Phoenix 2 beta discussion

[CFSworks]

Phoenix 2.0.0-rc2

The features of the new Phoenix 2 include:
Settings are stored in a single configuration file
No more managing hideously long command lines. Phoenix 2 loads them from a configuration file.

Multi-device mining from a single Phoenix process
You only have to launch one Phoenix process and it will manage multiple devices. The devices even share the same work queueing system for maximum efficiency.

Hardware autodetect (including hotplug support) and autoconfiguration
Kernels can now recommend optimal configurations for hardware. All you have to do is turn on autodetect and let Phoenix do its magic.

A better kernel interface supporting modern hardware miners
Kernels can now request full work units from the queue - ask your hardware vendors about making a Phoenix 2 kernel!

Complete JSON-RPC API
The RPC server listens on port 7780 by default, and is configurable through the config file.

Web control panel (not finished)
The RPC server also hosts a webserver. When the control panel is finished, you will be able to do first-time miner setup and long-term management through the web interface. If you are a web developer, consider looking at our PhoenixWeb Github repository and helping out.

Getting started
Download
You can find our repository on Github.
Windows users: see jedi95's post below. He is in charge of Windows builds.
Linux users can find the source code archive here.

Install
Windows users only need to extract the .zip into a new folder.
Linux users need to make sure all of the Phoenix 1 prerequisites are installed, and then run "./setup.py install" as root.

Configure
Windows users run the .exe to create a blank .cfg, then close it. (In the future, the configuration will occur through the web interface.)
Linux users create a .cfg, put "#!/usr/bin/env phoenix" at the top, and "chmod u+x" it.

An example configuration to get you started:

Code:

[general]
verbose = True
autodetect = +cl -cpu # The rightmost parameter takes precedence. This enables all OpenCL devices, except those that are CPUs.
backend = YOUR_URL_HERE # URL format is exactly as it was in Phoenix 1
backups = URL1 URL2 URL3 URL4 # Any number of backup backends (space-separated) to use in case the primary backend goes down.

[web]
password = rpc_password # Set an RPC password to keep people from messing with your miners.

# If you want to configure miners yourself, edit and uncomment this section:
#[cl:0:0] # Or whatever ID you want to configure.
#autoconfigure = True # Do you still want autoconfiguration?
#disabled = False # Do you want to disable the miner?
#bfi_int = True    # And any other kernel options...
#vectors = False # ...can go into this section.

Run
On Linux, simply execute your configuration file. It will launch Phoenix.
On Windows, run the .exe again

Please note: This thread is for discussing the development of Phoenix 2 only. Phoenix 1.x questions need to go to this thread and a new thread will be created when Phoenix 2 itself is released.

Donations gratefully accepted at: 1PHoenix9j9J3M6v3VQYWeXrHPPjf7y3rU
And there is a bounty for a web control panel here

[jedi95]

Windows binaries:
https://github.com/downloads/phoenix2/phoenix/phoenix-2.0.0-rc2.zip

Phoenix 1.7.x is now a legacy branch and will only see limited further development.

Kernels developed for Phoenix 1.x are not directly compatible with Phoenix 2, but the required changes are minimal.

[rjk]

If you don't have it already, proper support for RollNTime with expire=N would be a very useful feature for fast miners.

EDIT: Link to wiki page with list of miners and pools supporting various features: https://en.bitcoin.it/wiki/Getwork_support

[CFSworks]

If you don't have it already, proper support for RollNTime with expire=N would be a very useful feature for fast miners.

EDIT: Link to wiki page with list of miners and pools supporting various features: https://en.bitcoin.it/wiki/Getwork_support

We do have it. I suppose I should have put it in the features list...

[rjk]

If you don't have it already, proper support for RollNTime with expire=N would be a very useful feature for fast miners.

EDIT: Link to wiki page with list of miners and pools supporting various features: https://en.bitcoin.it/wiki/Getwork_support

We do have it. I suppose I should have put it in the features list...

Excellent. The unified queue manager is also an excellent new feature - one that most miners don't have and have needed for a long time.

[d3m0n1q_733rz]

Autodetect could use some work.  It would seem that it doesn't take vectors and optimum worksize into account.  How would one go about modifying the configuration file to set parameters manually?  I'm wanting to toss VECTORS8 in with a worksize of 64 while applying Dia's GOFFSET option if I can.  It would be nice to know the new layout.

[CFSworks]

Autodetect could use some work.  It would seem that it doesn't take vectors and optimum worksize into account.

That's jedi95's department. He's on it.

How would one go about modifying the configuration file to set parameters manually?  I'm wanting to toss VECTORS8 in with a worksize of 64 while applying Dia's GOFFSET option if I can.  It would be nice to know the new layout.

Forgot to put that in the first post. I updated the example config file to show how you can do it. You will need to know the device ID of your hardware (cl:PLATFORM:DEVICE) and you can override the autoconfigured settings (or disable autoconfiguration completely) from the config file.

[Bananington]

How would I have autoconfigure enabled on one out of two installed VGA's?

[CFSworks]

How would I have autoconfigure enabled on one out of two installed VGA's?

If you don't want to autodetect, you can do:

Code:

[cl:0:0]
autoconfigure = False
#settings here

[cl:0:1]
autoconfigure = True

If you do want to autodetect, you can just omit the one that has autoconfigure = True.

[d3m0n1q_733rz]

How would I have autoconfigure enabled on one out of two installed VGA's?

If you don't want to autodetect, you can do:

Code:

[cl:0:0]
autoconfigure = False
#settings here

[cl:0:1]
autoconfigure = True

If you do want to autodetect, you can just omit the one that has autoconfigure = True.

I'll give this a try.  I assume it's similar to passing the settings to the command-line based version?  Just all the little ducks in a row?

[CFSworks]

I assume it's similar to passing the settings to the command-line based version?  Just all the little ducks in a row?

Yes. The kernels are very similar, so the kernel options are pretty much the same as in Phoenix 1.

[Bananington]

In the example, it says "backend = UR_URL_HERE"; is that where I put my worker information?

[d3m0n1q_733rz]

In the example, it says "backend = UR_URL_HERE"; is that where I put my worker information?

That's where you put your worker and site info. combined.  Like http://My_miner_name:Miner Password@www.website.net:8332

[d3m0n1q_733rz]

Could we get a list of all accepted options?
Kernel selection
Worksize=
Vectors, Vectors4, Vectors8, Vectors16
etc.

[Bananington]

I can't get the miner to start.  
This is what I've got exactly in my cfg file. lol

Code:

[cl:0:0]
autoconfigure = True

[cl:0:1]
autoconfigure = True
backend = http://censored:censored@censored.com:8332

I also want to have GPU 2(Device #1) mine while GPU 1(Device #0) is not mining. But I also want to have GPU 1 set up to mine so that I can just enable it to mine by changing a value of some sort.

[d3m0n1q_733rz]

I can't get the miner to start.  
This is what I've got exactly in my cfg file. lol

Code:

[cl:0:0]
autoconfigure = True

[cl:0:1]
autoconfigure = True
backend = http://censored:censored@censored.com:8332

I want to have GPU 2(Device #1) mine while GPU 1(Device #0) is not mining. But I also want to have GPU 1 set up to mine so that I can just enable it to mine by changing a value of some sort.

Best bet would be to put the program in two different folders.  Close the program from the folder you have the cfg set for disabling GPU1 and then start the program for both GPU mining when you want to have them both working. 

[d3m0n1q_733rz]

[cl:0:0]
autoconfigure = False
BFI_INT VECTORS4 WORKSIZE=64

What's wrong with this portion that it's not giving me the correct settings?

[Bananington]

I can't get the miner to start.  
This is what I've got exactly in my cfg file. lol

Code:

[cl:0:0]
autoconfigure = True

[cl:0:1]
autoconfigure = True
backend = http://censored:censored@censored.com:8332

I want to have GPU 2(Device #1) mine while GPU 1(Device #0) is not mining. But I also want to have GPU 1 set up to mine so that I can just enable it to mine by changing a value of some sort.

Best bet would be to put the program in two different folders.  Close the program from the folder you have the cfg set for disabling GPU1 and then start the program for both GPU mining when you want to have them both working.  

My problem isn't the whole GPU thing right now, that was just a side note. My real problem is that I have NO IDEA what settings to use for where and for what. My miner won't mine due to PEKAC.

[CFSworks]

Could we get a list of all accepted options?
Kernel selection
Worksize=
Vectors, Vectors4, Vectors8, Vectors16
etc.

The Phoenix 2 core itself recognizes (not an exhaustive list):

Code:

# [cl:0:0] autoconfigure # Allow the kernel to choose its own configuration variables?
# [cl:0:0] disabled # Prevent mining on this device?
# [cl:0:0] kernel # Can be used to manually choose a kernel to mine on.
# [cl:0:0] start_undetected # Start even if autodetect doesn't find it?
# [general] autodetect # A list of rules for what devices to autodetect.
# [general] backend # The URL to mine on in the backend.
# [general] logfile # Set this option to log to a file.
# [general] queuedelay # Advanced - ???
# [general] queuesize # Advanced - ???
# [general] ratesamples # Advanced - number of samples to average for rate reporting
# [general] statusinterval # Advanced - how long to delay between statusbar updates
# [general] verbose # Enable verbose mode? (Shows debug messages)
# [web] bind # Bind the web/RPC server to a specific IP
# [web] disabled # Disable the webserver altogether?
# [web] logbuffer # Advanced - how many logs to remember in the getlogs() RPC call
# [web] password # The password necessary for web/RPC login (username is ignored)
# [web] port # What port should the web/RPC server listen on?
# [web] root # Advanced - The root directory for the webserver.

The phatk2/opencl kernels recognize:

Code:

# [cl:0:0] vectors # Enable two-way vectors?
# [cl:0:0] vectors4 # Enable four-way vectors?
# [cl:0:0] fastloop # Advanced - fastloop optimization for low aggressions
# [cl:0:0] aggression # Controls how hard Phoenix 2 hits the hardware
# [cl:0:0] worksize # Advanced - controls size of individual executions
# [cl:0:0] bfi_int # Enable BFI_INT optimization for Radeon cards that support it

[cl:0:0]
autoconfigure = False
BFI_INT VECTORS4 WORKSIZE=64

What's wrong with this portion that it's not giving me the correct settings?

Try:

Code:

[cl:0:0]
autoconfigure = false # Not actually needed since autoconfiguration disables by default when you supply your own args
BFI_INT = true # Also the boolean options aren't case sensitive.
VECTORS4 = true
WORKSIZE = 64

[Bananington]

Ok, I got things to work. Is there a way to have one GPU mine at one backend and have the other GPU mining at a different backend?