSON/JSON-RPC compliant. Truncating it works, but you'll lose shares if you add a 1.In any case, GBT is still the preferred protocol.
November 14, 2012, 03:24:04 PM
Since eclipse uses fractional difficulties and it seems they are less reliable on JSON could you accept whatever eclipse says to work on then if it has a decimal truncate it to an int and add 1?
I only ask because it seems like what you have done is best practices for getting reliable work. That would make less changes in the code so no need to internally change from int difficulties to decimal. Just it would require a change to the value on reciept.
You make the best miner. I would rather use CGMiner then stratum on eclipse so if it's not possible or doesn't get done I will still appreciate your miner.
Actually I see a good reason not to now that I think on it more. It seemed easy before but if I got a 1.9 difficulty I doubt 2 would be much worse but 1.09 would obviously cost me shares in theory. Unless the client could specify its difficulty it is working on that will cause an issue with not submitting work.
I only ask because it seems like what you have done is best practices for getting reliable work. That would make less changes in the code so no need to internally change from int difficulties to decimal. Just it would require a change to the value on reciept.
You make the best miner. I would rather use CGMiner then stratum on eclipse so if it's not possible or doesn't get done I will still appreciate your miner.
Actually I see a good reason not to now that I think on it more. It seemed easy before but if I got a 1.9 difficulty I doubt 2 would be much worse but 1.09 would obviously cost me shares in theory. Unless the client could specify its difficulty it is working on that will cause an issue with not submitting work.
November 14, 2012, 02:22:55 PM
LOL, does it look any better to you?
Nope, but I'm really not good with all this stuff.
November 14, 2012, 01:52:32 PM
I set it up, but it appears to take time for this problem to come up. I'll let you know when I have something.
Much appreciated! Occasionally nothing useful comes from the information, or windows decides to disable the debugger, but at least we have tried.Quote
cgminer-db.exe caused an Access Violation at location 63656e6e Reading from location 63656e6e.
Registers:
eax=00000000 ebx=070fd26c ecx=070fd72d edx=070fd72e esi=6f43203a edi=070fd77c
eip=772015de esp=070fd258 ebp=00002746 iopl=0 nv up ei pl zr na po nc
cs=0023 ss=002b ds=002b es=002b fs=0053 gs=002b efl=00000246
Call stack:
772015DE ntdll.dll:772015DE ZwRaiseException
771F014E ntdll.dll:771F014E KiUserExceptionDispatcher
070FD26C
070FD2BC
Registers:
eax=00000000 ebx=070fd26c ecx=070fd72d edx=070fd72e esi=6f43203a edi=070fd77c
eip=772015de esp=070fd258 ebp=00002746 iopl=0 nv up ei pl zr na po nc
cs=0023 ss=002b ds=002b es=002b fs=0053 gs=002b efl=00000246
Call stack:
772015DE ntdll.dll:772015DE ZwRaiseException
771F014E ntdll.dll:771F014E KiUserExceptionDispatcher
070FD26C
070FD2BC
I tried the version you reuploaded. Is that better?
Quote
cgminer-db.exe caused an Access Violation at location 63656e6e Reading from location 63656e6e.
Registers:
eax=00000000 ebx=0572d26c ecx=0572d72d edx=0572d72e esi=6f43203a edi=0572d77c
eip=777515de esp=0572d258 ebp=00002746 iopl=0 nv up ei pl zr na po nc
cs=0023 ss=002b ds=002b es=002b fs=0053 gs=002b efl=00000246
Call stack:
777515DE ntdll.dll:777515DE ZwRaiseException
7774014E ntdll.dll:7774014E KiUserExceptionDispatcher
0572D26C
0572D2BC
Registers:
eax=00000000 ebx=0572d26c ecx=0572d72d edx=0572d72e esi=6f43203a edi=0572d77c
eip=777515de esp=0572d258 ebp=00002746 iopl=0 nv up ei pl zr na po nc
cs=0023 ss=002b ds=002b es=002b fs=0053 gs=002b efl=00000246
Call stack:
777515DE ntdll.dll:777515DE ZwRaiseException
7774014E ntdll.dll:7774014E KiUserExceptionDispatcher
0572D26C
0572D2BC
LOL, does it look any better to you?
Yeah, this version is very unstable on my router. Got worse with the addition of GBT. Was segfaulting every day or two, now every few hours. From the logs, it seems to (mostly) crash after a new work thread is created.
November 14, 2012, 01:22:46 PM
I set it up, but it appears to take time for this problem to come up. I'll let you know when I have something.
Much appreciated! Occasionally nothing useful comes from the information, or windows decides to disable the debugger, but at least we have tried.Quote
cgminer-db.exe caused an Access Violation at location 63656e6e Reading from location 63656e6e.
Registers:
eax=00000000 ebx=070fd26c ecx=070fd72d edx=070fd72e esi=6f43203a edi=070fd77c
eip=772015de esp=070fd258 ebp=00002746 iopl=0 nv up ei pl zr na po nc
cs=0023 ss=002b ds=002b es=002b fs=0053 gs=002b efl=00000246
Call stack:
772015DE ntdll.dll:772015DE ZwRaiseException
771F014E ntdll.dll:771F014E KiUserExceptionDispatcher
070FD26C
070FD2BC
Registers:
eax=00000000 ebx=070fd26c ecx=070fd72d edx=070fd72e esi=6f43203a edi=070fd77c
eip=772015de esp=070fd258 ebp=00002746 iopl=0 nv up ei pl zr na po nc
cs=0023 ss=002b ds=002b es=002b fs=0053 gs=002b efl=00000246
Call stack:
772015DE ntdll.dll:772015DE ZwRaiseException
771F014E ntdll.dll:771F014E KiUserExceptionDispatcher
070FD26C
070FD2BC
I tried the version you reuploaded. Is that better?
Quote
cgminer-db.exe caused an Access Violation at location 63656e6e Reading from location 63656e6e.
Registers:
eax=00000000 ebx=0572d26c ecx=0572d72d edx=0572d72e esi=6f43203a edi=0572d77c
eip=777515de esp=0572d258 ebp=00002746 iopl=0 nv up ei pl zr na po nc
cs=0023 ss=002b ds=002b es=002b fs=0053 gs=002b efl=00000246
Call stack:
777515DE ntdll.dll:777515DE ZwRaiseException
7774014E ntdll.dll:7774014E KiUserExceptionDispatcher
0572D26C
0572D2BC
Registers:
eax=00000000 ebx=0572d26c ecx=0572d72d edx=0572d72e esi=6f43203a edi=0572d77c
eip=777515de esp=0572d258 ebp=00002746 iopl=0 nv up ei pl zr na po nc
cs=0023 ss=002b ds=002b es=002b fs=0053 gs=002b efl=00000246
Call stack:
777515DE ntdll.dll:777515DE ZwRaiseException
7774014E ntdll.dll:7774014E KiUserExceptionDispatcher
0572D26C
0572D2BC
November 14, 2012, 12:08:39 PM
Are there any issues with running multiple instances of cgminer as long as they are talking to different devices? Does cgminer hold any locks or open any pipes or temp files?
I do this without any problems as a way to point my BFL Singles to different pools than my GPUs.
November 14, 2012, 11:14:45 AM
Are there any issues with running multiple instances of cgminer as long as they are talking to different devices? Does cgminer hold any locks or open any pipes or temp files?
I do this without any problems as a way to point my BFL Singles to different pools than my GPUs.
Jump to: