Automatic Temperature regulation

Message boards : Questions and problems : Automatic Temperature regulation
Message board moderation

To post messages, you must log in.

Previous · 1 . . . 8 · 9 · 10 · 11

AuthorMessage
HAL9000
Help desk expert
Avatar

Send message
Joined: 13 Jun 14
Posts: 81
United States
Message 68606 - Posted: 29 Mar 2016, 18:24:16 UTC - in response to Message 68591.  
Last modified: 29 Mar 2016, 18:38:50 UTC

TThrottle may STOP working in the months to come.

Microsoft made driver signing prohibitively expensive we are talking $500 and more each year.

So right now we made the decision to stop the development for now, we might come up with a solution or the driver may keep on running longer than expected.

What is it they changed? A few years ago they stopped changed the $250 per driver submission to $0. So the only cost was the 3rd party code signing certificate.

EDIT: I see the Symantec code signing certificate is now $500/yr.
MS has a list of accepted code sign vendors some are less costly. Like global sign lists $219/yr.

Also a signed driver shouldn't stop working. They don't have an expiration date. Just a date when they were signed. Unless you are constantly updating the driver to work with the app code it should not require a new submission.
ID: 68606 · Report as offensive
Fred - efmer.com
Avatar

Send message
Joined: 8 Aug 08
Posts: 570
Netherlands
Message 68616 - Posted: 29 Mar 2016, 23:11:56 UTC - in response to Message 68606.  
Last modified: 29 Mar 2016, 23:12:18 UTC


Also a signed driver shouldn't stop working. They don't have an expiration date. Just a date when they were signed. Unless you are constantly updating the driver to work with the app code it should not require a new submission.

I'm not sure what's going to happen, there is a lot of confusion.
The certificate I signed it with will expire in a couple of months, it may stop working, or not that's the question.

Some notes say older drivers will be accepted by W10 nevertheless, but the driver wasn't signed with a timestamp, some notes indicate that it will stop working 29th of July 2016.

They made a special certificate off course more expensive for Kernel drivers.
https://www.globalsign.com/en/code-signing-certificate/ev-code-signing-certificates/
TThrottle The way to control your CPU and GPU temperature.
BoincTasks The best view of BOINC.
My other activities
ID: 68616 · Report as offensive
HAL9000
Help desk expert
Avatar

Send message
Joined: 13 Jun 14
Posts: 81
United States
Message 68617 - Posted: 30 Mar 2016, 0:01:15 UTC - in response to Message 68616.  


Also a signed driver shouldn't stop working. They don't have an expiration date. Just a date when they were signed. Unless you are constantly updating the driver to work with the app code it should not require a new submission.

I'm not sure what's going to happen, there is a lot of confusion.
The certificate I signed it with will expire in a couple of months, it may stop working, or not that's the question.

Some notes say older drivers will be accepted by W10 nevertheless, but the driver wasn't signed with a timestamp, some notes indicate that it will stop working 29th of July 2016.

They made a special certificate off course more expensive for Kernel drivers.
https://www.globalsign.com/en/code-signing-certificate/ev-code-signing-certificates/

I only did driver submissions to MS for ~10 years, but I can tell you that a signed driver never expires. When your code sign certificate expires that just means you can no longer use it to sign files.

When a driver is signed it works in the OS is was signed for and newer. So if a driver is signed for Windows 7. Then Windows 7/8/8.1/10 & Server 2008R2/2012/2012 R2 will all consider it signed. While Windows 2000/XP/Vista & Server 2000/2003/2008 will not.
ID: 68617 · Report as offensive
mike bonewits
Avatar

Send message
Joined: 9 Dec 15
Posts: 3
Message 69555 - Posted: 10 May 2016, 18:44:06 UTC

Any plans to port to linux?
I'm using Ubuntu 14.04.3 LTS Trusty Tahr
on a HP Pavilion g7 - 2269wm
AMD Quad-Core A8-4500M Accelerated Processor

As a retired QAA, I would be happy to test.
ID: 69555 · Report as offensive
Fred - efmer.com
Avatar

Send message
Joined: 8 Aug 08
Posts: 570
Netherlands
Message 69563 - Posted: 11 May 2016, 10:15:41 UTC - in response to Message 69555.  

Any plans to port to linux?

Linux means a complete rewrite, in other words building it from scratch.
An right now I'm busy with entirely.

Right now I'm busy with my other hobby http://borregowildflowers.com/
building an iOS app, that's months of work.
I can do only so much.
TThrottle The way to control your CPU and GPU temperature.
BoincTasks The best view of BOINC.
My other activities
ID: 69563 · Report as offensive
Barbara Mills

Send message
Joined: 30 Jan 13
Posts: 13
United States
Message 71657 - Posted: 17 Aug 2016, 21:55:05 UTC - in response to Message 68617.  
Last modified: 17 Aug 2016, 21:58:42 UTC

ahah, this totally explains something odd - about a week ago I had to upgrade to Win 10 - I sent FMER an e-mail asking about compatability of T throttle since his website didn't say anything about it...

MSN/Outlook Mail sent me a message stating that my mail to you would be "delayed" and that I didn't need to resend my message...I truly thought it was extremely odd at the time and I still have not seen a reply to my mail to Fmer...Me thinks big brother is monitoring mails...


Content of e-mail:

postmaster@mail.hotmail.com
Reply|
8/7/2016
boinc@fmer.com
This is an automatically generated Delivery Status Notification.

THIS IS A WARNING MESSAGE ONLY.

YOU DO NOT NEED TO RESEND YOUR MESSAGE.

Delivery to the following recipients has been delayed.

boinc@fmer.com
ID: 71657 · Report as offensive
Dr Who Fan
Avatar

Send message
Joined: 10 May 07
Posts: 1328
United States
Message 71658 - Posted: 17 Aug 2016, 22:30:25 UTC - in response to Message 71657.  
Last modified: 17 Aug 2016, 22:31:00 UTC

ahah, this totally explains something odd - about a week ago I had to upgrade to Win 10 - I sent FMER an e-mail asking about compatability of T throttle since his website didn't say anything about it...

MSN/Outlook Mail sent me a message stating that my mail to you would be "delayed" and that I didn't need to resend my message...I truly thought it was extremely odd at the time and I still have not seen a reply to my mail to Fmer...Me thinks big brother is monitoring mails...


Content of e-mail:

postmaster@mail.hotmail.com
Reply|
8/7/2016
boinc@fmer.com
This is an automatically generated Delivery Status Notification.

THIS IS A WARNING MESSAGE ONLY.

YOU DO NOT NEED TO RESEND YOUR MESSAGE.

Delivery to the following recipients has been delayed.

boinc@fmer.com

You have the WRONG INFO FOR SENDING EMAIL. You forgot the "e" on the beginning of the URL.

Correct web site address is: http://efmer.com/b/tthrottle
Correct email address is: boinc@efmer.com
Try resending it to the correct one noted above.
ID: 71658 · Report as offensive
Fred - efmer.com
Avatar

Send message
Joined: 8 Aug 08
Posts: 570
Netherlands
Message 71677 - Posted: 18 Aug 2016, 13:10:00 UTC - in response to Message 71658.  

[quote]ahah, this totally explains something odd - about a week ago I had to upgrade to Win 10 - I sent FMER an e-mail asking about compatability of T throttle since his website didn't say anything about it...

It should run just fine on Win 10, my development machine is Win 10.
Thanks for reminding me to update the download page to include W10.
TThrottle The way to control your CPU and GPU temperature.
BoincTasks The best view of BOINC.
My other activities
ID: 71677 · Report as offensive
Barbara Mills

Send message
Joined: 30 Jan 13
Posts: 13
United States
Message 71710 - Posted: 19 Aug 2016, 23:44:00 UTC - in response to Message 71677.  

Oops, my apologies on the email address...

You've pretty much answered my question here...and I"ve gone ahead and downloaded Tthrottle already - its been a lifesaver for my computer which tends to overheat without it...so I"m grateful for this program!!!

Til I found it I was thinking I might have to stop crunching because of the overheating issues.

Thank you so much.
ID: 71710 · Report as offensive
Walter Kraslawsky

Send message
Joined: 3 Sep 16
Posts: 2
United States
Message 72190 - Posted: 3 Sep 2016, 15:56:39 UTC - in response to Message 71677.  
Last modified: 3 Sep 2016, 16:07:13 UTC

I was looking for discussions on throttling cpu% because of heat problems and found your thread. It's nice to see that you have something useful. I might try your software on two PCs, but am focusing my attention on my much faster iMac since that is where I have the heat issue. That being said, I'm still chiming in here because I have some observations that might be useful. I am pretty sure any PC user will see the same if they have an Intel cpu.


I joined BOINC a few days ago and an hour after starting SETI@home I discovered my iMac was blasting hot air. I quickly located the BOINC cpu% setting and kicked it down to 50%, but that was also running uncomfortably hot, so I went to 25%.

Then I found TG Pro (a temperature gauge product that works for Macs). It showed me that when BOINC is set to throttle to something less than 100% (e.g. 25%), with 1 second snapshots, my four i7 cores would simultaneously cycle up to the redline max 212F, then down to green 150F, then up to 212F, then down to 150F, over and over and over. This on-off duty cycle corresponding to the selected cpu%; i.e. all four cpu cores quickly pinned to the redline 25% of the time and then very quickly cooled to green 75% of the time. TG Pro also showed that the fan speed shifted much more moderately, wandering up and down in the green zone (0% to 50% of max fan speed) at the 25% cpu BOINC setting, but would slide up and down between yellow (50%-75%) and red (75%-100%) at the 50% cpu BOINC setting. Obviously running 100% cpu also pinned the fan at 100% max speed.

The bottom line is that I found random snapshots of CPU temperature were useless to me for picking the BOINC setting. I had to use the fan speed instead (as well as just deciding what level of heat output worked for me). I don't know how other products work, but look forward to any product that can report a moving average of cpu temperature and fan speed, letting the user choose the average smoothing interval.

Edit: PS if anybody knows the BOINC on+off duty cycle, then let me know. I assume that would be the best smoothing interval.
ID: 72190 · Report as offensive
SekeRob2

Send message
Joined: 6 Jul 10
Posts: 585
Italy
Message 72208 - Posted: 4 Sep 2016, 6:48:09 UTC - in response to Message 72190.  

The problem of BOINC is, it runs at 1 second control intervals as are the science applications compiled to call and listen at that pace. When the client in development was changed to 1/10th of a second, pretty much all those science apps were not following in step **, so the code was reversed [cause a recompile requirement of all the projects was a bridge too far it seems], so here we are years later and still no proper temperature control, plus all sciences still in step pausing/running, instead of doing it asynchronous... heat all, cool all, heat all and so on.

** Did have an okay experience when testing, the load graph in task manager nice and even, and it would have worked on all platforms.
Coelum Non Animum Mutant, Qui Trans Mare Currunt
ID: 72208 · Report as offensive
PMH_UK

Send message
Joined: 24 Dec 10
Posts: 35
United Kingdom
Message 72213 - Posted: 4 Sep 2016, 11:36:04 UTC - in response to Message 72190.  

You could try limiting to 1 core or 25% of CPUs to keep temperatures under control rather than running all cores 25% of the time.

Paul.
ID: 72213 · Report as offensive
SekeRob2

Send message
Joined: 6 Jul 10
Posts: 585
Italy
Message 72214 - Posted: 4 Sep 2016, 12:42:07 UTC - in response to Message 72213.  

Here's what you get limiting cores in a 4770K 8-threaded CPU, [which is hard set in the control panel at 2.8Ghz as 3.4 would make it run to hot in our ambient temps], noting the CPU scales by itself, observed with CoreTemp and Open Hardware Monitor, with each increment allowed a minute to settle:

1 - 6 to 6.8 watts - 41 to 42C - 0.8Ghz
2 - 7 to 8 watts - 41 to 42C - 0.8Ghz
3 - 8 to 10 watts - 43 to 44C - 0.8 to 1.0Ghz
4 - 9 to 11 watts - 45 to 48C - 0.8 to 1.2Ghz
5 - 33.3 to 34.9 watts - 55 to 57C - 2.8Ghz
6 - 35.3 to 36.0 watts - 59 to 63C - 2.8Ghz
7 - 38.2 to 38.5 watts - 67 to 70C - 2.8Ghz
8 - 39.1 to 40.0 watts - 71 to 73C - 2.8Ghz

Fan is audibly speeding up as cores are being added to crunch. Obviously, the 4 cores and less results in a per-core production drop of 233% to 350%, system overall 2 or 8 cores a drop of 1400% in throughput.
Coelum Non Animum Mutant, Qui Trans Mare Currunt
ID: 72214 · Report as offensive
Raistmer

Send message
Joined: 9 Apr 06
Posts: 302
Message 72218 - Posted: 4 Sep 2016, 15:04:03 UTC - in response to Message 72214.  

Running 25% with all cores has another negative effect versus running only 25% of cores - bigger cache contention. If 8 processes share cache, per-CPU throughput drops quite considerably. That is, leaving 4 of 8 CPUs running doesn't mean 2-times total throughput decrease. Real decrease will be far less.
ID: 72218 · Report as offensive
Stevef9432203

Send message
Joined: 23 Jun 17
Posts: 1
United States
Message 79465 - Posted: 5 Jul 2017, 1:42:37 UTC

Hello,
Here is a link https://sourceforge.net/projects/boinc-gpucontrol/ tp a Source Forge Program which controls the GPU temperature in Linux for NVIDIA GPUs. We set a temperature target of 81C, and automatically suspend the GPU's work when this temperature is reached. The thresholds are easily changed in the code. I put enough comments so the code should easily be readable by anyone with at least some knowledge of shell scripting.

This is a first-pass at it, I tried it with both of my GPUs, and it works.



I set the cool-down threshold to be as short as possible to maintain the temperature, and at the same time, allow as much computing time as possible.

Thanks!

-SteveF
ID: 79465 · Report as offensive
Fred - efmer.com
Avatar

Send message
Joined: 8 Aug 08
Posts: 570
Netherlands
Message 79610 - Posted: 15 Jul 2017, 11:04:12 UTC - in response to Message 79465.  

A new update for the small temperature window.
It's now better scalable.

https://efmer.com/download-tthrottle and choose the beta version.
TThrottle The way to control your CPU and GPU temperature.
BoincTasks The best view of BOINC.
My other activities
ID: 79610 · Report as offensive
Fabio

Send message
Joined: 20 Aug 17
Posts: 1
Brazil
Message 80360 - Posted: 20 Aug 2017, 13:21:13 UTC - in response to Message 22485.  
Last modified: 20 Aug 2017, 13:25:41 UTC

Fred I don´t know if you are the right person to ask, but I´ve been looking for information about a Boinc Server software and didn´t find any. I also didnt find any e-mail that I could contact. As a user of Boinc for 3 years or so I started to wonder if there is any server software that I could register my projects to. I started The Materials engineering major in 2013 and now I use a lots of computational resources over the Physics dept to perform calculations in DFT, Hartee-Fock and many other technics to understand better the properties of the matter. If you could help me with any information I would be very pleased. Thank you.
ID: 80360 · Report as offensive
Fred - efmer.com
Avatar

Send message
Joined: 8 Aug 08
Posts: 570
Netherlands
Message 80361 - Posted: 20 Aug 2017, 13:25:16 UTC - in response to Message 80360.  

Fred I don´t know if you are the right person to as

??
https://boinc.berkeley.edu/trac/wiki/ServerIntro
TThrottle The way to control your CPU and GPU temperature.
BoincTasks The best view of BOINC.
My other activities
ID: 80361 · Report as offensive
Previous · 1 . . . 8 · 9 · 10 · 11

Message boards : Questions and problems : Automatic Temperature regulation

Copyright © 2024 University of California.
Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.