QMD's temporarily off line

DonutDonut Maine New
edited February 2006 in Folding@Home
I just read this over at the folding forums.http://forum.folding-community.org/viewtopic.php?p=121186&sid=0a629d66124c4530936bdf256148feca#121186

I would guess they'll run out what's on the servers now, then stop for at least a couple of weeks.

Comments

  • edited January 2006
    WoooBoyyy, all the Pee4 users are going to take a massive hit for a few weeks then. :eek:
  • LeonardoLeonardo Wake up and smell the glaciers Eagle River, Alaska Icrontian
    edited January 2006
    WoooBoyyy, all the Pee4 users are going to take a massive hit for a few weeks then.
    Please, please no sTinkers, no sTinkers! :sad2: :hair:




    Maybe....Double Gromacs? :D


    Oh...I just checked. My Intels now have a mix of yet to be completed QMDs, one Tinker, and two Gromacs. :shakehead What a shame. :shakehead I was so spoiled on nearly exclusive QMD production. Oh well, I would imagine my competition will be taking points hits as well.
  • edcentricedcentric near Milwaukee, Wisconsin Icrontian
    edited January 2006
    Tinker! Tinker! Tinker!
    I still have a couple of old TBirds folding. The only way that I can really make points is with Tinkers.

    Though all in all, the current mix of wus is working well for me. I have seen my points come back from a few months ago.
  • csimoncsimon Acadiana Icrontian
    edited January 2006
    I haven't gotten QMD's for a while here at work ...just mostly gromacs which is cool.

    One thing I've noticed about tinkers is that the 241 pointer folds faster than the 239er's. Who know's I just whish I had time to sit and download 1000 of them! =o)

    My dual xeon at 2.6g is getting 225 ppd right now and my opti at home is getting roughly 460 ppd. Is the difference really that drastic?
  • LeonardoLeonardo Wake up and smell the glaciers Eagle River, Alaska Icrontian
    edited January 2006
    It really seems like the dual Xeon ought to do better than that, but then there's the restricted data bus. I'll check when I get home today, but I think my Northwoods in hyperthreading are producing (without QMD) as much as your Xeon rig, and the Northwood rigs are only one CPU each. I'll check when I get home.
  • csimoncsimon Acadiana Icrontian
    edited January 2006
    Leonardo wrote:
    It really seems like the dual Xeon ought to do better than that, but then there's the restricted data bus. I'll check when I get home today, but I think my Northwoods in hyperthreading are producing (without QMD) as much as your Xeon rig, and the Northwood rigs are only one CPU each. I'll check when I get home.
    Another monkey wrench ...right now on the Xeons I am folding a p1136 tinker and a p1710 gromac ...241 & 116 pts respectively at 229 ppd. The Xeons are at 2.66ghz each.
    At home I know I have the 241 & 239 pointers on the Opti 175 @ 2.62ghz and production is at least 460ppd if not a little more.

    I appreciate your effort Leon.:respect:

    Here is a snap of the xeon production ...I'll post one of the opti tonight from home.

    edit:\ I've added the opti clip at 440 ppd.
  • profdlpprofdlp The Holy City Of Westlake, Ohio
    edited January 2006
    csimon, that sort of disparity between WU's is what drove me to setting all my rigs to the "endless" style.
  • csimoncsimon Acadiana Icrontian
    edited January 2006
    profdlp wrote:
    csimon, that sort of disparity between WU's is what drove me to setting all my rigs to the "endless" style.
    whatchew talkin' bout willis?

    I have both systems setup requesting endless wu's. Maybe not correctly ...LOL
  • profdlpprofdlp The Holy City Of Westlake, Ohio
    edited January 2006
    csimon wrote:
    whatchew talkin' bout willis?

    I have both systems setup requesting endless wu's. Maybe not correctly ...LOL
    You might want to recheck your settings. Since I went that route every WU I've received on 10+ rigs has been of either the 239 or 241 point variety. I used to suffer in silence through WU's that turned my A64 3500+ into a 500ppw weakling, but no longer... :necro:
  • csimoncsimon Acadiana Icrontian
    edited January 2006
    profdlp wrote:
    You might want to recheck your settings. Since I went that route every WU I've received on 10+ rigs has been of either the 239 or 241 point variety. I used to suffer in silence through WU's that turned my A64 3500+ into a 500ppw weakling, but no longer... :necro:

    You have any intels?
  • profdlpprofdlp The Holy City Of Westlake, Ohio
    edited January 2006
    csimon wrote:
    You have any intels?
    Nope. I might give the other WU's a go if I did, though. :)
  • LeonardoLeonardo Wake up and smell the glaciers Eagle River, Alaska Icrontian
    edited January 2006
    When I left home this morning, one of the CPU cores on my D820 rig was processing 56-point sTinker. All my Intels are set to "large packets - OK" and "-advmethods". These are dark times, I tell you!

    DOH! Speak of the Devil: my laptop here at work just finished a QMD...and just got assigned a, a...56-point sTinker! :hair:

    It's revenge of the Bartons! It's a Winchester conspiracy! It's...just plain nutso.

    ;D
  • csimoncsimon Acadiana Icrontian
    edited January 2006
    Leonardo wrote:
    It's revenge of the Bartons! It's a Winchester conspiracy! It's...just plain nutso.
    ;D
    it's it's it's ...loodachris! :csimon:

    Another observation ...my opti is more productive at 2.62ghz than at 2.66ghz. It may be that a mobo issue but I'm more inclined to think that the processor just won't perform properly if not supplied w/ enough voltage to adequate it's frequency. Make sense?

    Another note ...I'm finishing one 239 pointer and just got another so now I have 2. These are definitely slower than the 241's but I've also noticed that one core is a little slower that the other. I've seen them perform equally in the past but more than likely at a lower frequency ...which may have something to do with the issue previously mentioned.

    I'll get another shot of this after dinner and after I get a different wu.

    edit:\ I posted the opti clip in the above post. It's at 440ppd with 2 239's.
  • DonutDonut Maine New
    edited January 2006
    Just switched all 3 of my AMDs to do tinkers, my Xeon has 3 56 pointers and 1 116. Yay (sarcasm)
  • scottscott Medina, Ohio Icrontian
    edited January 2006
    Yes , these are trying times for Intel owners.

    I have 2 P4 3.0 HT and 1 1.6 P4 I have been running 5 QMDs on them for many months now. Well I was.... My daily production went from 2100-2200ppd to 1500-1600 ppd. I guess I will be slipping back in the standings a bit. Oh ya , that 1500 - 1600 ppd is with adding a neighbors 1700 barton to the fold. and it is doing about 110ppd. So on those 3 machines I took a 700 point daily hit. ouch !


    I wonder what this will do to Jon and Sally ?





    I think I will have a beer.


    Scott
  • LeonardoLeonardo Wake up and smell the glaciers Eagle River, Alaska Icrontian
    edited February 2006
    After the QMDs ended, my Intel CPUs started processing P1136 Tinkers (see, I didn't write "STINKERS", did I) and P17xx Gromacs. Per machine, the PPD fell about 60%. (The AMD 2800 Barton rig remains consistent with it's diet of Tinkers for a daily average of about 165.) The Tinkers and Gromacs have cycled through, and I've got a bevy of "GbGromacs" - all the ones I have are "Protein: 1258 p2102_lambda_5way_melt_0_00000". I've updated Electron Microscope III, but it is not calculating point data or giving the work unit's total value. I cannot find these 2102 units at Stanford's Folding site or at Folding-Community.org. If someone knows anything about these creatures, please post what you know, or better, a link.

    [EDIT] OK, I found the answer at the Folding@Home FAH project points summary page. The 2102's I'm working on are 414 points each. I haven't yet calculated the daily production. There is hope for spoiled me.
  • roland_dmproland_dmp Oklahoma New
    edited February 2006
    Leonardo wrote:
    [EDIT] OK, I found the answer at the Folding@Home FAH project points summary page. The 2102's I'm working on are 414 points each. I haven't yet calculated the daily production. There is hope for spoiled me.

    Hopefully your 2102's are running faster than mine! There's 100 frames in these bad boys, and I'm averaging between 55 minutes to an hour per frame! Hopefully that's just because I've been working on my computer for the last few hours - I'll find out in the morning for sure. At this rate I think I'd be happier knocking out more than a couple of Tinkers a day than waiting over 4 days for a 414 pointer to finish.
  • LeonardoLeonardo Wake up and smell the glaciers Eagle River, Alaska Icrontian
    edited February 2006
    Is anyone else with the "-advmethods" flag on getting these 2102 GbGromacs?

    I've got enough time data now to calculate production. My slowest Intel Folding rig, No. 2 is producing at about 100ppd for each CPU, virtual and real. The fastest rig, No. 4, is producing at about 150ppd for each core. That's a total of 300 points a day compared to the 700+ the QMDs were bringing in for that machine daily. I'd be interested to see what an X2/AMD DC can do with these. Someone at Extremetech was saying their Opteron 170 (@2.75GHz) was finishing work units at 33 minutes per - that comes out to about 180ppd/360ppd for two cores. Yup, the D820 is bested by AMD on this protein.

    EDIT: Wouldn't you know it. The EMIII site updated the protein data file just minutes after I did all the algebra computations. Oh well, never hurts to exercise the brain.
  • LeonardoLeonardo Wake up and smell the glaciers Eagle River, Alaska Icrontian
    edited February 2006
    Roland, my machines are ranging from a slow of (2X) 57min/frame to (2X) 40min/frame.
  • DonutDonut Maine New
    edited February 2006
    Yup, picked up a 2102 during the night. 70min per step!(%). 4 1/2 days for 414 pts. Mind you this is WITH sse enabled! Definately got spoiled with the QMDs.
  • lemonlimelemonlime Canada Member
    edited February 2006
    Welcome to the AMD folding world intel users :D

    I've been running timeless WUs for some time now. It would be nice if when QMD units return, they get that SSE2 issue for AMD's sorted out.
  • roland_dmproland_dmp Oklahoma New
    edited February 2006
    lemonlime wrote:
    Welcome to the AMD folding world intel users :D

    I'm on an AMD, just new to folding, I think this is my first 2102. I haven't checked to see what my other rig (which is a P4) is folding yet, but I'm guessing something similar.

    Leonardo, I was looking at some more times this morning, and my low was around 40 minutes, so it sounds about the same as yours. I need to get EM running on my machine, seems easier than manually calculating all the time ;)
  • LeonardoLeonardo Wake up and smell the glaciers Eagle River, Alaska Icrontian
    edited February 2006
    Leonardo, I was looking at some more times this morning, and my low was around 40 minutes, so it sounds about the same as yours.
    Yes, for a single core. The 820 rig has two cores, each running at that rate. Rigs 1 and 3 each have only one core (real CPU), but in hyperthreading have two instances of Folding running, each cracking a 2102 at about 55+min per frame. So dual core and hyperthreading are still advantageous over a single non-HT Intel single core or AMD single core. The AMD dual cores, at least the X2s, from what I'm reading at Extremeoverclocking, are outperforming the Intel dual cores with GbGromacs.
  • roland_dmproland_dmp Oklahoma New
    edited February 2006
    I didn't mean to venture so far as to say my AMD was as fast as your cpu Leo :) I was just comparing times for a single process. I dream of the day I can upgrade to AMD X2.
  • LeonardoLeonardo Wake up and smell the glaciers Eagle River, Alaska Icrontian
    edited February 2006
    I didn't mean to venture so far as to say my AMD was as fast as your cpu Leo I was just comparing times for a single process. I dream of the day I can upgrade to AMD X2.

    Oh, I wasn't trying to brag about anything, that's why I brought up the superiority in this case with the AMD X2s. I'm not defensive about any of my computer stuff (unless someone badmouths homebuilt computers, then I let 'em have it! ).

    It really is a refreshing change to see AMD come out on top. I've been an AMD fan ever since the K7 Athlons came out. My first computer build was an Athlon Thunderbird 900MHz clocked at 1100MHz (woo hoo, broke the inpenetrable 1GHz barrier!). The reasons I built my Intel Northwood machines is because that particular setup, hyperthreading with two instances of Folding, is one application where those CPUs have historically performed very well. The D820 dual core rig was so that I can remain highly competitive, and that I got the CPU for a song on Ebay.
    I dream of the day I can upgrade to AMD X2.
    I hear you! I'm hoping to upgrade my Barton 2800 rig next summer to M2...maybe quad core? :doh:
  • roland_dmproland_dmp Oklahoma New
    edited February 2006
    Leonardo wrote:
    Oh, I wasn't trying to brag about anything, that's why I brought up the superiority in this case with the AMD X2s. I'm not defensive about any of my computer stuff (unless someone badmouths homebuilt computers, then I let 'em have it! ).

    Nor was I insinuating that you were :D
    Leonardo wrote:
    I hear you! I'm hoping to upgrade my Barton 2800 rig next summer to M2...maybe quad core? :doh:

    I'm trying to squeeze everything out of my tbred B 2600+ as it is myself. For a while I thought about getting the best CPU I could for this mobo, but decided it would be best to just let it die and save up for a new board with PCI-E and SLI goodness. Given, that might take a while at my salary (which is currently $0). I can't wait for DDR2!

    OK, sorry :topic:
  • DonutDonut Maine New
    edited February 2006
    lemonlime wrote:
    Welcome to the AMD folding world intel users :D

    .
    Ahhh I see, just like old Pierre. All my AMDs and I add 1 little Intel...:shakehead :D

    At least I was able to try Qmd on an Opteron.:p
  • LeonardoLeonardo Wake up and smell the glaciers Eagle River, Alaska Icrontian
    edited February 2006
    We expect the analysis to take at least a few weeks (and potentially longer, depending on what we find). The upshot will be that we will be moving to second stage QMD WU's, which is very exciting to us. Vijay Pande
    Which is very good news indeed!
Sign In or Register to comment.