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 - NaN

Pages: [1] 2 3 4 5 6 7 8 ... 11
1
BitShares PTS / Re: AMD OpenCL Momentum miner
« on: February 09, 2015, 08:36:04 am »
That are the settings for clpts v2.0, that gave the best results for my rig:
  • -g 2 -a 0 for R9 280X
  • -g 3 -a 2 for R9 290 and Windows
  • -g 3 -a 4 for R9 290 and Linux

This took me from 4,879 to 6,099 c/m.

Will try it out for giggles on an Nvidia 970 later (not sure if it will run but will try).

EDIT: Tried for the GTX 970 just to see but would not run. Says no Open CL device found. Kinda weird because some AMD miners still pick it up and some don't.
Nvidia is not supported because I'm using AMD specific instructions and I don't know whether Nvidia supports SPIR (I don't think so). I did run my miner on a Nvidia GPU once, but it wasn't really fast and other miners were comparable. I don't own a Nvidia GPU, so I could not really test it, but perhaps the miner could run well on GTX 970 and later.

2
BitShares PTS / Re: AMD OpenCL Momentum miner
« on: February 08, 2015, 09:11:25 pm »
That are the settings for clpts v2.0, that gave the best results for my rig:
  • -g 2 -a 0 for R9 280X
  • -g 3 -a 2 for R9 290 and Windows
  • -g 3 -a 4 for R9 290 and Linux

3
BitShares PTS / Re: AMD OpenCL Momentum miner
« on: February 08, 2015, 08:42:50 pm »
@ NaN

https://bitcointalk.org/index.php?topic=896133.msg10397793#msg10397793

Not sure what binaries you are referring to that goes in the binary folder.

I just created the bat and ran it from the unzip.
You have to copy the gpuhash_* files of one Catalyst* folder into the binaries folder because the binaries of different Catalyst version are not compatible. Choose the Catalyst* folder according to your Catalyst driver version. If you have Catalyst 14.6 or 14.7 installed, then use the files in the Catalyst 14.9 folder.
EDIT 2: I fixed the README. I used the UNIX newline instead of the Windows newline

4
BitShares PTS / Re: AMD OpenCL Momentum miner
« on: February 08, 2015, 06:33:13 pm »
I've updated clpts to version 2.0, which has support for stratum pools and optimizations for different GPUs (see https://bitsharestalk.org/index.php?topic=2598.0)

5
BitShares PTS / Re: fast AMD OpenCL PTS miner released
« on: October 01, 2014, 04:56:58 pm »
I only got 20c/m with my msi 7850 1G card. 

driver 13.12  using catalyst13 minter
system: win7x64
card:  5  x  msi 7850 1G
[...]
Could you use GPU-Z to monitor RAM usage of the GPUs? If large amounts of dynamic memory are allocated then the low performance is caused by a driver bug. Also I am interested whether the problem shows up if the miner uses a single card. I have no idea whether a driver update would help.

6
BitShares PTS / Re: fast AMD OpenCL PTS miner released
« on: October 01, 2014, 04:51:30 pm »
Can someone show the proper setting for 290x with clpts 1.3?
I got around 6k cpm for 280x.But only around 5k5 cpm for 290x.
290x looks like the worse deal I've made.
Mario241077 seems to get best results with -g 3 -a 2 using a 290x (over 6400cpm, but I'm not sure whether it runs at reference clocks).

7
BitShares PTS / Re: fast AMD OpenCL PTS miner released
« on: September 21, 2014, 12:16:30 pm »
When I run 1.3 version, I got this error :

GPU-deviceIDs: 0
gpu_threads:   1
algorithm:     0

[MASTER] | Spawning 1 worker threads
[WORKER0] Hello, World!


Error during compilation of program for device 'Tahiti':

Frontend phase failed compilation.
Error: Compilation from LLVMIR binary to IL text failed!


Any idea how to fix this?

Still the same...
It start with verstion 1.2v2, 1.2 is working fine

GPU-deviceIDs: 0
gpu_threads:   1
algorithm:     2

[MASTER] | Spawning 1 worker threads
[WORKER0] Hello, World!


Error during compilation of program for device 'Tahiti':

Frontend phase failed compilation.
Error: Compilation from LLVMIR binary to IL text failed!
You have overlooked my previous post:
Your using the wrong binary. Most likely you have an older Catalyst 13.x driver installed. Download the *_Catalyst13 binary for your OS. If your driver is too old the miner will crash (Catalyst 13.12 and 13.11beta should work), but I recommend Catalyst 14.6 or newer because it's faster.

8
BitShares PTS / Re: fast AMD OpenCL PTS miner released
« on: September 20, 2014, 04:29:17 am »
@Mario241077: Thanks for sharing your performance numbers. Does the R9 290X crash if you use -g 4 -a 2? -a 0 and -a 1 use more than 1GB RAM per thread and will therefore crash with -g 4. -a 2 and -a 3 use ~900MB RAM per thread. P.S. You may combine the two instances by using -d 0-1 -g 2,4 -a 0,2 so that the 7970 runs with -g 2 -a 0 and the 290x with -g 4 -a 2. You will get separate stats for each card and you can compare the ART value (less is better) to find your optimal config because it has less variance than the cpm value and the share rate.

yes if i use -a 2, on this card is my on this card is my screen

following happens! the memory of my PC loads up, the CPU goes to full load and the screen freezes! After a while I can move the mouse and the batch close.

PS: two separate batchs per card is faster

I'll still post the results
Ok, I only tested my R9 290 without an attached screen.

9
BitShares PTS / Re: fast AMD OpenCL PTS miner released
« on: September 20, 2014, 04:27:04 am »
When I run 1.3 version, I got this error :
[...]
Error during compilation of program for device 'Tahiti':

Frontend phase failed compilation.
Error: Compilation from LLVMIR binary to IL text failed!


Any idea how to fix this?
Your using the wrong binary. Most likely you have an older Catalyst 13.x driver installed. Download the *_Catalyst13 binary for your OS. If your driver is too old the miner will crash (Catalyst 13.12 and 13.11beta should work), but I recommend Catalyst 14.6 or newer because it's faster.

10
BitShares PTS / Re: fast AMD OpenCL PTS miner released
« on: September 19, 2014, 11:38:23 pm »
@Mario241077: Thanks for sharing your performance numbers. Does the R9 290X crash if you use -g 4 -a 2? -a 0 and -a 1 use more than 1GB RAM per thread and will therefore crash with -g 4. -a 2 and -a 3 use ~900MB RAM per thread. P.S. You may combine the two instances by using -d 0-1 -g 2,4 -a 0,2 so that the 7970 runs with -g 2 -a 0 and the 290x with -g 4 -a 2. You will get separate stats for each card and you can compare the ART value (less is better) to find your optimal config because it has less variance than the cpm value and the share rate.

11
BitShares PTS / Re: fast AMD OpenCL PTS/NRS miner released
« on: September 15, 2014, 04:03:12 pm »
could you check a readme file file on -g parameter, clearly said i check the readme and don`t understand how to run(( i will use old one what have -t parameter
also it will be good to have an example.but in all archives.
also i would like to know what is current speed on amd 290? - my shows 4300
Performance numbers without details on OS, Catalyst version and configuration are pretty useless, but 4300 cpm is slow if the newest drivers are used and if the GPU does not run into thermal throttling. The Linux and Windows drivers have totally different characteristics.

I hope that I can explain the use of the -g parameter with some examples. It controls the number of threads which are run on each GPU. Let's assume that you used -t 0,0 previously. That means that you run 2 threads on the device with ID 0. Therefore the new syntax would be -d 0 -g 2. -t 0,0,1,1,2,2,3,3 is equivalent to -d 0-3 -g 2 and -t 0,0,0,1,2,3,3 would become -d 0-3 -g 3,1,1,2. Using the newest Catalyst version on Windows, I have observed that spawning 4 threads per R9 290 and -a 2 gives a significant performance boost (-g 4 -a 2). Please note that the -d parameter is optional. If it is not specified than all AMD GPUs are used by clpts. You can still use the -t parameter but people with many GPUs per rig reported problems and something like -t 0,0,0,0,1,1,1,1 instead of -d 0,1 -g 4 is not convenient.

12
BitShares PTS / Re: fast AMD OpenCL PTS/NRS miner released
« on: September 14, 2014, 08:45:17 pm »
I released clpts-v1.3. It resolves some issues and supports GPUs with 1GB RAM (see https://bitsharestalk.org/index.php?topic=2598.0 and read the README  ;D)

13
BitShares PTS / Re: fast AMD OpenCL PTS/NRS miner released
« on: September 08, 2014, 12:42:56 pm »
one of my rids only four 280X ,and it can run with  the command like this

clpts_x86-64 -o 6 -u xxx.xxx:xxx -t 0,1,2,2,3,3 -a 0

but when i change it to this

clpts_x86-64 -o 6 -u xxx.xxx:xxx -t 0,1,1,2,2,3,3 -a 0  or
clpts_x86-64 -o 6 -u xxx.xxx:xxx -t 0,0,1,1,2,2,3,3 -a 0

the program will crash and quit .

why ?
me too. Pls advise.
I'm working on clpts-1.3, which should fix this issue. But as workaround one could use two instances of clpts, i.e. the first instance running on GPU 0 and 1, and the second on GPU 2 and 3.

14
BitShares PTS / Re: fast AMD OpenCL PTS/NRS miner released
« on: August 20, 2014, 04:07:54 pm »
Can I run 2 instances with 270x?
Only if it has 4 GB GPU-RAM otherwise you are limited to a single thread/270x

15
BitShares PTS / Re: fast AMD OpenCL PTS/NRS miner released
« on: August 11, 2014, 04:14:26 pm »
Why i rig with 4x280X run at 11350cpm,
The detail of my bat file is :
Code: [Select]
clpts_x86-64  -o http://ypool.net -u xxx -p x -t 0,0,1,1,2,2,3,3 -a 0
Because i see all people can run 5000cpm/280x , who can help me to edit my bat file

Thanks
You are using an outdated version of clpts. Please download v1.2. It has a different command line syntax than older versions and you have to change your bat file to
Code: [Select]
clpts_x86-64  -o 6 -u xxx:x -t 0,0,1,1,2,2,3,3 -a 0, if you want to use ypool.net

Pages: [1] 2 3 4 5 6 7 8 ... 11