CPU preforming well under expectations, possibly CPU throttling.

Jake_-_

Prominent
Apr 24, 2017
14
0
510
So earlier this week I upgraded my AMD athlon 860k to and Intel i5 6600k. The main reason I did this was to have better performance but it was very underwhelming. My main game for testing the CPU was h1z1, which I know can be unstable but is the most CPU intensive game in my library.

WIth my AMD processor I was getting terrible performance, as expected, with ~40 fps at the lowest settings.
With the 6600k, I was greatly underwhelmed because I wasn't even getting 60 fps at high settings. I was getting an average of ~40 fps while max temp of my cpu was 46°C. Now this may seem unrealistic to get 60fps at high settings but one of my friends has the same processor and GPU (GTX 1060) as me and he is able to pull a stable 60 frames at Ultra settings.

I made a few posts to reddit but that was no help but one person said that it could be that my CPU was throttling, but I think that the max temp of 46 °C proved that wrong.
After this i decided to run a benchmark test and was very concerned.
UserBenchmarks: Game 63%, Desk 64%, Work 40%
CPU: Intel Core i5-6600K - 77.4%
GPU: Nvidia GTX 1060-6GB - 66.7%
SSD: OCZ ARC 100 240GB - 66.7%
HDD: Toshiba MK1254GSYF 120GB - 28.9%
RAM: Crucial BLS4G4D240FSE.8FBD 2x4GB - 70.1%
MBD: MSI B150A GAMING PRO (MS-7978)

Before people go bashing on that toshiba hdd, it was from a laptop and just stores pictures and videos.
Now I have no idea why my entire PC is preforming well below expectations and I was hoping you guys could help me out.
Thanks, Jake
 
Most people reporting to UserBenchmark (good site) with i5-6600K will be Overclocked. If you are stock clock its no surprise that you are 77.4% when they average 87.6%. Compare your stock clock to an i5-6600 no K (and no OC). It's comparable.

Your Nvidia GTX 1060-6GB hits 66.7% compared to an all card average of 71.8%. Is your card at the high end of the factory OC pool, or the low end ?

Your disk at 28.9% is close to the average rating of 29.4% for that type of disk.

Do:

1. verify that your CPU boosts to the correct frequency when running. It should be between 3.5 and 3.9 ghz at stock under load. CPU-Z or HWMONITOR will report frequency. If the frequency is low then we can look for why (e.g. windows power management settings or temp throttling or ...). From your score you will discover you are fine.
2. Reset your BIOS to defaults. SAVE. see if there is a change. This will fix any really bad settings for memory timing, PCI bus, etc. But from your scores you are already fine.

You have a nice config. The percents reported by Userbenchmark are not percents of what you should have with your hardware, they are percents of the best possible part.
 

jr9

Estimable
1. Try rerunning the benchmark with the storage HDD disconnected
2. If you didn't clean install then do that ASAP
3. Update BIOS and chipset drivers

I don't think it's thermal throttling at all, your temps are fine. GPU looks fine too. Seems like a software or OS issue.
 


Hi, what is the issue you are solving.? His CPU and GPU are performing where they should be according to userbenchmarks....

 

USAFRet

Titan
Moderator
userbenchmark is a particularly useless tool.

The way it presents the results has you thinking your PC sucks.
It does not.

Like this:
For a particular configuration, let's assume 100 is the absolute best.
Of all the PC's like that tested, they all run between 90 and 95. All very good, right?
But yours is at 91. Still very good, right?
userbenchmark will display that as "OMG! Way below expectations! Below 20%"
In reality, it is just looking at the difference between 90 and 95. Not 0 and 100.
But saying that your PC is running at 91% effectiveness is boring.

From those numbers, your system is running right at average, or a little above.
 

jr9

Estimable


With their data for example the Core i5-6600K there are 95,000 benchmark samples and the overall rating is 89.1% with 100 being the best or whatever arbitrary point 100% is. That number alone could be considered open to interpretation except if your processor scores 77.4% compared to 95,000 samples of the Core i5-6600K scoring an 89.1% average is it not safe to conclude your CPU is performing below average?

 

USAFRet

Titan
Moderator


"Performing way below expectations (16th percentile" is one thing
77.4% Very good is a different thing.

2 different 'numbers', for the same CPU in the same test.

If we assume 50% is 'average', his is at 77.4%. That would be "very good"
But of all the equivalent CPU's, his is in the 16th percentile.
By definition, not everyone can be "above average".


So which number do you believe?
"77.4%", or "16th percentile".

Hence, the utter unusability of "userbenchmarks.com".

If the "average" is 89%, 77.4% would not be, to my mind, "way below expectations".
And the single core of 83%, and the quad core of 92%....just leads credence to "it's working just fine".


For instance, this is my CPU, tested just now:
yQCqV24.png


What is the 'good number' here?
12th percentile?
Or 92%, 87%, 90%? Or "79% - Very good".

Useless and confusing.
 

jr9

Estimable
The percentile rating is derived directly from the percentage score directly using the same data against other 6600k chips not all chips I believe. If percentile is derived from vs all chips then that's another reason I never use the percentile graphs. It is entirely possible to be in the 16th percentile for your particular processor and also be 77.4 percent overall with 110% being a $1000 core i9. I'm looking more at the recorded percent vs the 95000 over benchmarks of that chip vs the chip in question.

If 50% as a benchmark score is the average relative to all processors tested, but the 6600k scores 50% that is not good because it's suppose to score closer to 90%. The 30% is terrible as a percentage but compared to the 4 benchmarks of the same HDD that 30% is actually normal as the average is about 30%.

I tend not to use the core ratings for CPUs on usermark because Intel and AMD processor cores don't run as fast as each-other so I'd expect higher core performance from an older Intel chip but overall Bench rating (the useful number) puts it lower as it's competing against chips like Ryzen with mores cores that are slower.

The percentile thing can indeed be misinterpreted and the graphics definitely give the appearance something is seriously wrong which is why I always look at bench % recorded vs bench % expected only. There's also the fact that you are competing against overclocked chips which skews the percentiles further but bench % is the main number and the percentiles are just a derivative of it.
 

USAFRet

Titan
Moderator


Exactly.
But far too many do.

Basically:
"All of the the equivalent CPU's are blindingly fast. Your just happens to be on the lower end of "blindingly fast" "
PANICPANICPANIC.