Ryzen 5 3600 turbo speeds

[AusWolf]

Hi,

I'm in the process of downgrading my Ryzen 9 5950X because I figured it would make sense with current hardware prices since I only ever use it for games anyway.

I bought a R5 3600 as a replacement because I liked the way it held an all-round 4200 MHz in the Techpowerup review. It doesn't do that for me, though. Instead, it fluctuates between 4050-4150 MHz depending on the load. Not a big deal, but it kind of bothers me, as I'd like to think the tests for the review were done with out-of-the-box settings.

What do I need to do to achieve the same stable turbo clock as in the review? I've got an ASUS Tuf Gaming B550M-Plus wifi motherboard with all auto bios settings, and a be quiet! Shadow Rock LP cooler coming later today - so far tested with the stock Wraith Stealth cooler, and a not so much better Arctic Alpine 64 Pro with similar results.

 

[tabascosauz]

Hi,

I'm in the process of downgrading my Ryzen 9 5950X because I figured it would make sense with current hardware prices since I only ever use it for games anyway.

I bought a R5 3600 as a replacement because I liked the way it held an all-round 4200 MHz in the Techpowerup review. It doesn't do that for me, though. Instead, it fluctuates between 4050-4150 MHz depending on the load. Not a big deal, but it kind of bothers me, as I'd like to think the tests for the review were done with out-of-the-box settings.

What do I need to do to achieve the same stable turbo clock as in the review? I've got an ASUS Tuf Gaming B550M-Plus wifi motherboard with all auto bios settings, and a be quiet! Shadow Rock LP cooler coming later today - so far tested with the stock Wraith Stealth cooler, and a not so much better Arctic Alpine 64 Pro with similar results.

Interesting choice in downgrading.

That review is a launch day review from July 2019, I'd regard the boost frequency data with a grain of salt. There is much that we didn't understand about Ryzen 3000 back then. There is no 3600 that is boosting to 4.2GHz all-core constantly on stock settings and a slim cooler like you have, the most you'll generally get is about 4.15 that decays. The 4050-4150 that you see is pretty normal. This is corroborated by the fact that although it stayed at "4.2GHz" on all cores, a true all-core OC still yielded only 4.125GHz @ 1.37, all in that very review.

I'm pretty sure this was all before HWInfo even had the Effective Clock metric, which is rather important since the "Core Clock" metric is pretty much bullshit on Ryzen without Snapshot Polling enabled. Yeah, it'll display 4200MHz on Cores 0 through 5, but it doesn't mean anything.

Of course, because it's PB2, it still depends on load. On any Ryzen 3000 or 4000 chip you can very well see sustained 4.2GHz all-core.........in a memory stress test like TM5. But in a CPU benchmark? No, that's not how Ryzen 3000 worked. If you're on Ryzen 5000 6/8 core, you'll see much higher all-core, but the design of Matisse didn't allow for that high of all-core out of the box. Matisse was more "up to x GHz", while Vermeer is pretty much "at least x GHz".

You could try working with PBO and a stouter cooler (large tower or 240mm AIO), but as with all PBO that initially exciting frequency will still decay slowly with time unless you have an absolutely insane cooler under which temps never rise. Alternatively, you can try all-core if the silicon is good. If it's a recent production 3600, you shouldn't have any problems hitting 4.2GHz all core at under 1.3V.

 

[WHDS]

It depends on the workload, my 3600 acts the same way. It's not realistic to achieve maximum boost clocks all the time, plus the extra 50-150 mhz won't provide any meaningful performance benefit

 

[nguyen]

Yeah even with a X63 NZXT Kraken AIO and PBO enabled, my 3600 never reach 4200mhz turbo boost, mostly between 4050-4150mhz like yours. Eventually I settled for 4.1Ghz all-core at 1.37Vcore with temp in the low 70s in games, seems like AMD reserve the higher bin silicon for 3600X and 3600XT

 

[AusWolf]

Interesting answers, thanks, guys.

Honestly, a couple MHz isn't a big deal, it's just weird that the CPU performs differently compared to expectations based on the tpu review.

As for my reasons to why I'm downgrading: I bought the 5950X with the intention to see how it'll perform in games in the coming years. I didn't expect hardware prices to rise so rapidly, but since they did, I got into my head that maybe a slower PC would serve me just fine with games at 1080p, and I could make some money for the summer holidays. Weird, I know, but I've never been the most reasonable person on the planet.

 

[Deleted member 205776]

1.37Vcore

You sure that much voltage is a good idea?

 

[nguyen]

You sure that much voltage is a good idea?

Still better than stock voltage with PBO I think, I have seen the voltage peak to 1.45Vcore in single core load. Anyways I have the Kraken X63 to keep it cool so no fret there, 1.37Vcore is perfectly safe for 24/7 if you have good cooling.

 

[kapone32]

The OP should be able to hit 4.4 GHZ with a 1.30 OC.

 

[Deleted member 205776]

I have seen the voltage peak to 1.45Vcore in single core load

Which is completely normal... how people still don't understand this is beyond me.

1.37v is way too much for an all core load on Zen 2 7nm, especially 4.1 all core where that chip would normally do 3.8-3.9 all core. But who am I to stop you from degrading your chip.

 

[freeagent]

My 3600XT does 4400 with 1.268v and 4500 with 1.3375v, I think under 1.3 would be better for all core all loads.

 

[Deleted member 205776]

My 3600XT does 4400 with 1.268v and 4500 with 1.3375v, I think under 1.3 would be better for all core all loads.

I decided to stop pretending like I know what the safe voltage for my 3900X is and just let the stock voltage fitness regulator do the thinking for me. 7nm doesn't tolerate voltage as well as 14nm. Running 4.1 at 1.37v seems like the best way to simultaneously get lower performance than stock and pave a clear path to the early coffin for your CPU.

I've found that the best way to overclock a Zen 2 chip is to get a high end cooler and good motherboard with solid VRMs - I have both of those, and they do a much better job at overclocking my CPU than I could ever do with manual OC... while keeping the CPU's safety features and voltage regulator still active, which they're not when you manually OC.

 

[Deleted member 178884]

https://www.reddit.com/r/overclocking/comments/ejd5c9
Even 1.3v can degrade Zen2, there's reports of down to 1.28v degrading Zen2 based processors without any issues - perhaps research before providing misinformation?

 

[nguyen]

Which is completely normal... how people still don't understand this is beyond me.

1.37v is way too much for an all core load on Zen 2 7nm, especially 4.1 all core where that chip would normally do 3.8-3.9 all core. But who am I to stop you from degrading your chip.

Yeah I don't do rendering or Prime95 all day, just gaming. It was a set and forget overclock since I gave that PC to nephew over a year ago, no problem so far, not that it matter LOL.

Seems like people like to stress test their PC with unrealistic testing condition like P95 or Furmark when all they do is gaming. All my overclocks are gaming stable only since all I do is gaming and that's it.

 

[Deleted member 205776]

Yeah I don't do rendering or Prime95 all day, just gaming. It was a set and forget overclock since I gave that PC to nephew over a year ago, no problem so far, not that it matter LOL.

Seems like people like to stress test their PC with unrealistic testing condition like P95 or Furmark when all they do is gaming. All my overclocks are gaming stable only since all I do is gaming and that's it.

Doesn't matter if all you do is gaming. The chip will degrade at 1.37v under load (which includes gaming), as explained above. 7nm is not 14nm. Hope your nephew won't wake up to a black screen.

 

[nguyen]

Doesn't matter if all you do is gaming. The chip will degrade at 1.37v under load (which includes gaming), as explained above. 7nm is not 14nm. Hope your nephew won't wake up to a black screen.

Nah degrading means it requires more voltage for a certain clocks, the chip won't die.
And no it won't degrade under gaming condition, that's taboo.

 

[thesmokingman]

Which is completely normal... how people still don't understand this is beyond me.

1.37v is way too much for an all core load on Zen 2 7nm, especially 4.1 all core where that chip would normally do 3.8-3.9 all core. But who am I to stop you from degrading your chip.

It's really surprising at this point, lol.

 

[tabascosauz]

Yeah I don't do rendering or Prime95 all day, just gaming. It was a set and forget overclock since I gave that PC to nephew over a year ago, no problem so far, not that it matter LOL.

Seems like people like to stress test their PC with unrealistic testing condition like P95 or Furmark when all they do is gaming. All my overclocks are gaming stable only since all I do is gaming and that's it.

There's no clear-cut two sides to the argument, but I will say that unless you exclusively play old games it has more relevance than you might think. BFV uses AVX, heats up Ryzens like a mofo. MW2019 uses AVX, and as we all know will not only hammer CCD1 while in game but any background processing such as applying an update or installing shaders runs basically close to max all-core AVX. We all know how much of a clock penalty the PB2 algorithm automatically applies to AVX, it's not without reason.

Personally I wouldn't run over 1.3V set Vcore, but a lot of the "degradation" stories are also unverifiable. Ryzen is an absolute bitch to stability test, and what you thought was "stable" for a week may very well drop worker threads tomorrow in P95. Average user sees that result and deduces "my CPU must have degraded", when in reality it still does take a fair bit of effort or negligence to degrade these CPUs, eg. an hour of 1.5V all-core thanks to Clocktuner.

But yes, otherwise, gaming is not an all-core workload. Only constant all-core workloads at high current draw will actually run the risk of degradation.

 

[Deleted member 205776]

Nah degrading means it requires more voltage for a certain clocks, the chip won't die.
And no it won't degrade under gaming condition, that's taboo.

Well, can't say I didn't try. Just trying to prevent a 3600 ending up the same way as my friends' did, when all he did was gaming. Didn't take long for that chip to say goodbye. Not entirely, but after a while, it produced some of the nastiest lock-ups, freezes, and BSODs. Not sure if the marginal performance differences are even worth chip degradation.

 

[nguyen]

There's no clear-cut two sides to the argument, but I will say that unless you exclusively play old games it has more relevance than you might think. BFV uses AVX, heats up Ryzens like a mofo. MW2019 uses AVX, and as we all know will not only hammer CCD1 while in game but any background processing such as applying an update or installing shaders runs basically close to max all-core AVX.

But yes, otherwise, gaming is not an all-core workload. Only constant all-core workloads at high current draw will actually run the risk of degradation.

Yeah degradation involves voltage, current and temperature. I mean enabling PBO alone already cause faster degradation because it allows higher current.

Well, can't say I didn't try. Just trying to prevent a 3600 ending up the same way as my friends' did, when all he did was gaming. Didn't take long for that chip to say goodbye. Not entirely, but after a while, it produced some of the nastiest lock-ups, freezes, and BSODs. Not sure if the marginal performance differences are even worth chip degradation.

Thanks but recently I sold my nephew's 2060S to miner (PC is mine) and slapped in a 1060, CPU is not doing much in games now . Also I set 1.37Vcore because I don't want to set high Load-line Calibration, because high load-line can actually damage CPU much worse than high Vcore, not that many people are aware of that.

 

[AusWolf]

Thanks, everyone, but I'm not planning on overclocking. I just want stable clocks with acceptable thermals - which brings me to my next problem. My be quiet! Shadow Rock LP has arrived, but this thing is still sitting at 50 C idle and hits 80 C as soon as I ask it to do something. I played around with the Windows power settings, disabled literally every performance increasing gimmick in BIOS (e.g. PBO, ASUS enhancements, etc.), still no effect. I never had this issue with the 5950X - maybe I should have gone with Vermeer again. *sigh* Any advice?

 

[thesmokingman]

Yeah degradation involves voltage, current and temperature. I mean enabling PBO alone already cause faster degradation because it allows higher current.

Wrong. Just by stating that you completely misunderstand the way the chip works. Like really man, why don't you read the link posted by L above instead of throwing out what you think is happening.

First PBO isn't even enabled at all. By default on auto... PBO is DISABLED. Thus you seeing 1.45v single threaded, it is one, NOT a high current situation and two regardless of whether PBO is on or off that is how the cpu is designed to operate. The silicon fitness monitor/controller will not fry the silicon it's meant to protect. /smh

 

[Kissamies]

Personally I gave up on manual overclocking on AMD when I had my previous build with R5 2600. The gains are so minimal when comparing to all the cons so I trust PBO more.

And my 3600 usually stays on that 4200MHz when gaming. Cooled with modified Eisbaer 240 so it's semi-customloop.

 

[tabascosauz]

Thanks, everyone, but I'm not planning on overclocking. I just want stable clocks with acceptable thermals - which brings me to my next problem. My be quiet! Shadow Rock LP has arrived, but this thing is still sitting at 50 C idle and hits 80 C as soon as I ask it to do something. I played around with the Windows power settings, disabled literally every performance increasing gimmick in BIOS (e.g. PBO, ASUS enhancements, etc.), still no effect. I never had this issue with the 5950X - maybe I should have gone with Vermeer again. *sigh* Any advice?

I dunno, if that's just on light load then that just sounds like a mounting/mounting pressure/contact issue. Especially if it's idling at 50C, without any noticeable background tasks going on. Ryzen idle is pretty annoying but it's easy to tell when it's actually idling and when it's running something in the background. New Be Quiet mounts on the DRP4 are an improvement but sometimes take a bit of fiddling compared to Secufirm.

I don't imagine the similarly sized Shadow Rock LP performs much worse than the NH-L12S, which is stout as hell and kept my 4650G at less than 65C at all times during the short time I had it. Renoir is slightly cooler but not by much. My NH-L9x65 now keeps the 4650G at under 70C at all times, I figure between Matisse and Renoir add about 10C to account for chiplets.

When I had my 3700X it crept closer to 80C under all-out CPU benchmarks under my NH-D9L, which is worse than my NH-U9S and performs about the same as the NH-L12S. IIRC the 3600 doesn't really run any cooler than the 3700X because it runs more wattage per core, and still takes up the 88W PPT and doesn't really "save" any PPT unlike the 5600X.

The 2019 Low-Profile Cooler Shootout: Crowning a Welterweight Champ | The Tech Buyer's Guru (techbuyersguru.com)

An all-core under 1.3V between 4.0-4.2GHz probably works out best for everyday thermals, but make sure it's not a mounting issue first.

 

[nguyen]

Wrong. Just by stating that you completely misunderstand the way the chip works. Like really man, why don't you read the link posted by L above instead of throwing out what you think is happening.

First PBO isn't even enabled at all. By default on auto... PBO is DISABLED. Thus you seeing 1.45v single threaded, it is one, NOT a high current situation and two regardless of whether PBO is on or off that is how the cpu is designed to operate. The silicon fitness monitor/controller will not fry the silicon it's meant to protect. /smh

How about you read that link instead .
Somehow you didn't read that the people who degraded their chips were stressed testing with P95 AVX, Linpack, etc for the hell of it....which are absolutely pointless for gaming PC. Furthermore those chips could have defects that are destined to fail no matter what the voltages are.
Oh well I will tell you when my 3600 degrade in the future and how long did that take.

FYI Der8auer also tested for degradation with a couple of Ryzen 5000, let see how it goes.

 

[Hemmingstamp]

3600 Silver sample owner here. Don't tinker with the thing manually. I pushed mine to 4.3GHz and it became unstable. Leave it at stock and use Ryzen Master to boost it.

I used ClockTuner to find out it was a silver sample. I was sure I could squeeze some more MHz out of the CPU only to find out it wasn't worth my time after all.