@bolapara: did you use the -D option?
We just tried hashkill on 5970 (which has the same cores as 5870) and speed was as expected: 620-630M/s at stock speeds. That is with -D applied
@mskwik The stale shares could indicate a problem (e.g bad BFI_INT replacement). What GPU do you have?
As a side note, I got ADL working and I can now correctly get GPU temperatures, activity percent and clocks. Wondering how to proceed with that thermal stuff: should I quit when a threshold is reached....or probably pause for a certain period....or completely disabling that GPU? Hmm...
Topic: hashkill - testing bitcoin miner plugin - page 19. (Read 90966 times)
yeah i also forgot to mention it failed for me after a few hours as well, i've reverted to the previous version, there was no communication problem it just stated 0/mash.
Tried it out overnight on deepbit and while it was faster it also ended up with roughly 10% stale shares. Not sure the long polling is working properly, is it supposed to give any indication when it gets a new block notification?
Thanks for that bolapara will give it a go.
i guess you removed the nvidia dependency or compiled it manually for the python pyopencl that is, i will have a go myself tonight.
If i can get phoenix working i will compare side by side
i guess you removed the nvidia dependency or compiled it manually for the python pyopencl that is, i will have a go myself tonight.
If i can get phoenix working i will compare side by side
PS @bolapara how did you get 434 on the 5870, i have one clocked at 850, what is yours at? please share.
Dedicated mining rig
Ubuntu 10.10
11.4 drivers, 2.1 SDK
1000 core, 300 mem clocks
100% fan speed
76C
phoenix 1.4 - VECTORS AGGRESSION=12 WORKSIZE=128 BFI_INT
I tested my 5870 just now
fresh install of ubu 11.04 x64, catalyst 11.4, sdk 2.4 = 370 mhash on hashkill.
i tried to install phoenix but trashed my system with that awful python-opencl dependency on nvidia driver, why oh why did they package it like that.
I will try to test some more tomorrow against phoenix after i re-install.
fresh install of ubu 11.04 x64, catalyst 11.4, sdk 2.4 = 370 mhash on hashkill.
i tried to install phoenix but trashed my system with that awful python-opencl dependency on nvidia driver, why oh why did they package it like that.
I will try to test some more tomorrow against phoenix after i re-install.
Apparently, on 5870 for some reason the generated binary is not optimal. If someone with such card (or 5970) is willing to help me test and fix that, please PM me.
Last version ran solid for me 3 days straight 24/7.
Tried the additional options normally my card settles on 283 after while, with the D & G options i can get it to settle at 285.
PS @bolapara how did you get 434 on the 5870, i have one clocked at 850, what is yours at? please share.
Tried the additional options normally my card settles on 283 after while, with the D & G options i can get it to settle at 285.
PS @bolapara how did you get 434 on the 5870, i have one clocked at 850, what is yours at? please share.
Damn...still that 5870 issue...hmmm wanna have one for tests 
Tried it out and you are about 90MH/s slower than phoenix 1.4 for me.
ubuntu 10.10 x64 - 5870
hashkill - SDK 2.3 - ~344MH/s
phoenix 1.4 - SDK 2.1 - ~434MH/s
ubuntu 10.10 x64 - 5870
hashkill - SDK 2.3 - ~344MH/s
phoenix 1.4 - SDK 2.1 - ~434MH/s
Fixed a couple of bugs:
* Progress indicator finally fixed
* Kernel reworked - there are separate codepaths, one for VLIW5 (interlaced uint2+uint to get best utilization) and another for VLIW4 architectures. Additional optimizations implemented.
* Added -D command-line option. This tends to increase speed at the cost of reduced desktop responsiveness (kinda like Phoenix AGGRESSION parameter)
* Additional marginal speedup can be achieved by using -G 3 option at the command line (or even -G4) - but that requires more memory and faster, multicore CPU
* the curl handles leak was fixed - no more "connection failed after half an hour of work" issues.
The code changes are confirmed to be incompatible with ATI Stream SDK 2.1 and 2.2. Please _DO NOT_ use older than 2.3 versions.
Not implemented yet:
* ADL thermal monitoring for ATI
* Failover extension (used in deepbit.net)
Download:
64-bit:
http://www.gat3way.eu/poc/hashkill-0.2.4-x86_64.tgz
32-bit:
http://www.gat3way.eu/poc/hashkill-0.2.4-x86.tgz
* Progress indicator finally fixed
* Kernel reworked - there are separate codepaths, one for VLIW5 (interlaced uint2+uint to get best utilization) and another for VLIW4 architectures. Additional optimizations implemented.
* Added -D command-line option. This tends to increase speed at the cost of reduced desktop responsiveness (kinda like Phoenix AGGRESSION parameter)
* Additional marginal speedup can be achieved by using -G 3 option at the command line (or even -G4) - but that requires more memory and faster, multicore CPU
* the curl handles leak was fixed - no more "connection failed after half an hour of work" issues.
The code changes are confirmed to be incompatible with ATI Stream SDK 2.1 and 2.2. Please _DO NOT_ use older than 2.3 versions.
Not implemented yet:
* ADL thermal monitoring for ATI
* Failover extension (used in deepbit.net)
Download:
64-bit:
http://www.gat3way.eu/poc/hashkill-0.2.4-x86_64.tgz
32-bit:
http://www.gat3way.eu/poc/hashkill-0.2.4-x86.tgz
^^ This means you don't have the SDK installed or you haven't done export LD_LIBRARY_PATH=/path/to/sdk/lib/ prior to running it.
You should also make sure the OpenCL runtime detects your GPU - by running the CLInfo sample from the SDK.
You should also make sure the OpenCL runtime detects your GPU - by running the CLInfo sample from the SDK.
When i run this: hashkill-cpu -p bitcoin xxxxxxxx:xxxxxxx:bitcoinpool.com:8334
[hashkill] Progress indicator will be available once Markov calculations are done...
[error] (bitcoin.c:86) This plugin is GPU only!
I get that error..
[hashkill] Progress indicator will be available once Markov calculations are done...
[error] (bitcoin.c:86) This plugin is GPU only!
I get that error..
Just wait, people, there are still lots of bugs I am working on 
A new release will be done in a couple of days, hopefully fixing them all. The reconnect issue is due to missing deinitialization of a curl handle and this will definitely be resolved. We still have problems with 6990 and this afternoon I had to rewrite the whole kernel (replacing uint4 with interlaced uint2+uint) to get that GPR thing working reliable on all VLIW5 cards.
A new release will be done in a couple of days, hopefully fixing them all. The reconnect issue is due to missing deinitialization of a curl handle and this will definitely be resolved. We still have problems with 6990 and this afternoon I had to rewrite the whole kernel (replacing uint4 with interlaced uint2+uint) to get that GPR thing working reliable on all VLIW5 cards. Works great @gat3way thank you for this, it improved my performance greatly from 220 mhash on Diablo to 267 mhash on your code! i did try your code from a few days ago same version number though 
and after a few hours it would loose the connection and just keep retrying, restart solved the problem, lets see if this new one lasts longer. PS on a 6870, just over clocked to 950 and now at 286 Mhash... Great Stuff!
PPS post your bitcoin address so people can make donations.
for those curious on app SDK 2.4, using this driver http://www2.ati.com/drivers/linux/ati-driver-installer-11-4-x86.x86_64.run
@gat3way - will give this a try again...
Those of us with 5970's tend to find 2.1 optimal (i.e., max Hash generation).
Made it further than last time!
Those of us with 5970's tend to find 2.1 optimal (i.e., max Hash generation).
Code:
[hashkill] Version 0.2.4
[hashkill] Plugin 'bitcoin' loaded successfully
[hashkill] Found GPU device: Advanced Micro Devices, Inc. - Cypress
[hashkill] Found GPU device: Advanced Micro Devices, Inc. - Cypress
[hashkill] Found GPU device: Advanced Micro Devices, Inc. - Cypress
[hashkill] Found GPU device: Advanced Micro Devices, Inc. - Cypress
[hashkill] This plugin supports GPU acceleration.
[hashkill] Initialized hash indexes
[hashkill] Initialized thread mutexes
[hashkill] Spawned worker threads
[hashkill] Successfully connected and authorized at deepbit.net:8332
[hashkill] Compiling OpenCL kernel source (amd_bitcoin.cl)[error] (ocl_bitcoin.c:923) clBuildProgram error (-11)
[hashkill] Attack took 4 seconds.
[hashkill] Bye bye :)
[hashkill] Plugin 'bitcoin' loaded successfully
[hashkill] Found GPU device: Advanced Micro Devices, Inc. - Cypress
[hashkill] Found GPU device: Advanced Micro Devices, Inc. - Cypress
[hashkill] Found GPU device: Advanced Micro Devices, Inc. - Cypress
[hashkill] Found GPU device: Advanced Micro Devices, Inc. - Cypress
[hashkill] This plugin supports GPU acceleration.
[hashkill] Initialized hash indexes
[hashkill] Initialized thread mutexes
[hashkill] Spawned worker threads
[hashkill] Successfully connected and authorized at deepbit.net:8332
[hashkill] Compiling OpenCL kernel source (amd_bitcoin.cl)[error] (ocl_bitcoin.c:923) clBuildProgram error (-11)
[hashkill] Attack took 4 seconds.
[hashkill] Bye bye :)
Made it further than last time!
Hello,
Yes, that's one of the bugs I have collected thanks to people that tested the alpha (related to an integer overflow). Another one found is related to missing deinitialization of certain curl handles that creates big problems after some time spent in mining. Another problem was related to improper BFI_INT replacement on 69xx cards (fixed now). Finally, the 69xx codepath is not optimal and I am now currently working on a separate vliw4 codepath that is best optimized for 69xx devices. Sorry for those, but your input was very helpful for me to identify and fix those issues. A new testing release will be ready soon with those problems resolved.
Another thing is that we're walking on the verge with those uint4 vectors...on my 6870 I'm getting 41 GPR usage currently. If that rises to 42 for some reason, performance degrades disastrously as the number of wavefronts/cu drops. I still need to find a way to reduce the GPR usage - cause on some other cards, the compiler is unable to generate code that keeps to 41GPRs thus generating slow-performing code. Since I am doing that by carefully reordering stuff, it's a bit wacky and not reliable at the moment...still need some work on that.
Yes, that's one of the bugs I have collected thanks to people that tested the alpha (related to an integer overflow). Another one found is related to missing deinitialization of certain curl handles that creates big problems after some time spent in mining. Another problem was related to improper BFI_INT replacement on 69xx cards (fixed now). Finally, the 69xx codepath is not optimal and I am now currently working on a separate vliw4 codepath that is best optimized for 69xx devices. Sorry for those, but your input was very helpful for me to identify and fix those issues. A new testing release will be ready soon with those problems resolved.
Another thing is that we're walking on the verge with those uint4 vectors...on my 6870 I'm getting 41 GPR usage currently. If that rises to 42 for some reason, performance degrades disastrously as the number of wavefronts/cu drops. I still need to find a way to reduce the GPR usage - cause on some other cards, the compiler is unable to generate code that keeps to 41GPRs thus generating slow-performing code. Since I am doing that by carefully reordering stuff, it's a bit wacky and not reliable at the moment...still need some work on that.
Hmm...perhaps an ISA dump would be useful to debug the problem.
You can do that by running export GPU_DUMP_DEVICE_KERNEL=3 prior to running hashkill (you need to be in a writable directory like e.g /tmp).
Then after say 30 seconds, stop execution (ctrl-c) and look for a file named bitcoin_.isa (e.g bitcoin_Cypress.isa). Please paste this file contents so that I have a look at it.
You can do that by running export GPU_DUMP_DEVICE_KERNEL=3 prior to running hashkill (you need to be in a writable directory like e.g /tmp).
Then after say 30 seconds, stop execution (ctrl-c) and look for a file named bitcoin_
I'm not the other guy, but I'm seeing some weird results. I can't tell *what* performance I'm getting out of your client -- it's showing me very strange results:
Code:
crunch@crunch:/tmp$ hashkill-gpu -p bitcoin xxx:xxx:deepbit.net:8332
[hashkill] Version 0.2.4
[hashkill] Plugin 'bitcoin' loaded successfully
[hashkill] Found GPU device: Advanced Micro Devices, Inc. - Juniper
[hashkill] Found GPU device: Advanced Micro Devices, Inc. - Cypress
[hashkill] Found GPU device: Advanced Micro Devices, Inc. - Cypress
[hashkill] This plugin supports GPU acceleration.
[hashkill] Initialized hash indexes
[hashkill] Initialized thread mutexes
[hashkill] Spawned worker threads
[hashkill] Successfully connected and authorized at deepbit.net:8332
[hashkill] Compiling OpenCL kernel source (amd_bitcoin.cl)
[hashkill] Binary size: 452144
[hashkill] Doing BFI_INT magic...
Mining statistics...
Speed: 0 MHash/sec [cur: 100%] [proc: 18] [subm: 14] [stale: 0] [eff: 77%] 82%]
Speed: 402 MHash/sec [cur: 28%] [proc: 18] [subm: 16] [stale: 0] [eff: 88%] 82%]
Speed: 236 MHash/sec [cur: 100%] [proc: 19] [subm: 16] [stale: 0] [eff: 84%] 88%]
Speed: 0 MHash/sec [cur: 100%] [proc: 19] [subm: 16] [stale: 0] [eff: 84%] 88%]
Speed: 6148914690576 MHash/sec [cur: 53%] [proc: 19] [subm: 16] [stale: 0] [eff: 84%]
(etc)
[hashkill] Version 0.2.4
[hashkill] Plugin 'bitcoin' loaded successfully
[hashkill] Found GPU device: Advanced Micro Devices, Inc. - Juniper
[hashkill] Found GPU device: Advanced Micro Devices, Inc. - Cypress
[hashkill] Found GPU device: Advanced Micro Devices, Inc. - Cypress
[hashkill] This plugin supports GPU acceleration.
[hashkill] Initialized hash indexes
[hashkill] Initialized thread mutexes
[hashkill] Spawned worker threads
[hashkill] Successfully connected and authorized at deepbit.net:8332
[hashkill] Compiling OpenCL kernel source (amd_bitcoin.cl)
[hashkill] Binary size: 452144
[hashkill] Doing BFI_INT magic...
Mining statistics...
Speed: 0 MHash/sec [cur: 100%] [proc: 18] [subm: 14] [stale: 0] [eff: 77%] 82%]
Speed: 402 MHash/sec [cur: 28%] [proc: 18] [subm: 16] [stale: 0] [eff: 88%] 82%]
Speed: 236 MHash/sec [cur: 100%] [proc: 19] [subm: 16] [stale: 0] [eff: 84%] 88%]
Speed: 0 MHash/sec [cur: 100%] [proc: 19] [subm: 16] [stale: 0] [eff: 84%] 88%]
Speed: 6148914690576 MHash/sec [cur: 53%] [proc: 19] [subm: 16] [stale: 0] [eff: 84%]
(etc)
With 2 5850s and a 5770, I expect to get about 600-700 MHash/sec. Here are the dumped ISA files:
http://dl.dropbox.com/u/694931/bitcoin_Cypress.isa
http://dl.dropbox.com/u/694931/bitcoin_Juniper.isa
(this is with AMD-APP-SDK-v2.4-lnx64)
Nice work, btw, I like the way it automagically finds all the cards and "deals with it", rather than having to run multiple copies of poclbm.
Hmm...perhaps an ISA dump would be useful to debug the problem.
You can do that by running export GPU_DUMP_DEVICE_KERNEL=3 prior to running hashkill (you need to be in a writable directory like e.g /tmp).
Then after say 30 seconds, stop execution (ctrl-c) and look for a file named bitcoin_.isa (e.g bitcoin_Cypress.isa). Please paste this file contents so that I have a look at it.
P.S. you would need ~ 5-10 seconds until speed peaks at maximum, it usually starts at lower speed and gradually increases. As for switches, you might try -G 3 and/or -D and see if it affects performance positively.
P.S 2: also please do not run the 32-bit version on a 64-bit system: it tends to be way slower. And (again) use SDK 2.3 or newer.
You can do that by running export GPU_DUMP_DEVICE_KERNEL=3 prior to running hashkill (you need to be in a writable directory like e.g /tmp).
Then after say 30 seconds, stop execution (ctrl-c) and look for a file named bitcoin_
P.S. you would need ~ 5-10 seconds until speed peaks at maximum, it usually starts at lower speed and gradually increases. As for switches, you might try -G 3 and/or -D and see if it affects performance positively.
P.S 2: also please do not run the 32-bit version on a 64-bit system: it tends to be way slower. And (again) use SDK 2.3 or newer.
I get 312 Mhash/s on both poclbm and phoenix, but only 208 with hashkill. Any useful information I can grab for you, command line switches to try, etc?
hashkill 0.2.4 x86_64, fglrx 8.801, LD_LIBRARY_PATH points to 2.3
hashkill 0.2.4 x86_64, fglrx 8.801, LD_LIBRARY_PATH points to 2.3
Jump to: