The Corsair User Forums  

Go Back   The Corsair User Forums > Corsair Product Discussion > Corsair Link

Notices

Closed Thread
 
Thread Tools Search this Thread Rating: Thread Rating: 83 votes, 4.87 average. Display Modes
  #1216  
Old 09-18-2017, 03:21 AM
red-ray red-ray is offline
Banned
red-ray's PC Specs
 
Join Date: May 2014
Location: England (GMT+1)
Posts: 7,148
POST ID # = 920423
red-ray Reputation: 81
Idea Could the sleep be bringing timers into phase?

Quote:
Originally Posted by 5kft View Post
As always, please let me know if you need any additional information or anything.
Hi K, Thank you for the report, I have been wondering what could cause this effect and how we could isolate the cause.

My current theory is that after the sleep all the timers end up in phase and my reasoning is as follows:
  1. The basic change for 5.23 was is to ensure that SIV was only running one thread, wait two ticks and then read the CPU ratios.
  2. I can't see how doing a sleep could affect #1 and suspect there must be some external factor.
  3. If other programs are running in phase with SIV then SIV may report higher speeds as a result of this, but in general this does not happen.
  4. Before the sleep other programs were running, but not in phase with SIV. Maybe one was, but not all of them.
  5. After the sleep all the timers will have expired so will be run bringing them all into phase with SIV.
  6. If they are updating every 2.0, 1.0 or 0.5 seconds they will stay in phase with SIV.
  7. When you exit and restart SIV it is likely that SIV will be out of phase with other programs.
  8. Did you have any other programs active and what happens if you exit them after the sleep?
  9. What happens if you specify -UPDATE=1.75 (-UPDATE=1.75:2:10) so that SIV updates every 1.75 rather than every 2.0 seconds? You could also try -UPDATE=2.25 (-UPDATE=2.25:2.25:10).
  10. What happens if you change the power scheme from Balanced to Power Saver after the sleep?
BTW As soon as the system awakes from sleep please press [Reset] on [CPU Speed Ratios] so it just contains the post sleep ratios. Same for changing the power scheme.

Do you know of a utility that will report all the currently active timers please?

Quote:
Originally Posted by 5kft View Post
From this I can infer that there isn't some other process in Windows that is causing clock rates to go to 4.9GHz.
Given #5 above I don't think you can.

Last edited by red-ray; 09-18-2017 at 10:55 AM. Reason: #5


1 members found this post helpful.
  #1217  
Old 09-18-2017, 03:38 AM
red-ray red-ray is offline
Banned
red-ray's PC Specs
 
Join Date: May 2014
Location: England (GMT+1)
Posts: 7,148
POST ID # = 920424
red-ray Reputation: 81
Unhappy Corsair have not commented

Quote:
Originally Posted by 5kft View Post
It's rather irritating as when the fans sweep through the curve then they turn off and back on (as previously noted in http://forum.corsair.com/v3/showthread.php?t=170330). I hope that Corsair fixes this soon .
Yes it is. I don't even know if Corsair have realised this as there are no posts from Corsair after my http://forum.corsair.com/v3/showthread.php?p=919108 post.

Have you reported this suspected firmware issue via https://support.corsair.com/? If not than I suspect you should do so.

I feel Corsair should fix the firmware ASAP, then make a firmware beta available to selected members so we can validate it and check that there are no regressions .


  #1218  
Old 09-18-2017, 05:53 AM
red-ray red-ray is offline
Banned
red-ray's PC Specs
 
Join Date: May 2014
Location: England (GMT+1)
Posts: 7,148
POST ID # = 920428
red-ray Reputation: 81
Confused From strange to even stranger

Quote:
Originally Posted by 5kft View Post
As always, please let me know if you need any additional information or anything.
Hi K, I have managed to create some similar effects on one of my systems. Please will you check if you get these effects please?

Quote:
  1. What I did was start both SIV + CPUZ and both were reporting sensibly.
  2. Next I slept the system and when I awoke it both were reporting 2.90 GHz as above. Note I pressed [Reset] as soon as the system awoke.
  3. I felt this leant weight to my theory about timers being in phase.
  4. Next I exited and restarted CPUZ and was surprised that both SIV + CPUZ then started reporting 800 MHz as below .
  5. If I have just SIV running then once I start CPUZ after the sleep then things also seem to become unstuck .
  6. I also found that changing from Balanced -> Power Saver -> Balanced also unsticks things .
  7. As to where we go from here I am open to suggestions .
  8. After some more testing then things seem to get suck on full speed for about 90 seconds and then drop back to 800 MHz as in image #3 .
Quote:

Attached Images
File Type: png RCK-1.png (205.4 KB, 428 views)
File Type: png RCK-2.png (202.7 KB, 697 views)
File Type: png RCK-3.png (224.0 KB, 431 views)

Last edited by red-ray; 09-18-2017 at 01:15 PM. Reason: Balanced -> Power Saver -> Balanced + speed drops afer 90 seconds!


1 members found this post helpful.
  #1219  
Old 09-18-2017, 10:18 AM
5kft 5kft is offline
Registered User
5kft's PC Specs
 
Join Date: Sep 2017
Posts: 47
POST ID # = 920441
5kft Reputation: 17
Arrow I like your theory

Quote:
Originally Posted by red-ray View Post
I have been wondering what could cause this effect and how we could isolate the cause.
As always, thanks for diving into this!

Quote:
My current theory is that after the sleep all the timers end up in phase and my reasoning is as follows:
  1. The basic change for 5.23 was is to ensure that SIV was only running one thread, wait two ticks and then read the CPU ratios.
  2. I can't see how doing a sleep could affect #1 and suspect there must be some external factor.
  3. If other programs are running in phase with SIV then SIV may report higher speeds as a result of this, but in general this does not happen.
  4. Before the sleep other programs were running, but not in phase with SIV. Maybe one was, but not all of them.
  5. After the sleep all the timers will have expired so will be run bringing them all into phase with SIV.
  6. If they are updating every 2.0, 1.0 or 0.5 seconds they will stay in phase with SIV.
  7. When you exit and restart SIV it is likely that SIV will be out of phase with other programs.
  8. Did you have any other programs active and what happens if you exit them after the sleep?
  9. What happens if you specify -UPDATE=1.75 (-UPDATE=1.75:2:10) so that SIV updates every 1.75 rather than every 2.0 seconds? You could also try -UPDATE=2.25 (-UPDATE=2.25:2.25:10).
  10. What happens if you change the power scheme from Balanced to Power Saver after the sleep?
BTW As soon as the system awakes from sleep please press [Reset] on [CPU Speed Ratios] so it just contains the post sleep ratios. Same for changing the power scheme.
  1. Understood. It sounds like the new implementation is simpler/more straightforward
  2. Got it. I wasn't sure if there was a different execution flow in terms of your threads/semaphores/etc. that might have occurred following wake. But given #1, I agree with you
  3. Agreed
  4. Agreed
  5. Yes, this could very well be the case
  6. Yes (typically programs update every 1.0 seconds or 2.0 seconds)
  7. Yes
  8. The interesting thing is that I tried this but I didn't notice a change. I will try this again tonight
  9. I will try this as well
  10. I will try this also

Quote:
BTW As soon as the system awakes from sleep please press [Reset] on [CPU Speed Ratios] so it just contains the post sleep ratios. Same for changing the power scheme.
Will do.

Quote:
Do you know of a utility that will report all the currently active timers please?
Not offhand, unfortunately I'll give this a search...

Quote:
Given #5 above I don't think you can.
I agree.

Quote:
  1. What I did was start both SIV + CPUZ and both were reporting sensibly.
  2. Next I slept the system and when I awoke it both were reporting 2.90 GHz as above. Note I pressed [Reset] as soon as the system awoke.
  3. I felt this leant weight to my theory about timers being in phase.
  4. Next I exited and restarted CPUZ and was surprised that both SIV + CPUZ then started reporting 800 MHz as below .
  5. If I have just SIV running then once I start CPUZ after the sleep then things also seem to become unstuck .
  6. I also found that changing from Balanced -> Power Saver -> Balanced also unsticks things .
  7. As to where we go from here I am open to suggestions .
  8. After some more testing then things seem to get suck on full speed for about 90 seconds and then drop back to 800 MHz as in image #3 .
  1. OK
  2. OK
  3. Agreed
  4. I have to admit that I laughed when I read this - i.e., laughed as in "what the heck???"
  5. Same response as for #4
  6. I will try this as well and will report back. (If this works then this deserves another "what the heck???")
  7. I totally hear you For lack of anything else offhand at this point, I've included a lame potential suggestion below...
  8. This was one of the first things I tried...I let the machine sit for quite some time following wake (probably 5-10 minutes). I noticed a bit of drop, but certainly not a fallback to normal operation (i.e., exit and restart SIV in that situation).

One quick thought...following wake, would it help to have SIV wait several ticks before continuing (i.e., so it is forced out of phase)? It seems that the "-UPDATE" test should help prove this out. Yes, I know...I know that this doesn't address root cause, but given #4, #5, and #6 immediately above, I'm not sure one can apply rational logic to this situation

I'll let you know how it goes!


  #1220  
Old 09-18-2017, 10:29 AM
5kft 5kft is offline
Registered User
5kft's PC Specs
 
Join Date: Sep 2017
Posts: 47
POST ID # = 920442
5kft Reputation: 17
Important Corsair...PLEASE fix the CLCP "fan goes to zero" issue

Quote:
Originally Posted by red-ray View Post
Yes it is. I don't even know if Corsair have realised this as there are no posts from Corsair after my http://forum.corsair.com/v3/showthread.php?p=919108 post.
We should keep bumping that thread, as it is the best consolidated location of information about the problem.

The thing is that CLCP works really well with SIV - complete control over everything (the Synthetics are great!). Everything works wonderfully, except for this ridiculous "fan goes to zero" problem.

I'll file a support issue on this as well.


  #1221  
Old 09-18-2017, 10:42 AM
red-ray red-ray is offline
Banned
red-ray's PC Specs
 
Join Date: May 2014
Location: England (GMT+1)
Posts: 7,148
POST ID # = 920444
red-ray Reputation: 81
Idea Check happens when both CPUZ + SIV are active when you sleep

Quote:
Originally Posted by 5kft View Post
I'll let you know how it goes!
Hi K, I am happy to hear you concur with #5.
  1. I feel your first check should be what happens when both CPUZ + SIV are active when you sleep.
  2. If when the system awakes both report full speed then maybe the system is really running at full speed and we need to somehow poke the system to get the speed to drop.
  3. I also checked with HWiNFO and it too reported full speed.
  4. I feel before I try and work around this we need to figure out what is really happening.
  5. I think re-running CPUZ + Balanced -> Power Saver -> Balanced may be red herrings and I just happened to do these after about 90 seconds.
BTW which time zone are you in please? I am GMT +1 (BST)

Last edited by red-ray; 09-18-2017 at 01:21 PM.


  #1222  
Old 09-18-2017, 02:14 PM
red-ray red-ray is offline
Banned
red-ray's PC Specs
 
Join Date: May 2014
Location: England (GMT+1)
Posts: 7,148
POST ID # = 920478
red-ray Reputation: 81
Happy Synthetics are great!

Quote:
Originally Posted by 5kft View Post
The thing is that CLCP works really well with SIV - complete control over everything (the Synthetics are great!).
Good, I suspected it would not take you long to start using Synthetics. I recall adding them for http://forum.corsair.com/forums/showthread.php?p=779634 and just realised this was over two years ago .

Since initially adding them I added the Ambient base and more recently the hold-off Delay to filter out short CPU core spikes. I would be interested to know how you using them so it would be good if you posted your [Synthetic Setup] panel.


  #1223  
Old 09-18-2017, 08:58 PM
5kft 5kft is offline
Registered User
5kft's PC Specs
 
Join Date: Sep 2017
Posts: 47
POST ID # = 920508
5kft Reputation: 17
Arrow results from sleep/wake testing

Quote:
Originally Posted by red-ray View Post
I am happy to hear you concur with #5.
After my latest round of tests I think I'm even more convinced

Quote:
  1. I feel your first check should be what happens when both CPUZ + SIV are active when you sleep.
  2. If when the system awakes both report full speed then maybe the system is really running at full speed and we need to somehow poke the system to get the speed to drop.
  3. I also checked with HWiNFO and it too reported full speed.
  4. I feel before I try and work around this we need to figure out what is really happening.
  5. I think re-running CPUZ + Balanced -> Power Saver -> Balanced may be red herrings and I just happened to do these after about 90 seconds.
  1. Test run results are detailed below.
  2. Actually I don't think we need to poke the system; patience seems to suffice
  3. Yes, the system appears to be doing things in background following wake, then this activity drops off (e.g., your experience with 90 seconds). Interestingly there generally isn't anything significant shown in Task Manager during this time, but something is clocking the CPU up to 4.9GHz (in both CPU-Z and SIV, and in HWiNFO, as you note). But at least they are all consistent.
  4. Agreed, however see detailed results below.
  5. Agreed. CPU-Z with or without Balanced -> Power Saver -> Balanced had no noticeable effect on my system; also the power mode didn't matter either after switching back to Balanced.

Okay, so to test this situation, here's what I did:
  1. Exit all obvious apps (all tray apps, etc.)
  2. Start CPU-Z, and start SIV using the defaults
  3. Sleep and wake the machine
  4. Both CPU-Z and SIV report 4.9GHz
  5. After a minute or so, CPU-Z drops to 800MHz, and SIV keeps bumping to 4.9GHz. So eventually they are out of sync, with SIV showing consistently higher clock rates than CPU-Z.
Next, I followed the same steps, however this time I specified the -UPDATE=1.75:2:10 qualifier for SIV in step #2. The result is that SIV tracks CPU-Z consistently. At wake, they both read 4.9GHz, and after a minute or so, they both drop to 800MHz. When CPU-Z clocks up or down, SIV reflects the same speeds.

Then I launched all the tray and background apps that I closed previously, and slept and woke the system. Again, SIV and CPU-Z operate in sync.

It sure seems like your timer phase theory is accurate!

Quote:
BTW which time zone are you in please? I am GMT +1 (BST)
I am in GMT-8, but we are currently in DST so we're currently GMT-7.


  #1224  
Old 09-18-2017, 09:19 PM
5kft 5kft is offline
Registered User
5kft's PC Specs
 
Join Date: Sep 2017
Posts: 47
POST ID # = 920509
5kft Reputation: 17
Default

Quote:
Originally Posted by red-ray View Post
Good, I suspected it would not take you long to start using Synthetics. I recall adding them for http://forum.corsair.com/forums/showthread.php?p=779634 and just realised this was over two years ago .
I'm so glad you added this facility; it's actually a VERY nice way of adding logic to various items!

Quote:
Since initially adding them I added the Ambient base and more recently the hold-off Delay to filter out short CPU core spikes.
Now that I have a CLCP installed, I added an physical sensor for air intake (i.e., ambient temperature), so now I have a sensor to assign to this item

I've been using Delay with my CLCP intake fans Synthetic, in order to add a level of hysteresis to the reported result.

I was wondering if you might have any thoughts on adding an option for a multiplicative adjustment in addition to the additive offset. E.g., if one wants to "scale" a completely different reported component temperature to better match the other component temperatures it seems that something like this could come in handy.

For example, I made a Synthetic for my intake fans that is a combination of my two GPU temperatures as well as another ambient internal case sensor. I offset the case sensor upwards a bit so that it reports higher when the case internal temperature is warmer, causing the fans to run to get cooler air into the case, even if the GPU temperatures are cool. It'd be great to be able to factor in my AIO coolant temperature as well, but the temperature scale is completely different from GPU temps and ambient case temperature. I could probably figure out an offset that would work for this (I haven't had time to think about this yet), but it seems that being able to scale the coolant temperature to match the other range more closely would be helpful.

Anyway, just a thought

Quote:
I would be interested to know how you using them so it would be good if you posted your [Synthetic Setup] panel.
Sure, I can do this as soon as I come up with something that isn't too silly


  #1225  
Old 09-19-2017, 03:20 AM
red-ray red-ray is offline
Banned
red-ray's PC Specs
 
Join Date: May 2014
Location: England (GMT+1)
Posts: 7,148
POST ID # = 920519
red-ray Reputation: 81
Question Do you think I should adjust SIV for sleep?

Quote:
Originally Posted by 5kft View Post
After my latest round of tests I think I'm even more convinced . It sure seems like your timer phase theory is accurate!
I am pleased to hear you concur. Do you think I should adjust SIV in light of this and if so how do you think I should adjust things please?

Quote:
Originally Posted by 5kft View Post
I am in GMT-8, but we are currently in DST so we're currently GMT-7.
Thank you, I can now judge when to reply to your posts. Setting this in your location would be handy.

Quote:
Originally Posted by 5kft View Post
Anyway, just a thought
  1. I guess you are asking for something like <current> = ( <actual> + <offset one> ) * <scale> + <offset two> .
  2. Am I on the right lines, what about the the ambient base and would an integer scale be OK?
  3. Some users already find SIV "too complicated" and I don't feel I should add to this unless there is a real benefit.
Quote:
Originally Posted by 5kft View Post
Sure, I can do this as soon as I come up with something that isn't too silly
Before I start extending the [Synthetic Setup] facilities then I would like to see your current setup.


1 members found this post helpful.
  #1226  
Old 09-19-2017, 11:13 AM
5kft 5kft is offline
Registered User
5kft's PC Specs
 
Join Date: Sep 2017
Posts: 47
POST ID # = 920542
5kft Reputation: 17
Default

Quote:
Originally Posted by red-ray View Post
Do you think I should adjust SIV in light of this and if so how do you think I should adjust things please?
Well, you already provide a perfectly sufficient facility for working around this (-UPDATE), which I personally am happy with using. If you felt compelled to adjust SIV to address this internally, you could control the behavior via a Qualifier, and have the default workaround to be disabled?

However I really don't feel comfortable having you add odd hacks to SIV. This is a problem, but not a big one, and there is an existing workaround. And most importantly, fundamentally the characterization of the system behavior is still more or less theoretical at this point.

Quote:
  1. I guess you are asking for something like <current> = ( <actual> + <offset one> ) * <scale> + <offset two> .
  2. Am I on the right lines, what about the the ambient base and would an integer scale be OK?
  3. Some users already find SIV "too complicated" and I don't feel I should add to this unless there is a real benefit.
  1. Yes, that is correct
  2. I think that <ambient base> could affect <actual>. <scale> would have to support fractional amounts to be useful, IMHO. This could be implemented via fixed point if you don't want to introduce floats
  3. Yes, I totally agree. To be clear, I was just "thinking out loud" here. I'll play some more with my setup and see if something like this is really necessary . (I've really only spent probably five minutes playing with Synthetics so far, since I got the CLCP installed)

Ray, thanks again for all of your work on SIV, and the time that you've spent going back and forth with me. I really appreciate it!!


  #1227  
Old 09-20-2017, 07:18 PM
red-ray red-ray is offline
Banned
red-ray's PC Specs
 
Join Date: May 2014
Location: England (GMT+1)
Posts: 7,148
POST ID # = 920669
red-ray Reputation: 81
Idea SIV 5.24 Beta-03 with [System Timers]

Quote:
Originally Posted by 5kft View Post
Ray, thanks again for all of your work on SIV.
Thank you for all your feedback. How are you getting on setting up [Synthetic Setup]?

I have started adding a [System Timers] panel to SIV, but at the moment it only lists timers created with CreateWaitableTimer(). I would like to add the ones created by SetCoalescableTimer() + SetTimer(), but they don't seem to be in the SYSTEM_HANDLE_INFORMATION block returned by NtQuerySystemInformation( SystemHandleInformation, ... ). Do you have any idea how to enumerate them please?

Quote:
Attached Images
File Type: png [System Timers].png (43.7 KB, 377 views)


1 members found this post helpful.
  #1228  
Old 09-21-2017, 09:38 PM
5kft 5kft is offline
Registered User
5kft's PC Specs
 
Join Date: Sep 2017
Posts: 47
POST ID # = 920810
5kft Reputation: 17
Default

Quote:
Originally Posted by red-ray View Post
Thank you for all your feedback.
You are very welcome!

Quote:
How are you getting on setting up [Synthetic Setup]?
I haven't looked further...work has taken literally all my time this week. Fun times. But I will keep you posted! My big thing previously was being able to map case temperature from the CLCP to the GPU range, but I'm not sure that's actually even important or necessary. But it's fun to think about!

Quote:
I have started adding a [System Timers] panel to SIV... Do you have any idea how to enumerate them please?
It's actually been forever since I've done Windows development (feel free to ask Linux questions however ). But perhaps SetTimer is "different" as it sets up the "synthesized" WM_TIMER messages sent to all associated WndProcs...? Unfortunately I have no idea how to query a canonical list of these timers from the system. But perhaps you could write something that could catch all of these in the system, like various Windows "spy" programs support? But I'm not sure if that would be very useful in the end, though

Could you find anything from SysInternals, perhaps? I haven't looked.

Apologies...I wish I could be of more help...


  #1229  
Old 09-24-2017, 02:38 AM
boeing253 boeing253 is offline
Registered User
boeing253's PC Specs
 
Join Date: Nov 2015
Location: New Zealand
Posts: 109
POST ID # = 921012
boeing253 Reputation: 20
Default Hard Drive temprature "intermittently NA"

HI Red-ray

I have an issue with SIV 5.23 (was also present in 5.22) but it only started happening not long ago. From what I noticed it was after installing my new graphics card as it has a big cooler which sits overtop SATA ports on motherboard.

Basically it's just 1 SSD, my Samsung 840 EVO 250gb. On the LCD panel I have all my connected HDDs showing what temprature they are at. Intermittently the 840 EVO will turn "Red and temprature becomes NA".

After a few seconds it returns back to normal.

I tested my 2TB external HDD (connected via USB) and unplugged it and it displays the same as the SSD and it appears the SSD has been disconnected somehow but this SSD is my Windows drive and nothing happens to the computer when this happens on SIV.

This SSD is now almost 4 years old and all other connected drives are operating normally so I wonder if this is a sign the SSD is starting to fail.

The samsung software shows the SSD as being "Good health".

Red-Ray what screenshots do you need me to provide on this issue for investigation?

Many Thanks


  #1230  
Old 09-24-2017, 05:22 AM
red-ray red-ray is offline
Banned
red-ray's PC Specs
 
Join Date: May 2014
Location: England (GMT+1)
Posts: 7,148
POST ID # = 921016
red-ray Reputation: 81
Idea I suspect the SMART read is failing once in a while - Please try SIV 5.24 Beta-06

Quote:
Originally Posted by boeing253 View Post
it only started happening not long ago
Thank you for the report and I suspect the SMART read is failing once in a while. I am wondering as follows:
  1. Did this start about the same time as you updated to Windows 10?
  2. Please post [LCD Panel Setup] so I can see how things are setup. After checking "it goes red" when the temperature can't be read. For SIV 5.24 Beta-06 I changed this colour from red to the default font colour.
  3. To confirm it's a read error I need to see [SMART Drives] from when the drive is reporting NA. This will also tell me which driver is being used.
  4. How often does this happen and how long before it starts reporting a temperature again?
  5. By default the temperature get's updated every 10 seconds, so it's going to report NA for at least 10 seconds.
Quote:
I get a similar issue with an NVMe disk when I boot Windows 10, but not with Windows 7 and below you should see there are retries on W10, but not on W7 .

Please try SIV 5.24 Beta-06 or later and post [Disk Status] once SIV has been running for >= 30 minutes. Do you still see NA temperatures?

Attached Images
File Type: png RCK.png (142.9 KB, 535 views)

Last edited by red-ray; 09-24-2017 at 09:24 AM. Reason: SIV 5.24 Beta-06


Closed Thread

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes Rate This Thread
Rate This Thread:

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT -4. The time now is 03:40 AM.


Powered by vBulletin® Version 3.8.7
Copyright ©2000 - 2020, vBulletin Solutions, Inc.