Ryzen Owners Zen Garden

[thesmokingman]

Probably. The curve optimizer doesn't make much of a difference from stock, maybe an extra couple hundred MHz on multi, but it drops the core voltages down a bit as well.

One actually uses the curve optimizer if you want to force it to waste more power.

 

[freeagent]

I played with it for the first time today.. it was in a different menu.. I didn't test any cores individually or anything, just typed in - 30 for all of them, and was doing 4850 with only 1.25 lol.. probably not stable.. that is what I need for a locked 4600. I chickened out.

 

[Muaadib]

'back to' 1.45v dram? what did you drop the DRAM to?

Apologies. Looking back, I think i misunderstood your point. I assumed you were saying 1.45v on the ram wasn't needed, and instead just increasing SoC voltage to 1.2v. I will see if I can reach 3200C16 when i get back home.

P.S. My XMP is 3000C16 so I'm already at tighter timings than it.

 

[tabascosauz]

Probably. The curve optimizer doesn't make much of a difference from stock, maybe an extra couple hundred MHz on multi, but it drops the core voltages down a bit as well.

Because of the inherently spiky single thread temps on Ryzen 5000 I really wanted to use Curve Optimizer. I'm still getting spikes up to 85C in COD and COD only (although 90% of the time CCD1 is still just 70-77C), it's not hitting 90 but the weather is still quite cool. Otherwise gaming temps are in the 60s and 70s, and MT workloads always around 70-72.

Took a look at CO. Realized that PBO needs to be on. Realized that most of the 2nd PBO menu under AMD OC is utterly broken despite it being a stable 1200 BIOS. Realized that PBO straight up ignores any power limits I or the motherboard set, while consistently nuking ST perf between 10-15% even without touching CO. Then gave up and decided it wasn't worth my time when nothing's actually wrong with how it is right now.

 

[Mussels]

Apologies. Looking back, I think i misunderstood your point. I assumed you were saying 1.45v on the ram wasn't needed, and instead just increasing SoC voltage to 1.2v. I will see if I can reach 3200C16 when i get back home.

P.S. My XMP is 3000C16 so I'm already at tighter timings than it.

nooo, you're meant to raise *both*

1.4v DRAM with 1.1V SoC is a universal good idea for overclocking on ryzen - the only thing that varies is how much SoC voltage, because it can vary a little between the generations

 

[8BitZ80]

Because of the inherently spiky single thread temps on Ryzen 5000 I really wanted to use Curve Optimizer. I'm still getting spikes up to 85C in COD and COD only (although 90% of the time CCD1 is still just 70-77C), it's not hitting 90 but the weather is still quite cool. Otherwise gaming temps are in the 60s and 70s, and MT workloads always around 70-72.

Took a look at CO. Realized that PBO needs to be on. Realized that most of the 2nd PBO menu under AMD OC is utterly broken despite it being a stable 1200 BIOS. Realized that PBO straight up ignores any power limits I or the motherboard set, while consistently nuking ST perf between 10-15% even without touching CO. Then gave up and decided it wasn't worth my time when nothing's actually wrong with how it is right now.

Have you tried using Eco Mode? Damn I can't remember, but there's a manual setting in there that you can edit to change the total CPU power consumption. Mine had a default value of 76 which limited my 5600X to a maximum power consumption of 76 watts. I'm pretty sure it works alongside PBO.

 

[Muaadib]

nooo, you're meant to raise *both*

1.4v DRAM with 1.1V SoC is a universal good idea for overclocking on ryzen

Thanks for the tip, I have some free time starting tonight so I will start with those settings and see where it goes. Will report back afterwards.

 

[Mussels]

Have you tried using Eco Mode? Damn I can't remember, but there's a manual setting in there that you can edit to change the total CPU power consumption. Mine had a default value of 76 which limited my 5600X to a maximum power consumption of 76 watts. I'm pretty sure it works alongside PBO.

you gotta turn PBO on, reboot, then eco mode shows up which basically has some preset PBO numbers

 

[tabascosauz]

Have you tried using Eco Mode? Damn I can't remember, but there's a manual setting in there that you can edit to change the total CPU power consumption. Mine had a default value of 76 which limited my 5600X to a maximum power consumption of 76 watts. I'm pretty sure it works alongside PBO.

Won't help me unfortunately, it just applies the 65W power limits. PPT isn't a problem, I can handle any MT workload up to the stock 142W limit, barely hits 70C. It's lightly threaded COD that's the problem because it boosts only the 2 best cores continuously at 4.4-4.8GHz. PPT only falls around 95-ish watts in game, and I'm not looking to cripple multithread performance elsewhere by limiting 12 cores to 90W.

And the board doesn't respect any other power limit I set because the PBO menu is broken.

Notwithstanding the obvious issue of PBO power limits being unenforceable, when I turn PBO on something's wrong with single thread perf. CPU-Z will run the MT portion normally (albeit boosts up to 167W and I can't even change that), but the ST bench is broken. The score just fluctuates wildly like a roller coaster until the test ends and it just settles on whatever fkin random number it's on at that second. When I turn PBO off, it goes back to normal.

 

[Muaadib]

So I tried various SoC/DRAM voltage configurations, all failed to post or BS'd. Guess I will just have to live with 3000@1.1/1.45v

Cant say I'm disappointed though. Quite happy actually since it saved me a potential ram upgrade.

 

[Gmr_Chick]

Hey peeps, have a question for you all. I've already looked at the HWInfo forum about Power Reporting Deviation but...I didn't really understand it. All I know is it sounds like anything 90% and above is considered "normal" and anything under that is considered under-reporting. Since getting this board, I've never actually seen the PR stat "in the green" so to speak. It's ALWAYS been in the red and though I've always wondered why (it's the only reading that's red so of course it stands out) I wasn't sure what it was exactly.

Anyway, here's a screencap I took during a session of Battlefront II. The lowest "minimum" for Power Reporting I've seen thus far (in the past few days) is a whopping 55.4% as well.

So, is my board playing unwelcome tricks on me? How do I stop it? Should I be overly concerned (ie think about a different board)?

 

[freeagent]

Put a hard load on it and it should drop to 0.. I wouldn’t worry about it. You don’t even see that anymore with zen3 I kind of miss it

 

[tabascosauz]

Hey peeps, have a question for you all. I've already looked at the HWInfo forum about Power Reporting Deviation but...I didn't really understand it. All I know is it sounds like anything 90% and above is considered "normal" and anything under that is considered under-reporting. Since getting this board, I've never actually seen the PR stat "in the green" so to speak. It's ALWAYS been in the red and though I've always wondered why (it's the only reading that's red so of course it stands out) I wasn't sure what it was exactly.

Anyway, here's a screencap I took during a session of Battlefront II. The lowest "minimum" for Power Reporting I've seen thus far (in the past few days) is a whopping 55.4% as well.

View attachment 195630

So, is my board playing unwelcome tricks on me? How do I stop it? Should I be overly concerned (ie think about a different board)?

The only number that matters is what you see during an all-core benchmark. Run Cinebench and see what you get during it.

70% at idle is not a good sign though as most boards idle higher and drop during benchmarks. The lower the number, the more the board is underreporting in an effort to surreptitiously boost performance. Because the board underreports current draw, the chip may be tricked into pulling more current than it thinks it is, which may or may not boost performance.

My B450 I Aorus Pro Wifi was one of the worst boards I've ever seen, with the deviation falling to 74% under load. ST and MT scores on my 3700X certainly were higher than on any other board. If your Cinebench scores seem consistently higher than others' 3600s at stock, the underreporting could be the culprit.

Longevity, I dunno. In theory, Precision Boost 2 is still in charge, so there should be no damage suffered to the chip. However, PB2 still relies on the board to tell it how much current it is pulling and the board pretty much lies to the CPU - in your case, by quite a bit. Ryzen is more sensitive to current than voltage and my 3700X did pretty much go downhill with time, although obviously it's unproven and a bit of a stretch to link it to that first board I had.

Normally for a 70-90% board I wouldn't consider changing the board just because of the deviation. Especially not for a 65w 3600.

Put a hard load on it and it should drop to 0.. I wouldn’t worry about it. You don’t even see that anymore with zen3 I kind of miss it

I certainly hope it doesn't drop to 0 lol, would the board even be reporting ANY SMU current draw at that point?

I've always had the deviation for any of my chips, Matisse, Renoir or Vermeer. Sounds like your HWInfo is due a Reset User Preferences, sometimes a newer version fucks things up and it needs it.

If you are upgrading from HWinfo V7.00 to V7.02 you should reset. Same if you are upgrading from any AGESA 1200 BIOS to 1201. The new AGESA does a lot of the sensors differently so HWInfo may be confused and drop or break half of them.

 

[freeagent]

Its been awhile since I have seen CPU PPT being reported like it used to for Matisse. I took a bios update and I havent seen it since. I do have reporting again in the new HWinfo, but its just current min max. It used to be all over the place for me lol 200% down 0 and back up. When I enabled LLC it calmed the eff down.

 

[Tomgang]

Switching from Intel to amd ryzen. I got a Ryzen 5 5600X a few days ago and have a Ryzen 9 5950X in order, but when that is coming. I have no idea of.

CPU can be seen here. It's not up and running yet as I am waiting for other parts to arrive.

https://www.techpowerup.com/forums/...h-related-purchase-thread.225885/post-4491280

5600X will be put in a ASUS ROG STRIX B550-I GAMING mini-itx board

5950X in a ASUS ROG CROSSHAIR VIII DARK HERO board. If it comes that is... It's all most just as difficult to get as proper GPU these days.

 

[Mussels]

Hey peeps, have a question for you all. I've already looked at the HWInfo forum about Power Reporting Deviation but...I didn't really understand it. All I know is it sounds like anything 90% and above is considered "normal" and anything under that is considered under-reporting. Since getting this board, I've never actually seen the PR stat "in the green" so to speak. It's ALWAYS been in the red and though I've always wondered why (it's the only reading that's red so of course it stands out) I wasn't sure what it was exactly.

Anyway, here's a screencap I took during a session of Battlefront II. The lowest "minimum" for Power Reporting I've seen thus far (in the past few days) is a whopping 55.4% as well.

View attachment 195630

So, is my board playing unwelcome tricks on me? How do I stop it? Should I be overly concerned (ie think about a different board)?

only relevant during all core load, so start up cinebench and read the current values then
Honestly who cares if it boosts a little higher or hotter, you can tweak PBO values since you're aware of it

 

[Gmr_Chick]

Values I got during a 10 minute run of Cinebench multi-core were 83.6 towards the end of the test, 78.3 for average and 57.2 for a minimum

 

[tabascosauz]

Values I got during a 10 minute run of Cinebench multi-core were 83.6 towards the end of the test, 78.3 for average and 57.2 for a minimum

Even when CB is looping there is still slight downtime between tests I think, might be where the 57% came from. Otherwise, 78-83% during the actual render test isn't terrible, wouldn't worry about that.

AMD says underreporting however significant still can't override internal voltage and current limits. I usually don't have the utmost confidence in AMD's claims as to longevity due to my 3700X, but this statement seems reasonable enough. Especially on a B550 board (some earlier B450 boards basically never respected PPT limits even with PBO off) and a 3600.

 

[Gmr_Chick]

Even when CB is looping there is still slight downtime between tests I think, might be where the 57% came from. Otherwise, 78-83% during the actual render test isn't terrible, wouldn't worry about that.

AMD says underreporting however significant still can't override internal voltage and current limits. I usually don't have the utmost confidence in AMD's claims as to longevity due to my 3700X, but this statement seems reasonable enough. Especially on a B550 board (some earlier B450 boards basically never respected PPT limits even with PBO off) and a 3600.

So I shouldn't worry too much about it? (I'm a bit of a worry-wort, worst-case scenario type -- partially because of my OCD -- so I tend to mentally exhaust myself thinking about all the shit that could go wrong with my rig)

 

[tabascosauz]

So I shouldn't worry too much about it? (I'm a bit of a worry-wort, worst-case scenario type -- partially because of my OCD -- so I tend to mentally exhaust myself thinking about all the shit that could go wrong with my rig)

I know how you feel. I went down that hole with Ford and GM's engines and trannys and as a result my next ride is probably a Model 3

If you tend to overthink things, just frame everything Ryzen within the philosophy of "if it ain't broke don't fix it".

No issues? Don't go reaching for that shiny new AGESA. No issues? Don't go wondering what PBO might have in store. No issues? Don't go reaching for that next 66MT/s jump up or the next 5ns tRFC step down.

I can abuse my HTPC all I want, but on my main where I count on its dependability day in and day out, the philosophy works like a charm.

In this case, it's working fine, so don't worry about it.

 

[Gmr_Chick]

I know how you feel. I went down that hole with Ford and GM's engines and trannys and as a result my next ride is probably a Model 3

If you tend to overthink things, just frame everything Ryzen within the philosophy of "if it ain't broke don't fix it".

No issues? Don't go reaching for that shiny new AGESA. No issues? Don't go wondering what PBO might have in store. No issues? Don't go reaching for that next 66MT/s jump up or the next 5ns tRFC step down.

I can abuse my HTPC all I want, but on my main where I count on its dependability day in and day out, the philosophy works like a charm.

In this case, it's working fine, so don't worry about it.

Thanks, Tabby! You've helped put my mind at ease

 

[freeagent]

I reinstalled windows, got the newest HWinfo, and put my CPU back @ stock settings with PBO and I see everything you guys do now

I think I will just leave it alone now..

 

[biffzinker]

I think I will just leave it alone now..

Your still OCing the RAM though?

 

[tabascosauz]

I reinstalled windows, got the newest HWinfo, and put my CPU back @ stock settings with PBO and I see everything you guys do now

I think I will just leave it alone now..

I think i might have misled you - I meant hit the Reset button in HWInfo settings, not Windows

Still it's good to start fresh esp after all the OCing youve been doing

 

[PaulieG]

I think i might have misled you - I meant hit the Reset button in HWInfo settings, not Windows

Still it's good to start fresh esp after all the OCing youve been doing

Way to try to make him feel better after a goof.