Ryzen Owners Zen Garden

[tabascosauz]

So it’s IF related? I started having the missing boot drive every other post before nudging the SoC voltage.

Didn't have it nearly as frequently as the two of you seem to suffer, and not on my boot drive. Probably about once every other month before I decided to lop off VDDG and VDDP a bit.

At one point I was considering the possibility that my SN750 was bad, but then I realized that it had to be IF. My boot drive (SX8200) has always been in the chipset-wired slot of all places and never failed me. It's my SN750 that I added later into the CPU-wired slot that always had problems.

This is how much I am winging it..

I have not touched any of those voltages

Also never lost my drives, ever.. crazy. Asus must just make it easy? They took the adventure away

Well they clearly didn't make it easy, because this only ever happens on my Asus board

 

[biffzinker]

Also never lost my drives, ever.. crazy. Asus must just make it easy? They took the adventure away

I doubt board brand has anything to do with how integrated everything is on the CPU now. Considering the nvme drive is hanging off the I/O die.

Edit: fixed post, typing on smartphone

 

[dgianstefani]

I don't really worry about cooling. On my EK AIO and good case airflow, I never see temps go over 80c on OCCT or Realbench, regardless of all core overclock or PBO/CO.


The IF is a good thing to explore. I have been pushing it pretty hard the last few days playing with some new ram.

That's because 80c is the throttle limit.

My system doesn't go above 65.

 

[freeagent]

I may need to review my cooling situation.

 

[PaulieG]

That's because 80c is the throttle limit.

My system doesn't go above 65.

I'm pretty sure the throttle limit is 90c or 95c. Can anyone confirm with a reliable link? I've set the manual throttle limit to 85c anyways. Honestly when I go back and look at screenshots, under the EK AIO my highest temp is 76c on OCCT and 74 on Realbench. The only time I reach 80c was with PBO enabled on a Cinebench R23 run where the vcore jumps up to 1.35v. Never over 74c on any bench run with all core manual overclocking under 4.7ghz.

 

[dgianstefani]

I'm pretty sure the throttle limit is 90c. I've set the manual throttle limit to 85c anyways. Honestly when I go back and look at screenshots, under the EK AIO my highest temp is 76c on OCCT and 74 on Realbench. The only time I reach 80c was with PBO enabled on a Cinebench R23 run where the vcore jumps up to 1.35v. Never over 74c on any bench run with all core manual overclocking under 4.7ghz.

Throttle limit yes, but ryzen boost reducing isn't technically throttling. If you want good boost frequencies, you keep it under 70c.

 

[biffzinker]

I'm pretty sure the throttle limit is 90c or 95c.

I thought I’ve seen the thermal throttle limit is 95c some where. Unable to locate the source.

 

[dgianstefani]

Throttle, by definition, means going lower than advertised clock. PBO speeds are what suffers. Having the mindset that "as long as it's under 95c it's fine" is what causes you to miss out on 2-300mhz.

 

[PaulieG]

I've found LLC plays a vital role on all core manual overclocking Ryzen. Anyone with info on how LLC works with PBO and CO?

 

[tabascosauz]

I've found LLC plays a vital role on all core manual overclocking Ryzen. Anyone with info on how LLC works with PBO and CO?

iirc you don't want higher LLC for PB or PBO, boost algorithm takes droop into account.

Honestly though, lower LLC usually delivers cleaner power when you measure it with an oscilloscope. Unless the out of the box droop is absolutely insane and unusable, I personally wouldn't crank LLC up too much, esp. on lower mid-range, doubled 4-phase ASP1106G VRMs with 50A Vishay parts.

It's the current that kills Ryzens so it's fine to have say 1.35V set/idle that droops down to 1.28V at load if temps and current are in check.

 

[PaulieG]

iirc you don't want higher LLC for PB or PBO, boost algorithm takes droop into account.

Honestly though, lower LLC usually delivers cleaner power when you measure it with an oscilloscope. Unless the out of the box droop is absolutely insane and unusable, I personally wouldn't crank LLC up too much, esp. on lower mid-range, doubled 4-phase ASP1106G VRMs with 50A Vishay parts.

It's the current that kills Ryzens so it's fine to have say 1.35V set/idle that droops down to 1.28V at load if temps and current are in check.

Of course I can just check it on my own, but do you know where on the continuum the auto setting is on LLC with Asus B550 boards? I don't generally trust auto LLC so I tend to set it at 3 or 4.

 

[tabascosauz]

Of course I can just check it on my own, but do you know where on the continuum the auto setting is on LLC with Asus B550 boards? I don't generally trust auto LLC so I tend to set it at 3 or 4.

I haven't played with all-core on my B550M TUF Wifi yet, but there's a review of the Strix-F and Strix-E where they use Level 4 LLC on the Strix-F for a 0.04V droop: ASUS ROG STRIX B550-E Gaming & B550-F Gaming (WiFi) – Page 8 – KitGuru

That's a pretty small droop so I guess Level 4 is starting to get aggressive

Not sure if they changed anything for newer Ryzen 5000 BIOSes. Definitely let me know how it goes; VRM-wise the TUF and the Strix-F should be near-identical. nevermind forgot the Strix-F is tripled

 

[Gmr_Chick]

Hey y'all. I'm extremely green when it comes to overclocking, but I'm wondering if doing so will help squeeze out a few more FPS in Battlefront II, or if OCing is even really worth it nowdays. Anyway, since I don't trust myself enough to play around in the BIOS for my board, I've just been using the "Game Mode" preset in Ryzen Master. I have a Corsair H100i RGB PRO XT AIO, but am in the middle of swapping it out in favor of my Cooler Master MA610P tower cooler. It's a long story, but I'm basically doing it because iCUE is a resource pig and I only use it to change the RGB lighting on the CPU block... Anyway, I'm using the stock cooler that came with my 3600 in the meantime and I'm just wondering what temps are considered safe?

 

[freeagent]

For max kick you right in the cores voltage?

I try to keep it under 85 worst possible case load you can imagine.

Its pretty tough to get up there unless you don't give a **** about cooling

 

[Space Lynx]

I have gone down to 1.270 volt, LLC Mode 6, 4500 all core no downclocking on my 5600x. only a loss of 125 mhz but much colder and lower voltage. 125 mhz won't matter in games. so pretty happy with this.

even on single core boosts i only get 4750 max, i did not win the silicon lottery sadly. but it's all good, i'm pretty happy with these cold temps and low voltage, and basically the same performance in games.

So, after getting a stable all core manual overclock of 4.7ghz @ 1.29v on my 5600x rock solid

if you are truly stable and not just talking smack, a 4.7 all core at 1.29v is a golden chip (what mode is your LLC at?), also in BIOS do you use "AMD Overclocking Mode" or some other kind of voltage mode to override the voltage? My mobo has like 4 to pick from, I never know which one to go with so I do AMD Overclocking, which has a override feature outside of the regular override one, I find it all extremely confusing why there are basically two of the same thing.. i'd leave it there and be happy. i have to use 1.360 volts to get 4625 stable, and it refuses to be stable above that regardless of voltage or temps.

you are pretty lucky to have such a nice chip.

 

[freeagent]

I manually set core vid per ccx as well as its over ride, both values being equal, and no LLC for this chip it doesn't seem to help nor appreciate it. I will give it some time though and play with it. I could not get 4700 with 1.25-1.325v. It will run sure, but I cant run everything. It is almost like it is programmed to not go over a certain speed continuously. I could boot into windows @ 4900 no problem too. But some guys are saying they cant run 4600 so maybe it still continues to be luck of the draw.. It can handle 2 sticks of ram pretty good. It is 3200C14 running at 4200 C16 1:1.. pretty crazy. It is everything stable so far. Since I updated my bios I have to work a little harder with 4 sticks, so I am just running 2 for now.

 

[Space Lynx]

I manually set core vid per ccx as well as its over ride, both values being equal, and no LLC for this chip it doesn't seem to help nor appreciate it. I will give it some time though and play with it. I could not get 4700 with 1.25-1.325v. It will run sure, but I cant run everything. It is almost like it is programmed to not go over a certain speed continuously. I could boot into windows @ 4900 no problem too. But some guys are saying they cant run 4600 so maybe it still continues to be luck of the draw.. It can handle 2 sticks of ram pretty good. It is 3200C14 running at 4200 C16 1:1.. pretty crazy. It is everything stable so far. Since I updated my bios I have to work a little harder with 4 sticks, so I am just running 2 for now.

i thought zen 3 had a cap of 4000 2000:1... that's what I am running out with my ram OC.

hmm interesting. well im just glad im stable with my ram where it is so im not going to mess with it. my 4625 all core was stable in everything at 1.360v... i think i will turn off LLC though, i think you are correct on that front.

im leaving it where it is for now. i love the extra extra cold temps im getting at 1.270 volt and strong fan curve, 4.5ghz versus 4.7ghz is really not noticeable in any real world scenario anyway, maybe a 5 fps gain in games at 1080p. but really its the screaming ram we have that is making the biggest difference for fps.

 

[freeagent]

i thought zen 3 had a cap of 4000 2000:1... that's what I am running out with my ram OC.

hmm interesting. well im just glad im stable with my ram where it is so im not going to mess with it. my 4625 all core was stable in everything at 1.360v... i think i will turn off LLC though, i think you are correct on that front.

im leaving it where it is for now. i love the extra extra cold temps im getting at 1.270 volt and strong fan curve, 4.5ghz versus 4.7ghz is really not noticeable in any real world scenario anyway, maybe a 5 fps gain in games at 1080p. but really its the screaming ram we have that is making the biggest difference for fps.

I know man, I thought I was going crazy lol.



I ran some linpack too but no pic.

 

[tabascosauz]

I know man, I thought I was going crazy lol.

View attachment 191132View attachment 191133View attachment 191134

I ran some linpack too but no pic.

Only thing being proved by running HCI is that your B-die is stable at 4200 16-16-16 (looks good, tighten tRC a bit). It says nothing about how your IF is doing. Only time and generally using your PC will tell you that. Use your computer, keep an eye out for weird behaviour, keep an eye out for disk/Windows errors and WHEA.

Still sounds like you're hitting a frequency wall, no point constantly butting up against it and benching a fat ton of current every day through your chip in AVX LinX.

As for not being stable in some workloads, that's why the stock chips downclock so hard in Prime95. They wouldn't last two weeks if they treated it like SSE all-core.

 

[freeagent]

Thanks, yeah that was an early run, it will run 286. Ok so I took it off the 1 speed 1 voltage thingy. And I think I may have found a secret button..

Edit:

Not sure if I could figure out how to get it over 4850, I tried increasing the value but nada. Is this where X570 comes into play?

 

[ViperXTR]

Worth updating to AGESA 1.2.0.1?
This contains the fix for L3 bandwidth test issue last time, is it only for that test or L3 in general i wonder

https://twitter.com/i/web/status/1364971430337740804

 

[Space Lynx]

Worth updating to AGESA 1.2.0.1?
This contains the fix for L3 bandwidth test issue last time, is it only for that test or L3 in general i wonder

https://twitter.com/i/web/status/1364971430337740804

yes this new bios is worth it. i upgraded my x570 tomahawk yesterday. its the first time i have ever been able to OC my ram to 4000 2000:1 2x16gb... IT USED TO LET ME, but then it wouldnt let me OC CPU or enable SAM, but this new BIOS allows me to do all 3 no issues at all, rock solid stable... AMD finally nailed it. at least for my setup.

 

[ViperXTR]

yes this new bios is worth it. i upgraded my x570 tomahawk yesterday. its the first time i have ever been able to OC my ram to 4000 2000:1 2x16gb... IT USED TO LET ME, but then it wouldnt let me OC CPU or enable SAM, but this new BIOS allows me to do all 3 no issues at all, rock solid stable... AMD finally nailed it. at least for my setup.

Some reports also say its running slightly cooler and removes the core clock cap for the boost? Its already available on my board, but still in beta, will probably just wait for the non beta release

 

[Space Lynx]

Some reports also say its running slightly cooler and removes the core clock cap for the boost? Its already available on my board, but still in beta, will probably just wait for the non beta release

mine is the beta variant too. it takes them a long time to release non beta. i can assure you its rock solid, our mobo's share same BIOS. i recommend you install it.

i can't comment on the boost, as i dont use boost i use an all core OC

 

[PaulieG]

I haven't played with all-core on my B550M TUF Wifi yet, but there's a review of the Strix-F and Strix-E where they use Level 4 LLC on the Strix-F for a 0.04V droop: ASUS ROG STRIX B550-E Gaming & B550-F Gaming (WiFi) – Page 8 – KitGuru

That's a pretty small droop so I guess Level 4 is starting to get aggressive

Not sure if they changed anything for newer Ryzen 5000 BIOSes. Definitely let me know how it goes; VRM-wise the TUF and the Strix-F should be near-identical. nevermind forgot the Strix-F is tripled

Nice review. Good to know I'm not the only one who doesn't like the way Asus does LLC on this platform. I'm probably giving it too much and driving my temps higher than they need to be. I'll be trying to nail the vdroop for each level, maybe this weekend.