Posts by FluffyChicken

41) Message boards : RALPH@home bug list : RALPH Version News! - Version 4.97 (Win/Lin/Mac) released! (Message 1074)
Posted 11 Apr 2006 by FluffyChicken
Post:
Carlos, you can get the older versions and newer of course from
http://ralph.bakerlab.org/download/
42) Message boards : Number crunching : Interesting credit differences between Linux & Windows wu\'s (Message 881)
Posted 16 Mar 2006 by FluffyChicken
Post:
It boils down to 'boinc' and I wouldn't bother looking into it.

Given the times for the rosetta are near identical is good to know though.


Anyways,
These are the two important parts of the information to look at
Measured floating point speed
Measured integer speed
for the computer, it is that and purely that that time the time taken that gives you the points.

Windows is well known to give higher values due to the way it's compiled (boinc client) you can manipulate thoose values yourself if you want as you can compile your own client.
Have a read here may give you more insight.
http://boinc.truxoft.com/

also do a search on either of the forums for credit and there's been plenty of discusion on proposed ways to solve it.

43) Message boards : Current tests : WinXP 64bit/AMD64bit-Support? (Message 702)
Posted 27 Feb 2006 by FluffyChicken
Post:
Hi,

I'm not a member of the project, but I don't think there will be a 64-bit version of most of the scientific projects, because all the "real work" is FLOATING POINT operations.

A much better options would be a SSE-enabled version of Rosetta, which should speed overall projects TeraFLOPS up by 3.5x - 4x times, as F@H experience has shown the last few years.


Assuming the type of coding is similar to F@H you may do, although over at FaD which was mainly single precision FP ops and was already optimised by intel at the initial stages, compiling for SSE or SSE2 gained little or no benefit. So SSE was always turned off (MMX may even have been turned off).

As for 64bit, well it would be a nice thing (as in run natively).
But you would think solaris and MAC-intel would come first as they are official BOINC platforms ;-)

... I think there is already this suggestion in Feedback as it's not a current test....
44) Message boards : Feedback : [website] - Add crosslinks between ralph/rosetta (Message 575)
Posted 24 Feb 2006 by FluffyChicken
Post:
Yes I noticed the Ralph to Rosetta after I posted, just need the Rosetta to Ralph one now

(being just in the news page is ok until the news goes of the front page)
45) Message boards : Feedback : Credit scores (Message 526)
Posted 23 Feb 2006 by FluffyChicken
Post:
First, it is of course a suggestion that they try something out. Why else would I post and ask for ideas and ease of integrating them.

Snake must see Ralph as something different to me, I see it as the testing and development side of Rosetta, again one in the same application and goal. to make Rosetta better.

By 'mess' I mean everyone jumping on the post without putting ideas toward it (which is what this topic has now been turned into :( )

All It was intended for was to gather the ideas together so it could be incorporated into the client for testing.

No where does it say credits are not being tested, the guidlelines just say credits and what becomes of them are not important (in the Ralph context), which I read as hey it doesn't matter if you loose loads of credits, they get zeroed dont work quite right or whatever and so anyone doing Ralph is here to improve 'Rosetta the project'.

I didn't say boinc suckz, screw you all ;) I just said that the current chosen implementation of credit scoring by Rosetta doesn't work and I gave proof AND an idea and then some more ideas with question (which have been answered). The initial suggestion was propesed due to the way Ralph was now

Shame I have to keep defending along the way things that will hopefully help Rosetta 'the project' along the way :(



No where in the guidliens does it say suggestions have to be exactly on what they are testing AND only on what they are testing.

[i]NOTE: This is a test project so credits* and other competitive statistics are not important. There will be times when no test work units are available.

1. Please do not abort work units.
2. Try not to cache too many work units since we are trying to test on as many different machines as possible.
3. Since the message board is the main feedback channel, please help to keep it manageable. Do not post repetitive information and only post constructive feedback.i]
* see above explanation

This is not repetative information, it is one topic on credit testing here related to new changes in the test client and it (was) constructive.


Did I say it needs changing now, no.

Just ideas <geez>
46) Message boards : Feedback : Mac Intel Application? (Message 525)
Posted 23 Feb 2006 by FluffyChicken
Post:
While it maybe a smaller percentage, BOinc is already set up for it.

They have just released new guidlines for compiling for the intel macs and is now the fifth official platform for boinc.

http://boinc.berkeley.edu/mac_build.html

Building BOINC Clients and Applications on Macintosh OSX

Written by Charlie Fenton
Last updated 2/22/06

This document applies to BOINC version 5.3.19 and later. It has instructions for building BOINC for Macintosh OSX, plus information for building science project applications to run under BOINC on Macintosh OSX.

Contents:

• Important requirements for building BOINC software for the Mac.

• Building BOINC libraries to link with project applications.

• Building BOINC Manager, BOINC Client and BOINC libraries.

• Building BOINC Manager Installer.

• Building project applications / upgrading applications for Macs with Intel processors.


.....

Building project applications

Upgrading applications for Macs with Intel processors

Apple began shipping Macs with Intel processors on January 10, and Apple expects to convert all its lines of computers to Intel by the end of 2006.

All future releases of BOINC will include "universal binary" builds for the Macintosh of BOINC Manager, command-line BOINC client and the boinc_cmd command-line tool. (Universal binaries contain both PowerPC and Intel executables in one file; the Macintosh OS automatically selects the appropriate one for that computer.)

The advantage of "universal binaries" is that you only need to have one copy of the application, and it will run on either PowerPC or Intel Macs, so users don't need to choose between two options. Since BOINC participants manually download BOINC from the web site, we will be providing BOINC in "universal binary" form.

However, participants do not manually download project applications; this is done automatically by BOINC. So there would be no advantage to combining the Intel and PowerPC versions in a single "universal binary" file, but doing so would double the size of the download.

So BOINC treats Intel Macs as a new, separate platform. BOINC previously directly supported four platforms: PowerPC Macs (powerpc-apple-darwin), Intel Linux (i686-pc-linux-gnu), Windows (windows-intelx86) and Solaris (sparc-sun-solaris2.7).

We have now added a fifth platform for Intel Macs (i686-apple-darwin).

As a temporary measure, projects can set their servers to deliver a copy of their current PowerPC application (renamed for the new platform) under the new i686-apple-darwin platform. The OS will run it in compatibility mode, emulating a PowerPC. (Apple calls this compatibility mode Rosetta, which of course has nothing to do with the Rosetta BOINC project.)

If you do this, be sure to give your native Intel application a higher version number when you do release it, so that clients will download it.

However, running a PowerPC application in compatibility mode has two significant drawbacks:

(1) Screensaver graphics do not work.

(2) Since it is running under emulation, your application will run at reduced efficiency. But the benchmarks are based on running native Intel applications. This may cause scheduler problems, such as uncompleted deadlines and inadequate credit for participants.

So it is important to make a native Intel application available as soon as possible.

It is very easy to add a new platform to your server with the xadd utility. For directions on how to do this, see these web pages:

http://boinc.berkeley.edu/platform.php

http://boinc.berkeley.edu/tool_xadd.php

BOINC supports all PowerPC Macs running OS 10.3.0 or later, and all Intel Macs. (The Intel Macs themselves require OS 10.4.4 or later.)

The easiest way to build your application for these two platforms is to build each one on its native platform. In other words, do your powerpc-apple-darwin build on a PowerPC Mac running OS 10.3.9, and your i686-apple-darwin build on an Intel Mac.

But Apple provides the tools to allow you to cross-compile your application on any Mac (PowerPC or Intel) running OS 10.4 or later. Here is how:

All BOINC software for Power PC Macs must be built using GCC 3.3 and MacOS10.3.9 SDK to assure backward compatibility with OS 10.3. If building a PowerPC application on an Intel Mac, you must also specify "-arch ppc" in the compiler and linker flags.

All BOINC software for Intel Macs must be built using GCC 4.0 and MacOS10.4.u SDK to allow cross-compiling. If building an Intel application on a PowerPC Mac, you must also specify "-arch i386" in the compiler and linker flags.

You can find examples of how to do this for two different kinds of configure / make scripts in the HEAD branch of the BOINC CVS tree at boinc/mac_build/buildcurl.sh and boinc/mac_build/buildjpeg.sh.

The lipo utility is used at the end of each of these scripts to combine the two binaries into a single "Universal Binary" file. You won't need to do that with you project applications, since you will be distributing them separately under the two platforms. But if you prefer, you can create a Universal Binary and distribute the same file for both i686-apple-darwin and powerpc-apple-darwin platforms.

Note that the BOINC libraries (and any third-party libraries) which you link with your applications must be built with the same configuration as the application itself. Follow the instructions earlier in this document to build the needed libraries.

Additional information on building Unix applications universal can be found here:

http://developer.apple.com/documentation/Porting/Conceptual/PortingUnix/compiling/chapter_4_section_3.html

and here:

http://developer.apple.com/documentation/MacOSX/Conceptual/universal_binary/universal_binary_compiling/chapter_2_section_7.html


For information on making your code work with GCC 4:

http://developer.apple.com/releasenotes/DeveloperTools/GCC40PortingReleaseNotes/index.html



And it seems we have an alpha tester ready and waiting :-)
47) Message boards : Cafe RALPH : Beta (alpha) testing (Message 440)
Posted 21 Feb 2006 by FluffyChicken
Post:
So what sort of thing do you look for or notice in your testing (generally in beta testing)


Me I normally find the useability problems which should be easiest to sort out (though normally takes the longest to fix)

Then it'll be messing about with the new functions to see if the work or not.

What does irritate me the most is when people don't describe the problem in detail with some logs (which I do sometimes myself :s) to see if it can be reproduced.
48) Message boards : Cafe RALPH : (DO NOT POST HERE) This is the Moderators Archive thread (Message 438)
Posted 21 Feb 2006 by FluffyChicken
Post:
hi m9 & colleagues,

I got an email that you'd moved a posting of mine from the guidelines thread to the feedback thread, which is fair enough if that is a better place to put it.

However, it seems that although the email says you moved it, in fact the thread has been hidden where it was before. Is this a bug, or did your mouse slip at the crucial moment?

Feel free to delete this post once my original posting has been found, or let me know if it has been lost in cyberspace and I will type it in again

R~~


Your post was moved and I checked to see that it arrived. The thread was deleted, as it was not a guideline. Check occasionally for Mod to Mod messages.


The problem comes from when it's moved/and hence deleted from the original location. The email we get points to the deleted(pre-moved) post and not the
post in the new location.

Is it possible (in the current software) to get the email to also link to the location of the moved post ?

If not could you also mention the section you moved it to when you post you comment (e.g. rather than just 'off topic', use 'moved to number crunching')

Although my experience comes from the main rosettta forum.
3 post in a row saying 'deleted, offtopic and the link goes to a delted post', a member can get rather pissy when that happens especially as all the people in the thread would have got it.
Mods have enough greif without that coming back at you at once.
Could certainly get newer members in a huff :)
49) Message boards : Feedback : Credit scores (Message 434)
Posted 21 Feb 2006 by FluffyChicken
Post:

If you have been around Distributed Computing since it's infancy, you should realize by now that a fair and equitable credit system is what keeps people at a project.

Sure, you get a few percent who are in it "for the science", but reality is that the biggest portion of the crunching power comes for the credits, and leaves rather quickly when the credit system goes awry.



On a production project you'd be right. That is why on Rosetta, the sister production project, David has put work in to correct errors that lose people credit.

On a test project, like this, we do not need very many people, and we particularly want the people who are interested in building up the science. So while you are still right that when the credit goes awry a lot of people will leave, the difference is that here it won't matter - it could even be an advantage if it leaves us with a higher proportion of those who will give constructive feedback over problems other than credit.

Like, a thousand crunchers is more than enough for a test project (so long as there is a good distribution of different speed boxes and different operating systems). If the only-in-it-for-the-credit crew self select themselves onto another project that is good news, not bad.

Personally so long as they stay on any BOINC project I'd be delighted - I guess the official project view is to hope they'd move over to Rosetta in particular.


Angus seems to understand, Rosetta/Ralph should be seen as one, but Ralph is just a test section to see if and check things work before it get put into production. (not a sister, brother, long distant relation, but one in the same project.)
What we do here directly effects Rosetta.

This is also a SUGGESTION section. It should not be distingued as a seperte identity to Rosetta. Credit certainly doesn't mean anything here at Ralph (and hence they should stop people from collectiong the xml stats files for places like boincstats, but they should still create them)

Here we can bash out ideas for improvment to the client without the mess associated with the main forum (rosetta).


But for thoose that think they are different and seperate project...
If we get the credit system sorted here at Ralph then we have a good chance it'll a lot better at Rosetta when it moves accross.
It does not matter if we bugger it up along the way in here ;-)

Priority, that's up to the developers :)




Seti@Home - Enhanced is testing a flops based measurement ?
Can it be used here ?
Is it easy to implement ?
Can it be tested ?
Does it restrict the boinc client version numbers ?

Dismiss my hour=xxxcredits unless someone can find a way to use it properly. (for some reason I was thinking fixed 'work done' length work units, not time which is completely different)
50) Message boards : Feedback : Credit scores (Message 421)
Posted 21 Feb 2006 by FluffyChicken
Post:
Hello,

Section title
Feedback
Comments, recommendations, etc.


So I posted a recommendation and it was relevant to the new time frame system of the workunits.

Although I just notice a flaw in my 8hrs=xxx credits ;-s wasn't too awake at the time.

No it is not a boinc issue as such, since boinc have options in place (and at least testing a bnew one) to over come or at least reduce it. (although it is a boinc community problem when one project gives out whatever points credit/cobbles the member would like)

Anyways although Ralph is not a credit project it is a place to test out solutions, Rosetta is not.
51) Message boards : Current tests : [idea] - version release/log in this section (Message 366)
Posted 20 Feb 2006 by FluffyChicken
Post:
When you release a new client or new feature for testing, put a new post up saying so.

Means we know what's happening and what to test and look out for next, also we can force an update at the time.


How long can it take to make a post ;-)
52) Message boards : Feedback : [website] - Add crosslinks between ralph/rosetta (Message 364)
Posted 20 Feb 2006 by FluffyChicken
Post:
Add a crosslink on ralph/rosetta between the two sites on the mian page.

Also maybe add one on the main forum as well (so people can be nosey and your self advertising for free)
53) Message boards : RALPH@home bug list : Same team on 20th and 21th rank in teams list (Message 363)
Posted 20 Feb 2006 by FluffyChicken
Post:
Hmm, looks like there is some mess in teams list. As I see it now the L'Alliance Francophone team (teamid=24) is also on 20 and 21 rank.

There is another weird problem with my team (Boinc.sk, teamid=34 -> probably the last team at this moment) it is not showing in the list of all teams. However it appears in list of unclassified teams.

For me it looks like bug in selecting which teams to show on which screen.



Rosetta did this as well,
my team moved from 21st to 20th (a long time ago now) and went missing for a while, while the ex-20th was now at both 21st and 20th

so I would be inclined to say it a boinc 'stats' problem and not specific to Ralph

Also not a newley created team thing

Ingleside's seems to answer it (I assume that's project web server cache as no amount of page requesting/forcing solved it)

Only thing I can think of would be to put a check in place on table updates for the 20/21 40/41 ... rank changes then do an adition refresh of the corresponing page, of course I have no idea if they could keep server loads down to what they like, but it is a noticable 'bug' (athough if it is intended I guess it's a 'feature' :)
54) Message boards : Feedback : Credit scores (Message 360)
Posted 20 Feb 2006 by FluffyChicken
Post:
I notice XtremeSystems moved to the RAC top on Rosetta (not Ralph) so I go t nosey.

Question,
How does Rosetta gauge it's work thorughput as you cannot be using Cobblestones/credit due to the technically pointlessness of it with Rosetta,

I for one claim more than the 'recommended boinc client' but that because I use some features it adds (truxoft)

But Looking throuhg XtremeSystems who are competative their second user has some high RAC's
(at least he show the computers)


Sempron 2800+
Measured FPU 3756.18 million ops/sec
Measured INT 11433.38 million ops/sec

Which way too large


One of my computers at least says its an optimised client
<core_client_version>5.3.12.tx37</core_client_version> but I know all do not.

my 3200+ (truxoft)
Measured FPU 2154.35 million ops/sec
Measured INT 6311.89 million ops/sec
my 3200+ (boinc normal)
FPU ~1500 region
INT ~4500 region


So how is the floating point benchmark comming along (or are you not bothering)
OR if his Time based workunits settles down well are you going to move over to blanket scoring, where 8hrs = xxx credit ?


At the moment the league tables are a shambles, I have seen it mentioned over the web (giving rosetta a bad name :( and it hugly distorts the overall 'boinc' tables as well. Also you're work rating cannot be used as they are not meaningful along with the part of your front site

Queued: 0
In progress: 757
Successes last 24h: 565
Users (last day ) :
323 (+16)
Hosts (last day ) :
431 (+27)
Credits last 24h :
12,310
Total credits :
44,638
TeraFLOPS estimate: 0.123


i.e. credit, teraflops parts :(



Previous 20



©2024 University of Washington
http://www.bakerlab.org