PrimeGrid

Gilthanis

[H]ard|DCer of the Year - 2014
Joined
Jan 29, 2006
Messages
8,718
It seems that AVG has been randomly flagging some of our executables as having an identity theft trojan, and removing them from the BOINC directory. It only happens occasionally and is likely a bug in AVG. The executables themselves are fine and pose no threat.

If you use AVG, I *strongly* recommend configuring AVG to ignore the C:\ProgramData\BOINC folder, and also disable identify protection. You can't tell the identity protection system to ignore a folder, so you have to completely disable it.

Or switch to a different anti-virus product.

http://www.primegrid.com/forum_thread.php?id=5908#79823

I figured others may run into this.

Edit: I created an account called HardOCPtest so that anyone whom wanted to test or contribute anonymously could do so.
To use your weak account key on a given host, find or create the 'account file' for this project. This file has a name of the form account_PROJECT_URL.xml. The account file for PrimeGrid is account_www.primegrid.com.xml.

Create this file if needed. Set its contents to:
<account>
<master_url>www.primegrid.com</master_url>
<authenticator>364983_c17d819be0e87f7596023e813b709c4c</authenticator>
</account>

teamdaily.php
 
Last edited:
Since we have PG challenges throughout the year and the challenge we are currently having supports more than one sub project for the first time, I wanted to remind people that PG gives bonus points to certain projects based on how long/short they run and possibly how important they are to get done at the time.

Here is the original thread on the topic: http://www.primegrid.com/forum_thread.php?id=5141#79006

However, that thread is not up to date. To see current bonus scoring all you have to do is go to your project preferences. When you select which sub-project to run, it shows you the credit bonuses, if any, that apply to each sub-project.
 
Posted yesterday

I just put a new version of all builds of the Genefer app into production on the BOINC server. This is BOINC app version 3.07, and corresponds to Genefer 3.2.6.

Its primary change is to correct a potential memory corruption problem that could occur when resuming from a checkpoint. This only has been shown to occur under specific testing conditions and we don't believe it was ever a problem in production, and we're not worried about the validity of past results. (It would have been caught by the normal double checking if it did occur.)

There may also be a small speed increase in some of the builds,

If you encounter any problems, please report them in the Genefer 3.2.6 testing thread. Thank you!
http://www.primegrid.com/forum_thread.php?id=6140#83310
 
There is currently a challenge going on with the SR5 sub project. According to some numbers from our EVGA brothers http://forums.evga.com/Prime-Grids-Solar-Eclipse-2015-Challenge-m2306586.aspx, it appears to run better with HT turned on. At least with SSD's in the system as that seems to be what most of the test information they shared had. I was hoping for a wider variety of SSD and non-SSD boxes to be tested as some people seem to think that SSD's level the playing field between HT on or off.
 
Funny you mentioned this. Was wondering why these WUs were taking so long as I remembered a post a in the Jersey Challenge saying "Turn HT off". I ignored it; wish I hadn't! Just turned HT off yesterday on the four quad-socket systems that are running the SR5 challenge and the difference was huge. 13+ hours down to 5:45. Now I can't help but wonder if this would have made a huge difference in the PPS/SGS challenge. :( BTW, these systems do have a simple 2XSSD RAID1, but they also have 256GB RAM. Watched the discs in Resource Monitor (not the greatest tool) and saw little activity; RAM is at 1%.

As with everything, YMMV.... in my case killing HT is the way to go.
 
Yeah.. I think there may be a lot of factors. The chips you have, the disk you have, but just as important the speed of the RAM too. I think RAM speed is more important than the Disk speed for the math projects like this. The SSD factor was one of the things I was interested in as there have been a few claims in the PG forums that their rigs performed just as well with HT on using SSD's. I have also seen some posts claim that if they limited the number of PG work units to half their threads (similar to turning HT off), that they could run other projects and not lose the efficiency at PG. So in essence gaining production elsewhere while maximizing their production at PG... I don't know how true that is, but was hoping for a lot more setups configs for the comparison. It really does go to show to run your own tests before the challenge begins.
 
On the front page

Putting the "Short" back in GFN-Short
The GFN-Short project has just about reached the end of the n=20 range of GFN tasks and is transitioning to n=17 tasks. The n=17 tasks are about 60 times shorter than n=20.

It will take a couple of months for all of the older n=20 work units to complete, so even after the n=17 tasks start going out it will still be possible to get the occasional n=20 task.

The new n=17 tasks will have a 4 day deadline.

For more details and discussion, please see this forum thread.
 
I'm currently re-burning in my 7970 (now with watercooling) on the Short GFN Tasks - which really are short now (5 min instead of 5 hours, so just about exactly 60x shorter). Temps are great - Card core and VRMs are 37-40C. If everything looks ok in a day or so I might try a GFN-WR and see how things go.
 
PrimeGrid added the AP27 sub project not too long ago. It has GPU work units. I have not ran any yet to get much detail on performance.
 
As announced in another thread the stats pages now list the relative ranking of GPUs in AP27, where fastest = 1 and everything else is lower than that.

1.000 1.000 GeForce GTX 1070
0.996 0.963 GeForce GTX 980 Ti
0.629 0.662 GeForce GTX 970
0.439 _.___ GeForce GTX 1060 6GB
0.206 0.197 GeForce GTX 750 Ti
0.153 0.168 AMD Radeon HD 7870/7950/7970/R9 280/R9 280X series (Tahiti)

Selected copy/paste above, first value is from primegrid stats, 2nd is what I worked out of my GPUs, normalising for my 1070=1. Obviously the relative scores will vary as I don't know where my 1070 sits relative to the primegrid average.

More interesting is 1060 6GB is quite a lot lower than the 970 which I wouldn't expect. Based on cores and typical clocks I'd expect the 1060 to be faster than a 970. Now, I don't know how many 1060 results there are, so it could just be one person with a slow configuration for all I know.

By multiplying the number of cuda cores, boost clocks and 2, you get the theoretical FLOPs. From that you can work out the relative placing expected of those cards:

1.000 GeForce GTX 1070
0.937 GeForce GTX 980 Ti
0.677 GeForce GTX 1060 6GB
0.607 GeForce GTX 970
0.606 GeForce GTX 1060 3GB
0.215 GeForce GTX 750 Ti

I do think if buying a new card optimised just for this project, the 1060 3GB would be an interesting choice, as even though it has fewer cores than the 6GB model, the cost per core works out lower. For me at least, cheapest 3GB is about 20% less than a 6GB, for 10% fewer cores.

In other testing, Pascal does show a performance per watt improvement over Maxwell, at roughly 25-33% less energy per unit done.

In team red there's only limited info to go on, but they're not fast at this application and gobble up a lot of power getting there. I don't see any Polaris results yet and I'm still curious if they are any different than previous generations. Still got a temptation to pick up a 460 or 470 to bench.

AP27 fastest GPUs
 
Gilthanis, you are my wingman for this SOB WU. What a coincidence!

Still crunching albeit slowly on this rig....
 
Seriously thinking about participating in the next month SoB PG challenge.... Nothing to speed up this bugger.
 
Use Intel...not AMD... that is about it. Faster RAM and over clocking may help. But they are what they are.
 
Gilthanis, you are my wingman for this SOB WU. What a coincidence!

Still crunching albeit slowly on this rig....
Completed this WU in 7.6 days. It takes longer than I expected since I'm using this rig for other crunching too.... Now, it's your turn.
 
Well... it is sitting at 81% and isn't due until the 1st of March. Depending on how often BOINC manager swaps projects... it may take all of that time to get to it. Luckily this is an Intel box. Amicable Numbers keeps taking all the threads away on a regular basis. But, I'm not too worried. Even if I miss the deadline, I will more than likely still complete it before anyone else...
 
Huh. That is weird. Option is checked.

Edit: Should say, default configuration - no app_config
 
Just ordered two EVGA 1070ti's that I will be throwing on Genefer tasks, as I am working on a Double Silver badge in that project and have ~74million points to go. Looks like the 1070ti's are about the best performance per watt out there now, which is important to me because my electricity costs here are ridiculous. I have had to shut down all my old 280X's and GTX770's because they are no longer efficient enough to run. Have been running just my 980ti and my 1080ti, so these two new GPU's should almost double my daily output for only 360 watts of additional power.
 
This card still rocks for Milkyway though in terms of PPD/W.
It does, thanks to its awesome Double Precision performance. My 280X’s got me to 50million+ points at MW several years ago, in fact.
 
thanks to its awesome Double Precision performance
I would just keep it for upcoming FB sprint :D.

I have one 7970 that got me 20M+ on a single card in MW. I plan on keeping this old card.
 
Just ordered two EVGA 1070ti's that I will be throwing on Genefer tasks, as I am working on a Double Silver badge in that project and have ~74million points to go. Looks like the 1070ti's are about the best performance per watt out there now, which is important to me because my electricity costs here are ridiculous. I have had to shut down all my old 280X's and GTX770's because they are no longer efficient enough to run. Have been running just my 980ti and my 1080ti, so these two new GPU's should almost double my daily output for only 360 watts of additional power.
These 1070ti's have impressed me so much, I just ordered two more. They are slightly faster than my 980ti in all Primegrid GPU tasks and only use 180 watts max, compared to 250W for the 980ti. Very nice. Decided to focus back on AP27 for a bit, in order to get Emerald in that project. Will get back to Genefer after that.
 
AP27 GPU WU comparison times (1 task per GPU - stock GPU/Mem clocks):

(GPU: avg time in min:sec)

1080ti: ~14:40
1070ti: ~21:40
980ti: ~24:45

I have also noticed that running VM WU's from both LHCDev@Home and Cosmology while crunching GPU tasks slows down the GPU tasks considerably. Looks like ~6 min slower on one of my 1070ti boxes. That sucks. I hate VM projects.
 
PrimeGrid: Upcoming REQUIRED changes to app_config.xml
If you don't know what app_config.xml is, or if you don't use it for PrimeGrid's LLR apps, you can ignore this notice. This doesn't affect you.
But if you are using app_config.xml on any of our LLR apps, we are changing the plan class of the LLR app versions, and therefore your app_config.xml won't work after the changes. For details, as well as what you should do right now, please see this message. Comments, questions, and discussion are also located in that thread.

https://www.primegrid.com/forum_thread.php?id=8750&nowrap=true#132260
 
New Project: Fermat Divisor Search
PrimeGrid is proud to announce that we have a new sub-project, the Fermat Divisor search.

This is a variant of our other PPS (Proth Prime Search) sub-projects and is specially designed to have a better chance of finding a Fermat Divisor.

Unlike most of our other sub-projects, this one will not run forever. It has a limited number of candidates to check, and once they're gone, the project is over. My best guess is that this project will last about three months.

It uses the same badge and stats as the other PPS sub-projects, so there's no new badge. But this is probably the best chance of discovering a Fermat divisor and earning that rare
fermat.png
badge.

For more information and discussion, read this forum thread or join our Discord server. 6 Sep 2019 | 22:18:39 UTC · Comment
 
Back
Top