Ryzen Owners Zen Garden

[tabascosauz]

A new BIOS setting got me all moist in my panties, i'll let you guys know if it works, or just a broken unlockable option

View attachment 273701
attempt 1: (it has a few choices)
nope
View attachment 273700


4408 has the new stuff, not 4403

Okay so that features doesnt do anything

set what i previously had and CO doesnt seem to be working oddly enough (and it WAS working)

1.275V and 75C says stock

I'm pretty sure 4303 is the 1208 beta and 4408 is the release. 4303 was released through forum thread only

An external clock gen isn't something you get for free in firmware. The board has to have that hardware. It provides slightly higher BCLK oc. I know the X570 Taichi clearly advertises one (Hyper BCLK or some name like that); people saying a bunch of MSI X570s have it. Not sure why Asus exposed it in software for all boards I think the C8E has one not sure about any other C8 or Strix

PCIe redriver is another piece of hardware that is not on all boards.

It seems that these options have existed under AMD PBS for some time, just hidden because they're irrelevant to most boards.

https://www.google.com/url?sa=t&source=web&rct=j&url=https://download.asrock.com/Manual/X570D4I-2T.pdf&ved=2ahUKEwiGyOfDiO37AhX7GTQIHa-DANkQFnoECCAQAQ&usg=AOvVaw3uTZQ6b1AeAzQLunofLMAd

 

[Mussels]

Tis all working as expected now
(And the clockgen was from the BIOS mod, the main issue is that if you go to even 100.01 you lose all boost/turbo and get locked to base clock)

Not something i'm sure is the BIOS change but might just be the CPU change, but i dont need as high an SoC voltage now and i'm tuning that in a lot lower

On the 5800x i needed ~1.13v on the SoC

I honestly didnt check and just copied the settings moving to the x3D
Re-testing on the new BIOS, and i only need 1.05v so far - testing lower to reduce wattages

voltage offset of -.061 is giving me full speed on effective clocks, seeing slightly higher wattages and temps than previous bios - but also slightly higher performance (as just about everyone is on this - people seem to love this AGESA, but its changed the curve offsets and people are needing to re-tune theirs)

Cant recall previous scores, but seems about right

 

[tabascosauz]

Tis all working as expected now
(And the clockgen was from the BIOS mod, the main issue is that if you go to even 100.01 you lose all boost/turbo and get locked to base clock)

Okay that makes more sense then. BCLK should be working just fine the normal way though, it's only changing +Fmax that will kick you to base clock.

AGESA 1208 is a bit of a challenge so far. I haven't dipped below 15k so far but it's running hotter and pulling more power at the same or lower score. I was doing about 15200 @ 4.45 with 100W and at 74C, now struggling to hit that same score @ 4.4 with 110W and 80C. Exact same BIOS settings.

Better than the 1207 experience at least.

Also game perf seems lower. But I haven't yet seen any MW2 stuttering on this BIOS. Can't decide what to do.

 

[Mussels]

BCLK has always broken turbo/boost on this board

Somethings definitely changed with the SoC voltages as I'm not totally stable at 1.025v - not 1.15v
This obviously helps temps and wattages - did they raise the defaults with an internal offset or something to help memory stability?

 

[Kissamies]

That audio problem of mine is purely CH VII Hero's fault. Put my old B550 board back to use and audio works just like it should.

Are there any other CH VII Hero users with audio problems?

 

[The King]

That audio problem of mine is purely CH VII Hero's fault. Put my old B550 board back to use and audio works just like it should.

Are there any other CH VII Hero users with audio problems?

Are these audio stutter issues? if so they are usually related to incorrect IOD voltages. In some cases increasing it usually helps with the stutters problems.
Different boards different BIOS may need voltage settings.

 

[Kissamies]

Are these audio stutter issues? if so they are usually related to incorrect IOD voltages. In some cases increasing it usually helps with the stutters problems.
Different boards different BIOS may need voltage settings.

Yeah, latency issues. I put the board back to the box but I also messed around with voltages without getting rid of the stutter.

 

[Mussels]

Busy day so havent finished testing yet

1.0v is boot stable at 3866, never was before
1.05v needed for actual stability, not 1.15v

still a drop, and i wonder if they really did add some secret offset to raise the voltages to boost stability (honestly the defaults always seemed too low
Current mission is to tune those voltages in as low as possible so i survive the coming 45C heatwaves

Are these audio stutter issues? if so they are usually related to incorrect IOD voltages. In some cases increasing it usually helps with the stutters problems.
Different boards different BIOS may need voltage settings.

^ This

Changing my SoC voltages today, running 1.0v worked quite well with 4 WHEA errors in 3 hours - but I had 3 USB disconnects in that time, and the onboard audio went funky

Definitely look into SoC and IOD voltages

 

[Mussels]

Ahah!
So solved the random "high wattage" power spikes my system was seeiung in HWinfo64 - purely a faulty battery cell in my UPS.
Was crashing out at ~300W load, with only one properly working cell it would spaz out and report briefy spikes of pretty much double what it was actually using

Currently gaming goes between 250 and 450W depending on how demanding the title is, with 275W being the 3090 and the rest between CPU and RGB

Also looks like i've got the reduced voltages sorted out, slow WHEA errors are a pain


Too low DRAM voltage resulted in black screen crash - 1.35v wasn't enough for 3866, 1.375v is
Too low SoC resulted in WHEA errors and the most bizarre USB issues - my keyboard wouldn't notice i'd released a key for example, so i'd hold A to walk left in a game then change to D, and randomly it'd keep going left for 3-5 seconds before updating

SoC used to be 1.15v but now it's fine at 1.06v

CPU doesn't throttle effective clocks at all with -30 CO and offset of -.06 so tada, maxed out 5800x3D with minimum energy

 

[AusWolf]

Ahah!
So solved the random "high wattage" power spikes my system was seeiung in HWinfo64 - purely a faulty battery cell in my UPS.
Was crashing out at ~300W load, with only one properly working cell it would spaz out and report briefy spikes of pretty much double what it was actually using

Currently gaming goes between 250 and 450W depending on how demanding the title is, with 275W being the 3090 and the rest between CPU and RGB

Also looks like i've got the reduced voltages sorted out, slow WHEA errors are a pain


Too low DRAM voltage resulted in black screen crash - 1.35v wasn't enough for 3866, 1.375v is
Too low SoC resulted in WHEA errors and the most bizarre USB issues - my keyboard wouldn't notice i'd released a key for example, so i'd hold A to walk left in a game then change to D, and randomly it'd keep going left for 3-5 seconds before updating

SoC used to be 1.15v but now it's fine at 1.06v

CPU doesn't throttle effective clocks at all with -30 CO and offset of -.06 so tada, maxed out 5800x3D with minimum energy

That just gave me a thought... what if my boot hangup issue with EXPO is actually down to DRAM voltage? I mean, it's a 1.35 V kit, but what if that's not enough? It's already proved that it can't do 1.1 V at JEDEC standard 4800 MHz. I have to increase it to 1.15 V to be stable. The only reason I'm doubtful is that the system is stable at EXPO once I do the double boot.

 

[Mussels]

That just gave me a thought... what if my boot hangup issue with EXPO is actually down to DRAM voltage? I mean, it's a 1.35 V kit, but what if that's not enough? It's already proved that it can't do 1.1 V at JEDEC standard 4800 MHz. I have to increase it to 1.15 V to be stable. The only reason I'm doubtful is that the system is stable at EXPO once I do the double boot.

Only know by trying it - issues like that with new sockets/boards aren't unknown, where they may have excessive droop at boot

Maybe try changing the LLC settings on the DRAM to be more aggressive?
It may need "warm up" time to get the voltages stable, DDR5 has those integrated voltage controllers now that work differently to before


Edit: yep they're given 3.3V and drop it down themselves. This could be like the memory-holes where certain voltage settings dont math out right and go weird - like 1.35v may have issues but 1.34 and 1.36 dont

 

[AusWolf]

Only know by trying it - issues like that with new sockets/boards aren't unknown, where they may have excessive droop at boot

Maybe try changing the LLC settings on the DRAM to be more aggressive?
It may need "warm up" time to get the voltages stable, DDR5 has those integrated voltage controllers now that work differently to before


Edit: yep they're given 3.3V and drop it down themselves. This could be like the memory-holes where certain voltage settings dont math out right and go weird - like 1.35v may have issues but 1.34 and 1.36 dont

That's interesting... and strange.

I'll have a look and report back my findings.

 

[Mussels]

That's interesting... and strange.

I'll have a look and report back my findings.

remember that you're sending avoltage request, its digital not analogue - so theres going to be dividers and multipliers and frequencies involved and some combinations will be more stable than others

with DDR5 being so new they're running on theory and not practical experience for what combinations work

 

[Mussels]

Zen3D go brrrr

NVIDIA GeForce RTX 3090 video card benchmark result - AMD Ryzen 7 5800X3D

And Vs people with the same hardware as me:

The faster people are getting higher clocks on the CPU, because they can do Bclk OCing and i cant. even 100Mhz makes all the difference there

And ofc i can raise the 375W limit to 450W on the GPU but seriously, outside of these few benchmarks whats the use

 

[Space Lynx]

Next month is going to be fun to watch unfold, I think AMD will retake the gaming CPU crown with Zen 7xxx X3D.

I would bet money that even Intel's upcoming 6ghz chips won't even be able to beat this coming up X3D chip, but we will see soon enough.

 

[AusWolf]

I've just updated my board to the latest BIOS from MSi (1.52), which brings AGESA 1.0.0.4 into the mix. I've enabled EXPO again to see if it helps with my double booting issue.

I've also noticed something strange:

• The RAM voltage is 1.335 instead of 1.35. I didn't notice this with the original BIOS. I did a TestMem5 just to be safe - it was successful.
• The SoC voltage sets to 1.36 V on auto. If I change it manually to 1.1 V, and then back to Auto, it sets to 1.2 V. Very strange behavior - like the board doesn't know what to do.

 

[Mussels]

Deleted post commenting on how good MSI boards are
respectfully and hilariously disagree

 

[Count von Schwalbe]

Deleted post commenting on how good MSI boards are
respectfully and hilariously disagree

There are times when I wish there was a "" reaction to posts.

 

[Mussels]

There are times when I wish there was a "" reaction to posts.

I'm not quoting deleted content or naming the poster, i dunno why they deleted

But, i have a super low opinion of MSI motherboards for a whole lot of reasons, not all ryzen specific

1. Only MSI had to release a second series of motherboards to get proper BIOS support (The MAX line with 32MB BIOS chips) - everyone else was fine with 16MB, but MSI wanted to sell more boards

2. They've had boards with overheating VRM's in every generation for years. The only brand doing worse is asrock, asus TUF are fighting hard too but staying in 3rd worst


We've had threads about it all here before

"The worst board we've ever tested, the MSI x570 pro carbon"
Crap like this is why ryzen has a poor reputation - 40W differences here with the same CPU (And yes the asus TUF is also garbage)

One example from another video is how they've designed things to give colder readings than they should with dishonest sensor readings, and slapping barely helpful heatsinks on top "Oh we got in trouble for hot VRM's so let's put material on the top to make them appear colder than they actually are"

All the other brands are within 5c, but not MSI - 25c difference because they've tried to hide how hot they'r really running

Later in that same video they showed this, with red bars indicating thermal throttling
With the MSI throttling at "just" 70c (or 90c measured on the bottom of the motherboards PCB)

They aren't exclusively terrible and at times appeared to learn their lesson: B550 for example they did quite well
Not leading the charts exclusively or anything, but solid over the entire range - and this led some people to think they would always be like this despite failing on all prior ryzen chipsets and various intel chipsets

And an older example from intel 11th gen, where you might be all "woo! MSI are running cold this time!"

Ahh no, they clocked way lower than everyone else. Nothing like a 300Mhz clock deficit despite low temps and the same wattage as everyone else
Look at the B560 tomahawk, ignoring the 65W limit and going to 265W but doing good for temps - that's the one they spam out to reviewers
Then look at the B560m bazooka, b560m pro, and pro-VDH - all throttle drastically
Theres no consistency with any of this, you never know if the boards going to ignore stock power limits and overclock, overheat and underclock, throttle and underclock, or

Every brand has bad boards. Every brand has a bad series.
MSI Are the only ones to consistently screw up not just their budget lines, but their premium chipsets too and it's always in new and exciting ways that make it a fucking gamble on what you're gunna get wrong

 

[AusWolf]

I'm not quoting deleted content or naming the poster, i dunno why they deleted

But, i have a super low opinion of MSI motherboards for a whole lot of reasons, not all ryzen specific

1. Only MSI had to release a second series of motherboards to get proper BIOS support (The MAX line with 32MB BIOS chips) - everyone else was fine with 16MB, but MSI wanted to sell more boards

2. They've had boards with overheating VRM's in every generation for years. The only brand doing worse is asrock, asus TUF are fighting hard too but staying in 3rd worst


We've had threads about it all here before

"The worst board we've ever tested, the MSI x570 pro carbon"
Crap like this is why ryzen has a poor reputation - 40W differences here with the same CPU (And yes the asus TUF is also garbage)
View attachment 275363

One example from another video is how they've designed things to give colder readings than they should with dishonest sensor readings, and slapping barely helpful heatsinks on top "Oh we got in trouble for hot VRM's so let's put material on the top to make them appear colder than they actually are"

All the other brands are within 5c, but not MSI - 25c difference because they've tried to hide how hot they'r really running
View attachment 275367

Later in that same video they showed this, with red bars indicating thermal throttling
With the MSI throttling at "just" 70c (or 90c measured on the bottom of the motherboards PCB)
View attachment 275370


They aren't exclusively terrible and at times appeared to learn their lesson: B550 for example they did quite well
Not leading the charts exclusively or anything, but solid over the entire range - and this led some people to think they would always be like this despite failing on all prior ryzen chipsets and various intel chipsets
View attachment 275371

And an older example from intel 11th gen, where you might be all "woo! MSI are running cold this time!"

Ahh no, they clocked way lower than everyone else. Nothing like a 300Mhz clock deficit despite low temps and the same wattage as everyone else
Look at the B560 tomahawk, ignoring the 65W limit and going to 265W but doing good for temps - that's the one they spam out to reviewers
Then look at the B560m bazooka, b560m pro, and pro-VDH - all throttle drastically
Theres no consistency with any of this, you never know if the boards going to ignore stock power limits and overclock, overheat and underclock, throttle and underclock, or
View attachment 275372



Every brand has bad boards. Every brand has a bad series.
MSI Are the only ones to consistently screw up not just their budget lines, but their premium chipsets too and it's always in new and exciting ways that make it a fucking gamble on what you're gunna get wrong

I don't know why Asus TUF is on your list. Sure, the brand wasn't the best at first launch, but it has come a long way since then. I've had the A520 (AMD), B550 (AMD) and B560 (Intel) micro-ATX TUF boards. The A520 is pretty basic, but gets the job done, but the other two are excellent boards. I wanted to get the B650 TUF for my current build, but it wasn't in stock anywhere, that's why I decided to give MSi another chance (I was burned with them so many times before). I wholeheartedly regret my decision. The only thing I like about this board (MSi Pro B650M-A) is that it remembers my RGB settings without having any bloatware installed (although it sometimes leaves all RGB disabled at boot for no reason). Oh, and the minimalistic, non-gamery design is quite nice, but that pleasure goes out of the window once I turn my PC on, unfortunately.

 

[Mussels]

Because a lot of the TUF boards are true, true garbage?

Those same VRM lists and videos have MSI and TUF as equal worst anytime asrock fails to show up to the party
Asus TUF x570 had nasty chipset coolers with no fan control forcing the entire rumour mill about AMD's new hot chipset that needed high RPM fans to cool them

I'm not too fussed when A series budget stuff has flaws, but when its top tier chipsets and priced highly - they cant give them out with VRM's that cant handle 80% of the product stack, or with false specs (things like screwing with reported wattages/voltages)

 

[The King]

I'm not quoting deleted content or naming the poster, i dunno why they deleted

But, i have a super low opinion of MSI motherboards for a whole lot of reasons, not all ryzen specific

1. Only MSI had to release a second series of motherboards to get proper BIOS support (The MAX line with 32MB BIOS chips) - everyone else was fine with 16MB, but MSI wanted to sell more boards

2. They've had boards with overheating VRM's in every generation for years. The only brand doing worse is asrock, asus TUF are fighting hard too but staying in 3rd worst


We've had threads about it all here before

"The worst board we've ever tested, the MSI x570 pro carbon"
Crap like this is why ryzen has a poor reputation - 40W differences here with the same CPU (And yes the asus TUF is also garbage)
View attachment 275363

One example from another video is how they've designed things to give colder readings than they should with dishonest sensor readings, and slapping barely helpful heatsinks on top "Oh we got in trouble for hot VRM's so let's put material on the top to make them appear colder than they actually are"

All the other brands are within 5c, but not MSI - 25c difference because they've tried to hide how hot they'r really running
View attachment 275367

Later in that same video they showed this, with red bars indicating thermal throttling
With the MSI throttling at "just" 70c (or 90c measured on the bottom of the motherboards PCB)
View attachment 275370


They aren't exclusively terrible and at times appeared to learn their lesson: B550 for example they did quite well
Not leading the charts exclusively or anything, but solid over the entire range - and this led some people to think they would always be like this despite failing on all prior ryzen chipsets and various intel chipsets
View attachment 275371

And an older example from intel 11th gen, where you might be all "woo! MSI are running cold this time!"

Ahh no, they clocked way lower than everyone else. Nothing like a 300Mhz clock deficit despite low temps and the same wattage as everyone else
Look at the B560 tomahawk, ignoring the 65W limit and going to 265W but doing good for temps - that's the one they spam out to reviewers
Then look at the B560m bazooka, b560m pro, and pro-VDH - all throttle drastically
Theres no consistency with any of this, you never know if the boards going to ignore stock power limits and overclock, overheat and underclock, throttle and underclock, or
View attachment 275372



Every brand has bad boards. Every brand has a bad series.
MSI Are the only ones to consistently screw up not just their budget lines, but their premium chipsets too and it's always in new and exciting ways that make it a fucking gamble on what you're gunna get wrong

I do agree with you mostly but you definitely not being fair putting this sort screw up on Msi only.

Asus and Gigabyte are very guilty of the exact same thing.

Look at the Asus prime and Aurous Elite boards here. I don't even look at Asus prime line boards anymore because if things like this.

 

[AusWolf]

Because a lot of the TUF boards are true, true garbage?

Those same VRM lists and videos have MSI and TUF as equal worst anytime asrock fails to show up to the party
Asus TUF x570 had nasty chipset coolers with no fan control forcing the entire rumour mill about AMD's new hot chipset that needed high RPM fans to cool them

I'm not too fussed when A series budget stuff has flaws, but when its top tier chipsets and priced highly - they cant give them out with VRM's that cant handle 80% of the product stack, or with false specs (things like screwing with reported wattages/voltages)

I'll give you the point on the X570. I still hold to my opinion that the other ones I've tried are great. On a side note, I also have the TUF 6500 XT, which is as good as a 6500 XT can be. It's super quiet, super overclockable, and runs super cool.

I do agree with you mostly but you definitely not being fair putting this sort screw up on Msi only.

Asus and Gigabyte are very guilty of the exact same thing.

Look at the Asus prime and Aurous Elite boards here. I don't even look at Asus prime line boards anymore because if things like this.
View attachment 275380

The Prime is their basic series, meant for office and HTPC use. I'd never pair one with a 3900X.

 

[tabascosauz]

1. Only MSI had to release a second series of motherboards to get proper BIOS support (The MAX line with 32MB BIOS chips) - everyone else was fine with 16MB, but MSI wanted to sell more boards

2. They've had boards with overheating VRM's in every generation for years. The only brand doing worse is asrock, asus TUF are fighting hard too but staying in 3rd worst

Every brand has bad boards. Every brand has a bad series.
MSI Are the only ones to consistently screw up not just their budget lines, but their premium chipsets too and it's always in new and exciting ways that make it a fucking gamble on what you're gunna get wrong

Weird that you say you're not generalizing against MSI, by generalizing about MSI's products the "new" MSI only came around after the X570 Pro Carbon fiasco and the release of B550 Unify, so it's a bit pointless to use Z390 MSI products to demonstrate how "bad" MSI is, isn't it?

Everything you've listed just points to the same conclusion, don't buy bottom of the barrel boards from any of the 4 vendors. I don't see how the B560 results point to MSI being "bad" either - good boards are good boards and bad boards are bad boards?

You seem to be of the belief that MSI should never be allowed to redeem itself despite overhauling itself from top to bottom after the Pro Carbon. So, by that logic, TUF will always be garbage (despite post-B550 TUFs receiving major upgrades), Strix will always be garbage (for essentially selling TUF at a marked up price for all of AM4), Gigabyte will always be garbage (because their customer support will always be worthless), and ASRock will always be garbage (for their cost-cutting on budget boards).

If you don't like them, it's perfectly ok to just say you don't like MSI, it's how I treat ASRock, instead of trying to convince people that MSI is objectively worse than anyone else.

 

[The King]

The Prime is their basic series, meant for office and HTPC use. I'd never pair one with a 3900X.

Still does not change the fact other boards basic series at cheaper prices outperforms it?

Overpriced junk then? My point exactly. Asus Prime boards are not cheaper here usually sell for more because of the Asus name tax on it.

Also goes to my point that ASUS screw up their boards as well same goes for Gigabyte and Asrock this includes Msi, who are not the only guilty party here and should not be singled out.