Print Page - I guess I understand the flags, now how to write a phoenix 2.0 config file?

Title: I guess I understand the flags, now how to write a phoenix 2.0 config file?
Post by: skeeterskeeter on February 26, 2013, 03:10:27 AM

So as most you might have seen I am having trouble with guiminer and getting it to run optimally. So I decided to try phoenix 2.0.

EDIT: I am running two 5830's a 7770 and a evga 610gt; all four support openCl, I guess I can use guiminer to find their device numbers too!

I see that most user set the flags as such
-k phatk DEVICE=0 WORKSIZE=128 BFI_INT VECTORS FASTLOOP=false AGGRESSION=13
or whatever they wish.

I get that these are flags for the device, but how do I get phoenix up and running to set this. I know it is used via command line, it sent me the message "looks like your new, edit the .conf file". So I am not sure as how to write it? I cannot find a guide how to write it, only different config values.

EDIT: Well I found this? It sure does make all these things scream when I run phoenix. So, this [cl:0:0] means For opencl miners with device ID 0 and ____ 0 do the mining as follows [...value setting...]

How can I tell which device is which, what is the the third number stand for?
What is a good, low low setting to start at for each of my cards?
- in guiminer the flags (-v -w512 -f100) seemed to let each card run at a decent temp. (though they seemed to not be getting many shares?; a little slower than the flags (-v -w128) and alot less heat)
It says queue size is the size of the queue, what queue? work queue? why would I want size one?

Code:

[general]
autodetect = +cl -cpu -cuda
verbose = True
backend = http://MyName_WorkerName:workerpassword@btcguild.com:8332
failback = 300
queuesize = 1
queuedelay = 5
statusinterval = 1
ratesamples = 10
logfile = log

[cl:0:0]
autoconfigure = False
kernel = phatk2
name = 5830_1
start_undetected = False
disabled = False
WORKSIZE = 256
VECTORS = true
VECTORS4 = false
BFI_INT = true
fastloop = false
AGGRESSION = 14
goffset = true

EDIT2: This works nicely
with a worksize of 1024 and agression 9 for a pure air cooled setup. Prolly needs to go even a little lower.

Title: Re: I guess I understand the flags, now how to write a phoenix 2.0 config file?
Post by: ssateneth on February 26, 2013, 06:11:12 AM

You have too many options/flags set that you probably do not even need to mess with. One example is you are declaring auto-detect, but you are manually declaring devices to mine on too?...
I crafted this config myself when I used phoenix 2 and it just -worked-.

Example of single-gpu setup

Code:

[general]
backend = http://15kZZVh28AA6Ya3232bCXC9FLMtYFnmHZd:x@mining.eligius.st:8337
verbose = true

[cl:0:0]
name = 5870
kernel = phatk2
AGGRESSION = 7
VECTORS = true
BFI_INT = true
WORKSIZE = 128

Example of multi-gpu setup

Code:

[general]
backend = http://15kZZVh28AA6Ya3232bCXC9FLMtYFnmHZd:x@mining.eligius.st:8337
verbose = true

[cl:0:0]
name = 5870 #1
kernel = phatk2
AGGRESSION = 7
VECTORS = true
BFI_INT = true
WORKSIZE = 128

[cl:0:1]
name = 5850 #2
kernel = phatk2
AGGRESSION = 12
VECTORS = true
BFI_INT = true
WORKSIZE = 128

[cl:0:2]
name = 5850 #3
kernel = phatk2
AGGRESSION = 12
VECTORS = true
BFI_INT = true
WORKSIZE = 128

Title: Re: I guess I understand the flags, now how to write a phoenix 2.0 config file?
Post by: skeeterskeeter on February 26, 2013, 12:49:44 PM

This was an example config I found perusing the interwebs. Idk he used whatever he did.

So how do I target each individual GPU? Do I set the two numbers after the "cl" in the device config sections to the numbers of the device that guiminer picks up? Say guiminer calls my 7770 capeverde [0-1] then in the config file i should use [cl:0:1] to hit the 7770?

Does verbose mean, "I am going to type the password in this file as plaintext"?

I got the 7770 and two 5830's running at 600Mhash collectively. Should be more, near 800, but not enough cooling right now.

Bitcoin Forum

Bitcoin => Mining software (miners) => Topic started by: skeeterskeeter on February 26, 2013, 03:10:27 AM