PrimeGrid
Please visit donation page to help the project cover running costs for this month

Toggle Menu

Join PrimeGrid

Returning Participants

Community

Leader Boards

Results

Other

drummers-lowrise
1) Message boards : Number crunching : Are Work Units Run on Cores or Threads (Message 131045)
Posted 5 days ago by Profile Rafael
To get the most out of one machine with 4 cores, 8 threads and 1 video card what would be the best setup. I have set video card to sieve which I believe needs 1 part of a core to feed it, how should I set my cores and thread to work for LLR work?

I have Xeon E3-1240 v3 4C/8T with Quadro K600 card

PPS Sieve uses very little CPU time, so you can ignore that. Higher n GFN also fillow this pattern, if you want to try to hunt for primes instead.

For the rest, it depends on what you want to do. You can hunt small primes, big primes or work on sieving, and they all have different ways to optimize your machine.
2) Message boards : Problems and Help : Sweet Spot for hardware? Cost/Work Unit? (Message 130011)
Posted 47 days ago by Profile Rafael

I am thinking to get a few used computers and have them work on PrimeGrid full time.

Ah, I see you have been bitten by the PrimeGrid bug. Wellcome to the family, most of us here are infected as well, and we love it!

:)


Is there a sweet spot with desktops that have a nice graphics card, but are not necessarily new computers - what do I look for? How to get the most bang for the buck?

Does GPU matter more then CPU? What about RAM, and hard disk speed?

Jokes aside, the hardware depends on what you want out of Primegrid.

-Most of the projects here are CPU only (and you can check that at your preferences page), so if you want to contribute to those subprojects, more CPU = more better. For the many LLR projects out there, you'll want an Intel CPU, and prefferably a 4th gen or higher one, as those have FMA3 and can crunch tasks much faster. Additionally, HyperThreading isn't of much use to those projects outside of maybe feeding a GPU, so worry about physical cores more than threads. With those 2 in mind, I suggest looking for PCs with an i5 4xxx.

-Now, as much as we love our CPU subprojects, there are those that are better crunched by GPUs, so let's focus on that. Generally speaking, Nvidia cards are more power efficient and thus preffered over AMD's. Additionaly, I'd say it's not worth it to go with anything below Maxwell (aka Gtx 9xx or 750ti) nowadays due to that same reason. Finnaly, cost is going to be your main concern here, and you really get value out of your money with a low cost Gtx 950 as much as you do getting an RTX 2070: figure ou how much you want to spend and go from there.

Those are the general guidelines, and if you want a balanced build, that's what you're looking for. But if you want to taylor to a particular subproject, we can tweak those for you. And if you want help figuring out what each thing does and why you may / may not care to work on it, feel free to ask, we'll be happy to answer.
3) Message boards : Problems and Help : Multi-thread, is which tasks? (Message 129625)
Posted 58 days ago by Profile Rafael
You said LLR but sieve also? principally pps sieve

The sieve apps do not support Multi-Threading. They do, however, benefit from Hyper-Threading if your CPU has it.
4) Message boards : Number crunching : Badges III (Message 129056)
Posted 79 days ago by Profile Rafael


And thus, my quest for an extra level on GCW Sv is over.
5) Message boards : Generalized Cullen/Woodall prime search : GCW-Sieve is SUSPENDED (Message 128408)
Posted 110 days ago by Profile Rafael
OFFICIAL 30 DAY NOTICE

GCW Sieve is shutting down May 1st


Need just a bit more credit for a badge? WUProp hours? Best hurry up and get it now.

Oof, that's gonna be tight. I need ~4.5M to reach Jade, but by my calculations, I'll only make it to 4M. I might need to create extra PCs out of thin air...
6) Message boards : Problems and Help : nVidia NVS 310 gpu (Message 128399)
Posted 110 days ago by Profile Rafael
I've got this in an OLD system; is it worth using?

I mean, I used to PPS Sieve with a Radeon HD 4450, it would take me 20~22h crunching non-stop to get a WU done, and I imagine you would be in a similar cenario with that GPU of yours.

In terms of power efficiency, no, it's not worth at all. But if you really, REALLY want that extra bit of crunching power (and assuming you can get it to work, ofc), I suppose you could try it.
7) Message boards : Seventeen or Bust : The SoB Double Check has begun (Message 128229)
Posted 116 days ago by Profile Rafael
How come running SOB on 2 cores and starting in the past few days 3 cores on a Intel(R) Core(TM) i5-6600 CPU @ 3.30GHz with 16GB RAM, running Windows 10 Core x64 Edition and I am about 80% of the time 1st.
Are all the big guns away ?

Multithreading more than halves compute time on a CPU like yours, and because you have to manually go out of your way to enable it, the few that do have a huge advantage over those that don't.

Additionally, people that aren't closely looking at their tasks usually stockpile a whole bunch of tasks, so it ends up being easier for people downloading one at a time to be 1st.
8) Message boards : Generalized Fermat Prime Search : First *megaprime* in GFN16 (or GFN15) series (Message 127559)
Posted 137 days ago by Profile Rafael
I wonder how many peoples with significant computing power (e.g. >20 cores total) are interested in this project. Required computing power still look big, and I need to make some estimations about optimal sieving ratio.

Currently we have 190'800 candidates to test. This number is decreasing slowly thanks to GPU sieving at high area, but I'm still working on the problem of sieving in middle area between 0.008P (current CPU sieving progress) and 1.815P (start of GPU sieving). I finally wrote a GPU-assisted sieving program, it's working fine in test environment (software OpenCL) but crashes NVIDIA drivers/compiler... So I still have work to do, may be try other versions of drivers... or just find an AMD card :)

Wouldn't it be easier to Run a couple P on the GPU and remove those candidates to then have less stuff to sieve on the middle area? If I understand correctly, it doesn't matter how many factors you have on the GPU side, but it does on the CPU, so maybe that's a way to cut the CPU portion of the sieve.
9) Message boards : Generalized Fermat Prime Search : First *megaprime* in GFN16 (or GFN15) series (Message 127458)
Posted 140 days ago by Profile Rafael
Good news everyone. I found that hacked version of our manual sieving GPU app can work with huge b values. But it does something strange. It could only find factors above 1,8P. I did a minimal range of 1P-2P and it produced nothing first. I thought that my assumptions for this hack were wrong and moved to other things. But the app kept running. Later I've occasionally switched to this window and suddenly seen factors...

To be exact, the minimum factor it found was

1814652271329281 | 1814570323409552^65536+1

All factors are correct, I'm checking them. I don't know what's behind this and why it's not working at lower ranges. Although I've ported this app, most of math stuff in it is a black magic to me. I even cannot be sure that it's not missing some factors, there is no way to verify. Anyway, this will allow us to sieve much deeper and remove many thousands candidates.

The bad new is that it's still an open question how to sieve up to 1,8P; it's too slow on CPU but there are lot of candidates waiting to be removed from this range.

Jim said small sieving had to be done with specialized software back when we restarted 15/16 sieve to accomodate for the new OCL app, so I'm pretty sure that behaviour is expected, though I have no clue as to why.
10) Message boards : Sieving : Multiple GPU systems (Message 126852)
Posted 152 days ago by Profile Rafael
On a multi GPU system, can I restrict Sieving to one GPU and let BOINC have the other?

Yes. The sieve app runs on a single GPU at a time, and you can tell which one to use. Meanwhile in BOINC, you can tell it to ignore a given device so that it won't use it.

On a multi GPU system, does Sieving 'automatically' use both GPUs?

See above. If you want to run on more than one GPU at a time, you'll have to launch the app individually for each of them.

On a multi GPU system where one GPU is much better than the other, does Sieving care?

The app doesn't care for how good a GPU you use, so long as it supports the necessary features. Whether it's worth to sieve or use the card for something else, that's up to the user to decide.

But generally speaking, I find that Nvidia GPUs are better at sieving and AMD's are better at crunching.


Next 10 posts
[Return to PrimeGrid main page]
DNS Powered by DNSEXIT.COM
Copyright © 2005 - 2019 Rytis Slatkevičius (contact) and PrimeGrid community. Server load 3.40, 3.47, 3.46
Generated 17 Jul 2019 | 4:44:47 UTC