Skip to content

HTTPS clone URL

Subversion checkout URL

You can clone with HTTPS or Subversion.

Download ZIP
Browse files

docs updates

  • Loading branch information...
commit 8d9b32a29269463a3770306318e209272306923b 1 parent 0d47e1c
@tomrittervg authored
View
2  README
@@ -27,6 +27,8 @@ with your temporary directory, and overall just read the stuff to see what you
have to change. I didn't make a note of everytime I removed my password and
put in PASSWORD for you to fill in.
+I recommend you read all the READMEs in the directories.
+
No Wordlists
============================================================
I'm not including my wordlists. Not out of secrecy, I just downloaded every
View
27 misc-notes.txt
@@ -1,14 +1,37 @@
The Amazon AMIs often have some old version of gcc, and you need to emerge/apt-get/yum install/whatever gcc44 and edit the makefile to use that.
+==============================
+
I was working off BOINC 6.11.1
svn co http://boinc.berkeley.edu/svn/branches/server_stable
http://boinc.berkeley.edu/trac/wiki/SourceCode
http://boinc.berkeley.edu/trac/wiki/BuildSystem
-
+==============================
+
You'll see I use xz compression in places. 7z doesn't allow streaming
compression. You can't decompress from stdin, you need to give an actual
filename. (So you couldn't 7zcat or 7zless like you can zcat, zless, xzcat,
and xzless.) I decompress in the job.xml files by wiring up stdin and stdout,
so I needed a compression function that could do so. xz gives great compression
-and the feature I needed.
+and the feature I needed.
+
+==============================
+
+I did some horrible shenanigans with my BOINC server, for reasons not terribly
+interesting to this presentation, where instead of having a platform like
+'linux' I made custom platforms for each individual CPU architecture.
+
+You don't have to do this. You should be aware of it though, so everywhere you
+see a bunch of directories 'barcelona', 'nocona', you can just replace them in
+your mind with 'linux'.
+
+If you want to do this anyway:
+1) define your platforms in project.xml
+2) Instead of _linux as a platform, put your custom one.
+3) When you comile the boinc client, after you ./configure but before you make
+ edit config.h and set HOSTTYPE.
+ #define HOSTTYPE "x32-nocona"
+
+
+==============================
View
8 patches/boinc-patches/README
@@ -23,7 +23,7 @@ client-app-running.patch
Again, causing 'command not recognized' errors, this time with
cudahashcat+
client-platform-tomfoolery.patch
- See TODO
+ See misc-info.txt
server-moreapps.patch
@@ -38,4 +38,8 @@ server-multithreaded.patch
server-builddirs.patch
client-builddirs.patch
- TODO
+ BOINC doesn't support an application that has subdirectories. I wrote
+ a _horrible_ couple patches that add support for this. Immediately after
+ doing so, someone remarked that most people just unzip the directories
+ on the client. I needed this for cudahashcat, and unzipping took 100MB of
+ files to 1, so I do that now. So you shouldn't need these patches.
View
13 sample-server/README
@@ -1,18 +1,7 @@
These are the files off of my BOINC server, and show my layout for the
applications, templates, and the dashboard I added.
-Q: Why do you have such strange platforms? Per-processor architectures?!
- Do I need to do that?
-
-A: No. I did that in the very beginning when I thought GGNFS performed
-wildly differently on different processor architectures. I never changed it.
-But I'll explain how to do this anyway.
-
-1) define your platforms in project.xml
-2) Instead of _linux as your platform, put the custom platform you defined in
- project.xml
-3) TODO
-
+Make sure you read misc-notes.txt
config.xml
config_aux.xml
Please sign in to comment.
Something went wrong with that request. Please try again.