Ryzen Overclock thread. JUST OC results, and hardware/settings

[Agent-47]

My vcore doesn't seem to droop much or at all as reported by Aida64. I'm using Auto-LLC settings though.

is that the X370 board or the MSI gaming pro (reading from your sig)? From what I know, AIDA64 does not report the SVI2 TFN voltage at the CPU input pins/VRM output, which is why you donot see any change.

I reckon the VRM is holding me back. I get decent temps even at 1.4v, but the VCORE drops below 1.32 making the system crash out.

 

[Crumpet]

Help a noob out, if my voltage is set at 1.39 in the BIOS but HWinfo is reporting a VID of 1.55v, does that mean I need to drop the voltage?

 

[Agent-47]

Help a noob out, if my voltage is set at 1.39 in the BIOS but HWinfo is reporting a VID of 1.55v, does that mean I need to drop the voltage?

VID is just what voltage the CPU thinks is needed to run stable, this is not of any consequence as this is not used anywhere as an input except BIOS initialisation after which BIOS will switch to the voltage you have assigned.

you need to watch out from VCORE (SVI2 TFN) under the CPU tree in HWiNFO64 at idle and load. thats what the CPU sees hitting it.

 

[Crumpet]

Awesome thank you, so on Auto this thing cranks up to 1.475. Wow Asus runs "safe" on auto considering I get it stable at 1.39 and haven't even tried to go lower yet.

 

[guskline]

Crumpet what cpu cooler are you using?

 

[Crumpet]

Crumpet what cpu cooler are you using?

NZXT Kraken X62 with the AM4 bracket adapter on a Asus Crosshair Hero 6.

Just did a succesful firestrike bench @4.15ghz 1.375vcore. I'm feeling pretty happy right now.

 

[Eric1987]

NZXT Kraken X62 with the AM4 bracket adapter on a Asus Crosshair Hero 6.

Just did a succesful firestrike bench @4.15ghz 1.375vcore. I'm feeling pretty happy right now.

Impressive. I can do 4GHz on 1.375 but windows bootable. How much vcore is safe 100%?

 

[malitze]

1.475v? you are a brave man. I went to 1.425v to get to 3.9Ghz semi-stable (45 mins of AIDA64, and then it crashed, PC was still running) and then did not push any harder.

3.7Ghz seems to be the sweet spot for me at 1.25 BIOS input with LLC1. VCORE droops to 1.2v minimum. also can be achieved via 1.265v BIOS and LLC2-- not sure which is the better option though.

What board are you on?

I'm on an ASUS Crosshair VI. And I don't think I'm that brave considering Auto VCore is 1.46 for 40x and on stock voltage goes above 1.5 for single core boost.

 

[Agent-47]

I'm on an ASUS Crosshair VI. And I don't think I'm that brave considering Auto VCore is 1.46 for 40x and on stock voltage goes above 1.5 for single core boost.

I see. I am on zen 1600. the BIOS does not detect the VID for me. its stays stuck at 1.231v no matter what the freq is. As a result, Auto does not do anything for me as the MB thinks that 1.231v is the limit. So i have to dial in the volts the old fashion way, which i do not mind.

 

[Crumpet]

Figured I should run AIDA64 for actual stability.

It lasted 1 second. Literally.

Bumped up the voltage. It lasted 3 seconds.

But, wondering if the problem is my unsupported RAM I unchecked the ram stress test in AIDA64's stability test and she's off and flying... Guess I may be stuck at 2133 for a while.

Lessons learned: Just because it benches, doesn't mean its stable.

 

[Agent-47]

But, wondering if the problem is my unsupported RAM I unchecked the ram stress test in AIDA64's stability test and she's off and flying... Guess I may be stuck at 2133 for a while.

could still be the CPU, as upping the RAM speed ups the IMC clocks of the CPU and so will always have an effect on the maximum attainable stable OC.

for example, i can reach 3.7 stable with stock voltage of 1.2v and 2133C14, but i need to bump the CPU volts to 1.25 and SoCto 1.05 to attain stability. I have a Samsung B RAM single rank.

 

[Crono]

Figured I should run AIDA64 for actual stability.

It lasted 1 second. Literally.

Bumped up the voltage. It lasted 3 seconds.

But, wondering if the problem is my unsupported RAM I unchecked the ram stress test in AIDA64's stability test and she's off and flying... Guess I may be stuck at 2133 for a while.

Lessons learned: Just because it benches, doesn't mean its stable.

Yeah, I was going to ask if you had run AIDA or had the build running at that speed for an extended period of time. Even with a "golden" sample you'd probably need more voltage to get it completely stable above 4GHz.

 

[Crumpet]

Yeah, I was going to ask if you had run AIDA or had the build running at that speed for an extended period of time. Even with a "golden" sample you'd probably need more voltage to get it completely stable above 4GHz.

15 minute stability test complete at what seems to be a "safe" OC for my system.

4.1ghz @1.395v vcore

It was the SOC voltage that was causing me bother, upped it to 1.05v and she flew like a dream.

(It almost sounds like I know what i'm doing... O-o, shhh, don't tell anyone)

Done for the night now, but i'll get her back up to my 4.15 or higher tomorrow.

 

[Soulkeeper]

Help a noob out, if my voltage is set at 1.39 in the BIOS but HWinfo is reporting a VID of 1.55v, does that mean I need to drop the voltage?

The 1.39 is your base clock voltage.
When it runs above the base clock or at turbo it goes up (that 1.55v is likely really happening).
IE: if I set ~1.3v it'll run at 1.416v even with turbo disabled and the freq locked at 4GHz

I'm currently using a +0.06v offest, which results in 1.4v at 4GHz
If I manually set 1.4v, it considers it a 0.175v offset

very strange bios behavior, hopefully this is made more clear later

 

[guachi]

R7-1700 3.7 GHz @ 1.25v

Hyper 212+ cooler

Gigabyte Gaming 3 motherboard

Max temp @ 100% on handbrake 65-67c with fans at about 60% (higher speed fans are louder but not really cooler).

I think I could do 3.8 @ 1.25v. Some folks have and I think, at least for cooler 24/7 operation, that 3.8 is a good speed. Though there are times I've set the CPU to 3.0 GHz in RyzenMaster at 1.1v and set it to shut itself off when done. I think that's more efficient. It's slower, but it uses less total energy.

 

[Agent-47]

So its the easter weekend. What better way to spend it than play with my new toy.

I have been toying with undervolting ryzen for the last 10 hour to try an understand the stability of ryzen with temp. This is what I did. I know I can hit stable 3.7 GHz on 1.25 v MB Vcore (1.2v after VDroop) at 57C (60% rpm). So I sat around with my fan controller and adjusted the RPM manually to hold the temps constant at a desired value. The minimum voltage required to be stable for 1 hr stress test @3.7ghz is listed below:
57C-> 1.25 v (1.2 v after vdroop)
55C -> 1.2375v (1.190v after vdroop)
52c -> 1.225 v (1.180v after vdroop)
50C -> 1.2125v (1.173v after vdroop)
48C -> 1.2 v (1.167 v after vdroop) 20 min stress stable.

I am impressed and upset at the same time. Should have gotten a better cooler!!! But it helps that I have 2 noctua and 2 cougar led fans and its not very loud. Decided to go with 1.225v as it requires 80% fan speed and barely audible.

But will try out 3.8 and 3.9 GHz stability as well tomorrow.

Just ordered couple of copper heatsink and a 40 mm noctua fan for the SoC VRMs which are left naked in the MSI b350m mortar to see if the Vdroop decreases.

 

[guskline]

Crumpet, I'm sticking with 4 Ghz. Voltage and heat too high at 4.1 on all cores.

My settings are Offset mode for vcore +,06824v, ram running at 3200 CL 14 with dram at 1.39V

 

[Agent-47]

The minimum voltage required to be stable for 1 hr stress test @3.7ghz is listed below:
57C-> 1.25 v (1.2 v after vdroop)
55C -> 1.2375v (1.90v after vdroop)
52c -> 1.225 v (1.80v after vdroop)
50C -> 1.2125v (1.73v after vdroop)
48C -> 1.2 v (1.67 v after vdroop) 20 min stress stable.

here are the minimum volts for 3.8 Ghz. (stress test 1 hr)

60C-> 1.3375v LLC1 (1.291v after vdroop)
55C-> 1.3250v LLC1 (1.275v after vdroop)
51C-> 1.3125v LLC1 (1.267v after vdroop) -- only stable for 15 minutes.

Decided not to test 3.9 Ghz as it requires 1.425 v LLC for stable and it makes me uncomfortable.

 

[Crumpet]

Crumpet, I'm sticking with 4 Ghz. Voltage and heat too high at 4.1 on all cores.

My settings are Offset mode for vcore +,06824v, ram running at 3200 CL 14 with dram at 1.39V

With your faster RAM we're probably running about the same overall speed anyway. There's not much of a difference in benchmark scores, I think I got 2 higher FPS in firestrike physics with the extra 100mhz

 

[3DVagabond]

What PSU's people are using would be useful info.

 

[DooKey]

What PSU's people are using would be useful info.

Take a look at my sig.

 

[3DVagabond]

Take a look at my sig.

That wasn't aimed at one particular person.

 

[Crumpet]

Corsair RM1000X

And it has more coil whine than a pig on a spring. (its going back on Tuesday)

If the replacement is just as bad i'm going back to EVGA, they were GOOD.

 

[IEC]

I use eVGA units. Mostly G2/P2.

I have a single Corsair RM650x refurb and it has (relatively mild) coil whine.

 

[Crumpet]

I use eVGA units. Mostly G2/P2.

I have a single Corsair RM650x refurb and it has (relatively mild) coil whine.

I hear this coil whine over the top of my 5.1 surround sound