[Zoropsis]

so far so good. slight uptick in vSOC. will observe over the week.

New AGESA seems rougher on the IMC and IF, had to bump both VSOC and VDDG a bit to pass with my previously stable settings.

 

[pershoot]

New AGESA seems rougher on the IMC and IF, had to bump both VSOC and VDDG a bit to pass with my previously stable settings.

What IF clock are you currently running?

 

[Zoropsis]

What IF clock are you currently running?

I'm running 8000/2200, previously was stable at 1.17v VSOC and 1v VDDGs but had to do 1.185v and 1.05v on the latest BIOS.

Could just be that I was on the edge with those voltages, but it went from passing 15-20h everything to failing 30mins Karhu runs.

ASUS board users have reported that VDDG was defaulting a lot higher than before on the new BIOS so I think it's the expected behavior.

 

[pershoot]

I'm running 8000/2200, previously was stable at 1.17v VSOC and 1v VDDGs but had to do 1.185v and 1.05v on the latest BIOS.

Could just be that I was on the edge with those voltages, but it went from passing 15-20h everything to failing 30mins Karhu runs.

ASUS board users have reported that VDDG was defaulting a lot higher than before on the new BIOS so I think it's the expected behavior.

ah, ya.
i was considering a slight vddg increase but havent touched it from auto this whole time (fclk 2067). i did increase it now a litle bit as i saw a very sparse usb audio / driver reinit error (which may have always been there but since im now chasing rabbit holes, hehe).

 

[pershoot]

ah, ya.
i was considering a slight vddg increase but havent touched it from auto this whole time (fclk 2067). i did increase it now a litle bit as i saw a very sparse usb audio error (which may have always been there but since im now chasing rabbit holes, hehe).

vSOC: sleep wakeup seems a bit better now (a little quicker); vDDG IOD / CCD: haven't noticed usb audio / driver reinit error(s). still observing...
vSOC: 1.2 -> 1.210
vDDG IOD / CCD: 0.903 -> 0.916

edit: testing with only vDDG IOD / CCD @ 0.909 (.906)

 

[BDBB]

Upgrading from a z390 + 9900k to a x870e + 9800X3D.

I haven't built AMD since uhh.. 1999? LOL

AMD Athlon 64 bit FX-60 Dual Core ! Socket 939 - It was my first PC I built.

Anyways... just ordered a 9800X3D and a Nova board, wanted Taichi or Lite so badly but, always out of stock and I caught the Nova in stock, so pulled the trigger. Maybe I'll regret that choice when I start overclocking since the Nova doesn't have the EBLK thing... I think AMD wants that for ocing? I don't know enough about AMD yet.

I do have a ASUS Xonar STX that I can't give up and I wanted the Nova for that x1 pci slot... but my case I've chosen (Hyte y70) is not going to let me use that card anyways due to the design of the slots on the back, big bummer because I'm sure the onboard sound on the Nova is **** tier.

I'm a bit nervous, I haven't used ASrock boards before, and I haven't messed with AMD in 20+ years... so I'll be spending alot of time here on this thread, reading and learning.

I could use some advice about programs. I see you guys using AIDA and ZenTimings programs... I'm not familiar with the AMD specific programs like this, could anyone provide a short list of "must have" AMD specific programs, system utility , over clocking type programs. I use CPU-Z, CoreTemp, Hwinfo64 etc etc but I'm more interested in learning about AMD specific things, preparing to OC this chip.

I need to look around for a OC guide for AMD, I don't know enough to tinker yet. The 9900k was so easy to OC, but I'm very familiar with Intel Bios settings.

I'll be here asking questions! I'm off to the DuckDuckGo machine to learn more.

Pretty much the exact same deal for me. The very first PC I built had a Core 2 Duo E8400 in it, and I've only built intel since then. This is my very first AMD build. I also ended up with the Nova rather than the Taichi. Just got it today and put everything together other than some case fans and one more NVME drive I'm waiting on. Luckily everything just worked and I was able to install windows without any issues.

A few things came up:
1. I couldn't find any type of setting in the bios where you can order the priority of boot drives. Is that no longer a thing?
2. I didn't know what the current best program to stress test a 9800x3d is, so I just threw Prime95 at it. Under load it was usually clocking around 4700-4900mhz rather than 5200mhz. Is that normal behavior? I read a review that said it was putting out 5200 all-core at stock.
3. Anyone had luck controlling the Nova's RGB lights? Admittedly this is the first time I've ever messed with RGB so it could be user error, but I downloaded SignalRGB and nothing I did would get it to control the motherboard's lights. Eventually I got SignalRGB to control the RAM (corsair), but the board lights kept flashing the default rainbow. Then I deleted signalRGB and downloaded openRGB and had the opposite problem: I gained some level of control over the motherboard but couldn't even see the RAM to adjust it. Having the motherboard lights at full blast is a huge problem because they are way too bright and they can be seen directly from some angles.

 

[ApolloX30]

@BDBB
1. You can set the sequence of boot devices ind the bios. The bios has a search function (or F4).
2. For which purpose?To see the highest clocks possible, use BoostTester, to heat up your room use Cinebench or Prime or Y-Cruncher.
3. Yes, I permanently switched them off in the bios

 

[renhanxue]

A cool thing about Asrock's AM5 boards is that they all support ECC. I like the idea of ECC, and I'd heard that the Hynix ECC sticks overclock basically like normal non-ECC ones, so I bought two Kingston KSM56E46BS8KM-16HA sticks to try my luck.

These are 16 GB single rank Hynix-A ECC UDIMMs, green PCB, no heatspreaders, no XMP or EXPO profile. They come with only their JEDEC stock config, 5600 MT/s and CL 46 at 1.1V, and since they're sold as single sticks of course they're not a matched pair. I enabled ECC and disabled PFEH (Platform First Error Handling, if you don't disable this the UEFI will swallow all warnings about errors corrected by ECC), and sure enough both Windows and Linux detect the sticks as multi-bit ECC with a 72-bit data width.

Then I started trying to OC them. 6000 CL30 at 1.35v just worked with zero fuss, which was honestly more than I expected. I did some experimenting with 2:1 mode, but while I can easily boot at 7600 MT/s, no matter what I tried all I managed to do was to prove that the ECC error correction and more importantly error reporting works, because within minutes of getting into Windows I always started to get WHEA warnings about corrected hardware errors (event ID 47).

1:1 mode was a lot more promising and these sticks are honestly way better than I could possibly have imagined. Currently I'm running them at 6400 CL30, with matched tPHYRDL's at 36/36 and Nitro mode 1/3/1, and I'd call this daily stable:

There's some more that can be done here, tRFC could be brought down a bit and ideally I'd want to run lower VSOC. Could also see if I can get GDM disabled and/or run Nitro 1/2/1. Still, very happy with these results already!


Also, overclocking with ECC enabled is kinda funny. You never see your stress tests fail, and you never bluescreen either. Instead, either you get WHEA warnings about corrected errors, or the entire system just freezes if there are uncorrectable errors.

 

[Neon Knights]

A cool thing about Asrock's AM5 boards is that they all support ECC. I like the idea of ECC, and I'd heard that the Hynix ECC sticks overclock basically like normal non-ECC ones, so I bought two Kingston KSM56E46BS8KM-16HA sticks to try my luck.

These are 16 GB single rank Hynix-A ECC UDIMMs, green PCB, no heatspreaders, no XMP or EXPO profile. They come with only their JEDEC stock config, 5600 MT/s and CL 46 at 1.1V, and since they're sold as single sticks of course they're not a matched pair. I enabled ECC and disabled PFEH (Platform First Error Handling, if you don't disable this the UEFI will swallow all warnings about errors corrected by ECC), and sure enough both Windows and Linux detect the sticks as multi-bit ECC with a 72-bit data width.

Then I started trying to OC them. 6000 CL30 at 1.35v just worked with zero fuss, which was honestly more than I expected. I did some experimenting with 2:1 mode, but while I can easily boot at 7600 MT/s, no matter what I tried all I managed to do was to prove that the ECC error correction and more importantly error reporting works, because within minutes of getting into Windows I always started to get WHEA warnings about corrected hardware errors (event ID 47).

1:1 mode was a lot more promising and these sticks are honestly way better than I could possibly have imagined. Currently I'm running them at 6400 CL30, with matched tPHYRDL's at 36/36 and Nitro mode 1/3/1, and I'd call this daily stable:

View attachment 2694462

There's some more that can be done here, tRFC could be brought down a bit and ideally I'd want to run lower VSOC. Could also see if I can get GDM disabled and/or run Nitro 1/2/1. Still, very happy with these results already!


Also, overclocking with ECC enabled is kinda funny. You never see your stress tests fail, and you never bluescreen either. Instead, either you get WHEA warnings about corrected errors, or the entire system just freezes if there are uncorrectable errors.

Very important information, thanks a lot. I guess i never tried , or better thinking of using ECC Rams - nice to know if i ever want to try it out.

 

[KedarWolf]

Windows 24H2 good for Cinebench 2024.

 

[pershoot]

Helpful hint for those that run a legacy GPU (that don't game so keep running with it, hehe). CSM enable + Legacy (Video OROM) will allow you to use UEFI bios with it. Note: iGPU will be inoperable in this state.

I had a modded BIOS on the card which the board apparently detected and wouldn't let me view BIOS with CSM off. Reverted back to stock VBIOS and that functionality is restored (I've turned CSM off; the BIOS does contain UEFI bits to allow for it).

 

[pershoot]

running with this for a few days / a bit; see how things go.
3rd_run, 4th_run, stability1, stability2. vDDG IOD / CCD back to Auto (0.903 reported), adjust tRTP, tRC, tRAS, tRFC.

 

[JustinJ]

Anyone pick up the new CL26 kits from G.skill? ( F5-6000J2636G32GX2-TZ5NR )

Curious how the 64GB CL26 kit will perform in the Nova.

 

[3838]

impressive, thanks for sharing all your information.wonder what their DDR5 6400MT/s* ECC Registered DIMM would do. Wonder if the D die on those would make a difference or not.

 

[koji]

Looks like 3.18.AS02 just got released on the asrock website.

 

[pershoot]

Looks like 3.18.AS02 just got released on the asrock website.

There seems to be a memory latency regression (comparing against AS01) for me; using Aida64 to quick-test / observe. I've notified support.

Attached (occt, 3rd_run, timings). cpu clock is higher on this run (due to occt being open) so it reduced latency, but when at bootup clock, its a bit higher (~5ns) then what's shown.
AS01: https://www.overclock.net/posts/29424332/

 

[PrimeLight]

Hi everyone! I'm still on bios 3.10 with my 9800x3d and 6200c28 ram. Are there any better and more stable BIOS versions or should I stay on this one?

 

[pershoot]

Hi everyone! I'm still on bios 3.10 with my 9800x3d and 6200c28 ram. Are there any better and more stable BIOS versions or should I stay on this one?

In my opinion, you should make an effort to keep on latest stable, so you can benefit from any fixes, patches, etc., unless there is a specific issue with your system when using them (which one should notify support on so they can work on remediating it, in that case). Latest stable is 3.16.

 

[Brolis]

Hi everyone! I'm still on bios 3.10 with my 9800x3d and 6200c28 ram. Are there any better and more stable BIOS versions or should I stay on this one?

If i were you, i would not go for 3.18 just yet
With 3.18 my system took longer to load, screen tearing with decent fps( no matter what settings i tried)
Decided to try and install older version and got 3.15 and my system runs like a BEAST( like it supposed to)
4090 and 9800x3d. I was puling my hear, thinking for what i pay such money if it performs so ****. thank god i came with this idea to roll back bios

 

[PrimeLight]

If i were you, i would not go for 3.18 just yet
With 3.18 my system took longer to load, screen tearing with decent fps( no matter what settings i tried)
Decided to try and install older version and got 3.15 and my system runs like a BEAST( like it supposed to)
4090 and 9800x3d. I was puling my hear, thinking for what i pay such money if it performs so ****. thank god i came with this idea to roll back bios

At the moment there is only one noticeable problem with the current BIOS - an error during reboot almost always, while MCR and PDM always on. Have they fixed it in the new BIOS? Also, my system freezes quite often when I launch any application, be it a browser or a game, it doesn't matter, I have had these freezes since the first day I have been using this PC with 9800x3d and x870e Taichi, on Intel Windows 11 Pro worked without freezes.