Thentilian's R5 1600 OC thread

[Thentilian]

I just bought my R5 1600 yesterday with an Asus Prime X370-Pro motherboard. I am able to get 3.8Ghz easily with it, I'm just trying to find the lower end of what voltage I need. So far it's looking like I might be able to get away with 1.29 or 1.3 volts, which makes me pretty dang happy... Only there is a catch...

As I have been using Asus real bench to find just that right amount of volts to keep it stable, when I go back into BIOS to change it up a notch, sometimes I come back into windows and the multiplier is set to 15.5 instead of 38. Only way I found to fix this is setting the profile to 'Asus Optimized', but then it tweaks the voltage itself, and hits 1.4+ volts, which isn't needed for this CPU to be stable. I set the profile back to 'Normal' and then it's a crap shoot if the multiplier stays at 38 instead of 15.5.

I am at a loss as to why this is happening. Any ideas would be appreciated.

EDIT: I'm using the latest 0801 BIOS. It was the first thing I did before I even put it into my case. I hooked it up on my test bed (AKA the motherboard box) to make sure everything was ok.

 

[PetrolHead]

If I'm not mistaken, OC3D experienced similar issues with their Asus X370 board. It's a bug in the BIOS and has to do with setting the voltage manually. If you set the voltage using the voltage offset instead of manually setting it where you want it, then the multiplier should stay where it is.

 

[Johan45]

It wasn't just ASUS boards having this issue. With AMDs new SOC structure there have been some odd outcomes that I have read over the net. With this multiplier drop being one of them

 

[Thentilian]

If I'm not mistaken, OC3D experienced similar issues with their Asus X370 board. It's a bug in the BIOS and has to do with setting the voltage manually. If you set the voltage using the voltage offset instead of manually setting it where you want it, then the multiplier should stay where it is.

Yeah, I'm using offset now to OC and it's working fine... Annoying as hell, but it works. I tried reflashing the BIOS, and it had the same issues. So it's offset OC for now until they fix this. I'm almost stable at 3.8GHz. Wasn't able to make 2 hours of Asus Realbench without it a crash yet.

It wasn't just ASUS boards having this issue. With AMDs new SOC structure there have been some odd outcomes that I have read over the net. With this multiplier drop being one of them

I think I might need to bump my SOC up to 0.925 or 0.950 to get stability. It only crashes during Realbench so far. It's handled gaming and videos so far. I think it might be a bit of a temp issue (only during Realbench) as well, as I only have the stock cooler until I get my AM4 bracket from Silverstone for my AIO water cooler. I hit 74 to 75C on the CPU Diode, but the CPU temp reads 65 to 68C at most. I'm not sure which one is more important to keep below 75C.


All I know is the learning curve is steep, since I haven't really dealt with many new BIOS, even when building Intel rigs, since I never OC a clients build... at least none have asked for that so far. Some prefer to do it on their own, fine by me, they blow it up it's on them! Anyways, last PC I personally owned was a Phenom II 965 based system, so the BIOS in that thing had all of 2 pages for NB, CPU, and another for RAM timings. This BIOS has so many options, custom P-States, etc, etc... I could really get in trouble messing around with some of those functions!

Anyhow, once I'm fully stable I'll post the numbers. For now I may need to back off to 3.7 Ghz, but we'll see. I know I'm super close at this point to being stable.

EDIT:

So far it's going well with Aida 64 stress test. The temps are not as good as I would like, but under normal conditions I see high 50's and low 60's when gaming, and I do not plan on doing heavy video editing until I get my water cooler bracket.

The Vcore goes from 1.297 to 1.330 while it's stress testing. I used an offset of 0.05625, hopefully this is as much as I'll need to get it stable since this stock cooler is about tapped out on me.

 

[Thentilian]

I was able to get stability with an offset of 0.06250 for a Vcore of 1.3V. I then worked on OC of my Corsair Vengeance RAM. It's XMP profile is 2400Mhz with 14-16-16-31 timings, I have overclocked it to 2800Mhz with 12-14-14-27 CR1 with a bump to 1.35V. The RAM is running cool and stable as well. Memtested while I slept for 8+ hours no errors. I tried for 2933, but the timings had to be 16CL for it to even boot to BIOS, so 2800 with tighter timings wins out. Here is a CPU-Z validation link: https://valid.x86.fr/maqvkz

 

[Johan45]

Are you stress testing with AIDA64?

 

[Thentilian]

Are you stress testing with AIDA64?

Yes, and Asus Realbench after that. I don't like Prime 95, as it loads way more heat than other stress tests. I game mostly, and do some streaming. I did a 720p 60FPS stream for almost 2 hours and it was smooth and no hiccups. I streamed Mass Effect: Andromeda. So it's a pretty taxing game on the CPU.

I did have a couple of BDOS yesterday. I'm thinking the RAM isn't quite as stable as the memtest led me to believe. Maybe not the RAM itself but the speed might of made the CPU unstable. It's hard to pin down since it's been fine with Asus Realbench. I'm probably just gonna have to do a 24 hour stress test to really see. 8 hours probably wasn't enough.

But this is the fun of overclocking. Always gotta figure out what the problem is and tweak it until it's just right.

 

[Johan45]

It is possible that what ever you were doing was just a bit more taxing than normal. Memtest is pretty good at weeding out memory errors so it could be CPU related. Programs like P95 do have there place and I still use them for stability testing to ward off thos "random" BSODs. I'm also a firm believer in to each their own but random unseen errors will eventually corrupt your windows install and require a full make over

 

[Thentilian]

It is possible that what ever you were doing was just a bit more taxing than normal. Memtest is pretty good at weeding out memory errors so it could be CPU related. Programs like P95 do have there place and I still use them for stability testing to ward off thos "random" BSODs. I'm also a firm believer in to each their own but random unseen errors will eventually corrupt your windows install and require a full make over

Yeah, I'm thinking maybe I didn't up the SOC voltage enough, maybe. I don't know, there is WAY more things to look after to overclock than my Phenom II. It's quite the jump in technical terms for me, from what I was used to, and how many different things you need to keep track of now when overclocking these new Ryzen chips.

I think I'll do a burn-in with P95 now that I made some more tweaks. I'll have to do it at night when the ambient temp is way down though. Too hot during the day in the desert where I live. This stock cooler is pretty good for what it is, but until my AM4 bracket shows up for the AIO water cooler I have, I don't want to push it any further, and if it's not stable, I'll just back off the OC til my bracket shows up.

Right now it's a learning curve, so I want to have some testing time under my belt when the bracket shows up, so I can really dig in. I'm not terribly worried if my install gets messed up. I can redo it again. I have my data backed up on a separate drive. It'll just cost me time to get it back up and running if that happens.

EDIT: Oh and I forgot to mention, it's my first really big attempt at OCing RAM past what it's specs are. I'm a bit more worried about that, but I know it's a risk, and I like to tinker. I pushed the RAM timings to 14-16-16-31, still seems to be performing the same as the lower timings. I guess I wouldn't notice much of a difference unless I was using a really memory intensive app. I also upped the SOC to 1.0v. I don't know if it's a bit overkill at this point, but nothing is running hot. The RAM is still cool to the touch even after 6 hours of Memtest hammering it overnight again.

 

[Johan45]

Why don't you list you rig in your Sig so we can see what parts you have. All you gave me were ram timings no speed or make of the ram which can be much more important. When you say lower timings what does that mean ?

 

[Thentilian]

Why don't you list you rig in your Sig so we can see what parts you have. All you gave me were ram timings no speed or make of the ram which can be much more important. When you say lower timings what does that mean ?

I just did my sig the other day, apparently it doesn't show up until I post something new, sorry. RAM is Corsair Vengeance DDR4 2400 14-16-16-31, that's the XMP profile. I have been able to test 2800 with those timings, and a bump in DRAM voltage to 1.35 and the SOC bumped up to 1.0v just to make sure it's stable. I tuned the CPU down to 3.625Ghz because 3.8 was requiring too much voltage for the stock cooler to handle. I was hitting 80+ on the CPU Diode and about 72 to 75 on the CPU with voltage around 1.37+ Now I have the voltage set to 1.26v on the CPU core. Ran P95 all night for 8 hours. I'll probably do a longer test to make sure it's rock solid. I was hitting 70C on the CPU Diode and 60 to 65 on the CPU.

I'm much happier with those temps and I can't really see any performance loss. I will try for higher OC once I get my AM4 bracket for my TD02-Slim AIO water cooler though.

 

[Johan45]

OK thanks. I wouldn't expect too much out of the ram. It's Hynix and Ryzen just don't like it so much. For most things the ram won't make too much difference any way. As long as it's stable that's the main concern.

 

[Thentilian]

OK thanks. I wouldn't expect too much out of the ram. It's Hynix and Ryzen just don't like it so much. For most things the ram won't make too much difference any way. As long as it's stable that's the main concern.

It's sticks I had on hand from a friend I traded with. Saved me quite a bit of cash considering how expensive RAM is right now... I'm running into an issue where memtest says RAM is good, but Intel Burn Test won't finish on High settings. Returns results like this: 4.213603e+009. That or some other wrong result. This is different than before when the PC would just go black and crash. It doesn't BSOD, just would go black and the PC would stay on, but wouldn't boot until I hit the reset button. I tried to loosen the RAM timings to something like 15-17-17-33, and it didn't even want to boot.

I think I'll just go with the 2400 speed, since I might only gain a couple of FPS anyways with the higher speed RAM, which is basically what I do on this machine, gaming. I do some video editing and streaming as well, so I just wanted to tweak the utmost out of my setup.

So, Hynix sucks for Ryzen? I want to get some Samsung IC's then? Which RAM modules are doing good with Ryzen right now? I should probably start another thread at this point to go over all this stuff, since I'm cluttering it up with results I think are good, then run something else and turns out it's not. The learning curve to Ryzen is whoopin' my butt at the moment!

 

[Johan45]

If we're gonna keep going I can move this and rename it just gimme a name
If you don't give me a name I'll make one up like Thentilian's R5 1600 thread

 

[Johan45]

Moved

 

[Thentilian]

If we're gonna keep going I can move this and rename it just gimme a name
If you don't give me a name I'll make one up like Thentilian's R5 1600 thread

Moved

Cool, thanks. I'm gonna keep tinkering with it, there is other things I haven't tired like the LLC stuff and the PWM presets, standard, optimized, etc... but I don't want to mess with them until I understand more of what they do, so I don't blow anything up!

Also, I reset RAM to it's XMP profile and seeing if Intel Burn Test will pass on high settings. If it doesn't then it's probably the CPU not stable, even though it did 8 hours of P95 last night... what a PITA this is. But I'm learning and it's fun.

EDIT: Just passed high settings on IBT. So, I guess the RAM was at fault. It would not pass a single time when set to 2800, even with the voltage as high as 1.37 on the DRAM. I know I could safely bump it to 1.4v but I'm not ready to push it that far yet. Although, I have to say the RAM has stayed very cool. Even after getting hammered by memtest it was only slightly warm to the touch.

Incidently, the temps on that IBT run were: 76.3C diode, and 73C CPU. Ambient temp being much higher during the day gives quite a jump in temps, but these are ok?

 

[Thentilian]

Ok, I have an article I'm writing on a Fractal Design Focus G case, so I moved my build into it to feature in the article. I tell ya, small case, but managed to squeeze my build into it.

Anyways, my AM4 bracket showed up for my SilverStone TD02-Slim AIO water cooler. I was in the midst of building it when it came in the mail... talk about perfect timing. I've been messing with the OC again, as I wasn't really pushing with the stock cooler anymore because of the temps I was getting. Now I'm hitting 67C on CPU Die and 63C on CPU after running IBT on High profile. I'm doing 3.9Ghz with 1.4 volts and LLC 1 with SOC on manual 1.0v seemed to help stabilize it.

Couple questions: 1. How much does the SOC help with stability? I only have my RAM at it's XMP of 2400 with 1.2 volts, but I even pumped that to 1.25 just to make sure RAM is not the problem with stability. 2. I ran IBT on high settings and it finished and went to open up chrome and the PC crashed... even though IBT finished successfully... anyone else encountering odd crashes after a stability test?

And I ran a P95 overnight with the core set to 3.95Ghz, and it crashed during the night. I was using a bit more voltage, 1.41 with LLC 2, which caused voltage to hit 1.45 to 1.46, which is becoming more than I'm ok with, so I eased back to 3.925Ghz, but that wasn't stable at 1.4 volts LLC 1 either. So, now I'm down to 3.9Ghz and encountered this.

So, looks like my 1600 just isn't a lucky chip, and temps are not an issue and I don't really want to pump more than 1.4 to 1.42 volts through it, as this chip might be in my rig for a long time, depending on what happens with Zen 2.

EDIT: Forgot to mention I grabbed the latest 0812 UEFI BIOS from Asus before doing this. Staid it's supposed to help with stability. Not sure if it is or not at this point. My chip is just probably a dud, and just gonna have to live with whatever I can squeeze out of it. Not even sure if 3.9Ghz is going to be stable yet.

 

[DaveB]

I needed 1.425V to get my Ryzen 5 1660 stable @ 4 GHz. It was running on an mATX ASRock AB350M and cooled with a Corsair H100. It was limited to 3.9 GHz on air using a Cooler Master Hyper 212 @ 1.4V - any higher and it got too hot. Don't know if there are any really "lucky" Ryzens since even the best only get another 100 MHz or so which is pretty meaningless performance wise.

 

[Johan45]

With a low ram speed like that your SOC voltage should be fine. I would set a static voltage of 1.0V though as sometimes the fluctuations can cause issues. Have you tried one of the other DOCP settings beside the XMP? XMP is intended for intel so some timings might not play nice. For a 1600 3.8-3.9 is pretty average so I wouldn't call it a dud just normal.
When overclocking it's best to start at the bottom and work your way up testing along the way. Helps identify where the stability issue may lay. as opposed to picking a clock and trying to adjust the voltage afterward

 

[Thentilian]

I needed 1.425V to get my Ryzen 5 1660 stable @ 4 GHz. It was running on an mATX ASRock AB350M and cooled with a Corsair H100. It was limited to 3.9 GHz on air using a Cooler Master Hyper 212 @ 1.4V - any higher and it got too hot. Don't know if there are any really "lucky" Ryzens since even the best only get another 100 MHz or so which is pretty meaningless performance wise.

I know I have the headroom as far as temps go now. Even IBT only pushes 67 to 70C on the die. I'm just really reluctant to push past 1.4 or 1.41. I might give it a go just to see if I can get 4Ghz stable, but as you said 100Mhz more isn't really going to do much for me but push better benchmark numbers, and I'm more concerned about stability, game performance, streaming, and video editing.

With a low ram speed like that your SOC voltage should be fine. I would set a static voltage of 1.0V though as sometimes the fluctuations can cause issues. Have you tried one of the other DOCP settings beside the XMP? XMP is intended for intel so some timings might not play nice. For a 1600 3.8-3.9 is pretty average so I wouldn't call it a dud just normal.
When overclocking it's best to start at the bottom and work your way up testing along the way. Helps identify where the stability issue may lay. as opposed to picking a clock and trying to adjust the voltage afterward

I am using DOCP settings, but I think it only has the one profile. The drop down only has the one settings in the menu. Currently, I am running them with 1.35 volts, 2800 14-16-16-31 CR1 timing. It passed an hour of memtest86, which doesn't mean much yet, I'm gonna run a full memtest burn tonight when I sleep to make sure the memory is stable. I did bump the SOC to 1.0v and vCore to 1.412 just to make sure I'm stable on the core. I'm not willing to go higher on the Vcore voltage for long term use anyways.

I set vcore to 1.4 at first then worked up from 3.8 and I could get 4.0 to boot, but the stability is wonky. As I mentioned I passed an IBT on high settings and immediately opened Chrome and the PC restarted... I had to scratch my head on that one. IBT on high passed, open Chrome, fail... haha

Then I've been tweaking SOC and vcore for stability. I need to do a full-burn in, but the early tests seems stable for gaming, internet and video watching. I just want to make sure, unlike last time when I thought it was solid, it was not. Also, I had a seating issue with my RAM. Seems these DIMM slots on this board are touchy. When I transferred to this new case, the BIOS only reported 1 stick working, and I was like ok, no wonder it wasn't firing up. So, hopefully this issue doesn't worsen or the motherboard might have to go back. I know it's not the RAM, they worked fine in another system my friend had for months without issues and they weren't overclocked or anything like that. Plus after I reseated them problem was fixed.

EDIT: Here is a shot from Aida64 of the RAM profile. Has a lot of different timings, but the highlighted one in the photo shows what it sets when DOCP is selected.