Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.


Messages - rots

Pages: [1]
1
BitShares PTS / Re: fast AMD OpenCL PTS miner released
« on: February 16, 2014, 11:40:55 am »
win7 64 amd13.12
280x 3100, but 290 only 2800?

I released a fast OpenCL PTS-miner for AMD GPUs. The performance is 3100+ c/m on a R9 280X and 1300+ c/m on a 7870, so that this seems to be the fastest released miner. Binaries are available for Windows and 64bit-Linux.
DOWNLOAD v0.1.x: https://www.dropbox.com/sh/d5qvec5kxlug1bm/FR3Y_78Eag connects to the beeeeer.org pool, which has 2.5% pool fees. The developer fees for this version are 7/367 ~ 1.9%.
DOWNLOAD v0.2.x: https://www.dropbox.com/sh/1ioh00043hqbsbk/wTptW3dr22 connects to the ypool.net pool, which has 5% pool fees. The developer fees for this version are 2.5% to make smaller pools more attractive. The developer fees are automatically transferred by ypool.net to the developer so the user's miner does not need to reconnect to support the developer.

UPDATE 10: version v0.2 of the miner is released, which supports the pool on ypool.net. If you want to connect to beeeeer.org you should use v.0.1.2. The miner does NOT support the 1GH-pool yet because the pool does not implement the latest xpt-protocol but 1GH-pool support is still planned. EDIT There was a typo in the README. The usage of v0.2 is <exe-name> -u <username>.<worker> -p <pass> -t <GPU DeviceID list>

UPDATE 12: I released a new version with large performance improvements. The syntax of the command line options of v0.1.x has changed and one has to specify the device IDs like in v0.2.x (see README). Furthermore, the command line option -a was introduced to select an algorithm. I guess that -a 0 or -a 2 should be fastest on R9 290(X) and -a 1 or -a 3 should be best for non-R9 290(X) cards.

known issues: The miner does not run if the installed Catalyst driver is too old. Multi-GPU support works for most users but sometimes the miner won't work if there are different cards installed. Then you have to run an instance of the miner on each card. The miner is not optimized to run on 5000 and 6000 series GPUs and needs about 1.2GB free GPU RAM.

NOTE 1: When you post errors I appreciate some information about the operating system (Windows or Linux / 32bit or 64bit is sufficient) and the used Catalyst driver. Furthermore you should suppress the section between the stars of the output of the miner excluding the miner version (at the moment it hasn't any relevance but it will become important in the future).

NOTE 2: Please make sure that you are not using an outdated Catalyst version and read the README.


I would be grateful for donations to speed up the development of a version for Nvidia GPUs and 1GH-pool support:
  PTS:  Pk3mHjZrW3HGmx5jMNaN1GhXT2WgXHjRCz
  BTC:  1AumJ5uzz1nuER7pBA6Bh4gNusaxhN85rc

2
BitShares PTS / Re: fast AMD OpenCL PTS miner released
« on: February 10, 2014, 02:42:58 pm »
My card:
270x 1450   -a 1
280x 2450   -a 1
290 2600?   -a 1(-a others is slower)

all cards not overclocked,seems 290 too low, is there some special setting need for 290?

I released a new version with large performance improvements. The syntax of the command line options of v0.1.x has changed and one has to specify the device IDs like in v0.2.x (see README). Furthermore, the command line option -a was introduced to select an algorithm. I guess that -a 0 or -a 2 should be fastest on R9 290(X) and -a 1 or -a 3 should be best for non-R9 290(X) cards.

3
BitShares PTS / Re: Open source optimized PTS CPU miner (BETA)
« on: February 01, 2014, 01:06:22 pm »
Any plan to release windows bineary file?

Following up on yvg1900's release of yam, I figured I'd improve the state of the art of the open source versions a bit:

https://github.com/dave-andersen/ptsminer

I haven't made it build yet on windows (it just needs to compile the avx2 assembly code - should be straightforward if someone wants to clue me in on how to appropriately invoke gcc there), but it should work on other platforms.  As a warning, I've only really tried it on avx2, since I'm a fan of Haswell.  THIS SOFTWARE SHOULD BE CONSIDERED A BETA QUALITY RELEASE.  At best. 

As with my GPU release, this one is based very directly on ptsminer, so it's tied to beeeeer for the moment.  I plan to fix that and let it be used with other pools in the near future, but that's going to take some dev work.  sigh.

There's a lot of optimization to be done, but this gets the basics as far as memory subsystem optimization, and bridges a lot of the gap between the old OSS version and yam M7i.  I haven't tried out M7j, mind you -- it's probably a bit faster still, but this release should bridge the gap considerably.

It incorporates the same optional, extendible 1% dev fee that the gpu miner does.  Prior ptsminer devs, if you feel like you should be in the list, please PM me and I'll get you added!

With gratitude to FreeTrade for the donation that kept me interested in hacking on and releasing this stuff, and to yvg1900 for some very engaging unofficial competition. *grin*

  -Dave

4
BitShares PTS / Re: fast AMD OpenCL PTS miner released
« on: January 31, 2014, 07:57:44 am »
@visterln have you over-locked? I can only get 1850 cpm on my R280x

Share found! (Blockheight: 46660)
Share found! (Blockheight: 46660)
Share found! (Blockheight: 46660)
collisions/min: 10189.2424 Shares total: 375
Share found! (Blockheight: 46660)
Share found! (Blockheight: 46660)
Share found! (Blockheight: 46660)
collisions/min: 10189.2000 Shares total: 378
Share found! (Blockheight: 46660)
Share found! (Blockheight: 46660)
Share found! (Blockheight: 46660)
Share found! (Blockheight: 46660)
collisions/min: 10183.2178 Shares total: 382
Share found! (Blockheight: 46660)
Share found! (Blockheight: 46660)
Share found! (Blockheight: 46660)
Share found! (Blockheight: 46660)
Share found! (Blockheight: 46660)
Share found! (Blockheight: 46660)
Share found! (Blockheight: 46660)
Share found! (Blockheight: 46660)
collisions/min: 10188.6765 Shares total: 390
Share found! (Blockheight: 46660)

1020 watts 5*280X
30% FAN
gpu load 99%
temp 62°C
mem used 1124M

5
BitShares PTS / Re: fast AMD OpenCL PTS miner released
« on: January 29, 2014, 02:36:52 pm »
No, ip is fixed. I restart miner can connect.

This means that the miner isn't able to connect to the server. That's a different problem. It could be a boost related problem, i.e. boost::asio::connect somehow fails. Did your IP change before that happens? I'm trying to fix it

6
BitShares PTS / Re: fast AMD OpenCL PTS miner released
« on: January 29, 2014, 01:27:31 pm »
with this version, after 2hours run, I still got
6 (0.42%), ST: 0 (0.00%)
failed to connect (system:10061). sleep 10 seconds until retry
[STATS] 2014-Jan-29 21:22:49 | 500.43 c/m | 0.02 sh/m | VL: 6223 (99.58%), RJ: 2
6 (0.42%), ST: 0 (0.00%)
failed to connect (system:10061). sleep 10 seconds until retry
[STATS] 2014-Jan-29 21:22:59 | 499.08 c/m | 0.02 sh/m | VL: 6223 (99.58%), RJ: 2
6 (0.42%), ST: 0 (0.00%)
[STATS] 2014-Jan-29 21:23:09 | 497.74 c/m | 0.02 sh/m | VL: 6223 (99.58%), RJ: 2
6 (0.42%), ST: 0 (0.00%)
failed to connect (system:10061). sleep 10 seconds until retry
[STATS] 2014-Jan-29 21:23:19 | 496.41 c/m | 0.02 sh/m | VL: 6223 (99.58%), RJ: 2
6 (0.42%), ST: 0 (0.00%)
failed to connect (system:10061). sleep 10 seconds until retry

It retry repeatedly, restart miner can work again


I updated the miner to version 0.1.1, which should fix with the "doing no work" bug in Windows.

@darkbane: the miner used a corrupted block with a wrong share target. therefore it is very likely that the miner will find zero shares in a reasonable time because it has a very difficult share target. it looked like as some buffers got corrupted. most likely windows missed some resources because i used many asynchronous events but v0.1.1 implements better code.

7
Suggest you give version number on zip file. Otherwise no one knows which version
is using.

We had a huge number of getwork connections that started to cause other connections to be closed. This is mitigated now, but we recommend to upgrade getwork miner to anyone who still runs it.

8
BitShares PTS / Re: fast AMD OpenCL PTS miner released
« on: January 28, 2014, 02:44:52 pm »
the mess on the startup are the messages of the compiler. i can suppress those messages but it makes debugging harder.

Quote
4th, If there's no network connection while starting, it will crash.
a miner without network connection makes no sense... but perhaps i can quit the miner more silently

sometimes when computer startup, it need time to re-acquire IP from DHCP server, but if miner startup at this time, it will crash. It's better to try to reconnect after a while, instead of crash.


Quote
[MASTER] 2014-Jan-28 21:43:00 | work received | sharetarget: 0000000100000000010
000000001000000000100000000010000000001000000
wow, that sounds bad. is your connection ok? Linux or Windows?

Win7 64bit, maybe connection not stable, like last issue, no one can gurrant connection 100% stable, it's better to handle it more robust.

Quote
second, I have 2 280x, it can't start 2 card in one instance, must run 2 instance
did you try to run the miner with <name of executable> <your address> 0 1?
if i use  Miner address 0 1, both cards not work. I must use Miner address 0, and Miner address 1 , seperately.

9
BitShares PTS / Re: fast AMD OpenCL PTS miner released
« on: January 28, 2014, 01:57:06 pm »
first, after startup, it shows
LOOP UNROLL: pragma unroll (line 160)
    Unrolled as requested!
LOOP UNROLL: pragma unroll (line 112)
    Unrolled as requested!
...............
but can still work

second, I have 2 280x, it can't start 2 card in one instance, must run 2 instance.

third, after a while, maybe minutes or hours, it seems meet some abnormal sharetarget, and stop find share anymore(from gpu monitor, gpu sitll 100% load)

0[WORKER0] share found: 16819923 <-> 25205019 #4 (1542) @ 24
[WORKER0] share found: 18873581 <-> 42439404 #5 (1542) @ 24
[WORKER0] share found: 46607814 <-> 26833283 #6 (1546) @ 26
[WORKER0] share found: 33826928 <-> 19127522 #7 (1560) @ 32
[WORKER0] share found: 21017147 <-> 36827130 #8 (1560) @ 32
[WORKER0] share found: 64941930 <-> 11055332 #9 (1562) @ 33
[MASTER] 2014-Jan-28 21:43:00 | work received | sharetarget: 0000000100000000010
000000001000000000100000000010000000001000000

4th, If there's no network connection while starting, it will crash.


Quote
Could just be the nature of PTS mining but Ubuntu build netted me 8 blocks to 1 in Windows version.  Is the 35 seconds mining for you getting messed up and all blocks go to developer?  Still getting this on start up in Windows.

No, that is because of bad pool luck while you were mining in windows. beeeeer.org is a small pool with only a few generated blocks per day (mostly less than 10), but is has low fees (2.5%). Your windows output seems normal.


@rots: did you encounter any problems or does the miner work well?

10
BitShares PTS / Re: fast AMD OpenCL PTS miner released
« on: January 28, 2014, 09:17:12 am »
same to you
win7 64bit 270x

latest build getting this on launch

Messages during compilation of program for device 'Pitcairn':

LOOP UNROLL: pragma unroll (line 160)
    Unrolled as requested!
LOOP UNROLL: pragma unroll (line 112)
    Unrolled as requested!
LOOP UNROLL: pragma unroll (line 174)
    Unrolled as requested!
LOOP UNROLL: pragma unroll (line 167)
    Unrolled as requested!
LOOP UNROLL: pragma unroll (line 221)
    Unrolled as requested!
LOOP UNROLL: pragma unroll (line 284)
    Unrolled as requested!
LOOP UNROLL: pragma unroll (line 112)
    Unrolled as requested!
LOOP UNROLL: pragma unroll (line 291)
    Unrolled as requested!
LOOP UNROLL: pragma unroll (line 112)
    Unrolled as requested!

Messages during compilation of program for device 'Hawaii':

LOOP UNROLL: pragma unroll (line 160)
    Unrolled as requested!
LOOP UNROLL: pragma unroll (line 112)
    Unrolled as requested!
LOOP UNROLL: pragma unroll (line 174)
    Unrolled as requested!
LOOP UNROLL: pragma unroll (line 167)
    Unrolled as requested!
LOOP UNROLL: pragma unroll (line 221)
    Unrolled as requested!
LOOP UNROLL: pragma unroll (line 284)
    Unrolled as requested!
LOOP UNROLL: pragma unroll (line 112)
    Unrolled as requested!
LOOP UNROLL: pragma unroll (line 291)
    Unrolled as requested!
LOOP UNROLL: pragma unroll (line 112)
    Unrolled as requested!

works each on it's own instance - will not work as devices "0 1" but 2 separate instances works.

2332c/m on 290X
1150c/m on 270X

4 instances running fine - 2 on 290X (1900/1214) 2 on 270X (970/568)

11
quite astonished that
6850 800cpm !!!
270x 650cpm
280x 1100cpm

6850 seems drugged!

Same error here...  :(
win7_x64, 2x R9280
Hope you can fix this!  :-*

Fixed the new implementation so it works on AMD. All that had " error: expression must have pointer-to-struct-or-union type" type errors, please download the new version (build6), or simply update the sha512vectorized.cl file in your already downloaded build5.

Thanks to all for the bug reports and for the patience.
girino.

12
I got 630cpm on 270x 2G, which is same to 7870, but only 150 on 7850 1G, what's the problem? Is that 1G memory slow down the cpm?

Pages: [1]