Would it be worth it to upgrade to a B550/X570 motherboard from a B450 motherboard?

[Mussels]

Your theory only holds true to antique flash drives that are slower than USB 2.0, not 2.0 itself.

stock on this board on the 5700X is broken, stuck on base clock for boost

That only happens if you manually change settings, if you alter Bclk for example or set a manual multi or voltage on certain setups, it disables boost.
Simply load defaults and/or update the BIOS if its too old for that CPU.

 

[Kodehawa]

That only happens if you manually change settings, if you alter Bclk for example or set a manual multi or voltage on certain setups, it disables boost.
Simply load defaults and/or update the BIOS if its too old for that CPU.

I'm on the latest BIOS available. I tried no PBO and I'm wrong though, base is 3.4GHz and it boosts to 3.6~3.8GHzw/o PBO.

Still kinda weird, Gamers Nexus got ~4.1GHz w/o PBO.

 

[Chrispy_]

What's your cooling solution? Is it still the old Hyper 212 in a closed PC case in a warm room? I'm going to assume Gamers Nexus tested with their usual air-conditioned, open-bench setup using an Arctic 360mm AIO, chances are good the CPU was at ~60C at 4.1GHz using ideal conditions for the AIO.

Default Ryzen 5000 precision boost is governed by a silicon health agent that aims to keep the silicon within optimal temperature ranges. If you're reaching 80C at 3.6-3.8GHz, then you're limited by the cooling.

Contrary to popular belief, Ryzen will not unconditionally boost to its thermal throttle limit on most boards. Those OC-focused boards override the CPUs own silicon health agent to get the most out of the overclocking potential. Vanilla boards with more standard firmware pay more attention to the silicon health agent on the CPU. When it reports that it's getting hot (so ~80C) It will stop asking for more power because it knows there's no additional cooling available. It's not throttling. It's just deciding that there's no point adding more voltage because it's already hot and more voltage will make it exponentially hotter.

If you crank your Hyper212 fan speed up to maximum and ensure your ambient air temperatures are as low as they can be (remove side panel and turn on air conditioning in the room, for example) then you'll probably see higher boost speeds. About 80C is the point at which the silicon health agent starts dialing things back...

 

[Kodehawa]

What's your cooling solution? Is it still the old Hyper 212 in a closed PC case in a warm room? I'm going to assume Gamers Nexus tested with their usual air-conditioned, open-bench setup using an Arctic 360mm AIO, chances are good the CPU was at ~60C at 4.1GHz using ideal conditions for the AIO.

Default Ryzen 5000 precision boost is governed by a silicon health agent that aims to keep the silicon within optimal temperature ranges. If you're reaching 80C at 3.6-3.8GHz, then you're limited by the cooling.

Contrary to popular belief, Ryzen will not unconditionally boost to its thermal throttle limit on most boards. Those OC-focused boards override the CPUs own silicon health agent to get the most out of the overclocking potential. Vanilla boards listed to the silicon health agent on the CPU more carefully. When it reports that it's getting hot (so ~80C) It will stop asking for more power and that's your boost limit. It's not throttling. It's just not boosting up to the voltage/power limit of the board because your cooler isn't keeping the silicon health agent happy enough to keep going further.

It's a Cooler Master Hyper 212 Turbo ARGB, dual fan tower cooler. Don't think they sell it in the US but it keeps the CPU rather cool. I've never seen it get over 80C, even with PBO at 120W PPT it goes to 67~71C, at stock clocks w/o PBO it barely crosses 51C.

Usually on a 10 minute test it settles on around 73C, depending on room temp (it's 25C now)

 

[Chrispy_]

It's a Cooler Master Hyper 212 Turbo ARGB, dual fan tower cooler. Don't think they sell it in the US but it keeps the CPU rather cool. I've never seen it get over 80C, even with PBO at 120W PPT it goes to 67~71C, at stock clocks w/o PBO it barely crosses 51C.

Weird. I'm not 100% sure how the silicon health agent determines when a CPU is too hot to start dialling things back, but 71C seems low to me.

What happens if you manually enter a 95C thermal target in PBO settings?

 

[Kodehawa]

Weird. I'm not 100% sure how the silicon health agent determines when a CPU is too hot to start dialling things back, but 71C seems low to me.

What happens if you manually enter a 95C thermal target in PBO settings?

I had to set it to 85C, if I set it to 95C the effective clocks start dropping to the ~4.21GHz range with PBO. I have 0 clue why lol

It also dials it back now, it starts at 4.35GHz and will slowly drop to 4.25GHz over a 10 minute test (even without reaching 75C)

 

[Chrispy_]

Install Ryzen Master if you haven't already.
That gives you more visibility on what's limiting your boost.

 

[Kodehawa]

Install Ryzen Master if you haven't already.
That gives you more visibility on what's limiting your boost.

With everything set to default. Seems to be PPT.

Interesting though, I wonder if GN's board set the default non-PBO PPT higher, or something.
Wasn't AMD's default PPT for 65W chips 88W, not 76W?

EDIT: Yep, it is

Default for Socket AM4 is at least 88W on motherboards rated for 65W TDP processors

Why the heck is this board setting it to 76W lol

With "motherboard" PBO limits (stupidly high, but boost still doesn't go over 4.3GHz):

Setting PPT manually to 115/80/115 has yielded me the best results at about 4.35GHz, even if it's constantly PPT limited.

 

[Mussels]

I'm on the latest BIOS available. I tried no PBO and I'm wrong though, base is 3.4GHz and it boosts to 3.6~3.8GHzw/o PBO.

Still kinda weird, Gamers Nexus got ~4.1GHz w/o PBO.

your hyper 212 isnt going to cool that CPU well, so it wont boost as high


Different loads clock at different boost levels, different boards alter that too

 

[Kodehawa]

your hyper 212 isnt going to cool that CPU well, so it wont boost as high


Different loads clock at different boost levels, different boards alter that too

It cools it rather well, for what I see at least. ~71C at 120W, 52C or so at default PPT without PBO.

But yeah, kinda interesting.

 

[Chrispy_]

Why the heck is this board setting it to 76W lol

76W is the correct PPT for the 5600, 5600X, and 5700X.
The old "PPT is 1.35x the TDP" rule seems to no longer be a guaranteed rule.

 

[DirtyBiker]

Keep the 450 upgrade your CPU get Ryzen 5 5600 or the Ryzen 7 5700 unless your video card is gimped with pcie 4X or 8X like some of the lower end model AMD Nvidia variants out today that require 4th gen pcie to run properly

 

[Dr. Dro]

It cools it rather well, for what I see at least. ~71C at 120W, 52C or so at default PPT without PBO.

But yeah, kinda interesting.

For a Hyper 212 on a closed case plus low end B450 motherboard, I think you're getting well expected performance.

The answer here doesn't seem to go after PBO: it's actually running your CPU through the curve optimizer so you can lower its heat output, leading to higher clock averages. Good luck.

 

[Mussels]

To be clear i'm not saying the 212 needs to be replaced - i'm saying temps will be weird with it, since you dont have a solid base you could have 3 cores cooled amazingly and 3 10c hotter if the have the alu over them - same for the IMC. Temps can be very uneven since the design was for older monolithic CPU layouts.


PBO and undervolting is the way to go, not more wattage. sustained clocks is better than tiny boosts that never stay there.

 

[Kodehawa]

To be clear i'm not saying the 212 needs to be replaced - i'm saying temps will be weird with it, since you dont have a solid base you could have 3 cores cooled amazingly and 3 10c hotter if the have the alu over them - same for the IMC. Temps can be very uneven since the design was for older monolithic CPU layouts.


PBO and undervolting is the way to go, not more wattage. sustained clocks is better than tiny boosts that never stay there.

Ah huh. That's interesting! I didn't know that.

Yeah, I've tried CO and it gives uhh, about 4.5GHz with a mix of -5 on the good cores and -15 everywhere else. Seems fine that way, but testing CO stability is pain I've heard.

 

[Dr. Dro]

Ah huh. That's interesting! I didn't know that.

Yeah, I've tried CO and it gives uhh, about 4.5GHz with a mix of -5 on the good cores and -15 everywhere else. Seems fine that way, but testing CO stability is pain I've heard.

If you want to ensure you've extracted every last drop out of the CPU, on a per core level, then yes. You can use CoreCycler to help you with this, though. Run it overnight

 

[Mussels]

Ah huh. That's interesting! I didn't know that.

Yeah, I've tried CO and it gives uhh, about 4.5GHz with a mix of -5 on the good cores and -15 everywhere else. Seems fine that way, but testing CO stability is pain I've heard.

It's a lot easier now with corecycler and ryzen master having per-core testing, you just need to make sure everythings on defaults before you use it (with the exception of if you're doing +MHz, set that one first)

 

[Kodehawa]

Ok so I did buy a new board (ROG STRIX B550-F GAMING WIFI II). Not for the reason I wanted to at first but because I'mma build a computer for a friend with other parts I have laying around and I decided to get a new motherboard and call this a completely new computer instead of having to build him a computer in a crappy A320 or something.

Wanna add a few notes:
- fTPM stutter is actually fixed now, on my old board I couldn't drag a window without it stuttering. Weird, it said it was fixed on the old one but nop, so system usage feels a LOT better now.
- My active speakers had a constant and annoying humming on my old board (like, I had to turn them off when I wasn't using my PC). No more.
- This board pushes more voltage to the CPU. Doesn't do anything, just of note.
- It's so much nicer to build on this board, geez.
- CPU/RAM overclocking is the same. What was stable on the old board is stable here (like RAM), no improvement.
- Built-in Wi-Fi and BT is pretty nice, specially BT for my Xbox One controller. I had a dongle, but it dropped out all the time.
- Both of my SSDs perform considerably better, even though they were PCIe 3.0. Dunno why. My secondary went from 800MB/s seq to 1200MB/s and my primary went from 3100MB/s to 3600MB/s
- Hey, it has 2.5G ethernet. My ISP is now selling 2Gbps internet, and it's not much more expensive than gigabit. No real reason to upgrade though, but I imagine they'll eventually upgrade everyone like they have always done (I started at 50Mbps in 2017 and now I have gigabit at the same price, they just slowly upgrade you... silently)
- I actually have a 3 pin ARGB connector now, but my cable doesn't reach it. Time to buy an extension lol
- If anyone here plays osu!, audio latency is CONSIDERABLY lower. To the point an osu! player might actually want to spend this much just to get that ~30ms of audio latency cut off.
- NVMe drivers appear on the UEFI. Yes, my older board decided NVMe drivers wouldn't show up anywhere on the UEFI. As if they didn't exist at all.
- Very pretty. At least to my liking.

I wouldn't say it's worth it on performance alone, but the audio is so much better that it kinda blew me away. My headphones are now about twice as loud lol, I have to use them at 30% now, could do 100% before. I wonder if the audio on the old board was just crap.

If you have a decent B450, stay on B450, but I imagine that was clear on this thread.

 

[Mussels]

Sounds like the old board was garbage, from all those problems

You sure it was fTPM stuter and not something like the VRM's?

- I actually have a 3 pin ARGB connector now, but my cable doesn't reach it. Time to buy an extension lol

I highly recommend NOT installing asus motherboard software, use openRGB instead. The motherboard header works fine with openRGB, or you can use any external third party controllers you want

 

[Kodehawa]

Sounds like the old board was garbage, from all those problems

You sure it was fTPM stuter and not something like the VRM's?


I highly recommend NOT installing asus motherboard software, use openRGB instead. The motherboard header works fine with openRGB, or you can use any external third party controllers you want

Yeah. It wasn't the best board out there for sure, it was like $80... in 2018. Not having a constant humming on my speakers is always good lol. I thought it was the speaker's fault for YEARS. Nope, was the motherboard. Can you believe I actually feel kind of odd without the constant speaker humming? You know, like 5 years of it was enough to make me used to it existing, lmao. Silence is too silent.

Yep, disabling fTPM fixed it, but then W11 complained. Can't be too sure tho, stutter was just less annoying, and games didn't really stutter. Could've been USB dropouts causing mouse input to cut.

Will do, OpenRGB is pretty good. I'm pretty aquantied with it.

 

[Dr. Dro]

That's awesome to hear. My old motherboard (B550-E) was a stupidly buggy mess for the longest time, but the latest AGESA updates for the Strix family (-F, -F II, -E and -XE) are rather solid. As solid as AM4 as a platform can be, anyway. ASUS just posted the AGESA v2 1.2.0.8 BIOS to their website, fixing a high-severity security exploit. You should flash that and call it a day.

 

[Kodehawa]

That's awesome to hear. My old motherboard (B550-E) was a stupidly buggy mess for the longest time, but the latest AGESA updates for the Strix family (-F, -F II, -E and -XE) are rather solid. As solid as AM4 as a platform can be, anyway. ASUS just posted the AGESA v2 1.2.0.8 BIOS to their website, fixing a high-severity security exploit. You should flash that and call it a day.

ty.
Yeah, I flashed the newest BIOS right out of the gate, when I first booted it up, so it's running 3002 now.
It's the v2 version, so at least I didn't need to use the flashback function to get it to boot, as it released after Zen 3 came out.

I found it kinda funny that one of the stages of the update was updating the LED firmware. It's doing a fine job of lighting up the ROG logo, I dare say, haha.

Also another note, I can actually get the CPU to draw more than 117~120W on this board. Well, its not a *good* idea with my cooler, but it was a motherboard limitation, seemingly. I got it to 139W, but that does nothing useful and only makes the chip hotter (got to a nice 89C after 30m of p95 lol), so I have it PPT limited to 115. Runs the same anyway and doesn't go over 75.

 

[Dr. Dro]

ty.
Yeah, I flashed the newest BIOS right out of the gate, when I first booted it up, so it's running 3002 now.
It's the v2 version, so at least I didn't need to use the flashback function to get it to boot, as it released after Zen 3 came out.

I found it kinda funny that one of the stages of the update was updating the LED firmware. It's doing a fine job of lighting up the ROG logo, I dare say, haha.

Nice. I know right. It's just the ROG logo, at least it was on the B550-E. It has some RGB headers, though, I don't know how relevant that may be, probably if you have a full-blown Aura Sync setup I suppose.

Your -F II board should be a "lite" version of the one I had, with the B550-F's 12-stage power configuration and features (missing debug code reader, etc.), but updated to look more like the -E and adding its WiFi module that the original -F didn't have. Should be a pretty solid motherboard, IMO.

 

[Mussels]

Could've been USB dropouts causing mouse input to cut.

That's usually SoC voltage related, the older boards often had lower voltages there or worse vdroop on them - but i've definitely ran into some super garbage boards (B450's from MSI specifically, but i've read about other brands just as bad) that were so badly underspec they had issues even on 65W CPUs

 

[freeagent]

Yeah the -F Wi-Fi II is a nice board. Doesn't have Intel Lan or Wi-Fi, and has a TB header. I was going to buy it, but was like 320 beaver bucks at the time. I ended up snagging the XE locally on sale for just a little more than that board was at the time. The XE a sweet board, kind of like the -E but better. I have the first gen -F running as my sons rig. Its been solid for me as I ran the piss out of it, now it lives a boring life playing Roblox driving a 5600X with some really crap Adatas.