At the moment I'm running PPSE on slower boxes, and Mega on my fast one. Forgot about GFN... I have to balance heat buildup and it isn't cold enough for me to leave that running full time!

CPUs: fast are on TRP-LLR, slow are on both SGS and PPSE
GPU: Genefer-short

g

To supplement my last post, I've chosen this strategy because I figure I have zero chance against the server-farmers for the "most primes" yellow jersey. So I hope to get lucky with a big hit for the green jersey (remote hope at best). On the mountain stage day I'll switch to PPSE or SGS just for fun, and then back.

It occurred to me that it might be possible to get some extra shots at "polka dot" by downloading a day or two's worth of short work right before, then disable network communication and crunch it all. After 00:00 on the appointed day, report it all, then continue to crunch as normal. But this scheme isn't so great after all, since for a prime to count for you in TdP, you have to be initial finder. Since the work is small, and you'd sitting on a big cache, chances are that a fair amount of it would get reported ahead of you. But not all of it... Probably not worth the effort except as a "thought experiment".

--Gary

Going with SR/5, but only on the i7 laptop for now.

I'm getting good run times 11,000 seconds per w/u on a 1.6 ghz dual core i7.
Running only 1 core, the laptop jumps up to 2.7 ghz.

I have my i5 w/ 750 & 770 out in the garage to keep the temps down.
I am sieving on the 770 using tpsieve-cuda for some PSA credit... doing a 15T run (that's a lot of sieving..) and running three prpnet dos windows in search of my own primes for R138.... Got 7 K's to discover primes for... Hopefully they come sooner rather than later. Current N is ~118,000... (that's starting from 100,000 with 8 K's... ;)
If I can discover another two primes, the candidate file will be reduced significantly and I could even sieve deeper at a faster rate...

When I do bring the i5 back online, I'm going to go with SR/5... it's heavily sieved, good credit, and the tests are relatively small. And, I need at least a silver badge ... :)

Neo

Couldn't resist trying it out and put my R9 280X on genefer short this morning. Looking like 5 hours a unit! Finding a prime in those would be really nice so I'll leave it going. Temps were around 75C though, and the rest of my box got hotter too.

Parts finally arrived this morning so I'll get a Haswell set up tonight.

To supplement my last post, I've chosen this strategy because I figure I have zero chance against the server-farmers for the "most primes" yellow jersey. So I hope to get lucky with a big hit for the green jersey (remote hope at best). On the mountain stage day I'll switch to PPSE or SGS just for fun, and then back.

It occurred to me that it might be possible to get some extra shots at "polka dot" by downloading a day or two's worth of short work right before, then disable network communication and crunch it all. After 00:00 on the appointed day, report it all, then continue to crunch as normal. But this scheme isn't so great after all, since for a prime to count for you in TdP, you have to be initial finder. Since the work is small, and you'd sitting on a big cache, chances are that a fair amount of it would get reported ahead of you. But not all of it... Probably not worth the effort except as a "thought experiment".

--Gary

PRPNet work isn't doublechecked, is it? You could easily do the above there, downloading and crunching a bunch of WUs, then saving them to upload on Mountain Stage day. If there is a prime in there, it will still be yours.

Well, my strategy almost paid off.

http://www.primegrid.com/workunit.php?wuid=421009949

I was the double-checker on the Mega PPS!

I actually downloaded it before Evelyn Chew, but my laptop's i7 at 1.8 MHz was much slower than her/his i5 at 3.4 MHz, and I turned it in about two hours later.

Also,

I think I read a post somewhere by Micheal Goetz that if you only have one stick of ram in your computer (laptop).. it's not dual channel memory.. it's single channel by default because there's only one stick of ram...

Don't hold me to that... And I haven't opened my i7 laptop to verify whether there's two sticks or not..

I can tell you that if you look at my SR/5 tasks, and their completion times, ... especially in comparison to my wingmen... I'm crushing them.

"If you're not first, you're last" - Ricky Bobby (Will Farrell) - Talladega Nights

;)

Neo
AtP

Gazzy:

Just a head's up... Some of your tasks are coming back invalid... :(

Not sure if you saw that or not.

Neo

You definitely need two separate memory sticks in slots that have the same color to take advantage of dual-channel. At least in every mobo I've seen in years. Preferably same brand, speed, timing. Note that the two with the same color are often not adjacent to each other.

Gary

Mattozan,

I use Core-Temp to determine what ghz my processor is running at, plus the temperature.

http://www.alcpu.com/CoreTemp/

It's especially useful on the laptop to verify whether the laptop is throttling due to temperature.

Try this program and see if your laptop is throttling. If so, set BOINC use to 25%

You may be able to crunch work units in less than 3000 seconds because your cpu will be running at 2.7 ghz (turbo 100% of the time) instead of 1.6 ghz or 800 mhz (fully throttled)

Odd to think that by only using 1 core you will crunch more workunits per day than if you ran BOINC at 100% or 50%...

:)

Happy hunting and may you never be a double checker again.

Neo

Nice job Doom!

...
Happy hunting and may you never be a double checker again.

I'm all switched over to PPSE for the mountain stage. Not that I'm expecting any miracles! I can't be around to hover over the keyboard right at the start, so I'm just running with minimal cache and we'll see how it goes.

--Gary

I've also switched to PPSE on all boxes ready for mountain stage. That reminds me, I got two to turn back on, plus I'll even throw in my laptop!

On average bigger farm = more chance but you never know, luck could triumph over brute force!

I just hope the server doesn't choke with all the prime hunters grabbing PPSE units with minimal cache all the time...

Laptop set up and running. Had to limit cores due to HT and then force some .net stuff through before I got full GPU power!

Also excavated an old box I couldn't get running. I suspected it was power related, but I tried every spare brick and nothing. In the end I had to take out the battery which reset something and now it boots! Looks like the last time I turned it on was July 2013! Started it off scanning for Windows Updates but that could take a while. Maybe I'll skip that until after tomorrow! Still, it gets me two more Core 2 era cores on the job.

Sometimes there is a jumper on the motherboard that one can use to do a "master reset of everything". But it sounds like your pull-the-battery trick did the job. Congrats.

"Windows Update". Yeah. I hate that thing. In the Linux (Ubuntu) world, the equivalent is "Software Updater". It seems like every few days there's something new to download; generally this is a good thing (I'm not really complaining) but there are three things that update that just petrify me to the point of lighting incense and waving sheep entrails over the box to ensure it works:

1: GRUB ("GRand Unified Boot-loader") software that runs at boot to allow choice of what OS to load. If this breaks, you get to re-install everything from removable media (i.e. wipe the machine)
2: NVidia video driver: if this breaks, you have no display. At least no nice desktop; you might get a black-screen displaying "login:". You get to "ssh" in from another machine and try to roll back. Hopefully you saved the previously-working install script. I've had to do this a couple of times.
3: The kernel itself. The minor updates work smoothly. It's the major version updates that cause trouble (usually with #2 above).

Sorry, I'm just venting now and rambling off-topic. I return you now to "The Tour".
--Gary

I know there's a jumper on the mobo that has the same clearing effect, but on this occasion since I knew I had the box for many years (2007!) I wondered if the battery might be flat so took it out to measure it. Was fine, put it back in and started working shortly after.

Have to say, I can't find the right power brick for this box. It is a "mobile on desktop" unit so is powered by a laptop style brick. The one I'm using it with isn't rated for the full power of the original, so I've decided to go to one core since the PSU ran borderline painful hot with both cores going! So I decided to let the other core mess around with Windows Update.

On Linux, I was only reading the other day, I believe the major updates aren't that major. They're driven by Linus' inability to count to big sub-version numbers! At the end of the day they're just numbers and don't have major differences.

Just got word I found a prime, although waiting for wingman to confirm before it gets published :) It isn't bigger than the ones already listed so no chance of the jersey for me unless there's another one afterwards...

I didn't catch any on PPSE (also tried a little SGS) on the 19th, unless I've got a "hidden" one in the pipeline still. I've moved back to TRP, still hoping against hope to snag a huge prime this month. I favor the conjecture work in general, so it's not a hardship. :-)

I note that one prime in the Tour was found by a user named "beercrates". What do you think... is that like "crates of beer" or a play on "Socrates"? I suppose I could just ask, but that would be too easy.

--Gary

I found one! :)

Me too!

http://www.primegrid.com/primes/?section=decimal&primeid=96679

scrolling through it...all i see is blonde brunette redhead

I only managed to be the DC of a prime on SGS earlier in the month. I didn't even get the task until the first one had already been returned, so not even a fighting chance :( Still 12+ hours to go, who knows!

- Iain
____________
Twitter: IainBethune
Proud member of team "Aggie The Pew". Go Aggie!
3073428256125*2^1290000-1 is Prime!

I have a different proof code to my 2 PPSEs in 2013. Do I have to do anything or will it automagically automagicalness?

Since the end of TDP is almost here, what are people's plans? I think my only chances of a jersey now are if any of my pending SoB or ESP units may still be prime.

I think I'll go back for a bit of badge upgrading shortly. Got all my LLRs to at least gold now, and could do same for the sieve. When sieves are gold I'll go back once again to get LLRs up to pink... and so it goes on.

I'm bringing SGS up to Gold... only have 12K left to hit gold,
I have one more Cullen to crunch to get to Silver... then I'll be at silver and above for all active projects....

Then I think I'll TRP and/or 321 .... I like the 321 search because it's heavily heavily sieved, and having 3 as the K keeps the FFT down... It's the most efficient project to crunch on CPU.

Have 2 more WR genefers to crunch to hit Turquoise ... so, I'll dedicate the next 11 days for that on GPU... then over to Primesearch Team for some manual proth/riesel sieving to bring PSA up to 20M. (this is how my RAC is so high as of late. I do a big chunk of sieving on my 770 (perhaps 5 days worth - and they dole out the PSA credit in chuncks thereafter)...

I do have three cores of my i5 working on my own project trying to find primes for 7 remaining K for the Riesel Base 138 conjecture (K*138^n-1).... I started at N=100,000 and this morning reached N=150,000. (been crunching for approx. 15 days)... And Gary Barnes over at CRUS(mersenneforum) said I had to get above N=178000 to hit the Top5000. I've sieved the heck out of this mini project so I'm going to continue with these three cores as a prime is overdue. P.S. this is the project I mentioned earlier that I offered anyone on the team the opportunity to crunch their own range of N from my candidate file... :)

Neo
____________

I'll be switching GPUs to GFN-WR OpenCL when the current "short" work finishes (should be before midnight UTC; I've already changed prefs). Trying things out for this GFN challenge in March.

As for CPUs, I have a load of TRPs which will finish in a few hours, and I've changed prefs to PPSE as a last-ditch attempt to at least get on the scoreboard for TdP. After that, I'm almost to Ruby on PSP LLR so I'll take care of that, maybe getting some January challenge clean-up work along the way. I'll probably run a bit on the PRPnet 27 challenge, though only off-and-on.

I took a noticeable bite out of the work needed for TRP LLR "Jade" during the Tour; though, still a *long* way to go for that. Plus, we've got a challenge on that project later this year.

Good work to all Rats on the TdP board (so far!): dyost, mackerel (on Mountain Stage day!), Mark Doom, Dave, KD7LRJ, Van Zimmerman, brinktastee. I hope I didn't miss anyone.

--Gary
____________
"I am he as you are he as you are me and we are all together"

87*2^3496188+1 is prime! (1052460 digits)
4 is not prime! (1 digit)

Whelp, TDP is over and I haven't found a prime :( Last year I found 2, a PPS and a PPSE. No such luck this year. I'm powering down my instances at vultr except for a 768MB to run some longer tasks and of course I'm leaving my 1GB web server up. The instance that is running SoB tasks should complete them within the day, then that one will be shut off too.


Good job to the rest of you!
____________

275*2^3585539+1 is prime!!! (1079358 digits)
Proud member of Aggie the Pew

Still 12+ hours to go, who knows!