1)
Message boards :
Number crunching :
Transit of Mercury Across the Sun Challenge
(Message 134669)
Posted 1425 days ago by Sebastian*
Sorry to have caused a major misunderstanding here, that was not my intention. :)
I didn't know that people go so far and run custom software at this point. Apart from optimized .exe files.
I still have to figure out how to convince Boinc on Windows to put tasks on numa nodes. And then stopping it from trowing tasks around from one node to the other.
Pavel, how many processors, and which ones are you running for this challenge?
|
2)
Message boards :
Number crunching :
Transit of Mercury Across the Sun Challenge
(Message 134640)
Posted 1425 days ago by Sebastian*
I need a little help guys. Check out the number one user for this challenge.
http://www.primegrid.com/challenge/2019_9/top_users.html
http://www.primegrid.com/hosts_user.php?userid=914937
He only has 2 2600K CPU, and one Q9400 CPU.
Could somebody explain what is going on here?
Is Bunkering allowed at this time? Or is just cheating going on?
Or is the Primegrid score system just broken?
I challenges like this, only the Workunits sent out, and returned within the challenge time count, right?
|
3)
Message boards :
General discussion :
i9 9900k causes noises when running Primegrid
(Message 130625)
Posted 1561 days ago by Sebastian*
I bought the 9900k especially for Primegrid. And as far as i have seen i get the sound as soon as i run one workunit at all 8 cores. No matter the subproject, if it supports multithreading.
I am very curious about the new Ryzens. The way it looks they should be comparable in AVX performance to the Intel CPUs. If that is true, i will try to get rid of the 9900k.
Still nobody with a 9900k?
AP is for GPUs, so no very helpful for the CPU ;)
|
4)
Message boards :
General discussion :
i9 9900k causes noises when running Primegrid
(Message 130616)
Posted 1561 days ago by Sebastian*
If it is coil whine then why does it not change when i use a Mainboard with a totally different VRM setup or a different PSU. The sound changes when i change the Memory Frequency.
Does anyone of you have a 9900k, so we could reproduce the problem i have?
And it is pretty damn loud for coil whine as well.
|
5)
Message boards :
General discussion :
i9 9900k causes noises when running Primegrid
(Message 130553)
Posted 1563 days ago by Sebastian*
Hi all, i hope you can help me, or at least point me to some new ideas.
Subprojects: The Riesel Problem LLR and other which allow to run one WU on all CPU cores at once. Set to 8 cores (HT on, simulate 8 cores).
My i9 9900k causes some high pitched noises when i run Primgrid on it.
I've been testing on two AsRock Z390 Tachi Ultimate and one MSI Z390 Carbon. I also have now the third i9 9900k (all P0 revision). I've tested with GPU and without, tried several PSUs, even very old ones, as well. Always the same noise. Watercooling (AIO) or aircooling makes no difference.
The only change of the noise was, when i changed the memory frequency.
The CPU runs quiet except when i run Primegrid. The CPU draws around 180W all at stock in the Bios. Prime95 draws up to 200W but all quiet.
I am in contact with Intel, again, because i sen't the first CPU to them. Now with the third CPU i've gone though all the first level support stuff again, and no answer for a week now. I've been asking for a R0 Revision CPU.
All my other Intel CPUs run well with the same setup. (4790k, 5665C, 3960X, 2600K...)
I am having the problem for half a year now, and i've no idea what else i could try.
My current hypothesis is, that the soldered heatspreader might act as an amplifier. But i won't delid any 9900k just to test it. If someone has a delidded one, it would be nice to test with it.
I have made 2 videos with the noise. But you have to watch them to the end (around 40 seconds each). At the end you hear the variations of the noises.
Dropbox links:
https://www.dropbox.com/s/99j5lrcjc7yicy3/20190528_194737.mp4?dl=0
https://www.dropbox.com/s/a6ipb7l0jvjoo77/20190528_194548.mp4?dl=0
List of what i tired so far:
- Three i9 9900k
-Onboard GPU, different PCI-E GPUs
-at least three different PSUs
-two AsRock and one MSI mainboards
-AIO Watercooling, Aircooling
-different Memory (noise changes with memory speed)
If the cpu is under load, you can even hear a difference in the noise when you move windows on the desktop around.
If some else has a 9900k i hope this person could help me to repoduce the problem
|
6)
Message boards :
Number crunching :
Halloween Challenge
(Message 121498)
Posted 1800 days ago by Sebastian*
Thank you for all the answers so far. :)
I've checked the actual time of the upload / report. And this seems pretty consistent.
On a quick check what version of Boinc people use, it seems that the Boinc version 7.14.2 is reporting incorrect run times. On older versions the time seem very consistent.
My best guess for now is that the latest Boinc version is just reporting false times. Maybe a calculating bug of some kind?
xii5ku wrote: - The L3 cache that your 4-core CPUs have is a lot smaller than llrWOO tasks like.
I run several Broadwell LGA 1150 CPUs which have a 4th level cache of 128MB. They are very fast, and perform better then any other quad core cpu with even faster frequencies. But the smaller the project the less of an impact it has.
There might be improved times of the huge L3 caches of the high core count cpus, when you run one WU on all cores, but i have no idea how much cache memory it takes for the llrWOO tasks.
Keith wrote: When using multi-threading through app_config any units you have waiting to go will count time towards the run time even if they are not actively being worked on. xii5ku wrote: This is misleadingly worded, IMO.
When multithreading is on, and more than 1 task is downloaded within a single request for new work, run time reporting when a job is completed is buggy.
You need to ignore the run times that you see in the results tables on the project web server, unless it was made sure that no more than 1 task was downloaded per request, in each and every request.
And this might be what causes my inconsistent run times. I have several tasks waiting, sometimes it is just one, on other times it is more. The run time shown is just several times the time of the lowest run time.
How can you make sure, that only one WU is downloaded per request, when already one is in progess? Is it even worth the trouble?
|
7)
Message boards :
Number crunching :
Halloween Challenge
(Message 121494)
Posted 1800 days ago by Sebastian*
Hello everyone,
does everyone else experience very inconsistent run times, while the CPU times for the Work Units is more or less the same.
Check my computers for this:
http://www.primegrid.com/hosts_user.php?userid=63648
I use all cores on each CPU (usually 4) to run one WU. There is no other Boinc Application or program running in the background. So i am very confused why my run times are so inconsistent. I am using the app_config file as posted in the thread.
Thanks for any help.
|
8)
Message boards :
Number crunching :
Winter Solstice Challenge
(Message 60683)
Posted 3937 days ago by Sebastian*
Well done Dutch Power Cows. Congratulations on winning this Challenge. I hope some people who joined you during the challenge will stay and keep up the good work :)
Thanks to everyone participating in this crazy challenge.
|
9)
Message boards :
Number crunching :
Winter Solstice Challenge
(Message 60657)
Posted 3938 days ago by Sebastian*
We just passed the 100 Million mark in the Challenge :)
|
10)
Message boards :
Number crunching :
The Autumnal Equinox Challenge
(Message 40735)
Posted 4391 days ago by Sebastian*
did as you told zlodeck. even did a bios update. Same problem again. Collatz doesn't have that prob, also minecraft is running fine. No crashes, and that on java...
How can i call a developer?
|
Next 10 posts
|
