Overclock.net banner

261 - 280 of 292 Posts

·
Registered
Joined
·
261 Posts
Might seem like an obvious question, but what do you guys use to stress the CPU temperature wise?
I tried blender (Classroom/BMW), prime 95 and AIDA and none of those go above 55 degrees, but I know it can get hotter since those all only push up less than 4600.
During 3dmark it went into the 70ties, but there isn't there something better to test that, something that loops and reliably heats it up? Heaven benchmark also left it around the 55 ballpark.
Cinebench R20 or R23 have been able to push my temperatures up.
 

·
Registered
Joined
·
127 Posts
Ah right, forgot to mention cinebench, same thing there in the 50ies, but I only tried R20, let me check R23.

--update
nope same thing. Ran a few minutes of R23, same behavior- then for comparison, just launching 3Dmark made one CCD go up to 70 degrees.
 

·
Robotic Chemist
Joined
·
3,592 Posts
nope same thing. Ran a few minutes of R23, same behavior- then for comparison, just launching 3Dmark made one CCD go up to 70 degrees.
That is very odd, Cinebench give me higher peak temps on the CCDs than 3Dmark does, r20 seems slightly higher than r23. :unsure:
 

·
Registered
Joined
·
202 Posts
Ah right, forgot to mention cinebench, same thing there in the 50ies, but I only tried R20, let me check R23.

--update
nope same thing. Ran a few minutes of R23, same behavior- then for comparison, just launching 3Dmark made one CCD go up to 70 degrees.
Whats the voltage and power usage during koad?
 

·
Registered
Joined
·
261 Posts
Ah right, forgot to mention cinebench, same thing there in the 50ies, but I only tried R20, let me check R23.

--update
nope same thing. Ran a few minutes of R23, same behavior- then for comparison, just launching 3Dmark made one CCD go up to 70 degrees.
What cooler are you using again? Maybe it is taking all the heat away.
 

·
Registered
Joined
·
127 Posts
Temps after a run of R23:
2489152


temps after opening 3d mark immediately afterwards:
2489153


after running timespy:
2489154


Timespy:
2489155

during R23:
2489156

Cooled by a custom water loop. (aquacomputer cuplex kroys next vario)
This block allows you to do some adjustments after mounting, to better adapt to your particular IHS.
So on the one hand I was looking for a stable sustained load in order to configure that.
They've measured up to 2 degrees improvement but I have no idea how to do it. Any load I put on it just naturally fluctuates enough to where I had no idea if I was improving or worsening temps.
I legitimately can't tell if it's working, but it makes sense on paper at least.

The other thing I wanted to do was simply get it hot to gauge the mount/loop is working fine. I am still working on getting the flow up, it's not where I want it to be atm.
But I can't tell what the max temperature is if I can't reliably trigger it, well apart from that peak when launching 3dmark, but that's not reliable, it doesn't always go that high.
Temps are overall fine I would say, but since I would do mostly games, it would be nice to push single threaded loads which seem to get hotter.
I have yet to play with CO, I would like to get a baseline first.

That said, I am a little worried at how much hotter CCD1 gets, 6+ degrees seems a lot.
 

·
Registered
Joined
·
202 Posts
CPU PPT maximum is only 140W and voltage 1.17v under load which is why temperatures are low and the CPU frequency is limited. With enhanced power target CB can go over 250W @ 1.3v @ 4600+Mhz. Then u will also see higher temps, probably around 80C with your cooling.

Otherwise u might see short temp spikes from 1.5v and 5+ Ghz in single core boosts.
 

·
Registered
Joined
·
436 Posts
Ah right, forgot to mention cinebench, same thing there in the 50ies, but I only tried R20, let me check R23.
You're powah and current restricted atm, and so, wasting time worrying. Just give it more (set PBO limits to enable, or motherboard, or set them manually to something like 200/150/200) and give it a try.
 

·
Registered
Joined
·
127 Posts
Sorry if that wasn't clear, I wasn't trying yet to overclock, just seeing how the base behaves.
That said (already posted in the ryzen stability thread) but I'll post it here too, one thing that is not clear to me is the pbo limits. I have seen gamernexus set these to motherboard limits, which are way beyond what the FUSE limits are. Assuming Fuse limits are indeed a thing, then why are people talking about different settings for TDP,PPT,EDC? You could set set everything to max and the chip would still not be at risk? Might thermal throttle at worst, but then you could lower TDP until it stops, and that would still not risk damaging the CPU
 

·
Cheesebumps!!
Joined
·
2,420 Posts
@GribblyStick when I had my 5900x (a rubbish one) your current temps are just my temps when I am using a 200w PPT limit for PBO..that is already boosting to 4.6ghz all core sustained load and a 5.05ghz Single core load..if you want to see how effective your cooling is, better do a 200W TDP limit via PBO on that chip and see how it cools it down, you don't have to worry with any risks since these chips will throttle the performance down when you are already getting near the Thermal limit set (92c) or you can manually set a lower limit than what is advertised in the same PBO menu..
 

·
Registered
Joined
·
1,477 Posts
Is it normal for CCX quality to differ so much on a single CPU (5900X)? The best core of CCX 2 is worse than the worst core of CCX 1. Is this lottery or done on purpose by AMD?

2489495
 

·
Registered
Joined
·
199 Posts
Is it normal for CCX quality to differ so much on a single CPU (5900X)? The best core of CCX 2 is worse than the worst core of CCX 1. Is this lottery or done on purpose by AMD?

View attachment 2489495
It's pretty common, that is more drastic than most from what I have seen though. As long as one CCD hits the top turbo numbers, then it's within AMD's spec.

You can use affinity codes to manually assign cores on launch for games and such.
 

·
Registered
Joined
·
1,477 Posts
Assigning affinities is not necessary from what I have seen. Windows 10 (Thread Scheduler) does this by itself when CPPC is active in BIOS, putting high load on the strongest cores and low background load on the weakest (1909 and 20H2 tested).
 

·
Registered
Joined
·
199 Posts
Assigning affinities is not necessary from what I have seen. Windows 10 (Thread Scheduler) does this by itself when CPPC is active in BIOS, putting high load on the strongest cores and low background load on the weakest (1909 and 20H2 tested).
That is true if you're not a heavy user. If you're encoding a HEVC movie and want an emulator to use your best cores for instance, you have to manually set it.
 

·
Registered
Joined
·
1,477 Posts
If you are running both at the same time then maybe. Windows will still put the most demanding threads on the strongest cores. You might be better off setting process priorities instead of affinities then.
 

·
Robotic Chemist
Joined
·
3,592 Posts
This is definitely intentional, the three X950X CPUs I have had all had much better cores on CCD0 than CCD1. It also makes a lot of sense given how the products work at stock.

Loads with a low thread count can boost higher on the higher quality cores but if all cores are loaded it would melt if run that fast (frequency is the biggest factor in power/heat), so it couldn't run the second CCD faster even if it was higher quality.
 

·
Registered
Joined
·
199 Posts
If you are running both at the same time then maybe. Windows will still put the most demanding threads on the strongest cores. You might be better off setting process priorities instead of affinities then.
An emulator like Cemu for instance wouldn't be as heavy a load per core as an HEVC thread though. Nor would a game.
 

·
Registered
Joined
·
1,477 Posts
But every time the Cemu or game threads max out their core Windows will automatically schedule them to a "good" core. By using fixed affinities you are then keeping other high load threads off the good cores when they would otherwise be free for use. I think that CPU priorities would serve you better then.
 

·
Registered
Joined
·
1,477 Posts
After lots of fiddling, testing, analyzing, measuring and wrapping my head around the whole thing I think that I now understand the mechanics and interoperability of PBO, Curve Optimizer, Vcore (offset), (maximum) frequency limit vs. single-core + multi-core performance.

The only thing that still confuse me are:

- The EDC (215 A) limit permanently hitting 100% during sustained load that is not even close to the TDC limit (210 A) or measured current (130 A).

- Some power read-outs of HWinfo that seem like either HWinfo errors of sensors reporting wrong values, specifically "Package Power" vs. "Core+SoC Power" (20 watts lower than Package Power) vs. PIn/POut (lower than Package Power).

Overall I am not convinced that messing with all these settings is worth over just enabling PBO in BIOS with either the AMD or mainboard preset. Even with CO offsets of -30 and a slight voltage bump (AMD overclock preset) the CB20 sustained multi-core increases only by 3.5 - 5% (average 4.65 - 4.7 Ghz vs. 4.5 Gz). And then you have to do stability tests.

Sustained single-core improvements (of maybe 2-3%) are even harder to achieve, because you have to use negative CO on the "best" cores, which are likely already running closer to the edge to begin with. This is where silicon lottery comes in again, which is what overclocking mostly is about anyway.
 
261 - 280 of 292 Posts
Top