[ANN][MTC] GPU Miner for Metiscoin - 35% FASTER on R9 280x

[girino]

do you have some  recommandation? i don't know  which one can  support to change multi-displaycard gpu-clock.

your driver should have come with such utility. You should look for it in the support forums for your card manufacturer...

[1715019767]

1715019767

[xx80mgopxx]

Getting

Code:

ERROR: -1001, ,No valid ICDs found
assertion "_MY_ERR_X == CL_SUCCESS" failed: file "xptMiner/OpenCLObjects.cpp line 159, funntion: OpenCLMain::OpenCLMain<>
0 [main] xptminer 1876 open_stackdumpfile: dumping stack trace to xptminer.

I really don't have any idea what I'm doing. Was I supposed to add everything from the two folders together into 1 folder then run? Because that's pretty much all I did, and also edited example file for my pools info. Thats it tho.

[girino]

Getting

Code:

ERROR: -1001, ,No valid ICDs found
assertion "_MY_ERR_X == CL_SUCCESS" failed: file "xptMiner/OpenCLObjects.cpp line 159, funntion: OpenCLMain::OpenCLMain<>
0 [main] xptminer 1876 open_stackdumpfile: dumping stack trace to xptminer.

I really don't have any idea what I'm doing. Was I supposed to add everything from the two folders together into 1 folder then run? Because that's pretty much all I did, and also edited example file for my pools info. Thats it tho.

Nope, you should not have messed with the folders! But the error does not seem to be that. Looks like it is unable to identify the vendor/model of your graphics card. Which card do you have? does it support opencl?

[girino]

NEW RELEASE, FASTER THAN EVER.

Hi all,

I just released a new build of this miner with several improvements, thanks to gigawatt. My current version is slightly faster than his, but this might depend on the machine you're running it. (Mostly because i copy AES tables from __constant memory space and he copies it from __global).

The binaries for windows, osx and linux are still available on the same old place:

https://www.dropbox.com/sh/2wv2awwgebpdo0m/STzOU5tlXa

Please run and enjoy.

[ziliao]

NEW RELEASE, FASTER THAN EVER.

Hi all,

I just released a new build of this miner with several improvements, thanks to gigawatt. My current version is slightly faster than his, but this might depend on the machine you're running it. (Mostly because i copy AES tables from __constant memory space and he copies it from __global).

The binaries for windows, osx and linux are still available on the same old place:

https://www.dropbox.com/sh/2wv2awwgebpdo0m/STzOU5tlXa

Please run and enjoy.

great job! i get about 48000k@6xGV-N660OC-2GD GTX660 2G DDR5,win7x64.but unfortunately，it only work a few minutes，then the miner error and quit,the computer crashed.gigawatt'sbuild get the same result。

[girino]

great job! i get about 48000k@6xGV-N660OC-2GD GTX660 2G DDR5,win7x64.but unfortunately，it only work a few minutes，then the miner error and quit,the computer crashed.gigawatt'sbuild get the same result。

Can you capture the screen or not even that? Maybe a picture?

[pavan@hosur]

Im using R9 280X but im getting 12900 kash only? i need to upgrade any things???

[girino]

Im using R9 280X but im getting 12900 kash only? i need to upgrade any things???

Well, the difference between this version and gigawatt's version is how we manage the lookup tables. Depending on how your card organizes its memory, his version will be faster. You can try his version here: https://bitcointalk.org/index.php?topic=452738

[~Money~]

Well, the difference between this version and gigawatt's version is how we manage the lookup tables. Depending on how your card organizes its memory, his version will be faster. You can try his version here: https://bitcointalk.org/index.php?topic=452738

on my 7870 it shows around 1.5% more hash rates but no significant changes in shares found

the only big difference i noticed was around 20% decrease in power consumption !!!

wot ever u did, it is a great work girino 

i am replacing all miners with yours in next few hours 

[girino]

Well, the difference between this version and gigawatt's version is how we manage the lookup tables. Depending on how your card organizes its memory, his version will be faster. You can try his version here: https://bitcointalk.org/index.php?topic=452738

on my 7870 it shows around 1.5% more hash rates but no significant changes in shares found

the only big difference i noticed was around 20% decrease in power consumption !!!

wot ever u did, it is a great work girino 

i am replacing all miners with yours in next few hours 

Probably copying from constant memory to local memory spends less energy than copying from global to local. The only difference, as I said, is that I use constant memory instead of global for the lookup tables.

[breakbeater]

can i use this miner behind a http proxy?

[girino]

can i use this miner behind a http proxy?

Not currently. I use and ssh client to an external machine in order to bypass firewall and proxy. (putty can work over http proxies)

[girino]

NEW VERSION RELEASED - Build 8

This time my focus was on networking, stability and 32 bits.

The networking code from ypool's latest xptminer was merged back. Several changes and tests were made to ensure stability (no more "0kh/s" after some hours, no more closing on network disconnect or server down).

Faster than ever, and two new hidden options: -s to set the step size (in powers of 2, defaults to 2^20), and -a to set variations on algorithm. Currently supports "-a 1" for lookup tables in constant memory and "-a 2" (default) for tables in global memory. -a 2 is faster in most systems.

The code for metis was reverted to a previous version since it is 2x slower on some NVidia GPUs. Binaries and code available, as always, on the links above.

Source code is available at:

https://github.com/girino/xptMiner

and Binaries (for windows only at present, will add linux later tonight)

https://www.dropbox.com/sh/2wv2awwgebpdo0m/STzOU5tlXa

Enjoy.

[Replex]

How do I set the step size( -s)?
How high can I go etc?

Also, great work! My GTX 660ti went from 21000 k/hash to 24000 k/hash so far!
With -a 2 it ran at about 20000 k/hash. With -a 1 it went 4000 k/hash higher!
Want to tweak the step size to see if I alter it even more!

[girino]

How do I set the step size( -s)?
How high can I go etc?

Also, great work! My GTX 660ti went from 21000 k/hash to 24000 k/hash so far!
With -a 2 it ran at about 20000 k/hash. With -a 1 it went 4000 k/hash higher!
Want to tweak the step size to see if I alter it even more!

Steps range from 15 to 27. (they are powers of 2, so 15 means that each step processes 2^15 hashes = 32K. The default is 19, which means  512K hashes in each step. I think your card would go best with -s 22 or higher).

[girino]

NEW RELEASE, 35% faster on R9 280x, SOLVED LOST CONNECTION PROBLEM WHEN YPOOL DOWN!!!

Build 9 tries to maximize use of local memory. Really fast on GPUs with fast local memory and lots of it (newer AMD cards).

If this version is slower in your GPU (NVidia or older cards), use option "-a 2" on the command line to revert to the old OpenCL code.

Also, no more freezing on connection lost. (useful on those times of ypool instability). Techinacl details bellow.

EVERYBODY SHOULD UPGRADE TO AVOID FREEZING!!! Very imoprtant now that ypool is being constantly DDoSed...

For those more into the techincal issues, the changes are:
1- use of local memory instead of global memory to pass hashes between the several hashing functions (keccak -> shavite -> fugue). This improves performance on machines where local memory is at least 64 * work_group_size + 8K.
2- Linux (and cygwin, for the matter) do not set an error on socket disconnect when calling recv. It just returns a 0 size. Since the original (windows code) was expecting that an error is sent on those cases, the linux code just kept trying forever to receive data in a closed connection.  So now, when checking for errors, it is also checking if return size is 0 to mark connection as lost. This solves the "freezing disconnect" bug that was haunting all of us.

Links for download are the same as always, source code is available at:

https://github.com/girino/xptMiner

and Binaries (linux 32 and 64 bits, windows 64 bits and OSX):

https://www.dropbox.com/sh/2wv2awwgebpdo0m/STzOU5tlXa

Enjoy!

[ziliao]

great job! i get about 48000k@6xGV-N660OC-2GD GTX660 2G DDR5,win7x64.but unfortunately，it only work a few minutes，then the miner error and quit,the computer crashed.gigawatt'sbuild get the same result。

Can you capture the screen or not even that? Maybe a picture?

yes,i copy it:
ERROR: -5, CL_OUT_OF_RESOURCES, if there is a failure to allocate resources requ
ired by the OpenCL implementation on the device.
assertion "_MY_ERR_X == CL_SUCCESS" failed: file "xptMiner/OpenCLObjects.cpp", l
ine 454, function: void OpenCLCommandQueue::enqueueReadBuffer(OpenCLBuffer*, voi
d*, size_t)

[girino]

great job! i get about 48000k@6xGV-N660OC-2GD GTX660 2G DDR5,win7x64.but unfortunately，it only work a few minutes，then the miner error and quit,the computer crashed.gigawatt'sbuild get the same result。

Can you capture the screen or not even that? Maybe a picture?

yes,i copy it:
ERROR: -5, CL_OUT_OF_RESOURCES, if there is a failure to allocate resources requ
ired by the OpenCL implementation on the device.
assertion "_MY_ERR_X == CL_SUCCESS" failed: file "xptMiner/OpenCLObjects.cpp", l
ine 454, function: void OpenCLCommandQueue::enqueueReadBuffer(OpenCLBuffer*, voi
d*, size_t)

Did you try the latest versions? Are you running another miner on the sabe GPU? This means the driver could not allocate enough memory. But memory is allocated only on startup. so if it runs for sometime, i can't see what's wrong.

[girino]

NEW POOL -- BETA TESTERS WANTED

For the eager ones: http://pool.girino.org/
Miners can be downloaded here: http://girino.org/metisminer/

Hi all,

As you know, whenever ypool is down (and it happened a lot last week) we are left without any way to mine (and the network stops confirming transactions, since no mining is done). So, to address this problem i developed this new pool, based on pushpool and MPOS, and i am trying to test it. But alone, it's like solo mining: completely useless. So please, pretty please, tho who can spare some of your time and miners, please mine in this new pool.

It uses a different protocol from ypool (getwork, while ypool uses x-pushtrough, that AFAIK is only used by them and was probably developed by them). So you will need to download a new miner.

IT IS STILL EXPERIMENTAL, I NEED BETA TESTERS SO WE CAN BREAK OUR FIRST BLOCK AND SEE IF EVERYTHING WORKS!

Please report any problems you find, send me any advice you have about running a pool, and anything you want, either at pool@girino.org or by PM here.

If everything works great, I'll be launching it officially and asking the dev to include it in the coin's page...

Thanks to all!

And please, test, test and retest! We need to make it work!

[breakbeater]

Great news, just pointed my "hamsters" at your pool.
The new version of miner does not support -d 0,1 flag? i must start 2 instances of minerd for every GPU.