Formula Boinc 2018 Strategy

So the following is all I had time for / can deal with right now...

3x - E5-2640 v3 (2P)
2x - E5-2678 v3 (2P)
2x - E5-2680 v4 (2P)
3x - Gold 5118 (2P)
1x - Gold 6126 (2P)
2x - Platinum 8160 (4P)

Have a few SB/IB 4P's, and some other misc 2P SB/IB/HW boxes available, just don't have time. Also, since this afternoon, seems I'm most likely flying to Seoul in the next couple days... so there's that.

The 4P 8160 I mentioned ~6 hours ago has spit out 100K in credit, so hopefully that's good? The other 4P is at approximately 50% of that.

Speaking of "hopefully" let's hope my ever-so-lovely proxy will behave itself!
 
Oh, you didn't hide your rigs. Better do so. Like to keep it sneaky too, lol.

Counted a total of 440C/880T running (y)
 
Last edited:
BTW, where's the real XS forum again? Seem to recall it was something a little odd.
 
we are on a good way ...
Screen Shot 2018-08-25 at 17.34.12.png
 
Sorry guys, I completely screwed up bunkering this time. Apparently yoyo doesn't allow multiple boinc instances so lost 2+ days work on 3 rigs.

On the bright side, we've now got 20 AWS platinums crunching. Thanks to Skillz, atp1916 and motqalden for your contributions towards avoiding my utter bankruptcy.

XS have got at least 2 bunkers brewing. StyM and OldChap have hardly registered a point in over a day.
 
I know we've seen 2 1m+ bunkers already (that we've chewed through).

If we get ahead of them by a million+, i am not sure if they can recover especially with fastgeek in play.
 
phoenicis said:
Sorry guys, I completely screwed up bunkering this time. Apparently yoyo doesn't allow multiple boinc instances so lost 2+ days work on 3 rigs.

On the bright side, we've now got 20 AWS platinums crunching. Thanks to Skillz, atp1916 and motqalden for your contributions towards avoiding my utter bankruptcy.

XS have got at least 2 bunkers brewing. StyM and OldChap have hardly registered a point in over a day.
Click to expand...
Dude, you never need to apologize to us for anything. Ever.
 
OK, as mentioned elsewhere, I've got 3 1950Xs looking for a good home with 3 dedicated [H] crunchers. They're all in good shape but I've only ever run them at 3.6GHz so no idea how well they overclock.

I'm asking for just $200 each to my wife's paypal to cover international FedEx shipping and to make her slightly less grumpy that she can't sell them on ebay. The only thing I would ask is that anybody interested be willing and able to build a TR rig swiftly to aid in our FB battle with XS.

So as not to contaminate this thread, please PM me if interested. Firm commitments only please, don't want this to be any more painful than it already is ;)
 
Doing awesome today:
upload_2018-8-25_19-23-6.png


I bought one of the chips from Phoenicis, maybe I can help out with a more powerful rig in the future. Now I need some recommendations on good RAM to get. PM me if you know of a set I should buy.
 
phoenicis said:
OK, as mentioned elsewhere, I've got 3 1950Xs looking for a good home with 3 dedicated [H] crunchers.
Click to expand...


Hell yeah! Thanks phoenicis all my parts are ordered and on their way! Can't wait to join the Threadripper club.


MN scout i am getting the Gskill tridentZ 3200. Model number F4-3200C16Q-32GTZRX <-- i am by no means an expert on memory compatibility but the "X" at the end makes this model Ryzen/TR specific suppored and it was only a few dollars more than the intel variant. No way could i afford something in the cas 14 variety as it would cost me $200 more for the 32GB size
 
atp1916 said:
phoenicis You are elite. Cheers.

The TR 1950x build will be:

ASRock X399 Taichi + Noctua NH-U14S TR4-SP3
G.SKILL TridentZ Series 32GB (2 x 16GB) F4-3200C16D-32GTZKW (Gonna go with another 32GB set in the very near future)
Click to expand...

Did you find confirmation that ram works with the 1950X and the board? I'm not settled on a board, but the memory compatibility has me all indecisive. Is there a special reason you picked that board?

I wish I could find better benchmarks on how the cas latency and memory speed affect different BOINC projects completion times. It is a lot of money going to a lower CAS and or higher speed.
 
Keep your fingers crossed for the last sprint of the year to be a GPU project. That way we won't have to divert any resources away from the WCG challenge.
 
Currently hitting TNGrid and PrimeGrid hard. Said screw the electric bill today and fired up my two HD7970's on MilkyWay, as well. I think I should be able to get somewhere around 1 million ppd with these two Double Precision beasts crunching. I'm running 3 GPU tasks per card and they are finishing at ~90-120 seconds each. I had forgotten how well these cards crunch MilkyWay. Looking to hit 100 million points (42 million to go), but may stay on it to pass up XS to get to 3rd in the marathon.
 
Well in the event that either one of them are chosen tomorrow you might as well bunker them now. You can always release the bunkers when the Sprint is announced if it's not either of those projects. I am bunkering CSG on my CPUs and Einstein@Home on my GPUs. If neither of those Sprints are chosen I will just upload what I have completed, abort what's not completed and start bunking whatever project is picked.

edit
Oh wait, it'll be announced in like 10 minutes. Nevermind.
 
All these sprints are killing me. I am more a marathon cruncher. I always want to support the team, but I am getting sprint fatigue.
 
After this one we'll have a 10 or 11 day break.

Team who wins the most Sprints will win overall.
 
Just started up Milky Way on my GPU and CPUs. Running 3 tasks at a time on my 1080 TI. Configured already from the last time we did this.
 
and a 128 task limit on CPUs, but it's based on the number of cores. My 20, 48 and 64 core rigs got 128 tasks each. My two quad core rigs got 32. So seems 8 tasks per core up to 128.
 
MN Scout said:
Just started up Milky Way on my GPU and CPUs. Running 3 tasks at a time on my 1080 TI. Configured already from the last time we did this.
Click to expand...

Mind sharing the config. I lost my old boinc install a while back and have nothing for this project
 
motqalden said:
Mind sharing the config. I lost my old boinc install a while back and have nothing for this project
Click to expand...

This is what I use. It results in one CPU thread being used with three GPU tasks running per GPU.

<app_config>

<app>
<name>milkyway</name>
<max_concurrent>0</max_concurrent>
<gpu_versions>
<gpu_usage>0.33</gpu_usage>
<cpu_usage>0.35</cpu_usage>
</gpu_versions>
</app>

<app>
<name>milkyway_nbody</name>
<max_concurrent>0</max_concurrent>
<gpu_versions>
<gpu_usage>0.33</gpu_usage>
<cpu_usage>0.35</cpu_usage>
</gpu_versions>
</app>

<app>
<name>milkyway_separation__modified_fit</name>
<max_concurrent>0</max_concurrent>
<gpu_versions>
<gpu_usage>0.33</gpu_usage>
<cpu_usage>0.35</cpu_usage>
</gpu_versions>
</app>

</app_config>

I am currently running 3 tasks per GPU on my 1080ti and HD7970's, 2 per GPU on my 980ti and 1070ti's, and only 1 task at a time on my ancient GTX770.
 
Last edited:
Thanks, 3 on my Titan Xp seems to run OK. Looks like my effective task time dropped around 30s.
 
I bet you could run 4 on that beast of a GPU. Try it and check your WU completion times. I bet it works out better with 4 running. Remember to also adjust CPU usage in the app_config to .25 if you try 4 WU's at a time.
 
You must log in or register to reply here.
Back
Top