zippytek
 

zippytek

rgarodnick
Fair enough.

While I stand corrected, my thought still holds true to only update bios when needed, either to fix something that's broken or for better stability. For those having issues on 2.x with ADL, is there much to gain by going past 1.15?

in my opinion, not at all. people want to get the benefit of hynix a-die, and this is optimized on 2.xx bios revs...however, alder lake a-die performance is not much better than m-die 1T. if you're going to buy a decent kit of a-die to use with last gen board and cpu, just get raptor lake?
 
there's very little argument for running the newer bios for alder lake, and the only one is a-die performance. a-die *and* m-die on raptor lake is head and shoulders above alder lake. so anyone playing the contrarian to your original point (1.xx is best for ADL) is just here to argue, or hastily updated thinking the latest and greatest is best for their use case without doing the brief research to learn why this is probably not the case.

Mr. Fox

rblaes_99
very happy with 2.05 bios.  great work EVGA!

Seems v2.05 is good for me, too. After flashing I found that I needed to cold boot from G3 (power disconnected or PSU switched off) in order to have P, E and ring ratios correctly applied. Simply turning off normally with power still connected to the board isn't enough and the CPU and ring ratios are capped for some reason. After a cold boot from G3 state everything functions correctly. Now, I just need a replacement board so I have more than one working BIOS chip and I think I will be good to go. My M-die performance and stability are definitely better on v2.05. The TG Delta A-die kit was junk and not stable on either of my systems. I have a G.SKILL A-die kit arriving today and I am hoping that works well.

Mr. Fox
So, the G.SKILL 7200 A-die looks like it might be a keeper. Not sure what was wrong with the Team Group Delta 7200 kit... maybe just a lousy silicon sample. Now all I need is my replacement mobo and I should be good to go.
 
Still need to tighten up some of the sloppy secondary "Auto" timings, but I don't have time to mess with it at the moment. Not too shabby for a $300 A-die kit. SA 1.350V, CPU VDDQ 1.400V, CPU VDD2 Auto, MEM VDD 1.490V, MEM VDDQ Auto. On the ODT Menu, RttWR 0/0, RttNomRd 48/48, RttNomWr 48/48, RttPark 40/40, RttParkDqs 34/34.
 

 
 

Mr. Fox

TheAffxct

Mr. Fox
So, the G.SKILL 7200 A-die looks like it might be a keeper. Not sure what was wrong with the Team Group Delta 7200 kit... maybe just a lousy silicon sample. Now all I need is my replacement mobo and I should be good to go.
 
Still need to tighten up some of the sloppy secondary "Auto" timings, but I don't have time to mess with it at the moment. Not too shabby for a $300 A-die kit. SA 1.350V, CPU VDDQ 1.400V, CPU VDD2 Auto, MEM VDD 1.490V, MEM VDDQ Auto. On the ODT Menu, RttWR 0/0, RttNomRd 48/48, RttNomWr 48/48, RttPark 40/40, RttParkDqs 34/34.
 

 
 

Out of curiosity, do you set the ODTs due to previous knowledge or did you adjust them all and test? I know QXE sets them as well. I’ve always been curious as to how you go about it.

My first exposure to ODT settings came about out of necessity due to what an abortion X570 and Ryzen 9 turned out to be, and trying to make myself a silk purse from a sow's ear. On an Intel system leaving them set to "auto" is usually fine. That is not true with Ryzen's bugginess and lackluster memory management. I start playing with those settings when my memory overclock is almost stable, and sometimes a lucky guess is good enough to make it fully stable. It is a shame that there is no meaningful and clear documentation for most firmware settings. I am not an expert, just an experimenter. I think the general idea is similar to using static voltage instead of adaptive. You trade off lower voltage and power consumption for better stability, performance and predictable behavior.

Mr. Fox
Yeah, the G.SKILL seems to be working well on v2.05.
 
 

 
 

mildvswild

Mr. Fox
Yeah, the G.SKILL seems to be working well on v2.05.

Can you share voltage setup for me?

Mr. Fox
Yeah, the G.SKILL seems to be working well on v2.05.
 
 

Mr. Fox

DerpFails
What Windows ver did you find best for wPrime?

I will only use 1.55, which is coincidentally the version packaged with Benchmate. I do not care for the newer versions of wPrime.

Mr. Fox

TheAffxct
I just wanted to ask if anyone on here has managed to stabilise 7800-8200. I mean like TM5 ABSOLUT, Karhu 10K+, or HCI 600%+ and with some or other IMC test and verification of reboot stability. I don’t ask this as a taunt or challenge, I actually need to gauge what this board can do.

I’m not far from grabbing a new chip; I’m just waiting for a used 13900K to pop up and I suspect it’ll happen when the KS drops in a few weeks. I’m contemplating selling the Dark to a friend if it has a hard limit that defeats its use case for me.

I know that 8000+ GB3 runs have been done, but I specifically mean for daily reboot stability. I really just need to know if anyone out there can confirm this. I don’t need any sort of IMC voltages, ODTs or anything other than just your word, or a basic screenshot of a reasonable amount of testing with the speed and timings, and verification of reboot stability and daily usability so I can get an idea of how much is left in the tank on this thing.

TIA if anyone out there would be willing to reply to me on this.

I don't have memory modules that are of good enough quality to run those clock speeds with respectably tight timings with 100% stability, and certainly not of adequate quality for anything like TM5 Absolut. I'm not sure how many people do, but I doubt there are many. I am finding a lot of inconsistency in silicon quality with DDR5 and more expensive doesn't always equate to better. I have purchased and returned about 8 retail DDR5 kits in the past 6 months, most of which were rubbish, and none of them have been as stable as my four generic naked green SK Hynix DDR5 16GB M-die modules with no XMP profiles. I have had a couple of retail kits that could boot and run benchmarks at 8000, but as soon as I start to tighten up loose timingsand bump up the voltage to correct the sloppy write, copy and latency performance they tend to crap out in stability tests. Even an extraordinarily good silicon sample is going to throw errors with the lousy factory thermal solutions they ship with. The garbage they call heatsinks actually cause them to run hotter than naked sticks with a fan blowing on them, and unless you can keep them below 45°C (I have mine on water) you are going to see errors in TM5 and MemTest Pro.