Thentilian's R5 1600 OC thread

[Keith Myers]

[Edit] Corrected Geardown setting

That enables me to run 3333 Mhz at 14-14-14-30-52-1T Geardown disabled for the G. Skill F4-3600C16D-16GTZ memory kit I'm running.

[Edit] Too late to edit post directly.

 

[Keith Myers]

I have my SOC set to 1.06v manually with no LLC. I have .13750 offset on CPU with 38.00 multiplier, and LLC 1, with optimized VRM settings. Maybe I'll play with those settings, I just didn't want to go aggressive and accidentally fry something!

You won't fry anything. What is your Core Voltage (SVI2 TFN) under HwInfo64 when under stress testing? From your signature, it looks like you have Vcore set at 1.41V for 3.9 Ghz. Try dropping to stock 1.35V with +0.01875V offset and LLC4 or LLC5. You should end up at Vcore SVI2 TFN of 1.3506V. See if that is stable under P95 or OCCT Linpack AVX. If not, try bumping the offset to 0.025V and try again. You should be safe up to 1.40V loaded at LLC5 read from Vcore SVI2 TFN.

What does Vsoc SVI2 TFN read under load on manual. It is probably down around 0.95 - 0.99V. Try LLC3. That should help the memory overclock along with setting BGS = On.

 

[Thentilian]

Both Auto and Optimized in the Power Phase settings utilize "phase shedding" per Elmor. That is, it turns off VRMs if the amp capacity is not needed. Unfortunately it does not respond fast enough to be stable when transitioning from idle to a high power state as when benchmarking, gaming or stability testing. Having them on all the time adds maybe 1-2° C. and maybe 1-2W in idle. Well tested by a lot of expert users. Also higher LLC settings have been proven with both digital multimeter measurements and O'scope screenshots to not cause overshoot or harm on the Prime X370 Pro. May not be the case with other motherboards. Extensive testing has been done at high load and stability testing to show that the voltage measured under the CPU socket on the caps, after the VRM output, is closest to what is set in the BIOS and read out by HwInfo64 SVI2 TFN readouts when LLC4 or LLC5 is used. This has been verified by several posters of high technical skill over at OC.net. HwInfo64 SVI2 TFN readouts are directly from the telemetry internal to the VRMs themselves. It is the closest to the actual delivered voltages other than directly measuring at the input to the CPU.

Ok, thanks. That's a huge help figuring this stuff out. It's all new BIOS stuff to me coming from a Phenom II. I have built plenty of Intel systems for people, but I never OC them, partly because I just don't know enough about that aspect of things when it comes to PCs. I have a new rig of my own that I can blow up (hopefully not ) instead of learning on a friends or blowing a clients rig up.. bad for business!

I did notice going to LLC 3 that I could lower the offset voltage quite a bit and it is almost stable. I just need to find the right new settings. And if leaving them on extreme is safe with THIS board, then I'm ok with a bit more temps and watts while idle if it means my OC is gonna be stable.

Another question, I'm confused on which readout is the actual CPU voltage in HWInfo. There is the CPU Core Voltage, (SV12 TFN) and lower down there is VDDCR CPU that reads higher... I assumed the VDDCR CPU voltage is the correct one from using LLC but the CPU Core Voltage via the SV12 TFN reads much lower. So, which one should I used when judging voltage that is actually going to my CPU? I have more headroom if the voltage is actually based on the CPU Core Voltage from the SV12 TFN. As you said, the SV12 TFN is from the VRM, so yeah I've been confused on which one was right and I erred on the side of caution with the higher voltage reading from the VDDCR CPU instead.

 

[Keith Myers]

Only look at SVI2 TFN reaouts

Ok, thanks. That's a huge help figuring this stuff out. It's all new BIOS stuff to me coming from a Phenom II. I have built plenty of Intel systems for people, but I never OC them, partly because I just don't know enough about that aspect of things when it comes to PCs. I have a new rig of my own that I can blow up (hopefully not ) instead of learning on a friends or blowing a clients rig up.. bad for business!

I did notice going to LLC 3 that I could lower the offset voltage quite a bit and it is almost stable. I just need to find the right new settings. And if leaving them on extreme is safe with THIS board, then I'm ok with a bit more temps and watts while idle if it means my OC is gonna be stable.

Another question, I'm confused on which readout is the actual CPU voltage in HWInfo. There is the CPU Core Voltage, (SV12 TFN) and lower down there is VDDCR CPU that reads higher... I assumed the VDDCR CPU voltage is the correct one from using LLC but the CPU Core Voltage via the SV12 TFN reads much lower. So, which one should I used when judging voltage that is actually going to my CPU? I have more headroom if the voltage is actually based on the CPU Core Voltage from the SV12 TFN. As you said, the SV12 TFN is from the VRM, so yeah I've been confused on which one was right and I erred on the side of caution with the higher voltage reading from the VDDCR CPU instead.

The VDDDR CPU voltage is read from the motherboard EC or embedded controller chip, the ITE8665E. It is about 3 times removed from reality in reporting the true cpu voltage. It has 20mv resolution accuracy to begin with because of its low LSB in the A-D conversion. It is also not sampled very often. Best to just ignore any voltage or currents coming from the EC chip. The only true voltage and currents are from the HwInfo64 SVI2 TFN readouts.

 

[Thentilian]

The VDDDR CPU voltage is read from the motherboard EC or embedded controller chip, the ITE8665E. It is about 3 times removed from reality in reporting the true cpu voltage. It has 20mv resolution accuracy to begin with because of its low LSB in the A-D conversion. It is also not sampled very often. Best to just ignore any voltage or currents coming from the EC chip. The only true voltage and currents are from the HwInfo64 SVI2 TFN readouts.

Ok, so I have way more voltage headroom than I thought. Probably why I wasn't getting anything stable past 3.8Ghz. I'm currently stable with 1.356v according to the SV12 TFN voltage readout. That's so confusing to have two readouts like that, that are SO far apart. The VDDCR CPU says I hit a max voltage of 1.439, and was why I stopped going higher in voltage. While the max voltage according to the SV12 TFN says 1.356. No higher than what is says the voltage should be with the offset in the BIOS.

EDIT:
Ok, so I cranked LLC up to 4, with extreme power phase, and using an offset of .150000. I was able to run IBT on a 'high' setting and was stable for the first time so far. Here's a shot of the HWInfo:


See the highlighted voltage, those numbers looked scary, so I didn't want to go higher, but under the SV12 TFN I'm not even at 1.4 volts yet. Hit 1.387v max. I'll need to confirm stability before moving on. Who knows might be able to get close to 4Ghz with these new settings.

 

[Keith Myers]

That is more of what I expected

Ok, so I have way more voltage headroom than I thought. Probably why I wasn't getting anything stable past 3.8Ghz. I'm currently stable with 1.356v according to the SV12 TFN voltage readout. That's so confusing to have two readouts like that, that are SO far apart. The VDDCR CPU says I hit a max voltage of 1.439, and was why I stopped going higher in voltage. While the max voltage according to the SV12 TFN says 1.356. No higher than what is says the voltage should be with the offset in the BIOS.

EDIT:
Ok, so I cranked LLC up to 4, with extreme power phase, and using an offset of .150000. I was able to run IBT on a 'high' setting and was stable for the first time so far. Here's a shot of the HWInfo:
View attachment 193728

See the highlighted voltage, those numbers looked scary, so I didn't want to go higher, but under the SV12 TFN I'm not even at 1.4 volts yet. Hit 1.387v max. I'll need to confirm stability before moving on. Who knows might be able to get close to 4Ghz with these new settings.

OK, that is more of what is to be expected. Everything looks fine to me with the SVI2 TFN CPU voltage. You might be able to push closer to 4 Ghz. Just be aware .... Ryzen hits a VERY steep voltage wall to get stability with incremental increases in CPU clocks as you get closer to 4 Ghz. Most people find it just isn't worth it in heat load and power usage cost for incremental performance increases. Once you get your CPU OC stable ~ 3.9-4.0 Ghz, it is then time to start working on your memory clocks. That gives the most performance increase for incremental steps in clock frequency or better sub-timings.

4 Ghz is generally accepted to be the max attainable CPU clocks on Ryzen because of the 14nm LPP process technology. That likely won't change until new process technology and Ryzen2. Also be aware that increasing the memory clocks increases stress on the IMC unit of the chip and you might find that with increased memory clocks that you need to bump Vcore again to be stable at your last stable CPU clock.

 

[Keith Myers]

You should be able to get your Corsair RAM to 2933 fairly easily with 0810 - 0902 BIOS. I would try the DOCP 2933 XMP settings and default timings for the sticks. If ProcODT Auto isn't stable, try 68 ohms. Vdimm ~ 1.37-1.40V.

 

[Thentilian]

OK, that is more of what is to be expected. Everything looks fine to me with the SVI2 TFN CPU voltage. You might be able to push closer to 4 Ghz. Just be aware .... Ryzen hits a VERY steep voltage wall to get stability with incremental increases in CPU clocks as you get closer to 4 Ghz. Most people find it just isn't worth it in heat load and power usage cost for incremental performance increases. Once you get your CPU OC stable ~ 3.9-4.0 Ghz, it is then time to start working on your memory clocks. That gives the most performance increase for incremental steps in clock frequency or better sub-timings.

4 Ghz is generally accepted to be the max attainable CPU clocks on Ryzen because of the 14nm LPP process technology. That likely won't change until new process technology and Ryzen2. Also be aware that increasing the memory clocks increases stress on the IMC unit of the chip and you might find that with increased memory clocks that you need to bump Vcore again to be stable at your last stable CPU clock.

You should be able to get your Corsair RAM to 2933 fairly easily with 0810 - 0902 BIOS. I would try the DOCP 2933 XMP settings and default timings for the sticks. If ProcODT Auto isn't stable, try 68 ohms. Vdimm ~ 1.37-1.40V.

I have my OC on the CPU stable with those settings. I'm currently working on 2933 with my RAM. I wasn't able to push it that much before, but with the VRM set to extreme and LLC 4 on both CPU and SOC, it's more stable and allowing me to clock it with 16-18-18-36 CR 1 timings. I'm pushing 1.395 volts so far, it's not quite stable, but almost there. It hasn't passed a custom IBT with 12024 MB used yet, but passes the standard 'High' profile.

 

[Keith Myers]

Try with sub-timings different than Auto

I have my OC on the CPU stable with those settings. I'm currently working on 2933 with my RAM. I wasn't able to push it that much before, but with the VRM set to extreme and LLC 4 on both CPU and SOC, it's more stable and allowing me to clock it with 16-18-18-36 CR 1 timings. I'm pushing 1.395 volts so far, it's not quite stable, but almost there. It hasn't passed a custom IBT with 12024 MB used yet, but passes the standard 'High' profile.

You are probably going to have to resort to sub-timing tweaks that are different than Auto to get Hynix stable at 2933 or greater. Read through the official-amd-ryzen-ddr4-24-7-memory-stability-thread thread.

 

[Thentilian]

You are probably going to have to resort to sub-timing tweaks that are different than Auto to get Hynix stable at 2933 or greater. Read through the official-amd-ryzen-ddr4-24-7-memory-stability-thread thread.

Thanks for the link. I have already been reading this site. I've been Trying timings of 16-18-18-36 CR1, with most of the sub-timings on auto. Just can't quite seem to get it stable. It seems stable and boots normally, but won't pass IBT on very high. I've had the kit as high as 1.41v but I'd rather not push higher than that. I'll continue to tinker and read further to see if anything will work.

 

[DaveB]

Ryzen 5 1600 now $169 at Microcenter with an additional $30 off any motherboard.

 

[Thentilian]

Ryzen 5 1600 now $169 at Microcenter with an additional $30 off any motherboard.

Yeah, just saw that. I got mine for 189. Always a better deal after you buy yours! Just the way it goes sometimes.

 

[DaveB]

I paid $189 too. Just thought someone else could benefit.

 

[Thentilian]

I paid $189 too. Just thought someone else could benefit.

Yeah, of course. It's a good price even at 189. 169 it's a steal. Good CPU. Yeah, it's not really an awesome overclocker, but it does what I need it to, and then some. And the performance to value is excellent.