Throttlestop Limits PCS LIMIT

[sk111]

Hi gang I was wondering if anyone can help with the below limits and any suggestions on what I should be doing

Post also here

Xeon Owners Club

Crazy voltage! I would have dropped it right off. 1.25 to start off. 4.6ghz really shouldn't need 1.55v. Yeah I noticed that too but that cpu he had is not listed anywhere it like a rabbit hole with that cpu he had heaps of issues with it barely getting to the operating system and what else is...

www.techpowerup.com

 

[unclewebb]

ThrottleStop Limit Reasons does not support your Xeon processor 100% correctly. I have never owned or used any similar Xeon processor for development purposes.

Is your CPU throttling during normal use or when stress testing? Post some screenshots of the FIVR and TPL windows so I can see your settings. Post another screenshot when stress testing and one of those boxes in Limit Reasons is red. Yellow boxes are not important if your CPU is not throttling. Red boxes show that throttling is in progress. That is important.

I need to see some screenshots so I can see what adjustments are unlocked and available.

The Intel definition is not too helpful.

Platform Configuration Services Status (R0) When set, frequency is reduced below the operating system request due to PCS limit

 

[sk111]

Thanks for taking the time. They are always in yellow never in red

Here they are at a stable configuration 43T (4.3Ghz) and they do not appear but if I push it to 45T (4.5 GHz) they do appear (in yellow) when I stress test or play a game. Temps are contained even with the standard cooler. Also with the 45T a game crushed but no issue at 43T.
Any ideas?
Below during the TS Bench

Just to mention this is the setup I am using and the below performance was during the 45T on

Please click the green human to continue - UserBenchmark

 

[unclewebb]

with the 45T a game crashed but no issue at 43T

I prefer using the Adaptive voltage setting with the voltage slider all the way to the left. It should say Default in the voltage box. This allows the voltage to automatically adjust to whatever speed you are running. If the default voltage curve is too high, you can use a negative offset voltage. For many Intel CPUs, you need to adjust both the core and the cache voltages together at the same time. A 45T multiplier might be near the limit of what an E5-1650 v3 can run at reliably. If you try to increase the voltage to increase stability, this will likely just trigger more throttling flags to start lighting up in Limit Reasons.

SpeedStep needs to be enabled if you want the Set Multiplier feature to work. As long as you are using the default Windows High Performance power plan, it is OK to have SpeedStep enabled. Old school users believe that SpeedStep needs to be disabled but it really doesn't.

I like it when UserBenchmark says that your computer is "Performing way above expectations". Lots of people hate UserBenchmark for a variety of reasons. The basic test score is only comparing your E5-1650 v3 to other user's E5-1650 v3 so that comparison is valid. The people you are comparing to are obviously not using ThrottleStop to overclock their CPUs to the moon like you are doing. Good work.

Can you maintain the full 44 or 45 multiplier while running Cinebench? ThrottleStop does a great job of accurately tracking any tiny change in the multiplier. If there is any throttling, it will show up as a reduced multiplier.

My 4th Gen mobile CPU is not 100% stable with the VCCIN set at 1.75V so I usually leave VCCIN at 1.80V. I have not tested a wide variety of CPUs to find out the best VCCIN setting. I know some users in some forums have come up with guidelines like VCCIN should be set 0.4V higher than the Vcore voltage or some other magic number like that. I prefer testing CPUs individually to find out what works best. Changing VCCIN can screw up the power consumption calculation within the CPU which can trigger throttling. Setting VCCIN low is not always best.

 

[sk111]

OK thanks I tried your method but couldn't get it right and was getting BSOD as soon as I was changing the voltage to Default.
With my method I had a stable version at 4.4GHz by lowering the voltage by 0.1V from the 4.3GHz however when I did run the Cinebench as per your suggestion it crashed midway in the multi-thread test.

Running the Cinebench at 4.3 GHz is very stable and no exceptions were given. Temps very contained (around 75C and max 78C the stock cooling is very effective even at 15% from the low) TDP max 137W (which is near the 140W manufacturers TDP). Results below performs really good given this is a budget system.

I may spend some time to figure out how to go to 4.4 GHZ stable by fine tuning further but given the tiny performance increase not sure it is worth it.

In terms of impact to the games from the original 3.5GHz version seems significant. In one game I tested it got 7% more average fps and 25% higher value for the minimum 1% FPS.

Somewhere I heard increasing the voltages of the cache will have a performance impact too but not sure as this is set already to the Default voltage and inherits the same VCCIN

The SpeedStep ticked or unticked didn't make any difference by the way.

I have done 1 more attempt with the following setup where the multiplier is 43T and SpeedStep ticked.

And then I set the ratios in FVR at 44 (so higher over the multiplier).

For some reason this seemed to work better giving max power 131.6 Watts (less than before) and at the same time the Cinebench performance improved a little vs previous run (7356 vs 7188).
This could be random but not sure. No exceptions reported either.

 

[sk111]

One question that I thought that may be beneficial to some games if 4 cores are operating at higher frequency say @ 48 (or above) and the rest 2 a@ 30 (or disabled) would it work and how can it be set in Throttlestop?

 

[unclewebb]

if 4 cores are operating at higher frequency say @ 48 (or above) and the rest 2 a@ 30

Most Intel CPUs have all of the active cores tied together using the same multiplier. You can have active cores and inactive cores but you cannot have two cores within the same CPU package actively running at two different speeds.

If you have the core C states enabled in the BIOS, when an idle core with nothing to do enters C3, C6 or C7, it becomes disconnected from the internal clock and it is disconnected from the voltage rail. That means it is sitting dormant at 0 MHz and 0 volts. That is as good as it gets. There is no way to improve upon that.

As long as the C states are enabled, you do not have to worry about micro managing the core speed. Intel CPUs do a great job of looking after themselves.

 

[sk111]

OK thanks I also managed to disable a couples of cores in the bios and reached 4.7GHz but not under much stress. I will experiment more with that.

Using now all cores and following your method of leaving the VCCIN to "Default" and playing with the offset I got it to run reliably at 4.4 GHz the full cycle of cinebench with new high score 7499 (from previous 7356)

Offset settings are as follows:

I also by pushing the offset voltage to 150 it would run the 4.5 GHz but it was giving me flashing red in Power Budge and EDP Current and Max Turbo plus temperatures were getting towards 90C and Power reached 150Watt (above the 140 manufacturers limit).

Overall though at 4.4 is very stable now with max temp at 80C

 

[unclewebb]

Power Budge

That is actually supposed to say Power Budget. The big wide font is hiding the last letter. No worries that there might be a budgie bird inside your computer.

I have only seen Xeon CPUs with a Power Budget limitation. Not sure how to fix or bypass this limit.

 

[sk111]

Hey thanks... haha yes the likelihood of a bird nesting there would be small unless it was fried
The only solution I think it is to make it stable with less current offset (hence less power consumption).
I will give it a go and report back if it is successful though probably would approach the limit of my cooling setup.