Question My new 1600af isn't boosting correctly.

[Socko001]

As i've seen recently, the 1600 af often boost normaly to 3600/3700 normaly without messing with much(bios, power plan, etc)mostly in games. I know that normaly the all core boost is often 3400, but in most reviews, most that I've seen it's reaching 3600 with no problem whit low voltage and low temps, even overclocking is easy with this chip, wich I wont do it, I just want the 3600mhz sustained and even undervolt it because of the ambient temperature of where I live.

Mine in ther other hand in games doesnt go over 3400 or 3425(stock bios, nothing messed with). I've tried a lot of things, from negative voltage, disabling core performance boost, power now and c-state power. it goes sometimes to 3600 or 3700, but very brieffly, and most likely a mistake in the hwinfo.

In bios(I use a Biostar b350et2, I live in Brazil, prices are not friendly here, so we buy what we can afford) the stock voltage for 3200 is 1.4, 1.44 and when I tried using Ryzen Master to set it to 3.6 1.23 volts it crashes.

I'm trying to narrowing down what is the problem, if the cpu(bad silicon lottery), my mobo, vrm's.

I mean I can be happy with the 3425mhz, but it's kind of sad to buy one product that can reach 3.6 sustained or more and receive what is eessentially a 1600 ae(and no its the AF version 12nm, pinacle ridge, the box says it, cpu-z says it).

And in terms of thermals, using stock cooler(wich shoudnt be a problem, not realy trying to push the chip) in stock settings reaching the 3425mhz, the temps are as high as 50, max 55, when trying to reach 3600 1.24v(it doesnt crashes) it goes to 70 easily.

Even trying to take prints ins't easy, voltages go up and down, clock as well,

Im using a:

1600 af(stock cooler)
biostar b350et2
16(2x8) 2400mhz
rx 570



Using hwinfo for comparisson. Sorry if any error in my english, as I said brazilian here, so not my mother language.

 

[Socko001]

What I did here to at least help me out is i set all cores throught Ryzen Master to 3400mhz to the lowest voltage I could 1.25v(could go lower 1.125v, but I didnt feel it would be stable and the gain would be meaningles) in stress i reached 64C,65C with a power draw of 80w, in games 49C drawing 40w to 50w. R20 gave a result of 2600 close to the 2750 that 3600mhz would give, no realy happy, but well, i think i need to buy a new board.

 

[Third-Eye]

You don't need a new motherboard, instead get a better CPU cooler like a 120mm tower to see if it helps. The CPU only boosts to certain speeds when it's under a certain temperature threshold. Leave it all on auto with the 120mm tower and see where it stands. On my Ryzen 5 2600 and Hyper 212 Evo, I get around 3.75-3.8Ghz all cores at 65c and max boost is 3.9Ghz.

 

[Karadjgne]

Ryzen are interactive boost cpus, not like most Intel which are performance boost. A Ryzen will only boost according to voltages and temps upto its predetermined power limit, which is governed by current.

So if your voltages are running high with auto settings, you'll hit the power limit much faster as the amperage needed to push the speeds will also be higher. P = V*A.

So the way to get a higher boost is tailor the voltages down, which also lowers temp. But it's got to work in concert. Some motherboards this works better without PBO, some it works better with. Most times just leaving the bios totally stock and just enabling PBO yields better and more stable results than any OC.

If a game like CSGO uses 2 cores, you'll get a higher boost on just those 2 than an all core fixed boost can afford.

And no, I think your cpu is boosting correctly, it's governed by its own limits and they start at just beyond 55-60°C. What's not working is any preconceptions you have a based on others results.

 

[Socko001]

Ryzen are interactive boost cpus, not like most Intel which are performance boost. A Ryzen will only boost according to voltages and temps upto its predetermined power limit, which is governed by current.

So if your voltages are running high with auto settings, you'll hit the power limit much faster as the amperage needed to push the speeds will also be higher. P = V*A.

So the way to get a higher boost is tailor the voltages down, which also lowers temp. But it's got to work in concert. Some motherboards this works better without PBO, some it works better with. Most times just leaving the bios totally stock and just enabling PBO yields better and more stable results than any OC.

If a game like CSGO uses 2 cores, you'll get a higher boost on just those 2 than an all core fixed boost can afford.

And no, I think your cpu is boosting correctly, it's governed by its own limits and they start at just beyond 55-60°C. What's not working is any preconceptions you have a based on others results.

I understand, thank you for the reply, the question I raised is that in similar boards the 1600af easily reachs 3600mhz. In my Bios i tried a lot of things, disabling core performance boost, powernow, c-states, reduncing voltages(as you said), it never went higher then 3425, wich lead me to believe it would be an VRM problem, what I did is what I said in the other coment, set to 3400, 1,125v. In stress never goes over 64C and gaming 49C.

I've read that the 1600af doesnt have PBO, the option to get to it(or in this case I think it would be XFR) is greyed out, so... Only option is the CPB, wich did lower the voltages, but not enough to, as I said, go higher then 3425.

 

[Socko001]

You don't need a new motherboard, instead get a better CPU cooler like a 120mm tower to see if it helps. The CPU only boosts to certain speeds when it's under a certain temperature threshold. Leave it all on auto with the 120mm tower and see where it stands. On my Ryzen 5 2600 and Hyper 212 Evo, I get around 3.75-3.8Ghz all cores at 65c and max boost is 3.9Ghz.

Thanks for the reply, the thing is, as my motherboard vrm's dont have an heatsink, they are cooled by the Wraith Stealth, if I used a tower cooler it would be very bad fo ther poor vrm's kkkk, As I said in the post the majority of people using the 1600af didn't use other type of air cooler, they do reach 3.6 at lower voltages just with stock cooler.

 

[Third-Eye]

Thanks for the reply, the thing is, as my motherboard vrm's dont have an heatsink, they are cooled by the Wraith Stealth, if I used a tower cooler it would be very bad fo ther poor vrm's kkkk, As I said in the post the majority of people using the 1600af didn't use other type of air cooler, they do reach 3.6 at lower voltages just with stock cooler.

Can you post a screen shot of the CPU voltage when in Windows? Maybe try a different program like HWInfo and post the sensor readings for the CPU and motherboard from the Sensors Status window. Obviously high voltage is going to cause issues, but I'm wondering if there might be some other issues as well.

Also, how many case fans do you have? What is your room temperature? You can't expect full 3.6Ghz boost on all cores under load. If you don't have good airflow your CPU is not going to boost that high under load.

 

[Socko001]

Can you post a screen shot of the CPU voltage when in Windows? Maybe try a different program like HWInfo and post the sensor readings for the CPU and motherboard from the Sensors Status window. Obviously high voltage is going to cause issues, but I'm wondering if there might be some other issues as well.

Also, how many case fans do you have? What is your room temperature? You can't expect full 3.6Ghz boost on all cores under load. If you don't have good airflow your CPU is not going to boost that high under load.

I'have been using hwinfo, i've heard it's the most consistent with Ryzen in the moment. The problem is, when in stock, no oc, just leaving to the mobo and cpu do it's thing, the voltage it's not consistent, it goes up and down very frequently, the general i've observed it's it goes to 1.3v, 1.35v, 1.4v and the goes down to 1.12,1.15v in stress leaving the clocks down to 3.0, 3.1. So it's not that easy to take a screen shot. The sensation I'm getting it's that I have and 1600AE, but it's not the case.

My case it's not the best with an airflow, but i've put one in the front(intake) and one at the back, and one at the top, it's quite cold here in Brazil this time of the year.

 

[Socko001]

Just to make it more understandable at idle, the voltages go to 1.3, 1.35, 1.4, I saw once 1.44, but at the same time goes as lower as 1. 08. At Idle the clocks stay 3.2, to 3.4, and sometimes 3.5, 3.6, but very rarely. When stressing in R20 or aida64, the clocks go down to 3.0,3.1 or even 2.98, the voltages lower because of that, but inconsistent as well jumping in the same way as in idle.

 

[Karadjgne]

In my Bios i tried a lot of things, disabling core performance boost, powernow, c-states,

You disable the core performance boost. That's not Precision Boost Overdrive, core performance boost is basically 'turbo'. Power now is for laptops basically, it's the dynamic frequency scaling, part of the eco settings that can limit boosts with battery power. C-states is your low power states for when the cpu is supposed to drop into idle, it'll go into sleep modes after no user input. With Intels that meant a serious drop in voltages, which could create instability if the dropped amount was more than the clocks needed.

In a nutshell, you are worried about what someone else has attained, not knowing anything about their pc in general but assuming that because they have a similar motherboard and cpu, you should get it too. This doesn't mention cooling, case airflow, ambient temps or the affects each or combined will have on temps and consequently boost levels. Ram also plays into the equation, higher speed ram offers more performance per clock, so boosts can be higher for lower temps.

You are missing a lot of information, not only about the other systems, but about yours too. You messed with settings in bios without really thinking about the affects they'll have on other settings. While a good attempt at trial and error, that lack of information is hurting you.

I'd worry far less about your current boost levels and do some serious research into OC, it's affects, it's needs, requirements etc as pertains to your cpu. OC is running the clicks at higher speeds than the stock set voltages. If you undervolt, that's the same as overclocking, just from a different perspective.

Learning what everything does, how it works, what it's for will give you insight as to what to expect, and what you'll need to overcome it.

 

[scout_03]

give full name and model of ram same with case also how did you set your ram before the overclock this m-atx board need more air to cool the vrm .

 

[Socko001]

You disable the core performance boost. That's not Precision Boost Overdrive, core performance boost is basically 'turbo'. Power now is for laptops basically, it's the dynamic frequency scaling, part of the eco settings that can limit boosts with battery power. C-states is your low power states for when the cpu is supposed to drop into idle, it'll go into sleep modes after no user input. With Intels that meant a serious drop in voltages, which could create instability if the dropped amount was more than the clocks needed.

In a nutshell, you are worried about what someone else has attained, not knowing anything about their pc in general but assuming that because they have a similar motherboard and cpu, you should get it too. This doesn't mention cooling, case airflow, ambient temps or the affects each or combined will have on temps and consequently boost levels. Ram also plays into the equation, higher speed ram offers more performance per clock, so boosts can be higher for lower temps.

You are missing a lot of information, not only about the other systems, but about yours too. You messed with settings in bios without really thinking about the affects they'll have on other settings. While a good attempt at trial and error, that lack of information is hurting you.

I'd worry far less about your current boost levels and do some serious research into OC, it's affects, it's needs, requirements etc as pertains to your cpu. OC is running the clicks at higher speeds than the stock set voltages. If you undervolt, that's the same as overclocking, just from a different perspective.

Learning what everything does, how it works, what it's for will give you insight as to what to expect, and what you'll need to overcome it.

Okay, thank you for the rather long and educated reaply, I apreciate it.

I admit not the more knowledgeable with OC's, i'm more of a adventurer of sorts, I know just a thing or two, not enough it seems. Well do you think the undervolt I did a bad idea? I mean higher clock(3.4), lower voltage(1.125v), good temperature(65C)?

To where shoud I proceed then? Just leaving at that and just accept? Or do you think with this board I could improve the 3400 limit? And if I could as I said, how shoud I proceed?

 

[Socko001]

give full name and model of ram same with case also how did you set your ram before the overclock this m-atx board need more air to cool the vrm .

2x8GB HyperX Fury DDR4-2400 CL15-15-15 @1.2V , just using it's stock, not overclocking it. The case is a Gamemax Expedition h605 with an adaptation I did to put a intake fan at the front.

 

[scout_03]

could you post a picture of the front panel mod seams your not having enough fresh air in to cool system aslo for ram did you activate xmp in bios you will gain 200 mgz .

 

[Socko001]

https://imgur.com/DQm1GRD

View: https://imgur.com/DQm1GRD

Not realy a mod, just opened the hole where the name of the brand was. I know that it's not what a good airflow would be, but im out of option in this area, I use the back one as intake as well, just the top fan that is outtake.

And already tried pushing to 2667, both my ram has a nanya b-die on them horrible to OC. The 2667 OC is not stable enough, some times is work as a charm, others not.

Even with the case open, wich I don't have a problem it still doesnt fixes my problems.

 

[Karadjgne]

You have a 1600AF. That's a 65w cpu. Not so much as that's TDP or thermal output, but that's the wattage limit that's self imposed by the cpu. As I said earlier, currant (Amperage) is the controlling factor for boosts. The harder the cpu has to work, the more current it needs. And it's going to stick to the 65w limit for duration.

What this means is voltages Will fluctuate as demand needs. If a boost to 3.4GHz needs 50A split up amongst the cores, 65w/50A = 1.3v. If the workload lowers or core count lowers, and consequent currant needs lower, voltages will go back up. 65w/45A = 1.45v. Just as if workload increases, voltages go down. 65w/60A = 1.08v. If the cpu would be unstable at the 1.08v, it'll drop the boost to compensate, so at 65w you'll be seeing a 55A load instead of the maximum 60A load you'd expect, and see a 1.18v vcore instead.

It's a dance, the cpu will balance itself between boost, load, voltages, current. It'll give you its best performance, but that might not be what you want or expect. Higher temps will also factor into the power limits, boosts will be cut back in appropriate stages, since higher voltage isn't the only thing that raises temps, higher speeds, higher currant also affect temps.

It's a Ryzen. Not an intel. Ryzens are dynamic cpus, they'll self adjust their own parameters to give you best performance. Intel gives you best performance, but expects you to adjust the parameters for it.

For example, people are often confused by Ryzen temps. They'll see 70's with a stock cooler, so dump it in favor of something better, then get frazzled because the temps didn't change. They'll reseat the cooler, waste more money on paste, hunt the internet for answers, then end up here all frustrated because they just spent $50 on a new cooler and got the same results, what did they do wrong. Answer is they thought Intel. Ryzen boosts are governed in part by temps. The cpu now has a more effective cooler, which can lower temps, so the Ryzen now has more room for a higher boost, and shoves the temps right back up to 70's. Prior, they got 3.4GHz stock, changed nothing but the cooler and forgot to look and see they now get 3.8GHz boost. Got their head so wrapped up in Intel standards, they forgot or never bothered looking to see they have a Ryzen, a dynamic boost cpu.

Better cooling means better boosts, upto the limit of the cpu, voltages, currants. Only after that point is reached will the cooler actually start lowering temps.

So let's say you get 80°C at 65w at 50A with the stock cooler. The cpu says that's as far as it's willing to go, so only allows a 3.4GHz boost. Swap coolers to a giant NH-D15S, the cpu sees closer to 50°C, but wait a second, we have headroom in temps so we can boost higher! So it does, and now the cpu is at 65w and 60A and 3.9HGz boost, but still has room in temps, but cannot boost any higher, so the original 80°C drops to 60°C. The cpu gets its cake, and gets to eat it, and still has room for secondses. A slightly smaller cooler would get the same results, same boost limits, down to the point where there's no room for secondses, at which point the temps will be higher. The efficiency of the cooler will determine temps, only after the cpu is maxed for boosts.

Generally.

 

[Socko001]

https://imgur.com/a/kF6wOwn

View: https://imgur.com/a/kF6wOwn

Just as a reference of what is my bios, and why we call it a "soviet bios".

 

[Socko001]

You have a 1600AF. That's a 65w cpu. Not so much as that's TDP or thermal output, but that's the wattage limit that's self imposed by the cpu. As I said earlier, currant (Amperage) is the controlling factor for boosts. The harder the cpu has to work, the more current it needs. And it's going to stick to the 65w limit for duration.

What this means is voltages Will fluctuate as demand needs. If a boost to 3.4GHz needs 50A split up amongst the cores, 65w/50A = 1.3v. If the workload lowers or core count lowers, and consequent currant needs lower, voltages will go back up. 65w/45A = 1.45v. Just as if workload increases, voltages go down. 65w/60A = 1.08v. If the cpu would be unstable at the 1.08v, it'll drop the boost to compensate, so at 65w you'll be seeing a 55A load instead of the maximum 60A load you'd expect, and see a 1.18v vcore instead.

It's a dance, the cpu will balance itself between boost, load, voltages, current. It'll give you its best performance, but that might not be what you want or expect. Higher temps will also factor into the power limits, boosts will be cut back in appropriate stages, since higher voltage isn't the only thing that raises temps, higher speeds, higher currant also affect temps.

It's a Ryzen. Not an intel. Ryzens are dynamic cpus, they'll self adjust their own parameters to give you best performance. Intel gives you best performance, but expects you to adjust the parameters for it.

For example, people are often confused by Ryzen temps. They'll see 70's with a stock cooler, so dump it in favor of something better, then get frazzled because the temps didn't change. They'll reseat the cooler, waste more money on paste, hunt the internet for answers, then end up here all frustrated because they just spent $50 on a new cooler and got the same results, what did they do wrong. Answer is they thought Intel. Ryzen boosts are governed in part by temps. The cpu now has a more effective cooler, which can lower temps, so the Ryzen now has more room for a higher boost, and shoves the temps right back up to 70's. Prior, they got 3.4GHz stock, changed nothing but the cooler and forgot to look and see they now get 3.8GHz boost. Got their head so wrapped up in Intel standards, they forgot or never bothered looking to see they have a Ryzen, a dynamic boost cpu.

Better cooling means better boosts, upto the limit of the cpu, voltages, currants. Only after that point is reached will the cooler actually start lowering temps.

So let's say you get 80°C at 65w at 50A with the stock cooler. The cpu says that's as far as it's willing to go, so only allows a 3.4GHz boost. Swap coolers to a giant NH-D15S, the cpu sees closer to 50°C, but wait a second, we have headroom in temps so we can boost higher! So it does, and now the cpu is at 65w and 60A and 3.9HGz boost, but still has room in temps, but cannot boost any higher, so the original 80°C drops to 60°C. The cpu gets its cake, and gets to eat it, and still has room for secondses. A slightly smaller cooler would get the same results, same boost limits, down to the point where there's no room for secondses, at which point the temps will be higher. The efficiency of the cooler will determine temps, only after the cpu is maxed for boosts.

Generally.

I realy am not trying to sound as an ignorant after all you typed, it's a class, thank you for taking your time. I'm trying to undertand, but since english is not my mother language and there are lots of math talk included I'm trying my best here. What I understand here is that the core boost in amds work involving frequency, load per core, voltage and current, as you just wrote.

I understand that this cpu limit is 65w, in hwinfo64 it shows 84w in stress stock settings. The voltages going up and down you alredy answered it moves like that because of what the load, temps and such is. realy smart.

What is giving me this sensation that is something's wrong is the high power draw, its normal? Because by the logic W/A=V if the W=84 and is receveing 50A then the voltage used would be 1,68v right, what would result in the clock going down to compensate for the extra 0.28 above 1.4? It's that what is happening here? If its, why its happening?

 

[Socko001]

I think I know what is the problem here. Inded part of the problem is the mobo, i've been researching about biostar motherboards, and in a video(a brazilian one) said that his b350(a little better than mine) gave too much voltage to the cpu, and when the voltage dropped it woudnt drop as much as it shoud, so a lot of the times the cpu would reach it's 65w tdp with a higher voltage then it normaly needed and thus the clock would come down to 3400mhz as it was happening to mine. Or lower when the load was a little too much for the task at hand(cinebench, aida64 etc).

Another culprit, not a major one, but a one nonetheless, stockcooler thermal paste. I have an 2200g bought last year it came with that more dense thermalpaste, this new 1600af came with another type of thermal paste a more simpler that was having a hard time kepping up thermals when Ocing. I exchanged with a silver based one I have, and take a quick look at this.

https://imgur.com/a/3HccOst

View: https://imgur.com/a/3HccOst


I've managed at least in games, didn't test in cinebench or aida yet, to put it 3600mhz@1.19v, i'm going to try and go lower, trying to keep the cpu power lower then the 65w tdp. It's 2 am right now, going to sleep, tomorrow will refine to realy get stable results in cinebench and other stress tests. As I said not realy interested in realy pushing to 3.9 or 4.0, not my intention. I will eventualy render a video em Davinci Resolve, but when that happens I will just let stock setting take over or if I can refine the results of this 3600mhz OC, I will use it.

Soon will post more results.

Thanks to all of you who helped thus far. I realy learned form you guys. Realy, thank you. Not done yet will post the stress test results, until then...

 

[Socko001]

Yep it tired me out, browsing the web it started giving me BSOD, I used a point of recovery and I think it fixed... Sigh, I'm tired and scared to go foward. Thanks everyone for the help and the knowledge you guys passed to me...

Bought a heatsink to put it in the board perhaps it will help me out a little.