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

Sponsored by:

Hosted at and sponsored by Rackspace.

Join PrimeGrid

Returning Participants

Community

Leader Boards

Results

Other

drummers-lowrise
1) Message boards : Generalized Fermat Prime Search : GFN Prime Discoveries (Message 121258)
Posted 18 hours ago by Profile Michael GoetzProject donor
We've found four n=15 (32768) primes!


On 19 Oct 2018 16:41:45, the 262,628 digit prime 103455878^32768+1 was found by CGB using the OCL2 (combined) app.

The double checker of this task was ostrik on 19 Oct 2018 19:54:10 using the OCL2 (combined) app.


On 19 Oct 2018 17:02:33, the 262,628 digit prime 103459020^32768+1 was found by CGB using the OCL2 (combined) app.

The double checker of this task was Theory on 19 Oct 2018 22:54:22 using the OCL2 (combined) app.


On 19 Oct 2018 20:26:21, the 262,631 digit prime 103480164^32768+1 was found by Skligmund using the OCL2 (combined) app.

The double checker of this task was bparsonnet on 19 Oct 2018 22:23:03 using the OCL2 (combined) app.


On 19 Oct 2018 23:52:42, the 262,633 digit prime 103495684^32768+1 was found by CGB using the OCL2 (combined) app.

The double checker of this task was bparsonnet on 20 Oct 2018 1:53:09 using the OCL2 (combined) app.


Congratulations!
2) Message boards : Number crunching : 2018 Mega Primes (Message 121257)
Posted 18 hours ago by Profile Michael GoetzProject donor
#273 is a GFN-17-Mega
3) Message boards : Project Staging Area : prpclient settings, not taking startupoption setting 6 now.... (Message 121248)
Posted 1 day ago by Profile Michael GoetzProject donor
Unfortunately I can't edit my old posts, is it possible for a mod to edit so my email addy doesn't show any longer in the posts? Its not the worst thing in the world, but may be better off not having it posted.


Done.

4) Message boards : Number crunching : Running GPU slows down CPU a lot (Message 121236)
Posted 1 day ago by Profile Michael GoetzProject donor
I suspect the task scheduling....


You may be right. It does make sense.

Later I think I'll do some tests with single threaded LLR and see how it behaves.
5) Message boards : Aggie The Pew message board (Message 121228)
Posted 1 day ago by Profile Michael GoetzProject donor
Guess here's as good as any place to wonder aloud some user practices.

So, I've recently started doing gfn 16. I noticed almost immediately that 99% of them were in pending status so I started looking at some of them and found that some users are doing 100's of units at a time. One user has several cards in one machine and has 900 16's in progress. That user was not alone. There were several users with lots of tasks in progress.

So a couple of things came to mind after seeing this. Are the users aware that if they find a prime there's probably less than a 10% chance it will be theirs? And because of all the units in progress my chance of finding a prime and being the finder goes up (at least just my opinion).

I'm not even concerned about the pending side on my end. I was just wondering. I know folks all look at things differently.

Cheers


Case in point:

http://www.primegrid.com/workunit.php?wuid=584297262

GFN-16 prime. Prime was found on a CPU, which took over three hours. Double checker ran the test in 5 minutes on a GPU. The prime finder finished almost 5 hours ahead of the double checker. They received the tasks within 3 seconds of each other.

Adding insult to injury, the CPU is a Sandy Bridge laptop, and the GPU is an AMD RX 460.
6) Message boards : Generalized Fermat Prime Search : GFN Prime Discoveries (Message 121227)
Posted 1 day ago by Profile Michael GoetzProject donor
We've found another n=16 (65536) prime!

On 18 Oct 2018 16:39:29, the 500,323 digit prime 43084562^65536+1 was found by StephaneD using the CPU app.

The double checker of this task was John Galt on 18 Oct 2018 21:11:34 using the OCL2 (combined) app.

Congratulations!
7) Message boards : Number crunching : Running GPU slows down CPU a lot (Message 121226)
Posted 1 day ago by Profile Michael GoetzProject donor
Right now, I'm running Cullen tasks on my CPU and GFN-17-Mega on the GPU.

System is a Haswell i5-4670K (4c/4t) with faster-than-stock memory DDR3, and a GTX 1060 6GB GPU.

If I'm not running the GPU, Cullen tasks take about 12 hours, and when I am running the GPU they take about 16 hours. The LLR tasks are running -t4.

That's a pretty big performance hit.

I did some stand alone benchmarks with the GPU running a 17-Mega.

With the GPU running:

C:\bench>cllr64.3.8.21.exe -t4 -d -q"16944075*2^16944075+1"
Starting Proth prime test of 16944075*2^16944075+1
Using zero-padded FMA3 FFT length 1792K, Pass1=448, Pass2=4K, 4 threads, a = 11
16944075*2^16944075+1, bit: 20000 / 16944099 [0.11%]. Time per bit: 3.258 ms.
Caught signal. Terminating.

C:\bench>del z*

C:\bench>cllr64.3.8.21.exe -t3 -d -q"16944075*2^16944075+1"
Starting Proth prime test of 16944075*2^16944075+1
Using zero-padded FMA3 FFT length 1792K, Pass1=448, Pass2=4K, 3 threads, a = 11
16944075*2^16944075+1, bit: 20000 / 16944099 [0.11%]. Time per bit: 3.450 ms.
Caught signal. Terminating.


Same test, this time with the GPU idle:

C:\bench>cllr64.3.8.21.exe -t4 -d -q"16944075*2^16944075+1"
Starting Proth prime test of 16944075*2^16944075+1
Using zero-padded FMA3 FFT length 1792K, Pass1=448, Pass2=4K, 4 threads, a = 11
16944075*2^16944075+1, bit: 20000 / 16944099 [0.11%]. Time per bit: 2.778 ms.
Caught signal. Terminating.

C:\bench>del z*

C:\bench>cllr64.3.8.21.exe -t3 -d -q"16944075*2^16944075+1"
Starting Proth prime test of 16944075*2^16944075+1
Using zero-padded FMA3 FFT length 1792K, Pass1=448, Pass2=4K, 3 threads, a = 11
16944075*2^16944075+1, bit: 20000 / 16944099 [0.11%]. Time per bit: 3.196 ms.
Caught signal. Terminating.


Notice that the idle-GPU 3-thread test if faster than the busy-GPU 4-thread test. Running the GPU costs the LLR task more than a full core's worth of performance.

However, when looking at task manager, the Genefer task is only using about 1% of the CPU (4% of a thread), and PrecisionX is using about 5% of the CPU (25% of a thread). In total, the GPU software is only using about about 30% of a thread's CPU cycles, and yet it's effectively slowing down the CPU by more than a full thread. Something else is happening here.

While I don't have the tools necessary to definitively see what's happening, I suspect it's our old bugaboo, memory bandwidth.

GPU programs spend a lot of time transferring data between main memory and video memory. Those transfers don't use much (or any) CPU cycles, but they do consume memory bandwidth. And, of course, LLR -- especially on large numbers -- uses a lot of memory bandwidth. It appears as if running the GPU has a significant effect on LLR performance above and beyond the use of a CPU core.

This has serious ramifications for all of us, specifically:

* Running the GPU will slow down LLR even if you leave a full core free.

* Running the GPU will slow down LLR even if you have hyperthreading enabled to service the GPU.

Of course, your mileage will vary. I suspect the that this effect varies significantly with the size of the number being tested. The GPU effect is probably much more pronounced with SoB than with SGS. How fast your memory is will make a big difference. What type of memory you have and the number of memory channels will probably also have a significant effect.

Is anyone else seeing similar results?
8) Message boards : Number crunching : 2018 Mega Primes (Message 121209)
Posted 2 days ago by Profile Michael GoetzProject donor
#272 will be a PPS-MEGA
9) Message boards : Project Staging Area : PFGW after LLR? (Message 121208)
Posted 2 days ago by Profile Michael GoetzProject donor
Thanks for the confirmation :)

I should never have stopped my boxes from auto crunching those... might have made the discovery myself :)

Now, what are the odds, another prime, is close by?


They seem to come in threes. :)
10) Message boards : News : Restricted access to statistics files (Message 121175)
Posted 2 days ago by Profile Michael GoetzProject donor
Hi Michael,

Have you noticed a drop off in the Gridcoin team stats since the restricted access went in?

Gridcoin is no longer accepting PrimeGrid as a 'whitelisted' project

Dad


I don't track team or user activity as a function of time, so I don't know. What I do know is that there's no noticeable difference in overall activity.

The Gridcoin devs are working hard to fix the problem. It's Gridcoin that caused this problem; locking down the stats was necessary to protect PrimeGrid. The ball's in their court to fix their architecture.

I just realized I've told the Gridcoin community much more about what's going on than I did the PrimeGrid community. Here's the scoop on what happened:

In early September I noticed that PrimeGrid was virtually unreachable for a few minutes. It turns out there was a DDOS attack against us, but it was a short lived attack. It only lasted a few minutes. Some investigation determined that this had been happening about once a day for a while, but since it was short in duration nobody really noticed.

By now you probably have guessed where this story is going.

Gridcoin needs our statistics, and collects them by having a large number of host computers each collect our stats files. At pretty much the exact same time. And these are really large files. While we have a fairly hefty server, and a decent pipe to the Internet, Gridcoin *is* a significant player at PrimeGrid, and all those hosts (over a thousand) downloading those large files at the same time was completely saturating our Internet connection.

PrimeGrid isn't the largest BOINC project, but we're up there, and our servers are more powerful than most. I'm not saying that to brag. The point is that if Gridcoin can effectively, albeit unintentionally, mount a successful DDOS attack against PrimeGrid's servers, imagine what Gridcoin is doing to all the other BOINC sites with much smaller hardware. Their only saving grace is that their stats files are almost certainly smaller than ours, and they may have fewer Gridcoiners (is that a word?) on their system. Nevertheless, this method of gathering the statistics is undoubtedly affecting many BOINC sites adversely. What is certain is that it has been making PrimeGrid's server unreachable for periods of time. That's unacceptable. Except for the fact that this is clearly unintentional, it would literally be criminal.

Back in September I reached out to Gridcoin. There's a fix in the pipeline. Things will hopefully be back to normal soon, minus the DDOS part.

We're not the bad guys here. Nor is Gridcoin. Everyone is working together to get this working.


Next 10 posts
[Return to PrimeGrid main page]
Copyright © 2005 - 2018 Rytis Slatkevičius (contact) and PrimeGrid community. Server load 3.05, 3.23, 3.71
Generated 21 Oct 2018 | 5:46:44 UTC