Downward trend of daily points

miragemirage
edited November 2006 in Folding@Home
Have you noticed that daily production rate has a steep downward trend since last 10 days. :nudge:
http://folding.extremeoverclocking.com/team_summary.php?s=&t=93

Can it be because of approaching Thanksgiving or recent discussion about the electricity bills? I have not been in the team long enough to know seasonal trends but if this is not a seasonal trend, let's keep folding :Rocker:

Comments

  • ThermalfishThermalfish Melbourne, Australia
    edited November 2006
    Ive lost access to over 30 CPUs in the past 2 weeks. :(
  • miragemirage
    edited November 2006
    Thermalfish wrote:
    Ive lost access to over 30 CPUs in the past 2 weeks. :(

    Sorry to hear that, are you working in an IT department or similar ?
  • ThermalfishThermalfish Melbourne, Australia
    edited November 2006
    mirage wrote:
    Sorry to hear that, are you working in an IT department or similar ?

    Yeah, at an audio/graphics/3Danimation college.
  • ShalimarShalimar Touching the Stars
    edited November 2006
    A lot of new work units "projects" have been released by dan ensign "over 70thousand i think", it is very well known that most of his projects take forever to fold "including the 5 way melts" with a less than fair points allocation for the time needed to fold alot of his work etc.

    The one thing i do not understand is that the projects that yield better folding times with bigger points allocations.....use much more ram.

    His projects use hardly any ram which leaves the question as to why he did not compile his work to use more memory which would "i think!" fold a lot faster.

    I am also getting some of those wu's that use no optimisations what so ever, they strictly rely on raw Megahrtz processing power period.

    "Just my 55cents worth"

    Shal
  • miragemirage
    edited November 2006
    Thermalfish wrote:
    Yeah, at an audio/graphics/3Danimation college.

    Nice job ! I hope you get those 30 cpus back soon. I do also build/maintain/administer Linux HPC clusters besides using them for computational research. But, there is no way I can run F@H in the clusters, they are always full with our own jobs.
  • miragemirage
    edited November 2006
    Shalimar wrote:
    A lot of new work units "projects" have been released by dan ensign "over 70thousand i think", it is very well known that most of his projects take forever to fold "including the 5 way melts" with a less than fair points allocation for the time needed to fold alot of his work etc.

    The one thing i do not understand is that the projects that yield better folding times with bigger points allocations.....use much more ram.

    His projects use hardly any ram which leaves the question as to why he did not compile his work to use more memory which would "i think!" fold a lot faster.

    I am also getting some of those wu's that use no optimisations what so ever, they strictly rely on raw Megahrtz processing power period.

    "Just my 55cents worth"

    Shal

    I have been getting extremely slow running jobs in the recent days. But they use at least 100MB memory each. Can you please tell how to identify those crappy jobs? Can you check out the other thread I started today about a job using 3DNow optimizations on a P4 Northwood. Do you think it is one of those jobs you were mentioning about?
  • ShalimarShalimar Touching the Stars
    edited November 2006
    mirage wrote:
    I have been getting extremely slow running jobs in the recent days. But they use at least 100MB memory each.

    If they are using roughly 128meg or around abouts that. Then it is more than likely that you are folding ribo's....yes they will fold a little slow but the points return is very fair for those wu's. They are much much better than the 5 way melts by comparison.

    As in they fold a lot faster than the melts and give better returns, but as with all proteins they will not be around for ever and another protein that needs extensive research will replace it.

    Keep in mind that whatever work you may get, then most everyone else will recieve the same. "Well that is what i was led to believe"

    Shal

    PS:The Downward trend of daily points is across the board affecting all teams/users etc. It affects some teams more than others because one team may have a larger majority of intel or amd cpu's etc.
  • SPIKE09SPIKE09 Scatland
    edited November 2006
    mirage wrote:
    I have been getting extremely slow running jobs in the recent days. But they use at least 100MB memory each. Can you please tell how to identify those crappy jobs? Maybe we can just kill them and request a new one. Can you check out the other thread I started today about a job using 3DNow optimizations on a P4 Northwood. Do you think it is one of those jobs you were mentioning about?
    Just a Random thought, please do not delete WU's it can cause big delay's to the project, take the Hit like the rest of us the 21xx wu's are a PITA and are dropping everyone's PPD.

    Spike
  • DonutDonut Maine New
    edited November 2006
    SPIKE09 wrote:
    Just a Random thought, please do not delete WU's it can cause big delay's to the project, take the Hit like the rest of us the 21xx wu's are a PITA and are dropping everyone's PPD.

    Spike

    I agree with Spike on this one, just put on the fake smile and pretend you like it.

    In my own observation, it seems like there are more non-bonus WUs being sent out. (like the 5-way "patty melts")

    Maybe somebody can clarify for me- aren't the points based on Stanfords reference machine putting out 100ppd? (I think it's a 2.8ghz P4)
  • ShalimarShalimar Touching the Stars
    edited November 2006
    Donut wrote:

    In my own observation, it seems like there are more non-bonus WUs being sent out. (like the 5-way "patty melts")

    Maybe somebody can clarify for me- aren't the points based on Stanfords reference machine putting out 100ppd? (I think it's a 2.8ghz P4)

    Yes it is a P4 2.8Ghz pc that the points awarded are based on.

    What i can not fathom though is that the patty melts take a lot longer to fold than say a ribo, with the returns being a lot less. :confused:

    I honestly do not understand how stanford structures a certain point value to a certain protein. I gave up trying to understand how it all works a long time ago, i just fold whatever comes my way.

    I think a massive problem will arise when they release the new folding client which i believe will be a step backwards. I could be wrong but i have heard that the new client may allow you to choose servers amongst other things etc. "If that is the case then i can see it now, they will have servers that are stretched to the limit."

    Oh well, only time will tell if it works or not.

    Shal
  • miragemirage
    edited November 2006
    Well, I hit a sensitive point I guess, that was a random thought since I started the thread to point out that the team daily points were down since last ten days. My own points are not down, let's forget about deleting the jobs. Shall we check our machines and make sure they are running properly?
  • SPIKE09SPIKE09 Scatland
    edited November 2006
    Donut wrote:
    I agree with Spike on this one, just put on the fake smile and pretend you like it.

    In my own observation, it seems like there are more non-bonus WUs being sent out. (like the 5-way "patty melts")

    Maybe somebody can clarify for me- aren't the points based on Stanfords reference machine putting out 100ppd? (I think it's a 2.8ghz P4)
    It's worse that that, the rating is actually supposed to produce 110PPD or 39PPD per Ghz. Unfortunately Dan Ensign's current crop of WU's do not use all the SSE instruction set's to the fullest.:banghead:
  • QCHQCH Ancient Guru Chicago Area - USA Icrontian
    edited November 2006
    mirage wrote:
    I have been getting extremely slow running jobs in the recent days. But they use at least 100MB memory each. Can you please tell how to identify those crappy jobs? Can you check out the other thread I started today about a job using 3DNow optimizations on a P4 Northwood. Do you think it is one of those jobs you were mentioning about?

    As a avid Folder interested in the project more than the points... PLEASE DO NOT DELETE Work Units. I am very disappointed with how the points are determined but as long as my kids and grandkids can be saved from some serious diseases... I'm happy. It's the best gift I can give them. :D
  • miragemirage
    edited November 2006
    QCH2002 wrote:
    As a avid Folder interested in the project more than the points... PLEASE DO NOT DELETE Work Units. I am very disappointed with how the points are determined but as long as my kids and grandkids can be saved from some serious diseases... I'm happy. It's the best gift I can give them. :D

    Since I strongly feel the same about the importance of the scientific findings, let me say, I feel a little offended with your reminder. :( It was completely my own decision to start folding for a good cause.

    PS: I have already modified my post accordingly.
  • the_technocratthe_technocrat IC-MotY1 Indy Icrontian
    edited November 2006
    QCH2002 wrote:
    As a avid Folder interested in the project more than the points... PLEASE DO NOT DELETE Work Units. I am very disappointed with how the points are determined but as long as my kids and grandkids can be saved from some serious diseases... I'm happy. It's the best gift I can give them. :D

    I agree.

    The points per day suck right now, but they suck for everyone. I'm sure some people will kill off low point WU's (maybe on other teams... :) ) but the points aren't really why I do this anyway...

    Eh, just keep folding, either the crappy WU's will get used up, or Stanford will change the points. Either way, it's all research that's getting done.
  • QCHQCH Ancient Guru Chicago Area - USA Icrontian
    edited November 2006
    mirage wrote:
    Since I strongly feel the same about the importance of the scientific findings, let me say, I feel a little offended with your reminder. :( It was completely my own decision to start folding for a good cause.

    PS: I have already modified my post accordingly.
    No offense intended. :D I am making another statement about my beliefs. There are those that would rather jump in here just to argue about "points are not the reason to fold."... I figured I would make the statement and avoid any similar posts.

    I know that many folders have decided to swap Work Units between PCs to maximize output. There are also many folders (not many, if any, here at Short-Media) that are ONLY in it for the points. They will "BORG" / Hack any PC in order to get it folding. These are the same people that feel it necessary to delete crappy WUs to maximize points. Short-Media is a family of folders... we want to do some good for society and do it with style!!! :vimp:

    Basically, I was making a political statement about those individuals, not inferring that you are one of those. To that end, I am sorry if I offended you. :respect:
  • miragemirage
    edited November 2006
    QCH2002 wrote:
    No offense intended. :D I am making another statement about my beliefs. There are those that would rather jump in here just to argue about "points are not the reason to fold."... I figured I would make the statement and avoid any similar posts.

    I know that many folders have decided to swap Work Units between PCs to maximize output. There are also many folders (not many, if any, here at Short-Media) that are ONLY in it for the points. They will "BORG" / Hack any PC in order to get it folding. These are the same people that feel it necessary to delete crappy WUs to maximize points. Short-Media is a family of folders... we want to do some good for society and do it with style!!! :vimp:

    Basically, I was making a political statement about those individuals, not inferring that you are one of those. To that end, I am sorry if I offended you. :respect:


    That is okay, no hard feelings :)

    At this point, may I ask a moderator to close this thread to avoid any more point counting discussions, please.
  • QCHQCH Ancient Guru Chicago Area - USA Icrontian
    edited November 2006
    I can...
This discussion has been closed.