[SOLVED] Ryzen 5 3600 boost issue

[aahmed12]

My ryzen 5 3600 won't boost past 3.9 GHz. When running games or performing a render test in cinebench r20, the cpu doesn't reach 3900Mhz most of the time, according to ryzen master. The cpu, is completely stock. No overclock, has been applied and I have CPB and PBO set to auto, in bios. I'm thinking this could be due to high temps, because towards the end of the cinebench R20 multicore test, my cpu reaches 92C, on average. My idle temps are usually between 40C and 50C, sometimes spiking up to the mid-high 50s.
My specs:
Motherboard: Msi b450 tomahawk
CPU: Ryzen 5 3600
Do I need to change cooler? Can somebody help me out on this?

 

[Barty1884]

All-Core boost is 3.6Ghz on a 3600..... 4.2GHz single-core, and I think 3.9GHz for a couple of cores.

With "high performance" enabled in Windows, you'd see 3.9-4.2GHz under good conditions, potentially a little higher due to the boost algorithm.

The biggest driving factor though, is temperatures. If you're seeing 92'C under stock operations, you're not going to benefit from any further performance, as the boost won't happen, and you may well at/approaching throttling territory (don't recall the specific number for Ryzen).

Your idle temps seem high. While you could run out & change the cooler, it's important to understand your specific situation first.

What are ambient temperatures in the room? Most of the figures you'll find to compare to are based on a typical ~22'C "room temperature".

What case are you using? How is airflow in the case? What is your fan configuration?

 

[TheSTRAFF]

Boost doesn't work that way.
Your base clock is what your CPU should be doing at 100% load with all cores working.
so running on 1 core at 30-99% should give you a boost of 4.2GHz.
the more cores being use will lower the boost clock. so all cores at 100% at 3.9GHz with a base of 3.6GHz seems right.

i have just ran R20 test on my machine -R7 2700 (AMD Cooler - everything at stock)
at idle temp : 30C
during test : 51C
base clock : 3.2GHz (Max Boost 4.1GHz)
during test : 3.4GHz

90C is high IMO

 

[drea.drechsler]

All-Core boost is 3.6Ghz on a 3600..... 4.2GHz single-core, and I think 3.9GHz for a couple of cores.
...

3.6Ghz is the base clock, that's what the clock will lower to on an extremly heavy all-core work load assuming at least stock cooling in a well ventilated case. It may go even lower if the cooler is bad or not well placed or case ventilation is non-existent.

4.2Ghz is indeed single core boosting it will only hit on light bursty loads if the system is properly set up... Setting it up right means starting with latest BIOS and chipset drivers and running the Ryzen Balanced power plan that comes with the chipset drivers.

3.9Ghz is about what you'd expect under a heavy, but not extreme, processing load hitting on 4-6 cores. If it's doing that at 92C then I'd not be surprised if better cooling would help.

The stock Wraithe is not terrible but not great either. Also make sure your case is set up properly with intake and exhaust fans to draw air through and out.

 

[Barty1884]

Setting it up right means starting with latest BIOS and chipset drivers and running the Ryzen Balanced power plan that comes with the chipset drivers.

The stock Wraithe is not terrible but not great either. Also make sure your case is set up properly with intake and exhaust fans to draw air through and out.

There's arguments for the balanced profile, absolutely. But high performance can (in some situations) maintain the higher clocks. I'm sure there's tradeoffs for this, of course.

The stock cooler is "sufficient" in most situations. I've seen the AMD stock coolers maintain the advertised single-core boost on all cores (specifically on a 2700 personally), but that does require good case airflow and reasonable ambient temps.

If the OP has restrictive case, minimal fans/airflow or high ambients (or a combination of any of these), then even a decent aftermarket cooler isn't necessarily a magic bullet to fix the issue, but should improve the situation some.

 

[drea.drechsler]

There's arguments for the balanced profile, absolutely. But high performance can (in some situations) maintain the higher clocks. I'm sure there's tradeoffs for this, of course.
....

I think the main thing about the Ryzen Balanced (or 1Usumus Balanced) is it lets it freely boost single cores to max clocks at idle and yet also readily put cores into C6 deep sleep as frequently as possible even when under load. And that much faster than Windows' power management works with Windows power plans.

I'm not sure it has a major effect on clocks during heavier loads, or performance there either.

 

[aahmed12]

Hi guys, thanks for the insight.

So, I cant measure my ambient temps at the moment, but I would guess they are around 25C, which i wouldn't consider unreasonable. I also run a ceiling fan almost all the time. I'm pretty confident that this isn't the cause of high temps, because my idle temps are pretty much always like this.

My case is an nzxt h510, which could potentially be the cause, as it has a fairly restrictive airflow compared to other cases in this category. Both of the included AerF fans are set to exhaust air, out the rear, and top, by default. Also, I forget to mention I run my cpu fan at max rpm, all the time.

Also, I've noticed my cinebench R20 scores are low (3243), which is what got me worried in the first place, thinking my cpu is underperforming

 

[Insane Potatoz]

The biggest driving factor though, is temperatures. If you're seeing 92'C under stock operations, you're not going to benefit from any further performance, as the boost won't happen, and you may well at/approaching throttling territory (don't recall the specific number for Ryzen).

Thermal throttle for Ryzen is 95 degrees I'm pretty sure

 

[Barty1884]

Thermal throttle for Ryzen is 95 degrees I'm pretty sure

Yeah, that sounds about right. So not throttling (although close), but definitely not boosting.

Also, I've noticed my cinebench R20 scores are low (3243), which is what got me worried in the first place, thinking my cpu is underperforming

Your Cinebench scores are always going to be low, relative to other 3600's which are boosting higher.

3243 sits between a 2600/2600X (~2800) and a "good" 3600 result (~3600), so it's not too surprising.
Underperforming vs it's outright potential, absolutely..... Given the scenario/temps in question though, it's a fairly proportionate score, IMO.

25'C is a little warm vs an 'average' ambient, but not dramatically so. You'd expect to run a few degrees warmer by virtue of the room being a few degrees warmer. But that alone doesn't explain the ~92'C peak.

With minimal intake (even with solely exhaust, you'll still 'pull' some air in), that's probably your biggest problem - if you have a GPU that's outputting a fair amount of heat, that'll compound the problem.

An aftermarket cooler could be beneficial here - as would intake fans. While the case will limit the fan's ability to do too much, even intakes that have to fight for the minimal ventilation available will still serve you better than none at all.

 

[aahmed12]

So I just ran the test again on AMD ryzen balanced (was using high performance before), and got a slightly better score of 3355 somehow. (I mean, I'll take it)

These are some screenshots I took during the test and after, monitoring usin HWinfo.

https://imgur.com/QFCloV9

View: https://imgur.com/QFCloV9
,

https://imgur.com/5D81kFR

View: https://imgur.com/5D81kFR
,

https://imgur.com/zypwuaj

View: https://imgur.com/zypwuaj


The first picture is towards the end of the test, and the second is after the test, in which the single core boost is working as expected:

Also, as you said 4.2 Ghz is single core boost, so is the behaviour in the last picture normal, where it will boost across multiple cores.

Is temperature the only factor for why my 3600 isn't boosting higher. Would enabling PBO etc, help in any way? I was thinking about buying some front intakes. Might pull the trigger. Also, is there any aftermarket coolers, you guys have in mind. How much performance, am I really missing out on?

 

[Barty1884]

Sanity check the numbers (temps and clocks) via RyzenMaster - never hurts to have a second opinion, although I believe the issues with HWInfo/HWMonitor and the like in capturing temps/clocks are long since fixed.

As a test, before running out & buying fans &/or a cooler, remove the font panel (or side panel) and see how your numbers look. I recall hearing the H510's front panel is not removable? But I may be making that up.

If you improve airflow this way, with the cooler with easy access to fresh air, and your temps don't improve dramatically, there are other issues to address.

I'd expect you to see a ~20'C drop with either panel removed... in which case, adding front intake should be sufficient.

In terms of an aftermarket cooler.... there are lots of offerings that won't break the bank. 212Evo is a tried & true, budget offering.... Shadow Rock3 would suffice, NH-U12S... lots of options.

 

[aahmed12]

Sanity check the numbers (temps and clocks) via RyzenMaster - never hurts to have a second opinion, although I believe the issues with HWInfo/HWMonitor and the like in capturing temps/clocks are long since fixed.

As a test, before running out & buying fans &/or a cooler, remove the font panel (or side panel) and see how your numbers look. I recall hearing the H510's front panel is not removable? But I may be making that up.

If you improve airflow this way, with the cooler with easy access to fresh air, and your temps don't improve dramatically, there are other issues to address.

I'd expect you to see a ~20'C drop with either panel removed... in which case, adding front intake should be sufficient.

In terms of an aftermarket cooler.... there are lots of offerings that won't break the bank. 212Evo is a tried & true, budget offering.... Shadow Rock3 would suffice, NH-U12S... lots of options.

Actually, the front panel isn't removable, but the side panel is. It's actually pretty late, where I live, so I'll update you guys on this tomorrow, after testing with the panel off.

 

[Barty1884]

Actually, the front panel isn't removable, but the side panel is. It's actually pretty late, where I live, so I'll update you guys on this tomorrow, after testing with the panel off.

Ok, I wasn't misremembering then...
Yeah, try the side panel. It'll help confirm if it's a lack of airflow.

If the temps are still quite high, pushing 80'C or higher, then at least part of the issue is likely in the mounting of the cooler (or thermal paste application etc).

 

[aahmed12]

Ok, so I've just run cinebench R20, with the side panel off, and see little to no difference in the final temps. The idle temps did lower by a few degrees, but at the end of the tes, I still get exactly the same temp, using both ryzen master, and Hwinfo, one after another. The score is also pretty much the same. Here's the results: http://imageshack.com/a/img921/9591/3uTrPz.png
So, restrictive case airflow, is definitely, not the cause of my high temps. Could it be an application problem, or is the wraith stealth, just not adequate enough, for such loads. Maybe I'm underestimating the ambient temps, but I imagine that wouldn't make too much of a difference

 

[drea.drechsler]

... Could it be an application problem, or is the wraith stealth, just not adequate enough, for such loads. ...

Application problems are more common than realized. I've seen where even experienced builders take an overheating cooler off to find they'd unwittingly left a plastic film protector on the heatsink base, for instance. It happens. But then, the wraith stealth really is a minimalist solution for cooling. An upgrade definitely can improve performance over long benchmark runs.

And about ambient...where in the world are you? is it coming to hot weather season there and you don't have refrigerated air to keep the room cool? season changes can have an effect as we get used to the temp change but thermally sensitive devices don't.

 

[aahmed12]

Application problems are more common than realized. I've seen where even experienced builders take an overheating cooler off to find they'd unwittingly left a plastic film protector on the heatsink base, for instance. It happens. But then, the wraith stealth really is a minimalist solution for cooling. An upgrade definitely can improve performance over long benchmark runs.

And about ambient...where in the world are you? is it coming to hot weather season there and you don't have refrigerated air to keep the room cool? season changes can have an effect as we get used to the temp change but thermally sensitive devices don't.

I'm located in pakistan. Islamabad to be exact. Currently, the outside temps in the day usually hover around 30C. It's usually hottest here, in May/June, so you could say, it's hot weather season. At the moment, I run a ceiling fan, in my room. I actually do have an AC unit for summers, but haven't started running it yet. (Probably will do in a few day's time). I think I'll probably go out and get myself a new cooler, after this whole pandemic is over, as that seems like the only probable cause for the high temps, I'm getting. I think I'll go with A Hyper 212 Evo or something of the like.

 

[Barty1884]

As on last roll of the dice, I'd remount the cooler.
Even with 30'C temps, your results still seem a little bit high - and the fact there's no significant change under load with the side panel off doesn't seem right.

If you're running the test in ~30'C with restricted air, then the same ambient with less restriction should improve.... might not be significant to the tune of 10-20'C, but your temps are almost identical?

 

[aahmed12]

http://imageshack.com/a/img921/9591/3uTrPz.png . These are the temps towards the end of the cinebench test, and yes they are pretty much identical. I got 92C with the tempered glass panel on, as well. Only idle temps, lowered by a few C.