Ryzen Owners Zen Garden

[Mussels]

Is it worth going to the previous AGESA?

It can bring the fTPM bug back, but if you disabled the need for that crap when installing 11 you're fine


CM keep changing the 212, every revision has drastic changes to the design - some are alu + copper heatpipes, some are pure heatpipes, some have alu or copper baseplates. The similar names trick people into thinking this $20 USD 212 is the one everyones raving about, when its not.

 

[freeagent]

My 212 Evo has massive gaps in it, circa 2012.

 

[Mussels]

(I left this open in my browser for like 8 hours, lol)

The Hyper 212 from 2007:
A solid slab of copper under the heatpipes.
It all goes downhill from here.

212 Evo: Copper slab gone, thinned back to just the heatpipes. That loss of solid copper hurts.

212x: This is the globally popular one that stayed with the close heatpipes that were effectively solid copper contact.

212 LED(white/turbo etc)
Flat in this image at least, but the spaces begin

.

Evo V2: This even comes in an LGA1700 version
More spaced out, but a terrible choice for multi chip CPUs
You can see the height difference between alu and copper

Black edition: (RGB variants the same) - It may have nickel plating on the bottom vs plain Alu, at least.
Gaps here exist but are very minimal

212 RGB black: They stop showing actual images and give you renders only

They sold a bunch of shitty products with the name of the original, relying on customer confusion to buy the wrong one

I skipped a good 10 variants just because they're all the same

 

[RJARRRPCGP]

I have a 6-pipe version expected to come my way. It's not a 212.

 

[freeagent]

I have a 6-pipe version expected to come my way. It's not a 212.

Hopefully it has a real cold plate and not some milled down pipes

 

[RJARRRPCGP]

Hopefully it has a real cold plate and not some milled down pipes

They aren't exposed pipes, it has what looks like a nickel plate. Like the copper plate of the beloved early-212s, but what is probably nickel it its place, it doesn't look like the usual aluminum.

 

[freeagent]

For you AM5 guys.. not sure if this will help you or not:

AM5 Secure Frame BLACK – Thermalright

www.thermalright.com

 

[RJARRRPCGP]

For you AM5 guys.. not sure if this will help you or not:

Looks like I won't be getting AM5 for a good while!

 

[dont whant to set it"']

@freeagent Will take into consideration if deliding.
Mine runs pretty well, tough a remount cpu cooler I have to chase because the cooler did not come with a screw driver (need long and thin, so it fits trough fin stack to tighten one nut). I used adjustable vice grips.

 

[RJARRRPCGP]

Vice grips? I fear that I'm going to be required to use those or similar, just to get the Hyper 212 Evo V2 mount off of my ASRock B550 PG Velocita! If I don't find the proper screwdriver or nut driver!

 

[AusWolf]

I've just noticed something strange with my new system. When I cold start it, the CPU and RAM debug leds stay on, and it won't boot. If I turn it off and on again, everything is OK. What could cause this?

 

[tabascosauz]

I've just noticed something strange with my new system. When I cold start it, the CPU and RAM debug leds stay on, and it won't boot. If I turn it off and on again, everything is OK. What could cause this?

Doesn't cold boot RAM training take a LONG time on AM5 right now? Like, minutes long?

 

[AusWolf]

Doesn't cold boot RAM training take a LONG time on AM5 right now? Like, minutes long?

The longest I've waited was around 5 minutes. The funny thing is, if I get fed up, and restart the PC, it works fine.

Should this happen at EVERY cold start?

 

[MagnyCours]

The time it takes for my system to reach the BIOS splash screen off of a cold boot is always 30 seconds. Something might be off with your system.

 

[Super Firm Tofu]

Doesn't cold boot RAM training take a LONG time on AM5 right now? Like, minutes long?

The longest I've waited was around 5 minutes. The funny thing is, if I get fed up, and restart the PC, it works fine.

Should this happen at EVERY cold start?

Normal cold boots should be about 30 seconds to the post screen. If you mean after a BIOS reset, or possibly removing power from the board/box, then that could be longer.

I had an old GB board on Sandy Bridge that did what's being described. Disconnect power from the box, and the first boot would hang, requiring a forced power-down by holding the button until the board shut off. After that, it was golden until power was removed from the box again.

Guessing, it's probably hanging on the memory init. (training, whatever) due to unstable settings from EXPO. Double-check SoC voltage - mine set 1.3v(!) just choosing EXPO. I manually changed back to 1.08v and it's been good since.

@AusWolf, are you on AGESA 1003 Patch A?

 

[AusWolf]

Normal cold boots should be about 30 seconds to the post screen. If you mean after a BIOS reset, or possibly removing power from the board/box, then that could be longer.

I had an old GB board on Sandy Bridge that did what's being described. Disconnect power from the box, and the first boot would hang, requiring a forced power-down by holding the button until the board shut off. After that, it was golden until power was removed from the box again.

Guessing, it's probably hanging on the memory init. (training, whatever) due to unstable settings from EXPO. Double-check SoC voltage - mine set 1.3v(!) just choosing EXPO. I manually changed back to 1.08v and it's been good since.

@AusWolf, are you on AGESA 1003 Patch A?

Yes, that's the first and so far only version that came for my board.

I'll check the SoC voltage as you suggested, but won't be able to report back my findings until tomorrow. I get these hang-ups literally during cold starts only.

 

[tabascosauz]

The longest I've waited was around 5 minutes. The funny thing is, if I get fed up, and restart the PC, it works fine.

Should this happen at EVERY cold start?

Sounds like bad RAM settings. I know AM5 is a bit slow but on DDR4 I had that experience once with a mystery meat Corsair LPX kit, took like 10 minutes to POST every time.

ASRock said they later reduced boot times, early AGESA is always a wild ride

 

[AusWolf]

early AGESA is always a wild ride

This is my main guess, to be fair. I think I read somewhere that AMD is planning on reducing memory training times with future AGESA updates.

Guessing, it's probably hanging on the memory init. (training, whatever) due to unstable settings from EXPO. Double-check SoC voltage - mine set 1.3v(!) just choosing EXPO. I manually changed back to 1.08v and it's been good since.

I've checked, EXPO sets it to about 1.3-1.38 V. I changed it to 1.1 V, we'll see if it does anything. Also, I found a setting that lets the motherboard skip memory retraining whenever possible. I've enabled that, too.

 

[RJARRRPCGP]

Disconnect power from the box, and the first boot would hang, requiring a forced power-down by holding the button until the board shut off. After that, it was golden until power was removed from the box again.

Sounds like CPU core OC'ing while CPU load-line-calibration is disabled. That happened to me on socket 775 on my Asus P5QL Pro with my Core 2 Duo E4500.
It happened one time after I forgot to enable CPU load-line-calibration. Had to retry to cold boot, but then was able to get into the BIOS, Windows and have stress tests pass.

 

[Mussels]

Looks like I won't be getting AM5 for a good while!

Me neither, but the silly heatspreader isnt why

Yes, that's the first and so far only version that came for my board.

I'll check the SoC voltage as you suggested, but won't be able to report back my findings until tomorrow. I get these hang-ups literally during cold starts only.

cold boot issues sound like the RAM or IMC settings are incorrect, and it then loads the JEDEC defaults and posts

You shut it down, it then tries to boot the bad settings, repeat. BIOS can show you say 3600 on the ram, but windows will show it's at 2133 when this sort of thing happens.

SoC is pretty likely to be all you need to change, but you'd wanna double check you've got XMP and no overclock/PBO settings anywhere (the various menus can conflict, you can do a per-CCX overclock, a curve undervolt and a static voltage and good luck knowing which one is active) - sometimes a clear to defaults is a good idea

 

[AusWolf]

cold boot issues sound like the RAM or IMC settings are incorrect, and it then loads the JEDEC defaults and posts

You shut it down, it then tries to boot the bad settings, repeat. BIOS can show you say 3600 on the ram, but windows will show it's at 2133 when this sort of thing happens.

SoC is pretty likely to be all you need to change, but you'd wanna double check you've got XMP and no overclock/PBO settings anywhere (the various menus can conflict, you can do a per-CCX overclock, a curve undervolt and a static voltage and good luck knowing which one is active) - sometimes a clear to defaults is a good idea

Nope, it's running 6000 MHz EXPO all day and night, confirmed in Windows.

Do you think PBO affects SoC voltage and I'd be good to go by just disabling PBO?

Me neither, but the silly heatspreader isnt why

Why would you need AM5 when you're on a 5800X3D?

 

[freeagent]

I don’t know anything about AM5, sounds like it could use some vddg iod voltage?

 

[Ruru]

The audio problems still persist. Hell, I'll try going back to Win10 if the problem is just a Win11 thing.

 

[tabascosauz]

Nope, it's running 6000 MHz EXPO all day and night, confirmed in Windows.

Do you think PBO affects SoC voltage and I'd be good to go by just disabling PBO?

Why would you need AM5 when you're on a 5800X3D?

PBO does not affect SOC voltage. If EXPO = XMP in function then setting that would automatically set VSOC as well, which is to be expected. 1.3V default VSOC........damn........reminds me of auto VCCSA going up to like 1.6V. Is it safe? Who knows?

I don’t know anything about AM5, sounds like it could use some vddg iod voltage?

There hasn't been too much OC coverage of AM5 (I guess they took one look at DDR5 OC on Raphael and were like nah ), but the voltage domains seem to have changed if you are familiar with what AM4 looks like.

I have seen a lot of reference to the new Vmisc rail, but not that much about how it behaves. iirc the VDDGs were previously subsumed under VSOC (except VDDP)? So this should actually be a pretty big change.

Sad to see that VDDCR_GFX hasn't been split off of VSOC (all of the Vdroop woes right now come from that fact)...

...but VDDCR_VDDM is interesting if it is what I think it is. It makes me think that with AM5 AMD is making preparations for a separate volt domain for Vcache (which they don't currently have), even if it still technically comes from VDDCR_CPU.

From the diagram it kinda looks like VSOC is now only responsible strictly for IO die elements, not any of the interconnect.

The audio problems still persist. Hell, I'll try going back to Win10 if the problem is just a Win11 thing.

Not much to lose to try it

get a nice clean Windows install while you're at it

 

[Count von Schwalbe]

makes me think that with AM5 AMD is making preparations for a separate volt domain for Vcache

One can only hope... Imagine a 7800X3D fully unlocked except cache!