Subbed

 

I am curious on where to get that PBO2 Tuner app, I know its from someone here over on OCN, but I also think its not publicly available.

 

No one told me that, I'll post it tomorrow

To not clutter the gaming results since it will be a lot of screen shots when complete, here is a shot of hwinfo during 16 thread run of cpuz bench to show VIDs etc being requested.
When interpreting temps, consider my coolant is 51f while this is running, room ambient is 84f, so on a ambient radiator based loop I would be at least +18.6c hotter on the cores, but more realistically about 24c hotter using a 10f air/water delta for ambient loop math (sorry for interchanging units, it's how I think):

Capture of the same during 1t boosting:

 

Nice info and testing! I wish they had allowed for negative CO, would have gotten a nice multicore boost (up to 6% if -30 CO is possible), also would liwered temps in SC scenarios.

 

Will probably be joining the club on release day.

 

My gb5 as comparison.

Micro-Star International Co., Ltd. MS-7C35 vs ASUS System Product Name - Geekbench

the difference for aes, speech and machine learning are huge.

 

Awesome initial write up. I’ll be jumping on the bandwagon as soon as it is released.

Quick question and hoping somebody can assist -
Does anybody know if the crosshair viii dark hero has an external clock generator and also supports the voltage suspension feature available in the crosshair viii extreme?

thank you kindly.

 

Awesome initial write up. I’ll be jumping on the bandwagon as soon as it is released.

Quick question and hoping somebody can assist -
Does anybody know if the crosshair viii dark hero has an external clock generator and also supports the voltage suspension feature available in the crosshair viii extreme?

thank you kindly.

it does have an external generator it can do what I'm doing, it is missing voltage suspension but voltage suspension is the least important of all the technologies to help overclocking using the Crosshair and x3d CPU. I'm using a Crosshair 6 extreme at the moment which is a 5-year-old Zen One release and that's what all the scores posted above were obtained with. I chose this over my Crosshair 7 Hero because, frankly, the C6E is a better board in almost every regard (boot times being the main exception).

Nice info and testing! I wish they had allowed for negative CO, would have gotten a nice multicore boost (up to 6% if -30 CO is possible), also would liwered temps in SC scenarios.

Curve Optimizer can be used using the pbo2 tuner but it doesn't need to be. That's why I mentioned it behaves differently than other Zen 3 CPUs, simply dropping the Curve for each core does not necessarily produce a better results like it does on all the other Zen 3s.
It's so low voltage stock that it ends up dropping voltage to a point where it no longer boosts properly and instead gets stuck at like 4 ghz when using more than -10 CO values for my sample. It's also not TDC/EDC/PPT limited through virtue of not hitting the limiters, so less power doesn't help anything.

The only time I've observed similar behavior on normal Zen3 is when PPT is cut so far down the CPU now doesn't have headroom to boost right, like PPT 50 on a 5950x.

 

Some good info for sure. I may grab one of these if I can manage to get it. Anyone know what time of day they will go up Wednesday?

 

I was shocked they shipped mine early, had no idea it was going to happen (might have been a mistake tbh), so I was unprepared and still waiting on the Crosshair 8 and second memory waterblock to arrive to do the 5950x comparison data hah.

I would see if you can find a place like B&H that has them on "Waiting list only" so you get a notification the instant they drop if you want one. Despite a lot of nay-sayers, I have feeling it's going to sell like Zen 3 originally did at launch...quickly.

 

Hmm also on a side note it seems Ryzen master now also has access to Curve Optimizer

https://www.tomshardware.com/news/curve-optimizer-heads-to-ryzen-master-for-zen-3-cpus

 

Hmm also on a side note it seems Ryzen master now also has access to Curve Optimizer

https://www.tomshardware.com/news/curve-optimizer-heads-to-ryzen-master-for-zen-3-cpus

View attachment 2556621

Good to know for those who use RM. I refuse to ever touch it again, but at least the option is out there.

 

PBO2 Tuner allows CO and PBO limits to be manipulated, do so carefully if you it, it doesn't behave quite the same as other zen3s do. For those who used a Zen 2 x cpu and a Zen 2 xt cpu, it boosts more like the xt variant. Flatlining target speed if temp headroom is available instead of floating like Zen 2x and Zen 3 Dual CCD chips.

"PBO tuner" is capable to move limits bellow AMD CBS supplied power limits which are for Skatterbencher

PBO Tuner on first usage needs to "reset" the values
It only works till AMD allows it
SMU Mailbox is locked in default state, but HSMP is identical and "not" locked ~ for now

RSMU has a firmware lock, which users of new chipset drivers, ryzen master or AGESA 1206C will get
Asus 1206B is questionable if it's nature is indeed B and not C ~ yet soft-toggles on patch B are defaulting on disabled, but are not hardlocks

Sample moves under same microcode as every other Vermeer B2 CPU
~ support for it should exist since before AGESA 1.2.0.3C but CO boosting curve behaves better after 1204 AGESA for all B2 units & cezanne
CPUID 00A20F12h not 10h

L3 cache does not heat up, used thick interposer onto cores, translates heat badly
4.5Ghz Freq should usually require near 1.2v - but sample can be indeed more hungry
L3 cache matches core Freq, but L3 cache does not match coreVID. The "reason of worry" is complicated.

 

Sample moves under same microcode as every other Vermeer B2 CPU
CPUID 00A20F12h not 10h

Yeah it looks like his sample was much more voltage hungry than mine or the lower temps I'm holding are impacting behavior.

I am using the recent Gigabyte published Chipset drivers on this OS build, published on 14 April, 4.04.11.742 WHQL here: X570 AORUS MASTER (rev. 1.1/1.2) Support | Motherboard - GIGABYTE Global

I'm on 1.2.0.6b now, 8503 Bios for C6E. What does the shift from 10h to 12h change in regard to behavior?

 

Leave 1 check in the Physical cores section, disable swap cores, use small data set instead of large and it'll find instability, almost guarantee it Large data set is cool, easy to run, and less far less harsh on the core as fair as error testing. More a FCLK/IMC test.

As I outlined above it will walk between cores, 1 at a time, and boost to 1t max speeds instead of testing at full load boost levels all the time as it is in your screen shot, which is much easier to pass when CO testing (and leads to false stability beliefs, which then manifest as "random restarts" in normal PC use).

 

Started fiddling with exploring going faster than 4795 1t/4690 16t and ran into the first sign of trouble, but also promise...

At 105.8 I ran into single core boost instability; it needs more voltage. This is where CO will help. Initially I bumped vcore offset 1 more notch, to 0.01875, but the same results, 3dmark 1t triggers a instant reboot, similar to how 5950x behaves with too much negative CO applied.

Backed vcore offset back down to to +0.0125 and working on minimal values of positive offset required for single core stability at 4813mhz 1t boost. 1.194 SVI2 is simply not enough at that speed on Zen 3.

Testing higher now and things are acting up more, will require more involvement to get stable without resorting to just shoving vcore offset at the problem. 107.4 up and running....

Now beating stock 5800x and many OC results I've seen of mediocre PBO only 5800x OCs... we're not done yet, not by a long shot

It has stopped flatlining all core boost at max multiplier, a sign PB2 is starting to run into some challenges with the power budget of 142/95/140. ////Now entering "you'll likely damage the CPU if you do this on AIR/AIO/Ambient water" territory.////

AMD Ryzen 7 5800X3D @ 4778.69 MHz - CPU-Z VALIDATOR

[fc8ph3] Validated Dump by OLDROG (2022-04-17 22:45:28) - MB: Asus ROG CROSSHAIR VI EXTREME - RAM: 32768 MB

valid.x86.fr

 

Started fiddling with exploring going faster than 4795 1t/4690 16t and ran into the first sign of trouble, but also promise...

At 105.8 I ran into single core boost instability; it needs more voltage. This is where CO will help. Initially I bumped vcore offset 1 more notch, to 0.01875, but the same results, 3dmark 1t triggers a instant reboot, similar to how 5950x behaves with too much negative CO applied.

Backed vcore offset back down to to +0.0125 and working on minimal values of positive offset required for single core stability at 4813mhz 1t boost. 1.194 SVI2 is simply not enough at that speed on Zen 3.

Testing higher now and things are acting up more, will require more involvement to get stable without resorting to just shoving vcore offset at the problem. 107.4 up and running....

Now beating stock 5800x and many OC results I've seen of mediocre PBO only 5800x OCs... we're not done yet, not by a long shot

It has stopped flatlining all core boost at max multiplier, a sign PB2 is starting to run into some challenges with the power budget of 142/95/140. ////Now entering "you'll likely damage the CPU if you do this on AIR/AIO/Ambient water" territory.////

AMD Ryzen 7 5800X3D @ 4778.69 MHz - CPU-Z VALIDATOR

[fc8ph3] Validated Dump by OLDROG (2022-04-17 22:45:28) - MB: Asus ROG CROSSHAIR VI EXTREME - RAM: 32768 MB

valid.x86.fr

Hello, I am impressed of your score. I tried bclk oc for months now and I am still working on it. Do you have bclk divider option in your bios? And do you have any usb dropouts / issues when running that high oc?

 

107.4 Results so far, as I mentioned earlier, AIDA 64 isn't the X3D's forte, but these numbers are a big improvement from stock:

 

Well that isnt even a bad Aida score tbh. Regular 5800x but thats kinda the limit for it at 3800mhz.

 

Think my frame of reference is skewed for AIDA from my 5950 lol, was shocked to see latency as high as it is at first on X3D, but the CPU speed is probably a big hindrance in that test.

 

Would you mind trying y-crunsher?

 

I just published on Hwbot Bloated windows 10, almost default install so efficiency is garbage but there they are.

I'll eventually get around to throwing high effort scores up, but not until people start posting ones that beat mine. Still carefully exploring behavior as there are some peculiarities involved vs X series SKUs.

Ultimate personal goal is 45.5x110 stable for 5005 mhz 1t in daily config. Not sure I'll get there without some rather extreme measures but I'm going to try.

 

AMD is supposedly working on it.

AMD Reportedly Working to Enable Overclocking on the Ryzen 7 5800X3D | Hardware Times

AMD’s Ryzen 7 5800X3D is one of the first CPUs from the company to launch with a locked multiplier, preventing proper overclocking. As usual, enthusiasts have found ways to push the chip beyond its marketed boost clock of 4. (classicsofttrim.com) 5GHz. By completely ignoring the multiplier and...

www.hardwaretimes.com

 

CO, they're going to officially add CO. What else they add is entirely anyone's guess, but don't hold your breath for PBO or multipliers over 45.5. I would expect them to unlock multipliers below though, ala non-k Intel CPUs.

I suspect the delay is moving CO outside the PBO menu or ensuring PBO options besides CO are disabled.

As it is, PBO "can" be enabled through trickery, but boosting is disabled so they've done shenanigans in the Microcode we can't see/access, trickery or not.

PS: All these cute cpuz 5000+ are gilding only at least so far. I have zero confidence they can actually run a heavy load unless on LN2, even then, somewhat unlikely, but not impossible for the brave to do it.

 

What's the trick to enabling PBO? Is this something that I could enable on an Asus B550-F?
(In your first post you mention PBO2 tuner works, so I assumed that was just through the bios).
Thanks in advance for the info.

 

Lol, ...preventing "proper" overclocking...

 

Indeed. If only they had have left multis under 45.5 open, and allowed DOCS to function daily setups with very high single and reasonable but safe all
core performance could've been very easily obtained without breaching 1.3. shame.

What's the trick to enabling PBO? Is this something that I could enable on an Asus B550-F?
(In your first post you mention PBO2 tuner works, so I assumed that was just through the bios).
Thanks in advance for the info.

No, sadly it won't enable functionality that doesn't exist in a bios already. PBO2 Tuner is software, linked on first page.

Does someone try what happens if boot with older BIOS (AGESA)?

Will be doing so sometime after the 21st.

 

Does someone try what happens if boot with older BIOS (AGESA)?

 

Enabling CO would belp, especially for those with poor cooling even though it may not work good below -10 acvording to one owner.

 

That is me, -10 or greater at stock BCLK was trigger boost malfunction, however, on hot systems that won't happen as they wont peg 44.5 solidly in the same way so it'll have a wider range of usefulness.

I hope the fTMP bug gets fixed first, tbh

Yeah, it's an issue. I just disable TPM, set bios to use discrete so it doesn't even look at fTPM, and stay on Win10. Tried 11, wasn't impressed with performance regression so swapped back.

Are there any cheap, as in <=$200, B550 or x570 motherboards with external clock generators?

None I am aware of. MB shopping for me involves 2 steps: 1 - Look at Asus product stack, 2 - Buy the board that has the feature set I want.


-----
3080ti seated, cobbled together loop assembled for 3 days of testing before the Crosshair 8 Extreme arrives and I do a full rebuild on both PCs, functional but ugly lol.

3080ti doesn't tolerate BCLK over 102.2 so far, not bothered to try and improve this until on C8E.

Scores going up today, it's...pretty good and more than expected 2080ti vs 3080ti scaling in some cases, will update 1st post. Initial low effort 3dmark scores are ranging from top 3 to top 5 compared to 5800x + 3080ti leaderboard at 102.2 bclk. I'm using daily clocks and the gigabyte vbios at this point, there's a good bit of headroom left for 3dmark scores.

 

Are there any cheap, as in <=$200, B550 or x570 motherboards with external clock generators?

 

I hope the fTMP bug gets fixed first, tbh

 

A little high quality test for others to compare to more normal gaming settings, 3440x1440 vsync off:
Settings:

Results:

 

Any owners care to test ram tuning vs stock? tcclaviger? Going from Zen2 to Zen3 I noticed that tuning ram didn`t matter that much (Zen2 often gave me up to 30% going from 3200cl16 xmp to 3800cl15 tuned B-die, Zen3 seldom gave more than 20%, probably due to larger L3 cache improving memory related performance). Any of you care to teest for instance SOTTR 1080lowest and see what CPU game avg you get stock vs tuned? My bet would be that you probably won`t get more than 10%, but I could be wrong.

 

Would point to Hardware unboxed day 1 review, he did 3200 and 3800 on the otherwise same config, as did skatter Bencher, showing the ram impacts.

I'll run a full "Auto" memory test at ~3600 of SotTR later to compare to the 1st post #s. Auto-3600 is probably what most normies/noobies set it at so seems most relevant (like Ryzen master docp/xmp doesn't exist in my world).

I read through the thread, and there is a lot of interesting results and feedback. Thanks for all the testing OP.



Question - Will you be benching games against your overclocks? E.g. stock X3D vs OC'd X3D.



It doesn't seem like OCing the chip has any impact over the general scaling that the L3 cache provides. But just curious to know how it holds up. Since most reviewers have covered nearly every popular game but without OCing, you may be able to generate some interesting data.



If its too much work, I understand

The OC does show gains, in every title I've tested except weirdo console ported AC Odyssey. OC vs stock is in 1st post pictures. Called it stockish because RAM is tuned so it's not stock.

Mostly I am OCing to gain back general computing performance outside gaming, which it's done as well as can expected for a very constrained OCing window.

SIC...

EDIT: Check Hardware unboxed videos, these notions are demonstrablly false.


X3D in 40 games and Core count vs cache size videos. Extrapolate using 12900k using ddr5 review vs 5900x gaming defecit and you'll see a massive gain from 5900x to X3D, most certainly not limited to low res.
40 Games tested
Core count testing
Cache vs core count

 

It's a pretty amazing cpu, the best @1080, given into account the low power consume and the good all-round performance for non-pro users. It has some edge at 1440 in some games but in the long run a cpu like the 5900x will be better suited for gaming @1440+ given the many cores. Games are going to use more and more cores and if someone like myself already play at 3440*1440 the next step is 4k.. And there you won't see any difference but core scaling if any.

Anyway, a great cpu to squeeze till the last the am4 platform, let's say anyone on a x370/470 or b450 with a 1st/2nd gen ryzen processor whose primary focus is gaming.

 

in the long run a cpu like the 5900x will be better suited for gaming @1440+ given the many cores. Games are going to use more and more cores

Games have aspects that aren't easy to parallelize and there are hugely diminishing returns after a certain point. There is also the fact that a large portion of games are cross platform where eight-cores are expected and negligible effort is made to parallelize them further.

By the time a 5900X is significantly faster than a 5800X, let alone 5800X3D, in more than a tiny portion of CPU limited games, Zen 3 will be several generations old and quite long in the tooth. I'm not going to be using my 5800X3D in a front line system anywhere near that long.

And there you won't see any difference but core scaling if any.

That depends on the game. I typically play at 1440p or 4k, and I have games where my 5800X cannot push enough frames to make my GPU the limiting factor. It's true that these are poorly optimized and unusually serial titles, but I cannot rewrite them, so the only way to get more performance out of them is to brute force things with faster CPUs and memory subsystems.