5950x underperforming, I believe

[Smettuh]

Hi all,

I don't expect to match the top results but I can't help myself and be bothered with my Cinebench R20 results and the core clocks I'm getting. They are too far off I believe, basically I am getting a MT score of a 3950x.

Clean install, all latest drivers installed.

My system specs:

5950x
ROG Strix X570-E Gaming
2 x 16 GB CL16 16-16-36
Arctic 360mm AIO cooler (reapplied this one to be sure)
980 Pro SSD
Slowish GPU

Cinebench R20 MT results: 9250 - 9350
All core clock: 3675 - 3700 mhz
Core voltages: 0.994V to 1.000V
Temps: 60-65 degrees (standard fan curves)
PPT 89% of 142W
CPU Power 94W

Also my single core clocks never reach the advertised 4900mhz. 4825mhz is the max during a long single core Cinebench run.

Don't want to use PBO or OC, I want out of the box performance basically.

Tried also with max cooling, temps remain below 60 degrees then. Clocks 25mhz higher (same for single core), Cinebench score more consistently hovering 9350.

Thanks for helping.

Some HWINFO Screenshots while running cinebench R20imgur.com" />

 

[Smettuh]

Power supply is the AX760i (platinum ce)
Power draw during the Cinebench R20 MT run below, quite some headroom left.
I run a GTX970, ordered an Asus RTX3080 on day one, still nothing .

I'm running the latest BIOS version on the ASUS website:
Version 3001
2020/12/08 19.97 MBytes
ROG STRIX X570-E GAMING BIOS 3001
"1.New CPU support
2.Add a ReSize BAR shortcut for AMD SAM support

Installed the latest chipset drivers: amd_chipset_software_2.10.13.408

Power plan is set to Balanced, but switching to high performance doesn't change anything.

Ryzen Master is set to the default control mode, no auto oc, pbo or eco mode on one of the profiles.

Some BIOS settings, only DOCP is loaded, rest is default.
Idle temps are high, but ok based on the voltage I'd say.
My case is the Meshify S2 with plenty of airflow (all Noctua case fans)

 

[Smettuh]

Odd. Your cpu core voltage and VID don't really move from 1v yet you're drawing a massive 120w+ avg for package. Your minimum clocks should be dropping below 3.7 as well.

Add to that your TDC/EDC are fully loaded and you've got really high memory usage.

Below shot is a 5950x I put in yesterday after r20 single and multi runs for comparison on power / current draw.

I don't know what the root cause is but I'd be tempted to do a clean install of Windows. What chip did you run before on this win install?

Keep in mind that I started recording hwinfo at the start of a MT Cinebench R20 run and I took a screenshot during the run. So that would explain the package power values that otherwise match your maximum I think?

This is already a fresh install btw, I had a 3700x before with a low end motherboard and RAM but selling them.
I did run cinebench early without many services installed and my score was a bit higher but not close to my expectations basically.
PBO results: 10800 +-
PBO results seems to be the result that I should get without PBO and with decent power and temperatures.

So rather high temperatures, hereby the HWinfo results from start to end (MT run 2 min) below.
Not sure if my AIO cooling is adequate. Should be one of the better ones for Ryzen. Did use Arctic freezer cooling compound from a few years ago (approx 4 years) but looked perfect.
Did already try to reapply.

 

[Smettuh]

This is with PBO enabled during an MT Cinebench run, I don't want them to downclock.
I think these clockspeeds are low for PBO enabled.
Seems to be related to my temperatures, who knows even my bad results without PBO are.
I will take pictures when I remove my Arctic Freezer 360 so you can determine if it was correctly mounted perhaps.

 

[Smettuh]

Reapplied the arctic freezer for a 3rd time, damn that cooler is a pain to install solo.
I immediately heard the difference, my fans are basically silent now while doing some casual work.

Results are still on the low end but this is with some minor services running.
Cinebench R20 MT 9550. Still low voltage (not more than 1.000V) but temps are hovering 55 degrees now.
PBO new score is 11020: Is this acceptable or also on the low end? All core clock was 4.4ghz with temps hovering 80-85 degrees.

I haven't checked the Ryzen DRM calculator, will try now.

 

[Smettuh]

Checked a few PBO results from reviewers etc and my PBO score seems to be better in line with the majority which makes me think it's BIOS related.

Temps indeed not very good when using PBO, could be a few reasons.
- Thanks to GamersNexus I'm running my radiator in Pull on top (should give better cpu temps then push on top) and ALL case fans exhausting.
Radiator on top because I can't install it 'the right' way in the front. See screenshot below.

- I probably do not have a great chip

- Case filters not removed

 

[Smettuh]

That's how it is usually done yes. However ....

After all the hype around incorrect radiator placement versus the pump (https://www.youtube.com/watch?v=BbGomv195sk&t=860s&ab_channel=GamersNexus) it was stated that if you cannot mount your radiator in the front with the tubes down that mounting it on top is better. In addition it's more optimal in that scenario for a pull config (just like pull in the front is better for your cpu temps).
You pull ambient temp air into your rad, not temps of your case like you would in push. The fact that hot air rises is basically irrelevant with case fans, it's a minimal difference.

I believe Steve in this regard, I could do a comparison this week .
The actual statement regarding my setup: https://youtu.be/tU7D6y_QYcI?t=1776

 

[Smettuh]

Yes had the cooler already and was initially attached to my 3700x.
Just using that cooler on my 3700x increased my Cinebench score, had a Noctua DH15 before.

For science, changed to a more common setup.
Front mounted rad intake, exhaust top/back.
Temps are basically the same as I would expect.

I applied Arctic MX-4 in a cross method, probably applied a bit much.
Could indeed try with new paste, Cryo is hard to get in Belgium I see. Would that matter a lot compared to (new) MX4 or 5?
Or do you mean this:
Thermal Grizzly Kryonaut 1 gr

I see a 4.5 degree difference between my CCX in a 4.375/4.4 all core clock. Not sure if that is ok.

Quite positive now that if i can get my temps down that my PBO score will be as expected.

 

[Smettuh]

Applied the Kryonaut, bit difficult to spread with the spatula, wasn't very sticky on the die. Maybe the surface purifier didn't dry enough ...
But eventually nicely spread I'd say.

PBO score increased with about 120 points to 11143. Temps are lower, 78-82 after a 5 min run.
Still toasty but it's running quicker than before.

I definitely notice my temps are lower when running stock settings without PBO.
I'm not going above 50 degrees and my score increased eventually from 9250 to 9850 (with some services running).

I guess it was all temperature related?!

I can live with these scores, thanks all .