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

Toggle Menu

Join PrimeGrid

Returning Participants


Leader Boards



11) Message boards : Number crunching : World Water Day Challenge (Message 154887)
Posted 99 days ago by Profile GrebulonerProject donor
I'll be making a much lower contribution, myself. My PG electricity budget is shared with my gas budget, so I'm down to my main CPU only (not even GPU) for daily crunching. I'll fire up my two X99 systems for this challenge because they have a good performance/watt ratio.

I love PG, but going to work is a wee bit more important.
12) Message boards : Project Staging Area : PRPNet Credit may be Irregular Until Further Notice (Message 154817)
Posted 109 days ago by Profile GrebulonerProject donor
Yes, it has been handled correctly. In the future, please keep threads on topic.


At, I have got 2,996,796 PSA credit, as available to the BOINC system.

At, I have got 59,852,608 (59,696,348+156,260 separated)

That's 56,855,812 in credit still missing and unchanged for about two months. That's irregular reported and on topic, if any.

PRPNet credit is calculated differently from Boinc credit, so what you receive as PSA is always much lower. (I don't recall all the different conversion factors)
13) Message boards : Number crunching : Dual Xeon Server dont use second CPU (Message 154566)
Posted 123 days ago by Profile GrebulonerProject donor
FMA3 = AVX2, the fastest instruction set that Broadwell-EP supported.

The other machine is an i9 9960X which is Skylake-X, one of the first architectures with AVX-512 support. AVX-512 is roughly 2x faster than AVX2/FMA3 at the same clock speed.
14) Message boards : Number crunching : Dual Xeon Server dont use second CPU (Message 154545)
Posted 123 days ago by Profile GrebulonerProject donor
we see now any sideeffects
11 tasks are prefered and are now ready and uploaded
run time1h 47min
all other task are now between 20 - 40 % ready
I have no idea why this happen.
Normally should all tasks run eaqual from 1 to 100% till there are ready

Any ideas ?

set now:Max # of simultaneous PrimeGrid tasks to 81
to see if it has any effect.

You're getting closer. (My advice only, you do what you want: Running LLR on hyperthreads is a waste of time, as it takes more than twice as long to complete tasks. Run 40 MEGA (1 per physical core at 1 thread) and it will be even faster and use less power.)

There are a few reasons they aren't completing all at the same time. First is that nothing is exactly the same, there are different bases and exponents, so workunits take different amount of time to complete, and over time it gets a bit out of sync, like runners on a track doing 20 miles. Nothing to worry about there. The other reason is that CPU resources are being shared because you're using all of your CPU's threads, which means the tasks have to fight for CPU resources to complete and things get out of whack. The last reason is that your computer is also doing other things, even if it's just Windows + Boinc, and that slows down certain areas.

On your question of frequency, there isn't as much info on the turbo states of ES cpus as there is on final retail, but with Xeons, the maximum turbo speed is based on the number of cores running at a particular time. 2.5GHz may be for all cores, or just a few. Without one of my own, I couldn't tell you. If temperatures get high enough, speed will lower until the CPU is back into its thermal happy place, even if that means going below base clock. You can test yourself using CPUz and Prime95, starting with 1 thread and recording frequency until you get to the full 20.
15) Message boards : Number crunching : Dual Xeon Server dont use second CPU (Message 154528)
Posted 124 days ago by Profile GrebulonerProject donor
When you see "0000" for an Intel processor number, it means the CPU is an engineering sample.

Hmmm, the things you'll find by dumpster diving behind Intel.

Those cpus only make it to the dumpster after being destroyed. ES server cpus are actually fairly easy to acquire as Intel sends thousands upon thousands to OEMs, major purchasing partners and others for real world bug testing and related component development (MBs, RAM, coolers, etc.) In advance of final silicon.
16) Message boards : Number crunching : Dual Xeon Server dont use second CPU (Message 154522)
Posted 124 days ago by Profile GrebulonerProject donor
i see a few things.

why is the processor reported as:
CPU type GenuineIntel
Genuine Intel(R) CPU 0000 @ 2.00GHz [Family 6 Model 79 Stepping 0]
Number of processors 190

if it's actually a "2 x XEON 20Core, 40 threads E52698v4" then BOINC should report 80 processors, not 190. not sure what is going on there.

and, it looks like you are using 64 threads on a PPS MEGA workunit. this is too many threads for such a small workunit. i would set your primegrid preferences "Multi-threading: Max # of threads for each task" to 2 or 4 for MEGA workunits.

When you see "0000" for an Intel processor number, it means the CPU is an engineering sample. In this case, the sSpec is QHUZ, and it isn't exactly an E5 2698v4, but it's close enough (in this case lower clock speed)

I see the threadcount has been updated, I remember there were tricks to fooling BOINC into thinking you had more cores for various valid reasons (particularly in the emerging GPU compute days).

Tasks are still running at 24 threads, and I think that's the biggest reason things aren't working out right for you, _heinz. If you have boinc set to 50%, it can't actually use 50% of the threads, because 24 is more than half of what is available (and bridging sockets for individual tasks holds quite the latency penalty).

The other side of that is that you are throwing away tons of computing power with that many threads and the system is probably sitting idle a lot trying to synchronize. MEGA only needs 1 thread per task on your CPU, and you will find you get similar (or even faster!) runtimes to what you're doing now, but while doing 40x the amount of work. My E5 2650v3 runs at a slightly higher clockspeed than yours, and its tasks are up to 10 minutes shorter running 1 thread each.
17) Message boards : Number crunching : Tour de Primes 2022 (Message 154194)
Posted 135 days ago by Profile GrebulonerProject donor
Woke up with 2 PG emails received at the same time, got very excited that I might have had simultaneous finds. They were actually a little more than an hour apart, though, but still, pretty neat.

With one a megaprime, I've reached my TdP goal of a mega and more primes than last year. Now on to some stretch goals!
18) Message boards : Number crunching : PG feature request (Message 154155)
Posted 136 days ago by Profile GrebulonerProject donor
I have (possibly temporarily) put this online. Check it out. Let me know if it doesn't work as expected, and especially if it causes the page to take longer to display. I don't personally have a mix of threads in different tasks, so I can't easily test it out myself.

Shows up as the correct value for my MEGA tasks, but just shows 0 for PPSE. GFN (GPU) is also 0, but that's a different animal.
19) Message boards : Problems and Help : GeneferXX Tasks hang on NVIDIA GeForce GTX 780 Ti (Message 154069)
Posted 139 days ago by Profile GrebulonerProject donor
That's good news... it's a load "shut down"
Thermal past check and adjust is a part of my cleaning routine.

How do you get the fans off of a gpu? I've replaced fans before but only by prying them off and being scared I was destroying the thing in the there a secret to getting them off like one does with a cpu for example?

Having done this a few times, myself: Find all the screws. There are several holding shrouds on (plus a few more for the fans), and they're all tiny.

Also, most common GPUs have at least one video on YouTube of someone doing it. Very helpful when the screw you're missing is between the heatsink fins.
20) Message boards : Problems and Help : Is there a table or chart that shows the optimal CPU cache for each project? (Message 154042)
Posted 140 days ago by Profile GrebulonerProject donor
Easy rule of thumb: buy the CPU with the biggest cache that fits in your budgets (initial outlay and power bill). In terms of currentishly available options, if you want it to last a long time, think AMD 5900X/5950X (64MB 12/16c) or take a chance on the supposed-to-be-released-already 5800X3D (96MB 8c). If you've got the money to burn, go high-core count Cascade Lake-X desktop or -SP Xeons, Ice Lake-SP Xeons, or Epyc 7xx3 (ordered by increasing cost). Honorable mention to Threadripper (up to 256MB cache), but it's stuck on last gen architecture.

Keep in mind that the desire for cache is to allow the CPU cores to run at their maximum potential without being held back by ram performance. If the cache requirement is met, performance can scale with cores. If work goes beyond the cache, then ram performance will become relevant.

A big caution with AMD CPUs: the L3 cache may not be unified. The 12/16 core Zen 3 models are two CCX, so are best seen as two lumps of 32MB, not one lump of 64MB. I'll have to say I've not done any testing on them since Zen 2, but to my understanding of the architecture there is no way to directly share data between each L3 cache segment (CCX) without going out to ram. Internal bandwidth of AMD CPUs is generally comparable to dual channel ram bandwidths, so it would get congested fast if multiple segments try to do so at the same time. It still shows some improvement but doesn't scale near where it could be had the cache been unified.

The upcoming 5800X3D does have 96MB of unified cache but only 8 cores. You have practically unlimited performance of those 8 cores. Note with such a large cache, you will probably get more throughput running two SoB tasks with 4 cores each than one task of 8, as it is more efficient the fewer threads you use on a single task. 3 tasks might even be higher throughput still but you'd have to find a way to manage 3+3+2 core split to try it, and it might not be great long term given how close it is to the limit already.

Can't remember where it popped up, but someone on here did do some tests with big tasks across CCX's and while it wasn't the best scaling, it was still noticeably faster (by throughput) than running 1 task per CCX and going to main memory.

I am very tempted by the 5800X3D, myself, but the sad reality is that I can pick up a used Skylake-X/Cascade Lake-X setup with more cores for less than it would probably cost, and having AVX-512+quad channel memory, would vastly outperform the AMD, even if the task is bigger than the cache.

Next 10 posts
[Return to PrimeGrid main page]
Copyright © 2005 - 2022 Rytis Slatkevičius (contact) and PrimeGrid community. Server load 0.61, 0.81, 1.24
Generated 29 Jun 2022 | 12:10:45 UTC