Originally Posted by danny9428
I think it’s more because very few of us even had the 3647 platform with the 3175X
Frankly none of us could speak of any due to lacking experience with this particular chip and 6-channel DDR4
I realized this after I posted - haven't seen many, if any, builds of this platform on this forum.
I am building a second rig with the 10900K and have all the parts except the CPU in hand - so hopefully OC'ing that will be more of a group-effort than this platform.
Originally Posted by J7SC
^^ That !
I do remember seeing YouTube vids (I thought by DerBauer, but not sure, could have been Linus) about the 'odd' mounting mechanism for Intel's 3647 socket CPUs being quite tricky and affecting memory channels, but I simply cannot locate them w/o several hours of viewing earlier vids. In any case, OCN's knowledge base is far more likely to be able to troubleshoot dual and quad channel than 6- or 8-channel RAM.
In addition, there's Level1Techs vid below about TRX40 AMD TR running 256GB quad-channel on 8 slots @3600...not the same as 6-channel 3175X or a 'dead RAM slot', but nevertheless underscoring oc-challenges with high-cap, all-slot RAM
Yes - finally figured out the issue. It was indeed a bad DIMM slot (DIMM4) - I got a new board and the 96GB was recognized instantly at stock settings. The only "issue" I'm having now is if I load XMP (4000Mhz @ 1.35V), it only recognizes 32GB. I figure it's a voltage thing since Luumi's videos on this platform showed him using 1.98V to get 3600Mhz which is downright insane.
I spoke to EVGA CS and they told me to try a kit that has been tested & verified with this MoBo and so I bought another RAM kit - same 4000Mhz as the Corsair Dom Platinum RGB so hopefully that works out of the box.
Originally Posted by Martin778
What's the VRM and PCH temp on the SR-3? I wonder if running coolant through the block wouldn't be mandatory when OC'ing?
Are those temps available through HW Monitor? Yes, that is one of the things about my builds - I have neither the interest nor the patience for water-cooling so the VRM water-block is just passively-cooled at this point. I confirmed with EVGA before buying the board that it is perfectly fine to use passive cooling for the VRMs even when OC'ing (not extreme OC'ing I presume).
Regarding VRM temps, I guess I'm running into yet another issue there - I turned off C-States in the BIOS and set the "Power Management" to "High Performance" however during Cinebench R20 (or R15), I see that the CPU downclocks to about 3.3Ghz (my OC is 4.50Ghz) - see pic. Can this be caused due to the VRM temps? If the C-States and Power Management settings are made to prevent down-clocking - why is it still doing it?
The SR-3 BIOS has a tool for "Stress Testing" and "OC Robot" - when I stress-test with this OC setting, it does NOT downclock at all. Only during Cinebench do I see it downclock (of course, I haven't tested the CPU with any other benchmark yet - do you guys recommend any other good programs for testing CPU OC?)
For my OC, this is what I have:
1.) V-Core: 1.20V
2.) Mesh: 30
3.) AVX Offset: 0
Everything else is on Auto - this is where I need help. What do I set Uncore voltage, VCCIO, VCCST, VCCPLL etc.?
Again, with the Asetek 690PX-LN AIO CPU cooler and passive-cooling on the VRM/chipset, there's only so much I can do but I want to optimize my settings as much as possible.
HWMonitor - BEFORE Cinebench R20 run:
HWMonitor - AFTER Cinebench R20 run:
the completed rig: