| Author |
Message |
|---|
|
|
|
There's a challenge on the 321 project on BOINC/PG starting in a few days... November 19, at 18:00 UTC. It runs 5 days. Talk about it here!
I'll be in with all engines running. Although, I have a personal grudge against this particular subproject. In the past, it seems like I've had more bad luck with things like power failures, network outages, invalid tasks, and so on, as compared to any other subproject. This is nonsense I know but that's what I recall. Hoping for better luck this time!
These are fairly long tasks... over 12 hours on a fast box, IIRC.
Cheers,
--Gary |
|
|
 Tyler Project administrator
Volunteer tester
Send message
Joined: 4 Dec 12
Posts: 1077
ID: 183129
Credit: 1,280,170,555
RAC: 7,054
                     
|
|
Hopefully I'll remember and throw in a few cores.. I need ~100k for next badge anyways.
____________
275*2^3585539+1 is prime!!! (1079358 digits)
 Proud member of Aggie the Pew
 |
|
|
|
|
|
The E8400 in my desktop will take ~48 hrs per WU.
I'm throwing in a cc2.8xlarge from AWS that should contribute 32 WUs in hopefully less time than that.
I just noticed that Amazon has introduced a new generation of compute instances that are Ivy-Bridge-based: the c3 series.
The c3.8xlarge is ~20% faster than the Sandy-Bridge-based cc2.8xlarge (judging by Amazon's own "Compute Unit" ratings of 108 vs 88), but the base spot price is almost 40% higher (~$0.36 vs. ~$0.26 per hour). The spot market for the c3 series is also more volatile, probably because there is more demand for them. So I stuck with the "legacy" cc2 system this time.
____________
Proud member of Team Aggie the Pew
"Wir müssen wissen. Wir werden wissen."
"We must know, we shall know."
- David Hilbert, 1930 |
|
|
NeoVolunteer tester
Send message
Joined: 28 Oct 10
Posts: 710
ID: 71509
Credit: 91,178,992
RAC: 0
                   
|
|
I'll be there.
Going to start a 321 right now to check the run time.
Looks like 45 hours per BIONC on an i7-4600
Neo |
|
|
|
|
|
I'm in - I ran a few WUs the other day to see what the runtimes were like. This is a 4770 @ 4.3, and with 3 cores running plus a (slow) graphics card sieving away, they took about 10 and a quarter hours each. I might disable my GPU for the challenge for a slight speed increase (you'd hope...). |
|
|
Dave
Send message
Joined: 13 Feb 12
Posts: 2829
ID: 130544
Credit: 954,747,840
RAC: 5,071
               
|
|
i7-2600K @ 4.3 = 14.3 hrs. 432000 secs (5 days) divided by 51500 secs = 8.3 units therefore 32 units over 4 cores. |
|
|
TylerProject administrator
Volunteer tester
Send message
Joined: 4 Dec 12
Posts: 1077
ID: 183129
Credit: 1,280,170,555
RAC: 7,054
|
|
10.5hrs each on my i5-2500k @ 4.5GHz
____________
275*2^3585539+1 is prime!!! (1079358 digits)
Proud member of Aggie the Pew
|
|
|
NeoVolunteer tester
Send message
Joined: 28 Oct 10
Posts: 710
ID: 71509
Credit: 91,178,992
RAC: 0
|
|
Update:
Looks like 24 hours instead of 45 hours... that's much better. :)
Neo |
|
|
|
|
|
I've got all of my preferences updated and have "no new tasks" set in preparation for the start which is in less than 15 minutes as I type this. With these longer tasks, a lightning-fast start probably won't matter, but hey, why not?
Good luck to all rats!
--Gary |
|
|
Dave
Send message
Joined: 13 Feb 12
Posts: 2829
ID: 130544
Credit: 954,747,840
RAC: 5,071
|
|
Last unit finished 17:59:19, 1st challenge unit downloaded 18:00:03. |
|
|
|
|
|
Well I've got my 'big rig' rolling, and a handful of other older machines (some of which may not finish a WU within the 5 days - we'll see!
- Iain
____________
Twitter: IainBethune
Proud member of team "Aggie The Pew". Go Aggie!
3073428256125*2^1290000-1 is Prime! |
|
|
NeoVolunteer tester
Send message
Joined: 28 Oct 10
Posts: 710
ID: 71509
Credit: 91,178,992
RAC: 0
|
|
I'm in with 4 Wu's.
|
|
|
|
|
|
I'm crunching too - I made a (short) trip home just to be there at 18:00. I wasn't as quick as Dave though, I was a whole 10 seconds late.
Go AtP! |
|
|
mackerelVolunteer tester
Send message
Joined: 2 Oct 08
Posts: 2460
ID: 29980
Credit: 442,802,854
RAC: 10,291
                 
|
|
Knew I forgot to do something when I got home from work. Crunching about to start... |
|
|
|
|
Well I've got my 'big rig' rolling
And by big rig, I'm of course referring to my hex-core Mac Pro, not (unfortunately) this bad boy. However, I did get access during the early user testing period today and ran some 118,080 cores jobs - that's right, using 100,000 processors for a single parallel application. Unfortunately the machine is turned back over to the regular users now!
____________
Twitter: IainBethune
Proud member of team "Aggie The Pew". Go Aggie!
3073428256125*2^1290000-1 is Prime! |
|
|
|
|
|
Late to the party but a whole 2 cores blazing away. Iain, thats some horses you almost saddled 😄 |
|
|
|
|
|
I'm in with just a couple cores. I'm heading down to the "big easy" for a few days, so most of my rigs will be powered off. |
|
|
|
|
And by big rig, I'm of course referring to my hex-core Mac Pro, not (unfortunately) this bad boy. However, I did get access during the early user testing period today and ran some 118,080 cores jobs - that's right, using 100,000 processors for a single parallel application. Unfortunately the machine is turned back over to the regular users now!
That's fun! What sort of analysis were you doing on it?
I've got a 32-core instance churning in Amazon's Oregon data center for the Leonids. Not nearly as big, but it is fun to see all of those WUs going all at once.
____________
Proud member of Team Aggie the Pew
"Wir müssen wissen. Wir werden wissen."
"We must know, we shall know."
- David Hilbert, 1930 |
|
|
Dave
Send message
Joined: 13 Feb 12
Posts: 2829
ID: 130544
Credit: 954,747,840
RAC: 5,071
|
|
My machine's now showing offline + I'm away from home so there's bugger all I can do about it. |
|
|
|
|
My machine's now showing offline + I'm away from home so there's bugger all I can do about it.
I hate not having Out Of Band management for my home machines!
____________
Proud member of Team Aggie the Pew
"Wir müssen wissen. Wir werden wissen."
"We must know, we shall know."
- David Hilbert, 1930 |
|
|
|
|
And by big rig, I'm of course referring to my hex-core Mac Pro, not (unfortunately) this bad boy. However, I did get access during the early user testing period today and ran some 118,080 cores jobs - that's right, using 100,000 processors for a single parallel application. Unfortunately the machine is turned back over to the regular users now!
That's fun! What sort of analysis were you doing on it?
Since it's the national HPC service machine, the user base is pretty diverse. Mostly I'm involved with computational chemistry applications e.g. http://www.cp2k.org, which is what I was testing out yesterday. Other things I'm involved in at present include Finite element modelling of bone growth and Fluid dynamics of two-phase flows.. But there's loads of other stuff going on including weather/climate, particle physics, earth science etc...
____________
Twitter: IainBethune
Proud member of team "Aggie The Pew". Go Aggie!
3073428256125*2^1290000-1 is Prime! |
|
|
Dave
Send message
Joined: 13 Feb 12
Posts: 2829
ID: 130544
Credit: 954,747,840
RAC: 5,071
|
|
In the cinema.
Put th i3-4 on 321s. It's HT as usually sieving. Tough.
I'll be alright on Sat. I hope. |
|
|
|
|
|
I've added some Sandy Bridge power to my Haswell. Not literally, that would require a highly customised motherboard. |
|
|
Dave
Send message
Joined: 13 Feb 12
Posts: 2829
ID: 130544
Credit: 954,747,840
RAC: 5,071
|
|
Wish I could add some SB power to my erm SB. |
|
|
TylerProject administrator
Volunteer tester
Send message
Joined: 4 Dec 12
Posts: 1077
ID: 183129
Credit: 1,280,170,555
RAC: 7,054
|
|
My SB is doing quite well, 10.5 or so hours per task. What is your haswell getting?
____________
275*2^3585539+1 is prime!!! (1079358 digits)
Proud member of Aggie the Pew
|
|
|
Dave
Send message
Joined: 13 Feb 12
Posts: 2829
ID: 130544
Credit: 954,747,840
RAC: 5,071
|
|
BSOD "Modification of system code or a critical data structure was detected".
The computer has rebooted from a bugcheck. The bugcheck was: 0x00000109 (0xa3a039d89f4df52f, 0x0000000000000000, 0x40e3235ec6c3c9bc, 0x0000000000000101). A dump was saved in: C:\Windows\MEMORY.DMP. Report Id: 112214-12324-01.
Slowed from 4.3 to 4.2. Not that it'll make much difference. |
|
|
TylerProject administrator
Volunteer tester
Send message
Joined: 4 Dec 12
Posts: 1077
ID: 183129
Credit: 1,280,170,555
RAC: 7,054
|
BSOD "Modification of system code or a critical data structure was detected".
The computer has rebooted from a bugcheck. The bugcheck was: 0x00000109 (0xa3a039d89f4df52f, 0x0000000000000000, 0x40e3235ec6c3c9bc, 0x0000000000000101). A dump was saved in: C:\Windows\MEMORY.DMP. Report Id: 112214-12324-01.
Slowed from 4.3 to 4.2. Not that it'll make much difference.
Could be a RAM voltage issue...
I find the following link very useful: http://www.overclock.net/a/common-bsod-error-code-list-for-overclocking
____________
275*2^3585539+1 is prime!!! (1079358 digits)
Proud member of Aggie the Pew
|
|
|
|
|
My SB is doing quite well, 10.5 or so hours per task. What is your haswell getting?
Just under 10 hours (9.8); your SB is at 4.5, isn't it? My Haswell is 4.3, 2133MHz RAM. The SB is running at stock speed (3.4)* but has 1866MHz RAM - that's taking 15.7 hours.
One 321 WU is taking about twice the time to process on the SB, I don't think that indicates a prime though, does it? It's only for SR5 and ESP if I remember correctly.
*It's my old machine which I gave to my parents, that's why it's running at stock. They might get overly worried if I "mess about with it" too much ;) |
|
|
TylerProject administrator
Volunteer tester
Send message
Joined: 4 Dec 12
Posts: 1077
ID: 183129
Credit: 1,280,170,555
RAC: 7,054
|
My SB is doing quite well, 10.5 or so hours per task. What is your haswell getting?
Just under 10 hours (9.8); your SB is at 4.5, isn't it? My Haswell is 4.3, 2133MHz RAM. The SB is running at stock speed (3.4)* but has 1866MHz RAM - that's taking 15.7 hours.
*It's my old machine which I gave to my parents, that's why it's running at stock. They might get overly worried if I "mess about with it" too much ;)
Mine is running at 4.5GHz, with 1866MHz RAM. i5-2500k
____________
275*2^3585539+1 is prime!!! (1079358 digits)
Proud member of Aggie the Pew
|
|
|
|
|
|
I gotta get me some of that fast RAM... the times posted above are kicking mine, which tend to be 13-15 hours at best on o/c'd i7's (fully loaded, HT off, mem clock is 1600, 9-9-9-24). GPU crunching is off.
@gazzyk1ns: I agree that a long-running 321 task does *not* indicate a prime. It definitely can for SR5 and SGS... not sure about ESP (but I don't think so). I'd check the Task Manager to ensure there's no "runaway" process in the background hogging CPU, and if not, maybe a quick suspend/resume on the task in question.
--Gary |
|
|
|
|
@gazzyk1ns: I agree that a long-running 321 task does *not* indicate a prime. It definitely can for SR5 and SGS... not sure about ESP (but I don't think so). I'd check the Task Manager to ensure there's no "runaway" process in the background hogging CPU, and if not, maybe a quick suspend/resume on the task in question.
Correct, for anything which is Riesel or Proth (including ESP, 321 …) the first test is deterministic, so the runtime is the same irrespective of whether the result is prime or not prime. For SR5 the initial test is a PRP test, if it is positive, then several further passes are done to confirm if the candidate is indeed prime. For SGS, if the first test indicates a prime (definitely a prime, not a PRP test), then subsequent tests are made of the twin (p+2) and SG (2p+1) candidates.
So a long-running 321 task definitely indicates some kind of problem, not a prime - sorry!
____________
Twitter: IainBethune
Proud member of team "Aggie The Pew". Go Aggie!
3073428256125*2^1290000-1 is Prime! |
|
|
|
|
|
Cheers for the confirmation - I'm not at that machine any more*, so I'll just see how it turns out and then link you to the result, to see if anything was amiss. I did have a thought - it might be BOINC giving a false reading because the WU was started and then put on hold (by BOINC, i.e. "waiting to run") when I remembered to disable HT after initially forgetting. Not sure if that's a plausible explanation or not, in any case, I think everything will be pretty much OK.
Famous last words.
*If I go for a midnight stroll, that will change. |
|
|
|
|
...
*If I go for a midnight stroll, that will change.
...
I *love* the midnight stroll. Sometimes if I'm just goofing off on my running "training", I'll go just a few miles very late at night. Quiet, cool, and no traffic. I have been bothered by the police a couple of times though.
AtP is doing very well with something over a day to go in the challenge. We're in a pitched battle with the Germans for 3rd vs. 4th, as I type this. EVGA and the Cows are in the rear-view mirror. Cheers all 'round!
--Gary |
|
|
Dave
Send message
Joined: 13 Feb 12
Posts: 2829
ID: 130544
Credit: 954,747,840
RAC: 5,071
|
I *love* the midnight stroll. Sometimes if I'm just goofing off on my running "training", I'll go just a few miles very late at night. Quiet, cool, and no traffic. I have been bothered by the police a couple of times though.
That leather miniskirt didn't suit you. |
|
|
NeoVolunteer tester
Send message
Joined: 28 Oct 10
Posts: 710
ID: 71509
Credit: 91,178,992
RAC: 0
|
|
I've got three more W/U's to return before the deadline.
I think I did OK for just having an i7 laptop with 4 cores. The runtimes for these 321 w/u's really went up when running all 321 w/u's...
Neo
____________
 |
|
|
|
|
I've got three more W/U's to return before the deadline.
I think I did OK for just having an i7 laptop with 4 cores. The runtimes for these 321 w/u's really went up when running all 321 w/u's...
Neo
Yes, that runtime increase is a CPU cache-miss effect. It can be somewhat mitigated by faster main memory. And, we're getting to the point in the challenge where it might be advisable to run fewer units concurrently so as to squeeze a last few in before the deadline.
@Dave: No, leather is too hot to run in. I think it was the bright pink spandex body suit. Or maybe the feather boa and diamond tiara.
g |
|
|
|
|
|
I have a further 10 WUs which should sneak in before the end, which would take me to:
Core i7-4750HQ 2.00GHz (Quad-core Haswell): 32 WU @ ~ 13 hours
Xeon E5-1650 v2 3.50GHz (Hex-core Ivy bridge): 54 WU @ ~12 hours
Athlon 64 X2 5200+ 2.70 GHz (Dual-core): 2 WU @ ~73 hours
Athlon 64 X2 3800+ 2.00 GHz (Dual-core): 2 WU @ ~97 hours
Pentium 4 2.40 GHz : 1 WU @ ~109 hours
And with a bit of luck, we can hold 3rd place in the team challenge!
____________
Twitter: IainBethune
Proud member of team "Aggie The Pew". Go Aggie!
3073428256125*2^1290000-1 is Prime! |
|
|
|
|
|
Gah, one WU just missed out on the deadline, but a good showing for the team overall in any case. Here's the results in full (barring any results declared invalid):
Rank Name Score
1 Sicituradastra. 9913271.84
2 Czech National Team 7975702.13
3 Aggie The Pew 5220486.53
4 SETI.Germany 4975527.25
...
12 Dutch Power Cows 1207017.87
Rank Name Team Score
1 zunewantan Aggie The Pew 3679008.42
51 Iain Bethune Aggie The Pew 307515.57
62 Gary Craig Aggie The Pew 238019.12
79 Walter Darimont Aggie The Pew 191265.20
106 mattozan Aggie The Pew 149093.70
112 gazzyk1ns Aggie The Pew 138974.74
124 ~~~(,__,)^'> Aggie The Pew 128753.45
162 brinktastee Aggie The Pew 81962.04
166 mackerel Aggie The Pew 81372.67
195 1998golfer Aggie The Pew 61224.62
199 NickBurrlock Aggie The Pew 57755.56
201 Dave Aggie The Pew 54613.74
236 Neo Aggie The Pew 40661.01
426 Rick Reynolds Aggie The Pew 10266.67
Congratulations to Z. Since you scored as much as the rest of us combined, I'd say we have a long-tailed distribution of scores ;) Go Aggie!
____________
Twitter: IainBethune
Proud member of team "Aggie The Pew". Go Aggie!
3073428256125*2^1290000-1 is Prime! |
|
|
|
|
Rank Name Team Score
1 zunewantan Aggie The Pew 3679008.42
51 Iain Bethune Aggie The Pew 307515.57
62 Gary Craig Aggie The Pew 238019.12
79 Walter Darimont Aggie The Pew 191265.20
106 mattozan Aggie The Pew 149093.70
112 gazzyk1ns Aggie The Pew 138974.74
124 ~~~(,__,)^'> Aggie The Pew 128753.45
162 brinktastee Aggie The Pew 81962.04
166 mackerel Aggie The Pew 81372.67
195 1998golfer Aggie The Pew 61224.62
199 NickBurrlock Aggie The Pew 57755.56
201 Dave Aggie The Pew 54613.74
236 Neo Aggie The Pew 40661.01
Don't forget The Tusked One!
426 Rick Reynolds Aggie The Pew 10266.67
Congratulations to Z. Since you scored as much as the rest of us combined, I'd say we have a long-tailed distribution of scores ;) Go Aggie!
Team Zunewantan would have placed 5th overall! Wow!
____________
Proud member of Team Aggie the Pew
"Wir müssen wissen. Wir werden wissen."
"We must know, we shall know."
- David Hilbert, 1930 |
|
|
|
|
|
What a great team effort and Z effort, congrats to Z for his 1st place finish and the team to our 3rd place. Highest team place I think??? (correct me if I'm wrong)
I'm sure Iain just over looked due to the distance I was down in the list -
Cheers Rick |
|
|
Dave
Send message
Joined: 13 Feb 12
Posts: 2829
ID: 130544
Credit: 954,747,840
RAC: 5,071
|
|
I would have been between Tim & Brink in an alternative life.
What I could really do with is a remote control mini robot arm that would press the PC's hard-reset button if I was away for a few days.
Well done to all - & lol @ long-tailed distribution :).
Variety of tasks going on this end now to make the numbers look pretty. Art with numbers... |
|
|
|
|
|
I'm leaving a few cores hanging around on 321 to do clean-up and work on the amethyst badge. To some degree though I've switched to PSP-Sieve for a badge upgrade there too.
Great work everyone... too bad it wasn't one of us that found the prime! I haven't seen who did the d/c.
--Gary |
|
|
|
|
I'm sure Iain just over looked due to the distance I was down in the list -
Cheers Rick
No Rick, I omitted you because I don't like you very much ;)
Actually I'm usually quite good at looking near the bottom of the list, after all I usually used to end up there... not sure why I missed it this time!
____________
Twitter: IainBethune
Proud member of team "Aggie The Pew". Go Aggie!
3073428256125*2^1290000-1 is Prime! |
|
|
|
|
|
Yup, top work everyone. I was away from my PC this evening and I think I was about one hour too late with another 3 WUs... Ah well, all I could have done if I was here was sit around having aborted one, waiting nervously to see if there was time for the last two I suppose.
I'm leaving a few cores hanging around on 321 to do clean-up and work on the amethyst badge.
--Gary
Same here. |
|
|
|
|
No Rick, I omitted you because I don't like you very much ;)
Arrrgggg ! My feelings are hurt !!!!!!!!!!
and of course you know I'm just kidding.
Cheers you blind rat!
edit: ps making this thread not sticky |
|
|
|
|
|
P.S. I added you in to my original list in the end!
____________
Twitter: IainBethune
Proud member of team "Aggie The Pew". Go Aggie!
3073428256125*2^1290000-1 is Prime! |
|
|
|
|
P.S. I added you in to my original list in the end!
Day late and a pound short! |
|
|
|
|
P.S. I added you in to my original list in the end!
Day late and a pound short!
the results in full (barring any results declared invalid)
I give up!
____________
Twitter: IainBethune
Proud member of team "Aggie The Pew". Go Aggie!
3073428256125*2^1290000-1 is Prime! |
|
|
|
|
I give up!
No you don't LOL |
|
|
NeoVolunteer tester
Send message
Joined: 28 Oct 10
Posts: 710
ID: 71509
Credit: 91,178,992
RAC: 0
|
|
Absent from team are "The Doominator" (Mark Doom) and Tarmo..
Are they just on a break or have they been gone for some time?
Neo |
|
|
|
|
|
Hmmm, I was wonderind about that the other day. Tarmo might have said something? I really can't remember.
Separately, turns out my old SB wasn't any help at all, as it appears to have returned nothing but invalid results. Strange for a CPU which isn't overclocked (any more). I suppose it could be the RAM but although "overclocked", it's just set to an official XMP profile. Not my ptoblem any more, anyway, like I said, it's at my parents' and I won't be using it for PG any more... |
|
|
|
|
|
I've flip-flopped around on projects the last few days, but am now back doing 321 clean-up on one i7 box. I'm wingman for one scoring WU now, and I still have one of my own awaiting validation.
We have a one-day SGS challenge coming very soon, so this will be a short-lived push.
--Gary |
|
|
Dave
Send message
Joined: 13 Feb 12
Posts: 2829
ID: 130544
Credit: 954,747,840
RAC: 5,071
|
|
I'm still doing all sorts here making the totals of units look pretty as they'll stay static for a while, but after about another week I start concentrating on 321s towards ruby which is what I should be doing. |
|
|
|
|
|
At one point maybe 6 months ago or so, I had one (or more) square badge of every possible color, simultaneously. It was quite a rainbow. I thought PSP Sieve was "stuck" at bronze until they reactivated it. If only I had run more AP26 when I had the chance... I've got credit there, but no badge. I'm splitting my resources between 321 and PSP Sieve for the next few days.
G |
|
|
|
|
If only I had run more AP26 when I had the chance... I've got credit there, but no badge.
That's the same for me and 321-Sieve. I have credit, but no badge. I think I didn't even know about badges back then...
If I'd at least done the minimum for a bronze, then I would have a bronze, a silver, a gold, a ruby, and an emerald in the Sieves...
____________
Proud member of Team Aggie the Pew
"Wir müssen wissen. Wir werden wissen."
"We must know, we shall know."
- David Hilbert, 1930 |
|
|
