P859 160 points ?

[KingFish]

I saw that I just got two of the 858's. I choked when I saw the 33 hours. I thought it had to be wrong. EM already needs another update.

KingFish

[scott]

What the heck is this ?
F@H log stats reports it as 160 points
MBM reports it as 32 points
It has a 33 hour time per WU, The NF7 has never taken more than 18 hours to do anything ? I sure hope it is 160 pts.

Scott

[scott]

I guess it is 160 points !!

[Leonardo]

Yeah, I thought something was wrong too! Protein 1000, 858 on my XP 2800+ system is 53:56 minutes per frame. Maybe that is too long? On my P4 2.8 rig I've got an 850 and a 258 running simultaneously at 18:18 and 41:10 respectively. These new proteins are something else.

[t1rhino]

OMG :hair:

[Necropolis]

3 of my machines here in work have them. 1GHZ PIII 160+ Hours.

[mmonnin]

All for of my machines got one. I am not going to turn in any work for over a day now.

There is an update for EM3 since 2.4.9. Just the file that keeps all the WU points info and stuff.

Merged 2 posted from the other thread into this one. Thats why KingFishs post is first. Goes by time stamp.

[witenoiz]

The slowest box I have (an old Athlon 700) got one of those 858's. It is going to take almost 2 weeks :mean: Oh well, it isn't due for 40 days. :rarr: And for that, I get penalized 40 credits. :screwed: I get 120 points. Jack

[Dragstk]

After reading this thread, I went and checked my machines. One of them has an 858.
It's taking around 30 minutes per step. But between every step is this listing
Timered checkpoint triggered
Anybody know what this means?
Thanks;
Dragstk

[fatcat]

maybe checkpoints within a frame so if something happens @ 93% of a 30min frame u dont have to do all 30mins again. im getting a timered check every 10 mins.

[Dragstk]

Thanks FatCat.
I guess this is a good thing. I was hoping it wasn't something going wrong with my machine

[KingFish]

With the newest version of FAH, you have the option of setting how often the work gets saved. By default it's set at 15 minutes. If you are having 30 minute per frame it will reach that checkpoint at least once per frame. You can set it to save more or less often if you like.

KingFish

[mmonnin]

My computers are pretty stable so I set it to 30 min. Saving the files takes away folding time and since they are up all the time all that does is hurt folding.

[qparadox]

On F@H logstats its listed as 120 points, anyone know what's up with this? Are there two different point values or is it a fubar'd program? Witenoiz also said his is only 120 points. Both of my palaminos @ ~ 1.6 GHz have them and the PPW is the same @ 120 points but will obviously be increased significantly if these things are really 160 points.

*edit it seems F@H logstats is mixing up the final deadline information and credits for this particular WU. w00t those two machines will be over 500 PPW if I get a steady stream of these suckers! (about 2 days / WU).

[mmonnin]

Should be 160 points.

[witenoiz]

I moved the 858 OFF my Athlon 700 and over to my P4 2.8 and moved the Gromac from that box down to the Athlon. Who knows? I'm not waiting to see how long this is going to finish step 4. Stay tuned, it's time for bed. Jack

[mcwc]

I just got one of these suckers on my notebook and it's gonna take it a while :banghead:.

[Black Hawk]

25mins per frame here on a 2400+.

[KingFish]

I've gotten inundated with these suckers now.

KingFish

[mmonnin]

I moved the 858 OFF my Athlon 700 and over to my P4 2.8 and moved the Gromac from that box down to the Athlon. Who knows? I'm not waiting to see how long this is going to finish step 4. Stay tuned, it's time for bed. Jack

Would be better the way it was.

[Straight_Man]

Would be better the way it was.

A P4 3.2 GHz does them in about 24.5-25.5 hours of folding while under load-- picking up email, surfing, Quanta Gold webdev, etc. Not TOO shabby. The Barton was taking 22 min per percentage, or 36-37 hours. I think in this case, the p858 Gromacs does better on the P4, but there is a speed difference also in that the 2500+ is running at 1.99 GHz (2010 MHz). The 858 and 859 are likely to be fastest on a GHz speed basis, irregardless of hardware platform. The Barton choked on a p859, which is also a 160 point dimer, so 859 might be tweaked a bit more for Intel SSE2 tunings.

John D.

[witenoiz]

Would be better the way it was. I think not - The boxes in question are JRock and j20 - before the move j20 was going to take over 300 hours! Now the 858 on JRock is about 21 minutes a step. Jack

[csimon]

KingFish ...what proc are you using on the Farm-4?

[KingFish]

All my farm specs are:
Biostar M7VQK Pro mobo
AMD 2400+ Thoro core
256 MB "el cheapo" ram (except for Farm-1, which has 512)
300W "el cheapo" power supply
coolermaster aluminum heatsink

[csimon]

there are big differences in frame times between Farm 1-4 ...I was just curious why so much difference. Doesn't add up.

[KingFish]

farm 1 was doing an 859, farm 2-4 were on an 858

some of the nodes had already turned in some wu before i got this so I didn't check the time per frame

KingFish

//edit
dug up the logfile
farm-1 was working on an 859 and was getting approx 24 min per frame
farm-2 was working on an 858 and was getting approx 28 min per frame
farm-3 was working on an 858 and was getting approx 27 min per frame
farm-4 was working on an 858 and was getting approx 28 min per frame

I didn't take into account the seconds so it appears that they were all in line

[mmonnin]

Sure it will take more hours on the Athlon than the P4 but look at points/hour. The P4 will be A LOT, I mean A LOT, better off on gromacs.

Look at the stats, everyones points the past 2 days or so are WAY LOW. I mean like half the production of normal. Wait til these start turning back in and see a huge jump in points.

I sent 2 in and a 3rd will go in in a little bit. Still got one on all 4 computers.

[edcentric]

I left my dial up TBird 1.6 running five wus when I left town last Sunday. I get home and two are still running......
With two running at once I am getting about 2.5 hours/frame. For a single one it would take about 5 days. 30pts/day is about right for this box.
Now, if I could only ask for these when I go traveling all would be fine.

[qparadox]

I woke up today and 7 of my 10 machines have em. Of course they're the 7 slowest machines but I'm too lazy (and wimpy) to risk transferring them when they're partway done. SM17 also has one but its gonna take over 3 days to complete!.

In other news I just bought an ASROCK board (no laughing) the K7S41GX because they were on for $49 and I needed to replace another MSI board with exploding caps (that makes 3 , gj MSI). Its seems like a decent board so far an even supports 5 bit FID (via jumpers unfortunately) but I'm not complaining. Its currently running an XP1700+ tbred A @ 11x166. Overclocking further seems to be a little flaky but once I find 5 spare jumpers I'll see what I can do with the multipliers :D. It might be an interesting board to look into for farms at that price point. It supports FSB333 / DDR333 Barton processors etc running on some SIS741 chipset.

[Leonardo]

I just bought an ASROCK board
;D

Oh, sorry, you said not to laugh. Silly me - I always though ASROC was an acronym for anti-submarine rocket! (Too much Tom Clancy reading.)

[Al_Capown]

My HT P4 has got 2 running right now. Getting 32 minutes per frame (per each console). According to my calculations it'll take roughly 2.22 days to finish :O. I'll be rakin in the points at least :-/

[qparadox]

We'll see if it lasts Leo ;). It certainly works better so far than those cheapo ECS boards which I always had IRQ trouble with in Linux. $50 cdn ($35 US) with a 5 bit FID and everything onboard isn't too shabby :D. Clock for clock it appears to be even with my KT333 boards.

[Leonardo]

I'm never against a bargain. Good luck with your board. Frankly, I hadn't heard of Asrock brand before. Makes me wonder if it's a value label of another board, giving the manufacturer opportunity to sell in another niche without having to engineer another product.

[Al_Capown]

I think, as geeky told me, they're made by asus. Basicly asus's line of budget boards.

[Geeky1]

Yeah, ASRock is ASUS' budget line of boards.

[MrBill]

p858_p53dimer858: 34 minutes on my Tbird 1.2@1.33

Not as bad as I thought it would be. :)

[Straight_Man]

The 858 on the P4 Northwood 3.2 ran about 11-16 (quite variable dependign on other load) min per percent (aka "frame"), average 4.35 frames per hour. On the Barton, well the Barton 2500+ at 2 GHz has done ONE, not enough for stats to average as box kept doing spontaneous cool temp shutdowns, but it is doing a fairly similar 859 at about 2.4 %\hour or 26 minutes per frame\percent.

John D.

[qparadox]

I've noticed that the 858 seems to like my Laptop P-M a lot more than my desktop XP's running 400-500 MHz faster. The larger size of the data (they increased the precision for that new core IIRC) like means that a large amount of cache results in the huge boost. Its terribly slow on my non-SSE machines resulting in a major decrease in PPW. (The XP's still fold faster but (Points / Time) / Mhz is 15-20% higher).

edit oops, for some reason FAHlog stats was telling me its a unit of type "100" so i assumed it was the new core. On second checking its parsing the project summary page wrong :/

Mr Bill does that tbird have SSE? I have a non-sse tbird 1.4 that's taking 1 hour 6 minutes / frame on the 858. That doesn't seem right ... is that tbird 1.55 or 1.56?

[mmonnin]

The new core (core_79) is not for this WU.

[witenoiz]

The grass I planted last week might be ready to mow before some of these things are complete. :rant: But, a few have finished, or at least something is going back. This is the talk of all the forums I've looked at. :type: It will all work for the best :confused: Ya Think? Jack

[profdlp]

My 1700+ (at 2000+ speed) has been munching one of these for a good long while. I've watched it since noon (12 hours now) and it's moved from about 60% to about 85%. That 160-point bump all at once will be a relief. I haven't called my dad or my daughter, but I suspect that they have tied into one of those, too. Not many points today, but look out down the road!

It looks like there are a lot of them out there - we haven't all been taking it easy this weekend:

[mmonnin]

It looks like that for a lot of the teams. Way low output the first 2 days they came out and then a surge like the 3rd day or so when they start coming back in.

[Necropolis]

Christ, I came in today to find 10 of my machines here have them. Oh well, now I have to wait a few days for them to finish (the machines here being PIII's)

[KingFish]

I've finally slogged through some of them and kicked them out the door. I have a feeling some will be back though.

KingFish

[profdlp]

Interesting the effect these big fella's have on your Folding. All rigs stayed the same, but my total WU's dropped in half while my point total increased almost 10%. (See weeks marked in red. The middle week was during the transition period when they just started to show up.)

[mmonnin]

Yep I went from WUs per day at somewhere in the teans to about 5-7.

[Straight_Man]

Very interesting-- with two boxes, I get less than 160 points every other day, then 300-400 in a day in a moving cycle depending on what I get-- average is about 180 points a day now for me. I am also getting p931 and p934 coded WUs, they take similar time frames to the p858 and 859s-- actually, the 858 is taking longest, then the 934, then the 859, then the 931(they are close to same time-wise, within 2-3 min per percentage across whole range of those four). I have not figured out the pointages for the 931 and 934s yet.

John D.

[profdlp]

Is anyone still getting the 160 point babies? I haven't seen one for a while now.

[qparadox]

I just finished my last one yesterday. My farm is now full of 6 520's after just finishing off a batch of 10 Tinkers :/. Production fell off nearly 40% during the run of tinkers, it hurt the p-m the worst which dropped over half (from 400 PPW to 140 PPW).

[KingFish]

I'm all clear of them at the moment. Been getting a lot of the p724_Abeta21-43-amberGS's though. What are those worth? Em-dc has it all scrambled due to the length of the name. I think it's 49 pts but not sure.

KingFish

[csimon]

I'm all clear of them at the moment. Been getting a lot of the p724_Abeta21-43-amberGS's though. What are those worth? Em-dc has it all scrambled due to the length of the name. I think it's 49 pts but not sure.

KingFish

724 171.67.89.152 p724_Abeta21-43-amberGS 47747 17.00 17.00 49.00 100 GROMACS Description vvishal

[edcentric]

My 1.4 TBird had two at once (I am on dial up). Check was set at 29 min. and it was hitting 4 between frames. About 2:15, or 9 days to do a pair of them.

[Dragstk]

No 859's for about a week. But I also had a small run of Tinkers. Including 1 Tinker that ran 20 steps, closed out, and I got credit for it.

[Leonardo]

My last one was about a week ago.

[dragonV8]

None of the p858, p859's for quite some time now. We're getting our fair share of p638, p639's and p520's and a variety of 11 pointers.

[mmonnin]

No 160 pointers in a while, had 3 tinkers past few days tho. 6:19 a frame on the p724.

[Dragstk]

As of this morning, all 3 of my machines have got p212_villinUR212, all taking about 21 minutes to run (ironic, huh?). Each one is worth 160 pts too.

[csimon]

the majority of my folders are getting p212's at the moment ...EMIII reports them as being 75 points but I'm sure that they were changed within the past 2-3 days to 160 points. :fold:

[Leonardo]

Intel Rig: two P520 BBA 5 @33pts each (EM calculator - 63.2pts/day ea)
AMD Rig: one P696 L939 @70.9pts (EM calculator - 52.4pts/day)

This is fairly typical. I think I need to double check to see if
-advmethods is on for the AMD's client.

Office machines? Hmm, don't know what they're running now; they seem to get (s)Tinkers even with -advmethods switched on.

[mmonnin]

Might want to check the performance fraction, if they are below 0.8 then no gromacs for them.

[Leonardo]

Yes, I'm imagining that's a problem. I've tried to enable hyperthreading in the office machines, but there is some kind of BIOS lock - or they're engineered (frigging CRIPPLED) from the factory to be non-hyperthreaded. Portions of these Dells' BIOS I can change; other areas will not change.

[profdlp]

I spoke too soon. Two of them just came in on out-of-town computers.<<===Make that three! :rockon:

:clap: :fold: :fold: