| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: FightAIDS@Home
Thread: Total CPU usage only at 83% on my dual core |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 26
|
|
| Author |
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
I just opened my windows task manager and WCG_Autodock is using anywhere from 28% to 30% of core #2 CPU AT 105,496K memory usage. System Idle Process: 20% CPU usage. F@H is at 50% of core #1.
The graph shows total CPU usage at 78% to 83%. What is the system idle process for? That is new to me. Shouldn't be anything else competing for CPU usage at the moment that I know about. My work units have gone from 4.5 hours of completion to maybe upto 13 hours in the last few days that I have took notice. Even mnore odd is the processor score still appears to remains normal at 217 to 222. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
There have been some changes, rbolo28. Check out the Member News forum.
The System Idle process does nothing - literally. It just fills in the time not used by other processes with no-ops. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Alright..fixed that with the tool. Read the post and all. Understand maybe some people might not like seeing 100% CPU usage but my dual core machine doesn't notice or see a difference at all. Even my old single core p4 had no problems.
Why not just include the option when installing and setting up the program? Have an option to use 100% CPU cycles by default or whatever percentage the user chooses. Maybe, provide a recommendation based on the detected hardware? I think this could waste alot of power and efficiency where a ton of users might not even know the difference between 80% and 100% CPU usage. I don't like having to use a special tool to get 100% usage that should be the default setting. I think the mass majority would have no problem with 100%..especially with the newer computers and dual cores now becoming widely available everywhere. If you can't keep up then pick another project or upgrade the computer! Gotta be a better solution...this is ugly to see. Most people aren't going to know enough to use that tool...even if they did...they probably won't. Catering to the weaker computers by DEFAULT isn't the answer in my opinion. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
It's a tough choice.
WCG considered it carefully. I'm sure they made the right decision. They will be watching developments carefully, and if they were wrong then I'm sure they will admit it and alter the default. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
The UD agent was throttled back to prevent CPU overheating. It has nothing to do with keeping other apps running smoothly. To help keep other apps running smoothly WCG raised the basic system requirements for HPF2. I think you might have those 2 issues confused.
The consquences of CPU overheating can be very bad. If someone's CPU fried WCG could have claimed it was their own fault because they didn't read the fine print. Unfortunately that would leave WCG looking rather callous. Think about it.... wouldn't people say they were just trying to squeeze out every last CPU cycle and to heck with the volunteer's CPU? By throttling back the CPU, WCG demonstrates that they put their volunteers first even if it slows down the crunching. I applaud WCG for doing the right thing and acting responsibly. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
I agree with the idea of being responsible and I've thought about the CPU heat issue myself. I originally had the stock CPU fan for my AMD X2-4400 and it ran like an aeroplane and caused alot of noise at 100% CPU usage. I upgraded to a liquid cooled system on my dual core and everything is fine now running 100% on both cores at 42c in the hot desert of New Mexico. My friend has a p4(2.8ghz) which I built that uses factory Intel heat sink and runs warm but nothing close to critical.
I do know that some people also freak out when they see 100% CPU usage and think something must be wrong. These are your average computer users of course...maybe even some so called experts also because they never heard of "distributed computing" and half don't even understand when you try to explain it to them because it matches nothing they learned in the text books. I've seen more intelligent programs which monitor hardware and CPU heat generated and might throttle back in accordance. None of this has yet made it into the distributed computing program world tho. Certainly not a black and white issue of course. Might take more intelligent design to squeeze out efficiency safely and responsibly. Hopefully, time will sort the issues out and reach the optimal solution for everyone. I personally prefer to see some big numbers, scores and ripping through these projects left and right to find CURES or better TREATMENTS...people are still dying and suffering regardless! |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
rbolo28,
Glad to hear you're staying  in New Mexico. We don't know what HOT means here in Calgary  It sounds like you have the tools and knowledge required to build and modify computers. I do too and it gives one a few more options when problems crop up. CPU running hot? No problem... cut another hole in the case and add another fan or 2. Or maybe a liquid cooling system like you did. When you get 100,000 average people with average computers purchased at average stores and they join a grid and try to do an above average computing job there's bound to be be a few problems. Time and good FAQs will cure all. I wish manufacturers would beef up the cooling systems they install. That would really help a lot. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
I think the real issue here is notification. Except for my team thread, I don't often look at the others in the forum so I had no idea about the cpu throttle. Found out about it by chance when testing some new software and had a look at task manager. When I saw cpu usage varying between 55 & 60%, my first thought was that the software I was testing did something to the system. Shut that down but % didn't change so I went to the forum here and saw the reason. Much relief! I got the tool and restored things to 100%. None of my systems have heat issues so running flat out doesn't bother me.
As I said, notification of some sort would have been nice. I'm sure that there are many, many members who have no idea what has happened because they run these projects passively, don't read the boards and probably have no idea what a task manager is or how to bring it up. In the meantime, taking longer to complete each workunit, 2 to 3 times longer, consumes more energy for the same result, while the effect on the cpu is negligable. Given the high cost of electricity, you really didn't do users a favor here. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
I agree, notification is the issue. WCG did a poor job of that and there was no need for it.
|
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
What would you like WCG to do?
Imagine the outcry if they sent out a mass email over a procedural change like this? I'm sure they will include it in the next newsletter (must be due soon, they seem to come out quarterly). Remember, they can't change the UD interface to alert of changes like this. If they could, they would have been able to implement the entire throttling thing much more elegantly. What other options can you think of? |
||
|
|
|