Question 5950x behaviour clarification

[stanzlavos]

Hi All

Configuration:

Asus ROG Crosshair VIII Dark Hero
5950x
G.Skill 2x16GB 3600MHz CL16 Trident Z Neo
WD SN850 1 TB
Zotax GTX1080 Amp Extreme
Corsair H115i Platinum RGB
Corsair RM1000i

I am using DOCP and thats about the only change in BIOS. No PBO.

Cinebench R23 Scores : MC -> 25.1-25.3K; SC -> 1.6K

During SC test, I can see that the active core reaches "Peak Speeds" of around 4.9 GHz (or a little under) and the temperature is around 67-70C (in Ryzen Master).
During MC test, I can see that the cores reach "Peak Speeds" of around 3.8 GHz and the temperature is around 62-63C.

Question: Does the numbers look OK ?

 

[Justinus]

70K? Daaaaaaaaaaaaaaamn what kind of crazy exotic cooling you got going there?

Assuming you meant celsius and not kelvin, those temps, frequencies, and performance numbers are pretty much expected. I have a custom water loop and fully stock CPU(memory and FCLK overclock only) I see around 26k multicore and 1,650 single core, but only on a super clean fresh windows install. Add some background processes and stuff and it can drop a bit.

 

[stanzlavos]

70K? Daaaaaaaaaaaaaaamn what kind of crazy exotic cooling you got going there?

Assuming you meant celsius and not kelvin, those temps, frequencies, and performance numbers are pretty much expected. I have a custom water loop and fully stock CPU(memory and FCLK overclock only) I see around 26k multicore and 1,650 single core, but only on a super clean fresh windows install. Add some background processes and stuff and it can drop a bit.

What Kelvin? What are you talkin about?

Thanks for the confirmation.

 

[Topweasel]

Seems normal. Most of the time the temp reported is mostly based around the hottest area of a CPU. So basically on SC thrashing you are seeing that single core getting most of the CPU package power feeding directly into it. Its actually easy on a cooler to cool a CPU when the power usage is being spread around the CPU so even if its using more power, more of the CPU is getting warmed, and no single core is getting fed as much power as the Single core in the Single Thread test when doing the Multithreaded test.

 

[Puffnstuff]

On my system with PBO enabled my 5950X was running around 160°F under load and with it disabled it idles around 105°F and peaks around 140°F under load. I've noticed that during gaming I see around 120°F peak temps so I'm happy.

I run an EVGA AIO 360 with CM ARGB fans and Arctic MX-4 paste that is not applied like frosting inside of a CM HAF 932 case. I don't think that I'm ever going to upgrade from it unless another high air flow tower appears on the market that is just too good to resist. Thus far I've used it the last 12 years and am satisfied with the thermals it produces having positive air pressure and all. I did swap out the 230mm side fan with 4 120mm CM argb fans which also doubled my cfm from the side.

I will say that with PBO enabled I was getting 200 fps in BF3 with settings maxed out at 3440x1440 freesync enabled as my monitor can scale beyond 144hz.

 

[stanzlavos]

I am using the stock thermal compound that came with the cooler. If I move to a different thermal compound (had my eyes on some Thermal Grizzly Kryonaut), will there be a substantial difference in the temperatures?

I am not planning to overclock immediately. In the near future, maybe I'd just enable PBO and play around with it a little. If so, should I re-paste? I was actually thinking of moving to a different thermal paste only if I decide to do some manual overclocking.

 

[Puffnstuff]

Yes replace the stock paste and remember that you are not trying to frost a birthday cake rather just filling in some microscopic surface imperfections. Once you replace it usually takes a little time to cure which various between compounds and altitudes.

PBO is super simple to use but once I did it my fans maxed out as the heat rose so it was a hot and noisy proposition. Leaving it off still yields great results and my game play is buttery smooth as my 5950x and rx 6900 xt black work very well together.

 

[stanzlavos]

Yes replace the stock paste and remember that you are not trying to frost a birthday cake rather just filling in some microscopic surface imperfections. Once you replace it usually takes a little time to cure which various between compounds and altitudes.

PBO is super simple to use but once I did it my fans maxed out as the heat rose so it was a hot and noisy proposition. Leaving it off still yields great results and my game play is buttery smooth as my 5950x and rx 6900 xt black work very well together.

How much of a temperature difference is expected if I re-paste?

 

[Puffnstuff]

It varies but if air flow was good beforehand and the original paste was a limiting factor then 10°F or more would not be unexpected.

 

[Golgatha]

Yes replace the stock paste and remember that you are not trying to frost a birthday cake rather just filling in some microscopic surface imperfections. Once you replace it usually takes a little time to cure which various between compounds and altitudes.

PBO is super simple to use but once I did it my fans maxed out as the heat rose so it was a hot and noisy proposition. Leaving it off still yields great results and my game play is buttery smooth as my 5950x and rx 6900 xt black work very well together.

For my setup, I found PBO to perform slightly (and I mean ever so slightly) worse than just straight up disabling it. I think the extra voltage causes more heat and thus more throttling. I'm using a Scythe Fuma 2 air cooler, so a AIO might handle the heat better. Either way, the extra wattage necessary doesn't really gain you anything in games anyway, as those are normally GPU limited.

 

[Justinus]

For my setup, I found PBO to perform slightly (and I mean ever so slightly) worse than just straight up disabling it. I think the extra voltage causes more heat and thus more throttling. I'm using a Scythe Fuma 2 air cooler, so a AIO might handle the heat better. Either way, the extra wattage necessary doesn't really gain you anything in games anyway, as those are normally GPU limited.

Even with a high end (Optimus) waterblock I have had absolutely zero success getting PBO to increase single thread or lightly threaded performance. I've got a Dark Hero so for straight multithreading loads I've dialed in a static all core overclock, but I still leave the normal settings fully stock because it never works out to increase performance.

 

[Golgatha]

Even with a high end (Optimus) waterblock I have had absolutely zero success getting PBO to increase single thread or lightly threaded performance. I've got a Dark Hero so for straight multithreading loads I've dialed in a static all core overclock, but I still leave the normal settings fully stock because it never works out to increase performance.

Same experience really. I've done all manner of tweaking. On my systems I enable DOCP and disable PBO, and this seems to get the best results every time.

 

[stanzlavos]

Thanks for all the inputs.

Once I run out of laziness, I'll change the thermal compound I guess.