X399 Asrock, where the hell is new Agesa bios!!

Dram Power Down - Disabled - made NO difference, still intermittent bluescreen/blackscreen reboot @ 3200MHz XMP.....yet 3066MHz XMP still perfectly stable.

Edited by lowdog - 08 May 2018 at 9:39am

SoniC wrote: lowdog wrote: [URL=][/URL] SoniC wrote: lowdog wrote: [URL=][/URL] datonyb wrote: lowdog do you have the option 'dram power down' in your bios ? (on the taichi its in the dram configeration option on oc tweaker) disabling is the desired option Yes that option is there and is currently set to AUTO.....just read up on it so I'll give it a try disabled. There is also an option called "Gear Down Mode"....know what that does??? it's set to auto/enabled as well. Geardown is also VERY important. If your RAM requires 2T CR then in order to increase compatibility/stability - disable it . Here some more explanation / definition on what is Geardown Mode: "Allows the DRAM device to run off its internally-generated ½ rate clock for latching on the command or address buses. ON is the default for speeds greater than DDR4-2667, however the benefit of ON vs. OFF will vary from memory kit to memory kit. Enabling Geardown Mode will override your current command rate." The last sentence is CRUCIAL if your RAM requires 2T CR. And here some benchmarks: https://community.amd.com/community/gaming/blog/2017/07/14/memory-oc-showdown-frequency-vs-memory-timings @ SoniC - Have you tried disabling "Dram Power Down" option in bios to see if it makes a difference? lowdog - yeah, every possible combination. Read my first reply to scythefwd here. I went so deep that I started to modify the BIOSes lol. so Yeah, I touched every setting, every switch possible, and tried every combination of BIOS / AGESA. To be frank I think you are very near to reaching 3066 stability... I would try to pump 0.2-0.3 V more into the RAM, increase the SOC voltage by 0.1-0.2 Volts, change the "VDDCR_SOC Load-Line Calibration" to 2 or even 1.

You must mean 0.02 - 0.03 V into the RAM eh which I have already tried, went to 1.4 V for ram even and went from current fixed 1.05 V SOC with LLC AUTO/5 to fixed 1.15 V SOC with LLC 2.....same scenario, intermittent instability @ 3200MHz.


Will just have to stick with 2933/3066 for now.


Can't remember where I read it but some were of the opinion with SOC voltage and cpu voltage was best to leave LLC low like AUTO/5 rather than extreme like 1 as it may have had a negative affect on the cpu in the long run......the voltage fluctuation at AUTO LLC 5 is actually the parameter in which the cpu is supposed to operate via AMD design so I've always tended to keep LLC loose and try to get the set voltage for stability set correctly first.

Edited by lowdog - 08 May 2018 at 2:58pm

SoniC wrote: lowdog wrote: ? You must mean 0.02 - 0.03 V into the RAM eh which I have already tried, went to 1.4 V for ram even and went from current fixed 1.05 V SOC with LLC AUTO/5 to fixed 1.15 V SOC with LLC 2.....same scenario, intermittent instability @ 3200MHz. Will just have to stick with 2933/3066 for now. Can't remember where I read it but some were of the opinion with SOC voltage and cpu voltage was best to leave LLC low like AUTO/5 rather than extreme like 1 as it may have had a negative affect on the cpu in the long run......the voltage fluctuation at AUTO LLC 5 is actually the parameter in which the cpu is supposed to operate via AMD design so I've always tended to keep LLC loose and try to get the set voltage for stability set correctly first. Yeah I meant 0.02-0.03 on the RAM side and ~0.1 on the SOC side. Have you also tried to play with ProcODT?? Normally you can tune the LLC to your liking (especially to see if the LLC is causing the instability) when you are moving in a safe area of voltage.? It has been perfectly explained by Gamers Nexus here > https://www.gamersnexus.net/guides/3118-what-is-load-line-calibration-llc-for-overclocking But yeah... leaving LLC on 1 for a longer time AND staying at an upped voltage would have a negative impact on your CPU. However fine-tuning LLC can prevent instability caused by voltage fluctuation going too low. On my X1950 I have noticed that it spikes to 1.41 on CPU BY DEFAULT, and I remember that TR needs actually 0.1-0.15 less Volts compared to AM4 (X1800 reached but not exceeded 1.37) so I fine tuned the voltage with an -0.1 offset (currently running up to 1.34-1.35 which is completely fine for me).

But those vcore spike to the cpu when on auto volts are not to the whole cpu but per core and very temporary as per AMD spec......read it's better for the cpu to run cpu stock and have XFR with voltage spikes that set for example 3.9GHz cpu overclock with a constant 1.34v, the constant volts will have a more negative affect on the cpu over time than the in spec behaviour of stock volts with offset spikes.


I can run my cpu @ 4Ghz with 1.355 vcore set and LLC auto/5 which gives a real vcore measured at core via DMM of around 1.33v (can't remember exactly but near enough) but I just leave it auto most of the time cause I couldn't be fuged ...just wish I could get more out of the RAM ffs.

Edited by lowdog - 08 May 2018 at 3:48pm

lowdog wrote: Dram Power Down - Disabled - made NO difference, still intermittent bluescreen/blackscreen reboot @ 3200MHz XMP.....yet 3066MHz XMP still perfectly stable.

Given that the PC seems to fail at idle, have you tried disabling the C-states to see if it has any effect?

So it seems we need to wait... Like I've said -- it will come eventually. "in near future" intrigues me... As far as I know the release cycles of ASRocks BIOS/UEFI we should get one in a week or two (leaked, JZ - I am looking at you my friend! & count me in as a beta tester  ) and till the end of month as a stable @ASRock website... 

Edited by SoniC - 09 May 2018 at 2:46am