• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

Ryzen Owners Zen Garden

Toothless said:
Can't hit 3600 on the memory since I'm running four dual-rank dimms.
77b9059b2c.jpg
Click to expand...
3400 is pretty good and that latency you have is pretty good. I've run into the same wall albeit on threadrippers.
 
She aint no sports car, but she has some curves. This thing is like 3 steps behind a regular Vermeer. But it is hella smooth in game. So weird. But in Aida, the copy speeds with 4 sticks smoke 2 sticks running 2K 1:1. Latency seems to be about .5ns off.. probably because I was messing with my saved profile. Things seems to get a bit dirty in uefi if you make a lot of changes vs just entering your settings once. It could just be me though :D

Screenshot 2022-10-12 210943.png
 
Gaming results:
I reset HWinfo after I was already in game and took the screenshot before quitting so the average stats here are true gameplay averages, no idle time.

80 minutes of DRG, 4K ultra 140FPS (DX12, Unreal engine 3)
Peak system wattage of 380W, 32" monitor included.
CPU peaked at 63C, 69W
CPU averaged 48C 56W
3090 was in the 200-250W range (seriously, the gains from it going to the 375W limit are totally not worth it)

It's the little things like average core count being 3 threads average, 5 at most that show that 6 core CPU's are definitely viable for gaming.

1665630617746.png
 
I may be returning my 7950x and Asus X670E-E.
Random Bluescreens the past few days. Tried a few things (different Memory, uninstalling software, full reinstall of OS, etc)
Frankly I spent too much to beta test this stuff.

A quick google seems im not alone in these blue screens.
One thing i have not tried is windows 10.

sure is fast though, when it works :)
 
Guess I'll share my 32gb two stick setup. It could use some more tweaking...

3800-aida.png

3900-zent.png
 
thesmokingman said:
Yea, those are dram calc suggestions, they don't work. It won't even boot with trdwr at 12.
Click to expand...
Yes because tCWL need to be 16 it wont boot with CWL 10 and low RDWR like 8 or 9

This is Micron RAM runing CL 16 CWL16 and RDWR 8. Your Samsung B-die RAM should do much better timmings than this.
What voltage are you running? At least 1.45V?
ZenTimings_Screenshot CB 3800 CL16 Tm5 Pass.png
 
Last edited:
Psychoholic said:
I may be returning my 7950x and Asus X670E-E.
Random Bluescreens the past few days. Tried a few things (different Memory, uninstalling software, full reinstall of OS, etc)
Frankly I spent too much to beta test this stuff.

A quick google seems im not alone in these blue screens.
One thing i have not tried is windows 10.

sure is fast though, when it works :)
Click to expand...

The King said:
Are you running the latest BIOS? I think one was released 2 days ago.
rog.asus.com

ROG CROSSHAIR X670E EXTREME | Gaming motherboards｜ROG - Republic of Gamers｜ROG Global

Designed for the broadest range of demanding fans, the ROG Crosshair X670E Extreme features 20 + 2 power stages, DDR5 support, five M.2 slots, ROG Gen-Z.2, a USB 3.2 Gen 2x2 front-panel connector with Quick Charge 4+ Support, dual USB4®, PCIe® 5.0 and onboard Wi-Fi 6E.
rog.asus.com rog.asus.com
Click to expand...

Also new Chipset drivers released today.


But to your comment, yeah, it's going to feel like beta for a while.
 
977a52b762.jpg


Time to fix CPU clocks.. Took out half the ram. Guess I can live on 32GB.

Edit:
16c706ec07.png


And last edit, done with memory.

dfef35f93a.jpg
 
Last edited:
Super Firm Tofu said:
But to your comment, yeah, it's going to feel like beta for a while.
Click to expand...

And this is what makes me so disappointed. Even now the AM4 platform has bugs that are long since known and completely unacceptable for a platform that has been around for so long.

The media's massive inertia towards publicizing and getting AMD to fix that EDC bug eating up to 10% of MT performance on every CPU with PBO enabled in the past 3 AGESA releases (almost a year!) is baffling, if it was Intel people would be screaming that Intel nerfed overclocking or whatever. It's especially nasty on the 65 W spec processors because it limits them to just 90 A of current, and it really hurts the 5950X too.

I love my CPU but this platform can be such a bother sometimes. AM5 is definitely going to have a long way to go... because its 5 year old predecessor does!
 
Soooo.........apparently it's not just Asus, and all 1207 BIOSes are just worthless garbage?

Moved back into the Unify-X and the A.70 BIOS is also dogshit like the Impact's 4201 BIOS. Tanks the clocks, tanks the scores, yes Kombo Strike is functionally -30 (checked in PBO2 Tuner) but does weird shit and is essentially useless when I already use PBO2 Tuner......and randomly lags, just like both Asus BIOSes do...

The only annoying thing about using the previous A.50 1206 BIOS is that MSI apparently forgot that VDDG was two separate rails.

Wondrously, there's a fully available PBO section in the A.50 BIOS, and apparently it works for setting limits and scalar. Unfortunately, Curve Optimizer is nowhere to be found so I still have to use PBO2 Tuner. I'd have to go to the A.70 BIOS for Kombo Strike.......and I won't do that, especially since A.70 doesn't have the PBO menu so we're right back where we started.

How hard is it to have both PBO and CO in the BIOS, AMD? We still have control over both even with you hiding shit. You can still impose hard limits like you do on Fmax. Just give us back the menus

Can also confirm, pre-1206 BIOSes all lock the 5800X3D to base clock (3.4GHz).
 
Last edited:
Super Firm Tofu said:
Also new Chipset drivers released today.


But to your comment, yeah, it's going to feel like beta for a while.
Click to expand...

Yep, tried new bios 0705, new chipset driver, fresh OS install.
Clear CMOS, Different GPU, Different memory, disabling PBO/Boost, etc.

Before tearing this thing down and sending it back, i figured i'd install windows 10 pro and see how that goes, so far its survived the install and has been up about 2 hours (Longer than it was staying up before without a BSOD)
 
tabascosauz said:
Soooo.........apparently it's not just Asus, and all 1207 BIOSes are just worthless garbage?

Moved back into the Unify-X and the A.70 BIOS is also dogshit like the Impact's 4201 BIOS. Tanks the clocks, tanks the scores, yes Kombo Strike is functionally -30 (checked in PBO2 Tuner) but does weird shit and is essentially useless when I already use PBO2 Tuner......and randomly lags, just like both Asus BIOSes do...

The only annoying thing about using the previous A.50 1206 BIOS is that MSI apparently forgot that VDDG was two separate rails.

Wondrously, there's a fully available PBO section in the A.50 BIOS, and apparently it works for setting limits and scalar. Unfortunately, Curve Optimizer is nowhere to be found so I still have to use PBO2 Tuner. I'd have to go to the A.70 BIOS for Kombo Strike.......and I won't do that, especially since A.70 doesn't have the PBO menu so we're right back where we started.

How hard is it to have both PBO and CO in the BIOS, AMD? We still have control over both even with you hiding shit. You can still impose hard limits like you do on Fmax. Just give us back the menus

Can also confirm, pre-1206 BIOSes all lock the 5800X3D to base clock (3.4GHz).
Click to expand...

The worst thing is, the 1.2.0.7 bios for my B550-E is actually not bad, and it's had the memory training problem fixed (board stuck at 0D unless I try to power it on twice), but the EDC bug just ravages my MT performance, so I have to choose between computer turning on at the first try and no stutter with fTPM enabled, or having my full CPU performance. It's something like up to -300 MHz vs. the 1.2.0.3 C BIOS.

I can't even, and honestly with AM5 out and performing commercially poorly, what incentive does AMD even have to fix a bug that eats previous generation performance, reducing the gap between Zen 3 and Zen 4? I'd say none.
 
Psychoholic said:
Yep, tried new bios 0705, new chipset driver, fresh OS install.
Clear CMOS, Different GPU, Different memory, disabling PBO/Boost, etc.

Before tearing this thing down and sending it back, i figured i'd install windows 10 pro and see how that goes, so far its survived the install and has been up about 2 hours (Longer than it was staying up before without a BSOD)
Click to expand...

The instability at stock under Windows 11 would probably be a return for me. That's was the mental line I drew when I bought my AM5 stuff. I expected it to be a bumpy ride, and I can deal with funky stuff in AGESA, but if it wasn't going to be usable out of the box it was going back.

If I hadn't bought from Newegg with their 'replacement only' policy on the board and CPU, I probably would have returned it and stayed with the Alder Lake I was using before. No complaints, it's just nothing special.
 
Super Firm Tofu said:
The instability at stock under Windows 11 would probably be a return for me. That's was the mental line I drew when I bought my AM5 stuff. I expected it to be a bumpy ride, and I can deal with funky stuff in AGESA, but if it wasn't going to be usable out of the box it was going back.

If I hadn't bought from Newegg with their 'replacement only' policy on the board and CPU, I probably would have returned it and stayed with the Alder Lake I was using before. No complaints, it's just nothing special.
Click to expand...

I agree, i expected a bumpy ride as well but yeah, constant BSODS are a no go.
There have been lots of reports of random BSODs on 22H2 (Regardless of platform) so i just tried to install windows 11 22H1, it didn't even make it through the installer without a "Critical_Structure_Corruption" BSOD.
Windows 10, for whatever reason is fine so far.

I remember my 3900X system being a little funky when it was new, but i dont remember random reboots or bsods.
My alder lake system was fine since launch, glad i didnt immediately sell it :)
 
Mussels said:
sleep mode related issues are almost always RAM if it wont wake, and SoC/chipset voltages with USB issues. Voltage is drooping when it wakes up, not enough to reset the specific device and it 'hangs'

You can easily try an externally powered USB hub, 2.0 or 3.0 - it'll help prevent a lot of issues since the external power will keep the devices 'awake' the entire time
Click to expand...
It doesn't happen only after resuming though. It happens on a cold boot too.
 
Psychoholic said:
I agree, i expected a bumpy ride as well but yeah, constant BSODS are a no go.
There have been lots of reports of random BSODs on 22H2 (Regardless of platform) so i just tried to install windows 11 22H1, it didn't even make it through the installer without a "Critical_Structure_Corruption" BSOD.
Windows 10, for whatever reason is fine so far.

I remember my 3900X system being a little funky when it was new, but i dont remember random reboots or bsods.
My alder lake system was fine since launch, glad i didnt immediately sell it :)
Click to expand...

I can confirm zero issues with my X670E setup on windows 10. Don't know what MS is doing with windows 11 but all I hear about are the random bugs.
 
Mussels said:
Yes, four USB 3.0 5Gb ports are direct to the CPU. These are the most reliable and have the best power delivery (best one to use my Rift S on, regardless of which ryzen CPU was in these boards)

As said tho, they can suffer disconnects if attached devices are erroring. PCI-E risers and unstable memory were common causes, it's why some people have never ending issues but others are problem free - because it's usually caused by an external piece of hardware
Click to expand...
So my X470GTQ has 4 USB 3.1gen1 5 Gbps ports on the rear IO (and headers for another 2), in addition to 2 3.1 gen2 ports on the rear IO and headers for 3.0 and 2.0, which presumably are all connected to the chipset. According to the manual the 5 Gbps (which presumably are connected directly to the CPU but I cannot find anything in the manual to confirm that) ports are split into two pairs, one pair below the PS/2 connector and one pair below the ethernet/RJ45 port, so I decided to unplug some stuff and try using those ports instead. So I plug my Cherry keyboard into one of the ports below the PS/2 connector (the other one is in use by my ALFA USB WiFi adapter) and it doesn't work; I replug it and it still doesn't work. ALFA USB WiFi in the port above it is working fine though. Nothing related to this shows up in the Linux dmesg output. I plug it into another of the 5 Gbps ports, one part of the pair below the ethernet port and it works immediately and it shows up in the dmesg output. So much for the magic reliability of these supposedly directly-to-the-CPU ports. I don't know what the f is wrong with this system/platform but I am not amused (and to be clear this is on the latest 1.2.0.7 AGESA from May 2022; no OC, no XMP even and you can see my PSU and RAM in my system specs).

IMG_0921.JPG

IMG_0924.JPG

IMG_0922.JPG

During my trip to the UEFI, I saw that there are also USB configuration settings to enable/disable 3.1 gen1 and gen2, which are currently both set to auto. I wonder if I should try disabling gen2.
 
Last edited:
The King said:
Yes because tCWL need to be 16 it wont boot with CWL 10 and low RDWR like 8 or 9

This is Micron RAM runing CL 16 CWL16 and RDWR 8. Your Samsung B-die RAM should do much better timmings than this.
What voltage are you running? At least 1.45V?
View attachment 265282
Click to expand...
Hijacked your settings again. Previously been using your 3600 CL14 profile since we have basically the same kits of Micron E-die DR. This one allowed me to finally achieve stability on 1900 FCLK. Thanks! Latency in Aida64 is at around 57.3 ns, with reads at 54700 MB/s. It passed full run of TestMem5 with Extreme @anta777. No WHEA errors, no crashes when gaming.
 
Alright so it was NOT because of the voltages as some crackling was still present. Oh boy as I tried LatencyMon and some googling, looks like disabling HPET finally did something.

Before disabling HPET, the bars went red (hella high) from time to time, right after the program recognized a crackling in the audio. I guess I can finally call this resolved (with auto voltages & PBO +200). Maybe I can now fine-tune the RAM back to 3466 on next boot.


1665746008351.png
 
Goddamn
Moved my 5800x to the ITX system, and it refuses to post with XMP.
The infamously shitty corsair LPX strikes again... That ram is gunna go to my x370 system if it wont behave
Edit: 1.1v didnt fix it, wow this RAM really loves to disapoint. It worked on the 2700x, but had issues on the 1400 and 1700 back in the day, too.

Lenne said:
Alright so it was NOT because of the voltages as some crackling was still present. Oh boy as I tried LatencyMon and some googling, looks like disabling HPET finally did something.

Before disabling HPET, the bars went red (hella high) from time to time, right after the program recognized a crackling in the audio. I guess I can finally call this resolved (with auto voltages & PBO +200). Maybe I can now fine-tune the RAM back to 3466 on next boot.


View attachment 265457
Click to expand...
HPET is off by default in most AMD BIOS, might explain why you've got it and most people dont? glad you found it, HPET is seriously old and problematic
 
@Mussels 1.1V VSOC? Sometimes XMP can fail to boot @ 1.35V increasing VDIMM to 1.36V-1.4V can help.
 
Yep swapped from the LPX to the vengeance pro - exact same timings mind you - and insta booted. Didnt even notice i'd changed RAM, and booted with the settings the LPX failed to run at.


I'll have to check the SPD and XMP settings for the secondary timings and see if they're too tight or something, and if i can edit them...

The King said:
@Mussels 1.1V VSOC? Sometimes XMP can fail to boot @ 1.35V increasing VDIMM to 1.36V-1.4V can help.
Click to expand...
1.10 and 1.20 didnt work either, the LPX ram has broken XMP settings for ryzen, always has. It's from the early days when it had its shittiest zen compatibility.

So i can see this myself later:
The working vengeance pro:
3200
16-18-18-36
Trc 75
Trfc560
 
You must log in or register to reply here.
Back
Top