also @ TechSpot: LG Display announces 5.0-inch, 1080p AH-IPS display for smartphones

TechSpot

TechSpot News Products Downloads Forums

Collaborate in the cloud with Office, Exchange, SharePoint, and Lync.
Microsoft Office 365. Pay-as-you-go starting at $8/user/month. Begin your free trial now.

Choosing the right project...

Discussion in 'General Discussion' started by Julio Franco, Sep 9, 2002.

Thread Status:: Not open for further replies.

Page 1 of 2

Julio Franco TechSpot Editor

Before we launch Team TechSpot officially, I believe we should get together and decide what's going to be the best project we could join.

First thing to consider I believe is if the project is really helpful. For example, SETI is cool and has one of the best databases of 3rd party tools, etc., in fact I was extremely happy with it returning results for work units in very little time compared to the overall average (4 hours instead of 10!).

In the other hand, you can't compare the search for extraterrestrial vs. cancer cure!

UD is nice, it has the best looking utility of all and isn't as consuming as SETI GUI, however the problem I'm seeing with it is that you can't have WU (work units) cached, you can't resume packages (so if you restart your PC you go back to 0%) and WU take quite a longer time than SETI, which is a bad thing because... you can't resume!

So unless we have a lot of members using broadband and leaving their PCs turned on all day, we don't have much of an opportunity.

That's where the team's second priority comes in... room for competition. As you probably are aware of, many other hardware sites compete for the top spots in all these distributed computing projects... while I'm not looking pure recognition, I do believe some healthy competition does help the cause and I would love to see TechSpot Team scaling positions and making it's way to the top.

Having considered that...

I would like you guys to help me choosing any of the computing efforts before we go live with it. I know you have been TS regulars for quite some time and you will follow us wherever we head to in the future concerning this project, however regular users will be much harder to convince.

Do not try making big sacrifices for now in neither of the projects as we could switch in little time and I do know it's annoying going back to zero.

I'm currently downloading clients for Folding@home, I will see how those turn up and keep you updated.

Any help or experience related to the different options we have would be kindly appreciated.

Julio Franco, Sep 9, 2002

#1
svtcobra TechSpot Paladin

I did Seti and the cancer cure screensaver for awhile over at Anand's team. I felt that the cancer cure screensaver was a better reason for me to leave my computer on 24/7 than to do math computations.

Im in for either one. I think the cancer screensaver one is best because it is for a good cause and it isnt eating up cpu cycles when your actually using the computer.

svtcobra, Sep 9, 2002

#2
Vehementi TechSpot Paladin

I miss SETI.

I know, I like the whole Cancer research thing, but the UD client needs some work :dead:

For example, I have 28k, the WUs for UD are 800-900kb. These take a few minutes to download. In comparison to SETI's 300-400k-ish WUs, which downloading 10 of is easy...Did I mention you can cache SETI WU's?

I don't know about TTS' competition in UD, but in SETI I don't think we would have a chance. There are single people with about 100 computers running SETI 24/7 (Team Anandtech's Robor for example - recently broke the 100,000 WU mark) on some teams. But, since everyone's basically doing UD, I don't think we have the option of SETI anyway. And with Papa posting the link and info for the UD team in the SETI thread... Grrrrr!

UD rated my CPU 2/3 slower than a P4 1.5GHz, what's with that? it's more than 2/3 faster than a 1.5GHz...so I dunno if I want to do it...And they include available disk space in the system rating.

Anyway, I know I rehashed some of what Julio said, but that's all I like and don't like.

Vehementi, Sep 9, 2002

#3
SNGX1275 TechSpot Special Forces

My personal opinion is the cancer one. But I'll probably go with whatever is the majority on our site. Espically when winter comes around and my ground floor will be averaging 65 degrees F (very good for my athlon).

SNGX1275, Sep 9, 2002

#4
LNCPapa TS Special Forces

I vote for UD Cancer because it hits a bit closer to home. Not that anyone I know has cancer, but if a cure is found wouldn't it be nice to know that we might have had something to do with it? What if afterwards someone we know finds out they do have it and now there's a cure.

It's true the UD client needs some work, but it's young compared to SETI and as it picks up in popularity more tools will spring up (hopefully some clients for other OSes.)

As far as the overall rating thing, I don't really think that matters. My main machine is rated at 173, but I think that's just some bogus number generated by a combination of several of the machine's stats. I don't think it had too big an effect on the actual time it takes to process a unit except on the processor section. That takes a while to update as well so you can't tell right from the beginning.

Keep in mind that I do still run SETI on my Macs - only because they don't have any UD clients yet. As soon as they do I think I'll switch them - SETI is super slow on them anyway. I also run a single instance of SETI CLI on each of my duallie machines - I just started this since UD isn't SMP capable/aware.

P.S. - I think the GUI version of Win32 SETI is SMP capable - on my duallies it uses 100% of my CPU time except for the first few seconds.

LNCPapa

LNCPapa, Sep 10, 2002

#5
Julio Franco TechSpot Editor

Any comments on Folding? I've put UD to sleep for a few hours while I try the Folding@home client, I'm running the Windows GUI.

I would run the command based utility however I read somewhere that the GUI doesn't even take 1% of the CPU time, so should be the same... it has been running for quite a few hours now and it's only at 19%...

Anyway, I've got to test it more - Folding seems to be very popular too (nothing like SETI though), haven't found any 3rd parties yet and no caching apparently either, however packets do resume...

http://folding.stanford.edu

Oh and BTW, Folding does have SMP, Linux and Mac support. If some people can test it, it would be nice to have more opinions other than mine :grinthumb .

Julio Franco, Sep 10, 2002

#6
ss1 Newcomer, in training

Folding@home gets my vote as well.

ss1.

ss1, Sep 10, 2002

#7
Phantasm66 Newcomer, in training

Personally, my feelings on aliens are as follows:

1)Of course there are aliens out there in the vastness of space. The universe will be teeming with life.

2)Why the Hell would they come here?

3)Who cares anyway?

Curing cancer, on the other hand, is a much more worthwhile cause.

So I vote for UD.

Phantasm66, Sep 10, 2002

#8
Per Hansson TechSpot Server Guru

Well, for 56k users, and for any OS other than Windows the Distributed RC5 CLient would be a good choice, it allows you to choose how many packets you like to download, and it allows resuming...

Though hacking 64bit encryption isn't really "well used computing time" to put it in that way...

Seti which has been mentioned above feels a bit like RC5, ok, so we find an Alien signal coming 2000 galaxies away, but what are we going to do with that?

Then atlast we have UD, which the people at out IRC channel has started a small unoficcial team of, and while it's cause is one of the best, running the client is a whole other thing...

Just like everyone else has said... Though as it gets older it will probably offer more features so if we begin runnig it now we should have an edge regarding competition when the client gains more popularity...

So my vote is for the UD Cancer agent...

Per Hansson, Sep 10, 2002

#9
Didou Bowtie extraordinair!

I vote for UD too. If aliens are smart, they prolly won't touch Earth with a 50 foot pole.

The UD caching issue is a bit problematic but looking at their forums, it seems it has been brought to their attention quite a few times allready.

Just for clarification, it doesn't restart at 0% when you close it. According to the UD FAQs, here's what happens when you quit the process. ( UD Forums FAQs )

Note... the agent will only save what you have done through the last completed molecule. It will resume whenever you restart it at the beginning of the molecule you are currently processing.

Here is a post by Rymond H. Earle...

It takes a variable length of time to finish a work unit. Although many of them can be processed in under an hour on a fast machine, others can run for days if one or more very complex molecules exist within the work unit.
If you select the THINK client's information page ((i) icon on the main screen), you'll see the current molecule name of at the top left of the largest section. The name has the following general format: 123-456-xxx, where xxx can either be blank (for the root molecule), or be a number between 1 and 100 (indicating the nth derivative of the molecule being processed).

Each work unit to be processed, that you receive, includes a set number (100/200) of core molecules. Each of the 100/200 core molecules has a name (123-456 in the case above). The client can form up to 100 derivative molecules from each of these 100/200 core molecules. Each molecule plus any of its 100 derivatives can have up to 1,000,000 conformers. The first derivative molecule (represented by the -xx) is -1, the second is -2, and the last is -100. When derivative -100 is processed, the processing for that molecule (123-456 in the case above) is completed, and a checkpoint may be written. Whether a checkpoint is written or not at the end of the processing of a given molecule or molecules is dependent upon how quickly the processing for that molecule or molecules has proceeded. On quickly processed work units (i.e. under one hour) I've seen the client not take a checkpoint for several molecules in a row. You can generally judge when a checkpoint has been taken by the sound of a little disk activity that occurs right around the processing of the -100 derivative.

Regarding de novo structures that you will sometimes see being processed when you're watching the graphical display. Sometimes a molecule may have extra potential if it's slightly altered in some way. The client attempts small variations or substitution of atoms to increase the number of molecules being evaluated. These newly altered molecules are the de novo derivatives, or structures, which are also processed by the client.

The completion per centage is non-linear (i.e. it does not represent wall clock time), but rather indicates the number of completed core molecules within the set of 100/200. This is because the client has no way to determine how complex a core molecule will be until it has been processed. Therefore when you note that the completion per centage has not changed for a while, first check that the time is being updated each second on the main screen and the legend says "Primary task is executing...". If you find these conditions, switch to the information screen to see if the derivative number, the conformer count, the de novo structure count, or the hit count, is increasing. It depends on how fast a machine you have and what else it is doing at the time as to how quickly these numbers will be changing.

If you need to stop and restart the client, try to do so as quickly as possible after the client has written a checkpoint (derivative -100 for the current molecule was processed) so that you'll duplicate as little processing time as possible when you restart the client. When it is restarted, it will not redraw the current molecule, but will wait until it begins processing the next one before updating the display. When this is the case, you'll see a message near the top of the information page that states "Structure not ready for display".
This is a normal condition and the message is nothing to be alarmed at.

The molecules in the Superoxide Dismutase protein target in particular "conform" very well. That's why there are often a lot of hits associated with them and why some of them seem to take so long to crunch. Those that do are considered to be Long Running Molecules (LRM)! If you're processing an LRM (in a Superoxide Dismutase protein or otherwise), stopping the client is generally not advisable if at all possible, because you can lose hours or even days of processing time when restarting it.

Taken from Project FAQ's: Cancer Research Project

Q. My Task Execution Progress bar is frozen at a particular percentage. When I close the United Devices Agent and restart it, it stays at that percentage but I lose some time. Is the Agent frozen?

A. If you look at the top of the THINK screen, you will see a display like SOME-Molecule-E843F-02. The name of the current molecule that the agent is working on is SOME-Molecule-E843F. The -02 indicates a derivative molecule, formed from the original molecule. In this case, the agent is working on the second derivative of the SOME-Molecule-E843F molecule. There can be up to 100 derivatives of each molecule in a work unit.
Currently, the checkpoints in THINK are set after every successfully completed molecule (not its derivatives). If you turn off your PC, or restart the agent, you may be stopping it from completing the current molecule. When you restart it is going back to the beginning of that complex molecule. This is why it seems you are losing time. We are working on a fix, but in the meantime try allowing the agent to run for a longer period, and keep an eye on the conformers to ensure that they are not looping during this run period.

It's likely that THINK task is not stuck. The % indicator is not a linear time indicator; it counts the percentage of molecules in the work unit that have completed processing. The reason that the time on the task bar continues to increment even though the % counter stays the same is that the agent is still doing matches with the protein and the same molecule-the molecule (or one of its derivatives) is just complex. That is why the conformers are still rising - it is still analyzing. We have seen some cases of molecules taking several days to complete!

The time it takes to process a work unit varies depending on the Members' system and the molecules' complexity. All molecules are different. Some molecules may take a few moments to be processed by the agent, while others can take several minutes, or even hours depending on the speed of the machine. Since the % indicator increments by molecule, it will stay the same for long periods as the agent works through these more complex molecules.

Didou, Sep 10, 2002

#10
Per Hansson TechSpot Server Guru

Originally posted by TS | Julio
UD is nice, it has the best looking utility of all and isn't as consuming as SETI GUI, however the problem I'm seeing with it is that you can't have WU (work units) cached, you can't resume packages (so if you restart your PC you go back to 0%) and WU take quite a longer time than SETI, which is a bad thing because... you can't resume!

I just tried this an UD did resume when I restarted my comp...

Per Hansson, Sep 10, 2002

#11
Phantasm66 Newcomer, in training

If it comes down to technical aspects playing a part in the decision I move that the speed of people's connections be a more prime motivator than OS or platform.

My reasons for this are:

1)Even although I have a 512mbps cable connection, I would hazard a guess that the vast majority of members are dial up users, and

2)That even although I have multiple operating systems, the vast majority are PC Windows users.

The main criteria then should be that its low bandwidth friendly. If its not non-Windows compatible then sorry but I don't think that should be a primary factor.

Before someone bursts in with cries of "That's not fair I have Linux / Mac OS" or something, please consider that that is essentially just as selfish as me saying "That's not fair I have cable so stuff you 56k lamerz...."

Phantasm66, Sep 10, 2002

#12
Phantasm66 Newcomer, in training

From a "good cause" perspective, surely curing cancer is the most worthwhile cause put forward so far.

1 in 4 of us will get some kind of cancer at some point in our lives. Anyone who has lived with someone suffering from cancer will know about the almost unimaginable pain and suffering that this can cause. And the high mortality rates cause immense grief and sorrow in the friends and relatives left behind when cancer prematurely kills.

I don't give a crap about aliens, etc.

Phantasm66, Sep 10, 2002

#13
csimon Newcomer, in training

I highly recommend Folding@home ...for more info check out our team site here ... www.teamicrontic.com ...we always welcome some real competition!!!:grinthumb

csimon, Sep 10, 2002

#14
muddocktor Newcomer, in training

Folding@Home

I've also come over from the Icrontic forums to ask you to try Folding@Home. Folding is run by Vijay Pande, who is an Assistant Professor of both Chemistry and Structual Biology at Stanford University. This program is run from Stanford and doesn't have ties to private enterprise, who might be tempted to use the results of this project to develop drugs or cures that would only be released though themselves.

Here are some links to the project:

Stanford Folding@Home hompage:
People at Stanford involved in the project:
The scientific background of folding:
The results page:
Finally, the download page:

As you can see on the downloads page, there are several versions of Folding@Home for Windows, Linux and Mac OSX. I presently use the console version in windows as it has the least issues. The GUI version runs fine also, unless you have older video drivers and I've also heard that 3dfx card have had issues in the past with the GUI version.

I hope that you folks will give serious consideration to the Folding@Home project. It is a long-term project and some of the work that you do now might just save your @ss later on in life.

muddocktor, Sep 10, 2002

#15
Elcarion TechSpot Paladin

UD gets my vote. Linux and Windows support is a plus.

Elcarion, Sep 10, 2002

#16
Vehementi TechSpot Paladin

They could only wish!

Vehementi, Sep 15, 2002

#17
Julio Franco TechSpot Editor

Things seem to be working nicely with UD IMO, we will probably stick with it and go public soon. I read that Icrontic thread, very friendly members over there, nice indeed.

In the other hand however I wanted to comment on those "website size" comments mostly made by our own members. The fact they have a larger forum community and more distributed computing participation has really nothing to do with a site being more visited. Each site has their own niche and we do our stuff pretty well... if you ask me I would say that TechSpot doubles Icrontic traffic - that is, around the whole site :grinthumb then again, this is a purely informative post, I really like Icrontic guys and their work, nothing personal.

Julio Franco, Sep 16, 2002

#18
Rick TechSpot Staff

I consider Techspot one of the larger non-corporate hardware sites... I know we generate massive traffic despite our more humble forum size.

Perhaps some acknowledgement on the front page of our distributed UD efforts will bolster membership. Right now, I am considering just letting LNCPapa single handedly outdo them.

Rick, Sep 16, 2002

#19
Vehementi TechSpot Paladin

LNCPapa is a one man team :rolleyes:

I'm just kidding of course, he's only our top performer

Vehementi, Sep 16, 2002

#20

(You must log in or sign up to reply here.)

Page 1 of 2

Thread Status:: Not open for further replies.

TechSpot | Technology News and Analysis
Copyright © 1998-2012, TechSpot, Inc.
Forum software by XenForo™
TechSpot is a registered trademark
All Rights Reserved