| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Active Research
Forum: OpenPandemics - COVID-19 Project
Thread: Disc activity |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 15
|
|
| Author |
|
|
Yamgalf
Cruncher Sweden Joined: Aug 31, 2010 Post Count: 6 Status: Offline Project Badges: 
|
I have a few computers running WCG.
The older ones has HDD (not SSD). The CPU load is not 100% all the time, jumping up and down a little. It seems like when the HDD is busy, it can't support the CPU fast enough with disk operations. The OpenPandemics seems to do more disc activity compared to other WCG projects. When running other WCG projects, disc activity is less and CPU load is 100% all the time. The HDD is a 500GB WDC with 43% occupied space. Computer has 16GB RAM, swap file exists but is not used. CPU is a i7 2600K. Questions: Has anyone observed this? Is it a good idea to replace the HDD whith SSD? |
||
|
|
tcora
Cruncher Joined: Feb 24, 2009 Post Count: 5 Status: Offline Project Badges: 
|
I don't see any real significant disc activity (MacOS 10.15.7) that correlates w/ WCG processes.
|
||
|
|
Sgt.Joe
Ace Cruncher USA Joined: Jul 4, 2006 Post Count: 7846 Status: Offline Project Badges: 
|
One question I would ask is: What is your operating system ?
----------------------------------------I run a I7-2600K with a 1tb hd running Linux Mint. It runs a mix of OPN, SCC and the occasional ARP. My system is consistently pegged at 100% cpu. I see the occasional dip to 99% when I am typing, but that is it. Cheers
Sgt. Joe
*Minnesota Crunchers* |
||
|
|
Yamgalf
Cruncher Sweden Joined: Aug 31, 2010 Post Count: 6 Status: Offline Project Badges:
|
I'm running Ubuntu 16.04.
|
||
|
|
Jim1348
Veteran Cruncher USA Joined: Jul 13, 2009 Post Count: 1066 Status: Offline Project Badges:
|
It is easiest to monitor the writes on Windows 7 64-bit with SsdReady (free version is OK). I am running OPN on six (out of eight) virtual cores on an i7-4771, with the other cores free. I am seeing writes to disk of about 60 GB/day. That is higher than most WCG projects, which are usually only 20 or 30 GB/day, but not excessive for my SSD, a Samsung 850 Pro. I use the Samsung Magician utility, which includes a write-cache that helps reduce the writes to the SSD.
I have several machines on Ubuntu, and use iostat to measure disk writes there. Try "iostat 3600 -m", and it will give you the writes per hour (3600 seconds) in megabytes. Ignore the first reading, since it will be the cumulative writes since the last reboot. As for the CPU, it is at 99+ percent for both Windows and Ubuntu. You would have to have a serious write bottleneck to reduce it below that. |
||
|
|
IBM01902
Cruncher Joined: Aug 13, 2017 Post Count: 11 Status: Offline Project Badges: 
|
If you're looking at disk activity when tasks first start up, I'd say that's normal.
5 minutes or more and the disk activity should be very light, so look somewhere else. WCG/Open Pandemics is a fairly light disk and memory tax. None of my stuff is a new as your old stuff. One Boinc parameter you could relax is the allowable time between checkpoints. Look at Options, Computer Preferences, Other, Request Tasks to Checkpoint at Most... I make this 600 seconds because I tend to let the computers run for hours without interrupt. I may lose 10 minutes at the end of a day per task. If you run like a student, and the computer is only crunching when the screen saver is on, and there are frequent computer off states during the day, this might cause you to never make any progress because you're always losing work from interrupts. The Options/... path might be slightly different on your version of Boinc. Also look at Suspend When Non-Boinc Useage is above nn%. This is as it states, Non-Boinc, it doesn't have to be just your keyboard/mouse, you may have other processes causing Boinc to pause. These processes should be evident in the System Monitor sorted by cpu useage. Lastly, maybe just live with it. If the tasks are completing close to the estimated time, don't drive yourself crazy. Good Luck |
||
|
|
Yamgalf
Cruncher Sweden Joined: Aug 31, 2010 Post Count: 6 Status: Offline Project Badges:
|
I have used iostat, thanks for the advise.
----------------------------------------Just normal ammount of data is read/written. I have halso changed the checkpoint time, no differce. The computer is running 24/7 and the screensaver is turned off. S.M.A.R.T does not report any problems. The HDD has been running for 5 years. I will replacet it soon. I vill also upgrade ubuntu to a later version if it does not help i will live the problem since it's not a big issue. Thanks for all advises and ideas. [Edit 1 times, last edit by Yamgalf at Nov 9, 2020 3:08:13 PM] |
||
|
|
KerSamson
Master Cruncher Switzerland Joined: Jan 29, 2007 Post Count: 1684 Status: Offline Project Badges:
|
Long time ago, there was discussions regarding "HDD vs SSD".
----------------------------------------I operate since years multithreading machines with HDD and SSD and, based on BoincTasks, I did never notice a real performance difference between HDD and SSD running machines. All machines are between 99.2% and 99.8% CPU performance. Because no real performance difference is noticeable, I usually put the BoincData directory on a HDD and I keep the system on SSD, avoiding to over use the SSD. I assume that the performance problem you experience is probably related to a service outside of Boinc. Cheers, Yves |
||
|
|
Sgt.Joe
Ace Cruncher USA Joined: Jul 4, 2006 Post Count: 7846 Status: Offline Project Badges:
|
All of the work units here are cpu bound, not disk bound. The only project I recall as having an inordinate amount of disk activity was the Clean Energy Project, which had a huge amount.
----------------------------------------Cheers
Sgt. Joe
*Minnesota Crunchers* |
||
|
|
|