Heavy CPU throttle.

[plumel]

Hi guys,

Just went on userbenchmark to check the general state of my PC, and have found that I need to have a very minimal amount of processes open to get a decent score for my GTX 1060.

If I open OBS, or have Windows Desktop Management running, it puts me to the 5th percentile for my GPU. If I don't have those things running, and I just have discord, chrome and steam running, I get put in the 74th percentile.

Just wondering if there is anything I can do about this.
Also a quick side question - Is it bad to have 1 monitor plugged into my GPU and the other plugged into my CPU?

Many thanks,
Daniel

 

[13thmonkey]

Your monitors are plugged in fine.
What temps is your CPU running at?

 

[plumel]

Your monitors are plugged in fine.
What temps is your CPU running at?

Normally the temps are around 60-70 degrees. low 80s at a push. This is under load

 

[13thmonkey]

and what CPU is it? and what clock speeds are you seeing. (no idea why I didn't ask that earlier).

 

[plumel]

and what CPU is it? and what clock speeds are you seeing. (no idea why I didn't ask that earlier).

Sorry for the late reply. It's an i5-3570k clock speed is 4.2GHz. 1.6GHz when idle

 

[13thmonkey]

what is the actual clock speed when under load? It doesn't seem to be a throttling issue if you are at 4.2Ghz underload.

There's something else going on, software running in the background, malware, miners etc.

 

[plumel]

what is the actual clock speed when under load? It doesn't seem to be a throttling issue if you are at 4.2Ghz underload.

There's something else going on, software running in the background, malware, miners etc.

under load the clock speed is 4199ghz. just checked.

I have no miners, I'll run a quick virus scan with anti malware bytes but im confident it wont show anything.

EDIT: To confirm - No viruses.

 

[13thmonkey]

and task manager recognises 4 cores?

 

[plumel]

and task manager recognises 4 cores?

Yep, seems to.

 

[13thmonkey]

right then, two routes of action at this time.

1. figure out how bad this is, and then see whether it is important, the 5th percentile might only be marginally the average, or could be skewed large numbers being OC'd.
or
2. assume that some piece of software is taking more than it should do, and figure out which it is, by disabling one at a time, and using task manager to assess load. Ideally benchmarking should be done with a lightly loaded system anyway.

 

[plumel]

right then, two routes of action at this time.

1. figure out how bad this is, and then see whether it is important, the 5th percentile might only be marginally the average, or could be skewed large numbers being OC'd.
or
2. assume that some piece of software is taking more than it should do, and figure out which it is, by disabling one at a time, and using task manager to assess load. Ideally benchmarking should be done with a lightly loaded system anyway.

You're right. I'll try take some other benchmarks and see where my 1060 is coming out to with and without the software running.

I'll report back once I've taken some!

Edit: Scores on Unigine Heaven Benchmark 4.0 are 1621 with no processes and 1597 with some processes (chrome, spotify, steam) running.

Does this seem reasonable? I also noticed the GPU power wasn't going above 70% most of the time. CPU usage was around 20-50%.

 

[plumel]

right then, two routes of action at this time.

1. figure out how bad this is, and then see whether it is important, the 5th percentile might only be marginally the average, or could be skewed large numbers being OC'd.
or
2. assume that some piece of software is taking more than it should do, and figure out which it is, by disabling one at a time, and using task manager to assess load. Ideally benchmarking should be done with a lightly loaded system anyway.

My score in 3DMark Fire Strike, if this helps clear anything up: https://www.3dmark.com/3dm/22502803

I got this one earlier though, any potential reason for the inconsistency? https://www.3dmark.com/3dm/22501797

 

[13thmonkey]

I simply don't know, the core clock on the lower one is lower, but it's at an idle state (or one state above), so that could be based on the precise point it was read at, if it were truly that low, it'd be much worse as a score.

comparing it to the rig below, 1060 vs 970, is about evens, the 1060 being slightly better perhaps. The i5 vs the i5 they are within 5% of each other, and the scores are similar. The 7% swing is larger than I expected, but no idea what causes it.

All in all, it's OK, probably just a wayward process. If you happen to have a spare disk lying around you could try a bare windows + drivers + updates + firestrike install and see how it performs then? That would pin it down to hardware or not hardware?

 

[plumel]

I simply don't know, the core clock on the lower one is lower, but it's at an idle state (or one state above), so that could be based on the precise point it was read at, if it were truly that low, it'd be much worse as a score.

comparing it to the rig below, 1060 vs 970, is about evens, the 1060 being slightly better perhaps. The i5 vs the i5 they are within 5% of each other, and the scores are similar. The 7% swing is larger than I expected, but no idea what causes it.

All in all, it's OK, probably just a wayward process. If you happen to have a spare disk lying around you could try a bare windows + drivers + updates + firestrike install and see how it performs then? That would pin it down to hardware or not hardware?

Thanks for the help. I'll try with a fresh windows some time in the near future.