X370 Fatal1ty Gaming Professional - [BIOS v2.x]

parsec wrote: yoonie wrote: Well, I figured it out.  Either I'm misremembering or something changed (probably the former) because I could have sworn that the fan curves responded to tdie temps and not the socket temps but I guess not. Since the socket temp stays in such a small range that explains why the fan speeds barely change.  As a follow up, in general should I be monitoring the socket or die temps?  AMD said tjmax for R7 is 75C but I'm not sure which reading they're referring to.  This is my first AMD system since the Athlon so I'm still getting used to their terminology. What version of HWiNFO are you using? Mine looks very different than yours (v5.51-3135). AMD does not have a CPU socket temperature (thank Gawd!) on Ryzen.  The CPU reports two temperatures Tctl (T control) and Tdie. Tctl has a +20°C offset programmed into the CPU by AMD for their stock CPU coolers, for use with the Ryzen 5 series processors, apparently. Tdie is being monitored for the fan speeds. Are you calling the Tctl the socket temperature? When stress testing using AIDA64, which shows both Tctl and Tdie, I was at ~80C (Tctl) during the test, without any throttling of the CPU speed. AIDA64 is showing this for the CPU temperature limit (HTC Temperature Limit, Tctl.) Who knows if this is correct?

Edited by yoonie - 20 Apr 2017 at 9:17pm

I forgot to mention that I am cooling with a single liquid loop.

It is a typo. It should be 1.192v. I verified the CPU v fluctuations with CPUZ. It is definitely not stable using v1.93D and v2.0. 

I wasn't looking for Ryzen compatible memory.

Two things:
I LET Intel Burn Test run 10 iterations of memory stress testing and hwinfo showed a max VRM of 52C from the IRF whatever it is sensor.  I have 2 of those listed and the VRM temps look inverted however they must be different at the voltages and other measurements DO differ,  Not at all horrible AFAIK.

Apparently the AGESA for P2.00 BIOS IS INDEED 1.0.0.4a and NOT 1.0.0.3 as identified by various means according to ASROCK support.  As to why it show .3, that varies from AMD 'forgot' to update the identifier string to various other reasons.  uCode 111C IS also backing this up as it is SUPPOSED to be the .4a release uCode INCLUDING the FMA3 'fix'.

That said memory support is utter crap for me w/P2.00 and I HALFWAY want to go back to 1.93D HOWEVER I SUSPECT that ASROCK REMOVED it for a VERY GOOD reason. AGESA .5 CANNOT get here FAST enough!

[EDIT]
I suppose that I should've mentioned this WAY back in the thread, but I'm using a Corsair Carbide R600C INVERTED ATX case WITH an H110I AIO cooler.  The radiator is mounted in the BOTTOM of the case drawing air INTO the case, the case fans are as default, which IIRC are 2 rear drawing air in, and one front pushing out.  My eVGA 1.1kW PSU is top mounted(inverted) pushing air out.  GPU is an eVGA 1070 FTW(don't ask, I got it SUPER cheap, ty newegg as they screwed me on my ORIGINAL RAM order that I ordered WITH the GPU o.w. I would've ordered a 1080+ for the discount that I got... c'est la vie...)

I'm REALLY LIKING this inverted setup case as it runs pretty damned well cool, however it has DRAWBACKS only 2x 3.5" bays, 2x 5.25" bays and 3x2.5" ('SSD') bays, BUT IF AMD release an HEDT Naples based CPU this thing will be immediately demoted to a local server alongside the 3930k system that it's replacing.  TBH I'm REALLY hoping for HEDT as the paucity of PCIe lanes is already killing me, MUCH more than I had originally thought... really the current chipsets/CPUs are more like midrange Intel and I probably will build a naples based system IF AMD do NOT DO HEDT focused simply because of paucity of PCIe lanes, and dual channel memory.  My original thought this time around for a desktop anyways was to seriously consider using server/workstation parts.
[/EDIT]


Edited by cutterjohn - 22 Apr 2017 at 7:29am