Question Going for my first overclock what should i be aiming for with my hardware ?

[fendertele]

Hey everyone im planning on doing my first overclock ive been doing a lot of reading up and got a little confused with conflicting approaches and opinions on safe volts etc.

This is my setup:

ASRock B450M-HDV R4.0
Amd Ryzen 3 2200g
Corsair CMK16GX4M2B3200C16 Vengeance LPX 16 GB (2 x 8 GB) DDR4 3200 MHz
Corsair CP-9020102-UK CX550M 550 W 80+ Bronze Certified ATX Po Wer Supply Uni

The three things im most concerned about is.

People keep mentioning i showed check the VRM temp however i cant see any VRM temp in bios or HWinfo etc... despite reading up the motherboard does have it ?

Even if i get a stable OC and everything seems good no crashes etc... that im using too high of a volt so being able to know the maximum volt i should use would be good so i can drop the OC to keep within this.

And the final one OC the CPU and IGPU or just the IGPU and RAM as i've read this APU gets best results with a IFPU overclock along with RAM however the RAM timings need to be tight.. which is where im confused i got 3400 at 1.4 16,18,18,18,38 however i dont know if thats tight or not.

Any tips or even just known working settings would be great.

thanks.

 

[DrMrLordX]

You've got Raven Ridge, which is similar enough to Summit Ridge that I might be able to help you out.

First off, you've only got a 4c/4t chip. The TDP for the CPU itself is actually going to be lower than the 65W rating since I think the iGPU is included in the TDP. I would not worry too much about the VRMs. The worst part about them is that they may introduce some pretty significant vdroop and other inconsistencies with clean voltage delivery to the CPU. You're not going to get a top OC with that board. Given the CPU . . .it's not a big deal.

Anyway. What operating system do you run? If you use Ryzen Master, you can probably experiment a bit and make things easier.

The max voltage I would recommend with stock cooling would be in the range of 1.35v actual. Use a utility like CPU-z to detect what voltage you're giving the CPU actually during a workload. Note that due to your VRMs, there's going to be a lot of vdroop. Overshoot could get pretty bad as well. I would stay well away of the "normal" voltage limit of 1.4v vcore for CPUs of this generation.

Worry about the iGPU and RAM later. If you're going for an iGPU OC as well, you may actually stress out for VRMs. They are not beefy.

 

[UsandThem]

I thought you already went your first overclock?

http://www.portvapes.co.uk/?id=Latest-exam-1Z0-876-Dumps&exid=threads/random-beep-noise-coming-from-pc-speakers.2578488/

 

[fendertele]

I thought you already went your first overclock?

http://www.portvapes.co.uk/?id=Latest-exam-1Z0-876-Dumps&exid=threads/random-beep-noise-coming-from-pc-speakers.2578488/

yeah that was a an amateurish attempt i basically just increased volts to up the speed without looking into it more.... so this is technically me doing a proper one now with timings etc after a bit of long long reading,.

 

[fendertele]

You've got Raven Ridge, which is similar enough to Summit Ridge that I might be able to help you out.

First off, you've only got a 4c/4t chip. The TDP for the CPU itself is actually going to be lower than the 65W rating since I think the iGPU is included in the TDP. I would not worry too much about the VRMs. The worst part about them is that they may introduce some pretty significant vdroop and other inconsistencies with clean voltage delivery to the CPU. You're not going to get a top OC with that board. Given the CPU . . .it's not a big deal.

Anyway. What operating system do you run? If you use Ryzen Master, you can probably experiment a bit and make things easier.

The max voltage I would recommend with stock cooling would be in the range of 1.35v actual. Use a utility like CPU-z to detect what voltage you're giving the CPU actually during a workload. Note that due to your VRMs, there's going to be a lot of vdroop. Overshoot could get pretty bad as well. I would stay well away of the "normal" voltage limit of 1.4v vcore for CPUs of this generation.

Worry about the iGPU and RAM later. If you're going for an iGPU OC as well, you may actually stress out for VRMs. They are not beefy.

Well my plan was to keep the cpu at stock speed or just a slight increase to 1.37 and get my igpu high as i had read this is the way to go with my Apu high ram speed and tight timings to get the best out of the igpu and a leave the cpu as is or just a slight increase.



these are the temps i can see in hwinfo.

cpu tctil tdle
cpu soc
apu gfx
motherboard
cpu
auxillary
auxtin1
auxtin2
auxtin 3
cpu peci
drive temp
gpu temp.

If i knew what each temp shouldnt go above along with all the volts safe settings i could just plug it all into the mobo and then work my overclock round it safe knowing whatever i finally achieve wont damage anything. if that makes sense ?

 

[DrMrLordX]

@fendertele

First off, one of the aux temps is your VRM temp. It would really help if you could show us a screenshot of those temps to help us figure out which one is your VRMs. Usually, at desktop, VRMs on a cheap board with little board cooling will be sitting at 40-50C.

Also you wouldn't necessarily raise vcore if you plan on only overclocking the iGPU. That's going to go off different voltages, notably vSoC. It would help if you could show us a screenshot of your UEFI voltage options.

 

[fendertele]

You've got Raven Ridge, which is similar enough to Summit Ridge that I might be able to help you out.

First off, you've only got a 4c/4t chip. The TDP for the CPU itself is actually going to be lower than the 65W rating since I think the iGPU is included in the TDP. I would not worry too much about the VRMs. The worst part about them is that they may introduce some pretty significant vdroop and other inconsistencies with clean voltage delivery to the CPU. You're not going to get a top OC with that board. Given the CPU . . .it's not a big deal.

Anyway. What operating system do you run? If you use Ryzen Master, you can probably experiment a bit and make things easier.

The max voltage I would recommend with stock cooling would be in the range of 1.35v actual. Use a utility like CPU-z to detect what voltage you're giving the CPU actually during a workload. Note that due to your VRMs, there's going to be a lot of vdroop. Overshoot could get pretty bad as well. I would stay well away of the "normal" voltage limit of 1.4v vcore for CPUs of this generation.

Worry about the iGPU and RAM later. If you're going for an iGPU OC as well, you may actually stress out for VRMs. They are not beefy.

@fendertele

First off, one of the aux temps is your VRM temp. It would really help if you could show us a screenshot of those temps to help us figure out which one is your VRMs. Usually, at desktop, VRMs on a cheap board with little board cooling will be sitting at 40-50C.

Also you wouldn't necessarily raise vcore if you plan on only overclocking the iGPU. That's going to go off different voltages, notably vSoC. It would help if you could show us a screenshot of your UEFI voltage options.

Hi thanks i can defo print my temp is see off of hwinfo and post it not sure how to take a screen shot in bios though what i will do is take a pic off my phone and hopefully it looks decent enough.

If you gimme 5 mins ill be right back with it.

So yeah from what i can tell when its all auto my vcore sits at 1.4v so i dropped that to say 1.37 and set the cpu to 3700 would this then allow a much greater igpu overclock?

anyways back in 5.

 

[fendertele]

edit

 

[fendertele]

 

[fendertele]

edit

 

[extide]

Wow, if those 3 AUXTIN1-3 temps are your VRM, then that's all she's got. I wouldn't push it any more than that. That is surprising that they would be that hot sitting idle, though. Might want to reach in there and see if they are actually that hot to the touch...

 

[fendertele]

 

[fendertele]

Wow, if those 3 AUXTIN1-3 temps are your VRM, then that's all she's got. I wouldn't push it any more than that. That is surprising that they would be that hot sitting idle, though. Might want to reach in there and see if they are actually that hot to the touch...

Edit : The Auxtins dont move so was thinking a fake reading, the Auxillery does move but were talking 1c most... so at least its a moving reading could this be the VRM ?

 

[extide]

But those auxtin temps dont move at all ? surely a fake reading ? are we not thinking auxiliary is the vrm temp ?

It's hard to say -- the Auxiliary is at 26C which is basically room temp. I would reach in there and touch the VRM's (ground yourself) and see. If they burn you, then it's the 112/113C temps, lol, if it's not hot at all, then it's probably the 26C one.

 

[DrMrLordX]

@fendertele

Okay, first off, those AUXTIN temps all seem bogus. The first AUX temp is basically case temp, probably. What you should do, before you overclock anything, is run a benchmark like CBR20 or Prime95 and show us what happens to the AUXTIN temps during the run. I'm guessing they won't change, or they may go to some crazy reading that makes no sense.

Since your board has no VRM cooling, you would probably need an IR thermometer to get temp readings off them.

 

[fendertele]

btw just so its known my pc isnt overclocked at all right now.... just the ram set to xmp profile to allow me to get 3200 as it defaults at 2200 without the xmp profile ill go run the prime 95 and be back soon.

 

[fendertele]

Used Aida as i had it installed.... the Auxillary moved 1c... the Auxtins didnt move however i think at one point or another ive seen one of them move 1c also however could just be my eyes playing tricks but i know for sure Auxillary moves 1c from time to time.

 

[fendertele]

Eariler when i was tinkering i had my pc set as follows.

Ram xmp profile to get 3200 1.1 SOC and 1.350 DRAM
Cpu 1.4v @ 3700
Gpu 1.3gfx @ 1600

And ran aida it didnt crash and the highest temp it hit for cpu was 71c... is this a good indication for how high VRM would be getting as under stress 71 aint too bad ?

 

[DrMrLordX]

@fendertele

Okay. The AUXTIN temps are garbage. The first AUX temp is your case temp.

Given your temps during an AIDA run, which is not that intensive, I would say that you need to be very careful with overclocking anything, including the iGPU. You don't have much thermal headroom left. Raven Ridge shouldn't go over 72C I don't think. Your CPU and SoC temps are pretty high, and you would have to raise vSoC (NOT vcore) to overclock your iGPU. You can try running your RAM faster or with tighter timings, but i can't promise you it'll work well. That could improve graphical performance a bit.

 

[fendertele]

@fendertele

Okay. The AUXTIN temps are garbage. The first AUX temp is your case temp.

Given your temps during an AIDA run, which is not that intensive, I would say that you need to be very careful with overclocking anything, including the iGPU. You don't have much thermal headroom left. Raven Ridge shouldn't go over 72C I don't think. Your CPU and SoC temps are pretty high, and you would have to raise vSoC (NOT vcore) to overclock your iGPU. You can try running your RAM faster or with tighter timings, but i can't promise you it'll work well. That could improve graphical performance a bit.

Thats disappointing.... strange that when i oc the cpu and gpu to 3700 and 1600 and ran aida and furmark the max my temp ever reached was 72c...... my pc idles at 34c.

ill defo be ordering a IR thermo to test vrm.

So just so i have this right a fully overclocked pc cpu,igpu and ram all being stress tested be it memtest,aida,furmark or benchmark hitting max 72c is a bad thing ? i was considering in the 90s being a scary temp and 72c not bad for full stress testing?

 

[DrMrLordX]

Thats disappointing.... strange that when i oc the cpu and gpu to 3700 and 1600 and ran aida and furmark the max my temp ever reached was 72c...... my pc idles at 34c.

If you didn't increase voltage, then that is normal. Though your default voltages may be very high which is why you are having such high temperatures. And actually looking back at it, it looks like the max temp for Ryzen 1xxx/Raven Ridge was 95C, not 72C. 72C was for older CON cores.

Did you say your default vcore was 1.4v during benchmarks? Use CPU-z to monitor this. Nothing else. Make sure you monitor voltage during a benchmark, since idle voltage doesn't mean much. 1.4v is actually too high. You can afford to lower that if that is in fact your voltage during a benchmark.

ill defo be ordering a IR thermo to test vrm.

Careful, they're kinda expensive right now.

So just so i have this right a fully overclocked pc cpu,igpu and ram all being stress tested be it memtest,aida,furmark or benchmark hitting max 72c is a bad thing ? i was considering in the 90s being a scary temp and 72c not bad for full stress testing?

I do not recommend Furmark at this point.

If you want to stress test your iGPU, I would recommend 4k-optimized Unigine Superposition or 1440p GFXBench Aztec. Run them maybe 3 times to see if they'll clear.

For testing memory, I personally like y-cruncher, though that's only once you've established that the CPU is stable. Memtest86 is fine, and there's another one, but I keep forgetting the name. You can also try Prime95 Blend for awhile, but again, only after you have established that the CPU is stable.

For CPU, I prefer Prime95 SmallFFTs for that generation of CPU.

 

[fendertele]

If you didn't increase voltage, then that is normal. Though your default voltages may be very high which is why you are having such high temperatures. And actually looking back at it, it looks like the max temp for Ryzen 1xxx/Raven Ridge was 95C, not 72C. 72C was for older CON cores.

Did you say your default vcore was 1.4v during benchmarks? Use CPU-z to monitor this. Nothing else. Make sure you monitor voltage during a benchmark, since idle voltage doesn't mean much. 1.4v is actually too high. You can afford to lower that if that is in fact your voltage during a benchmark.



Careful, they're kinda expensive right now.



I do not recommend Furmark at this point.

If you want to stress test your iGPU, I would recommend 4k-optimized Unigine Superposition or 1440p GFXBench Aztec. Run them maybe 3 times to see if they'll clear.

For testing memory, I personally like y-cruncher, though that's only once you've established that the CPU is stable. Memtest86 is fine, and there's another one, but I keep forgetting the name. You can also try Prime95 Blend for awhile, but again, only after you have established that the CPU is stable.

For CPU, I prefer Prime95 SmallFFTs for that generation of CPU.

I want to thank you for your patient replies after we stop speaking i researched why my temps were high unclocked... this was due to a setting that is called core boost which would give me crazy settings when required like a temp oc constantly when doing things.

This is probably why my cpu was at 1.4v by default and why i would see my cpu clock hit 3700 at times... so i turned off c state and the boost option in bios and did my overclock from there.

I did the ram first getting it up from the 3200 xmp profile to 3400 by just setting everything in timings to default when i compared both in membench there was a tiny 5 second increase but an increase non the less... and of course ran memtest for 4 fails with no issues.

Then i did a very tiny cpu underclock by fixing the frequency to 3700 as this was what my pc would bump up to anyways when understress default but lowered from 1.4v to 1.375v.

Next i got my IGPU up to 1600 with 1.11 soc and 1.175 gfx volt.

Ran unique benchmarks at 1080p high with no crashes with fps between 10-15 througout and highest temp 68c.

Ran Aida stress test for 30 minutes with everything selected bar the GPU and Drive highest temp 69c.

When idling my CPU temp hits around 34c and normal browsing movies etc a high of around 42c.

I also took an old CPU fan and installed it in silent mode and placed it facing my VRM modules well i think its the 3 cyclindars with the 4 blocks right next it at the side of the cpu ? well i picked them there is other cylenders all over the board but no blocks next to them.

ITs probably not making a hell of a difference as i have it on a low setting as it was noisy... but my cpu fan is in performance mode.

So so far looks good ill also run a few of those other test just to be sure specifically prime to be very sure its good.

Those volts look more reasonable long term ?

CPU 3.7 @ 1.375

GPU 1600 @ GFX 1.275 Soc @1.11

Ram 3400 @ Dram 1.360

 

[DrMrLordX]

First, putting a fan over the VRMs is probably a good idea.

Secondly, your temps look okay.

Third, your voltages look like they're in a safe range.

If your'e happy with the OC, just lock it in and go with that. You can possibly tweak for more performance on your RAM (see: Ryzen Memory Calculator), but the improvements to be had will be minimal. Tighter timings won't help your iGPU that much, and I doubt you'll get much more clockspeed out of that RAM on that IMC. Anything over DDR4-3466 is nearly impossible. I would not recommend shooting for more than 3.7 GHz on your CPU since the amount of power it takes to push past that speed on Summit Ridge and Raven Ridge will not be worth the trouble. You might be able to tune for lower voltage on your CPU if you want to try. Just slowly lower voltage until you start seeing instability, then bump it back up a few millivolts.