A few days ago my newly procured hardware arrived - amongst it a memory kit from G.Skill (4x 8GB DDR4-4000 CL17-17-17-37 XMP, 1.35V, B-Die) (F4-4000C17Q-32GTZR)
, since my older memory kit (G.Skill 2x 8GB DDR4-3200 CL14-14-14-34 XMP, 1.35V, B-Die) (F4-3200C14D-16GTZR)
sucked at memory OC despite also being Samsung B-Die... a possible reason is, since it's only a 2 DIMM-Kit but my motherboard has a T-Topology memory layout favouring a 4 DIMM setup, the OC potential could automatically been compromised (besides
the fact, that this former kit is binned for 3200 MHz only). Another possible explaination is, that my CPU IMC is probably garbage. So immediately after the hardware's arrival, I first...
1.) cleaned my case, motherboard, CPU-Cooler and case fans with a vacuum cleaner, removed the heatsink of my Noctua NH-D15 air cooler in order to access the DIMM slots (a part of the bulky heatsink was right on top of the DIMMs) and...
2.) cleaned the top of my CPU - I did NOT remove it - and the bottom of the Noctua heatsink from the residues of thermal paste (which is a CM Mastergel Maker Nano) with isopropyl alcohol. Then...
3.) I removed my former 2 DIMM-kit from the DIMM slots and immediately inserted all 4 DIMMs of my NEW memory kit - now, all 4 DIMM slots are occupied, which is the way to go for the T-Topology layout of my motherboard. After that...
4.) I re-applied thermal paste (again CM MGM Nano) on top of my CPU and then re-installed the Noctua heatsink. Next...
5.) I reconnected all cables and ports I had to unplug during the cleaning and memory install process, then switched on the PSU's power button and finally started the booting process.
While performing the above mentioned 5 steps, I meticulously took care not induce electrostatic discharge.
At first, the system didn't boot at all, but after I eventually decided to clear CMOS/reset BIOS, I could finally boot into the BIOS menu. There I recovered my previous BIOS settings and temporarily set the new memory to DDR4-3200 CL14 (Voltages manually set: 1.35V/1.05V/1.15V for VDIMM, VCCIO and VCCSA, respectively) alongside my usual 2nd & 3rd timings tweaks. The CPU remained unchanged @ 4.8GHz (Uncore: 4.5GHz) All-Core AVX Offset "0" @ 1.325V LLC5 (and the remaining settings in the "External DigiPower+ Control" sub-section all maxed out, as they already were since a long time). Being finished with that, I could successfully boot into my Linux Live-System and a while later, I plugged in the LAN-Cable into the Ethernet port of my motherboard (the NIC is a Intel I219-V
) and from then on, I could successfully surf in the internet for hours without any issues!
After logging of/shutting down my Linux Live-Session, I booted to W10 and likewise, I had flawless internet connectivity for hours!
Immediately after being done with W10, I returned to the BIOS menu and started my first overclock with the new memory kit:
at first, I set memory frequency to DDR4-4000, the primary timings to 16-16-16-48 and Command Rate to "2" - alongside that, I increased voltages initially to 1.4-1.45V for VDIMM and to 1.175V/1.225V for VCCIO/VCCSA. The BCLK:Memory ratio was 100:100, as always. In the End, the maximum voltages I tried out during my overclocking attempts were 1.475V/1.225V/1.275V VDIMM/VCCIO/VCCSA.
The Good News:
After a few dozens of failed POSTs, Retries, Resets and Adjustments (Voltages, Subtimings), I could finally get that setting to work. The positive news is: at that point, I could already tell, that the new memory kit is definitely better
than my former 2 DIMM-Kit - memory training before POST took longer *but* eventually always succeeded (or almost always?), whereas my previous kit only succeeded in the beginning (and on later reboots/POSTs failed to POST or POSTed inconsistently) or didn't POST at all no matter how long I waited for it or my system/applications/games crashed, once DIMM temps exceeded 40°C. I didn't have yet the opportunity to do much stability testing with the new kit, but in contrast to my previous kit, it didn't crash a single time
in Windows 7 (Win7 is my still my primary OS, W10 is secondary - it's a Dual-Boot setup) or after hours of gaming, despite temperatures of all 4 DIMMs already reached 41-42°C in the latter workload!
Of course, much more thorough long term stability testing is necessary, but from my current point of view, the new kit is definitively more stable than the former kit and for the first time, I could lower 2nd & 3rd timings @ DDR4-4000 speeds and then successfully POST and
run the system stable for hours! With my previous kit, when lowering 2nd & 3rd timings in DDR4-4000 memory frequency, I couldn't even get it to POST, let alone get it to boot or stay stable...
(The Bad News)
Like written above, I could successfully boot into both Windows 10 and then Windows 7 and play pretty stressing games on the latter OS for hours without any issues in my first overclocking attempt with the new memory BUT
already on the *very first boot* into Windows 7 with this new setup, I noticed the network adapter wasn't working properly
regardless of the fact that the system was stable. At first, I didn't even think the cause was the network adapter at all, because occasionally (or even frequently) I get internet connection losses from anywhere between just 15-30 minutes to up to a couple of hours, because of the crappy quality management and/or maintenance work of our ISP. I always know, when
the cause is the ISP, because other devices also lose internet connection including the devices of other family members! So, I assumed the cause was again the same as always and would soon subside from itself, and didn't care about it - instead, I played my games in Steam Offline mode for hours without crashes. However, when I came back to my computer 7-8 hours later, the network adapter was still not working
- still with a constant rotating circle around it's symbol in the taskbar of Win7. When hovering the mouse over the symbol, it says: "Unidentified Network - No Internet Access"
I then checked my other devices and the devices of my family members for internet connectivity and was surprised: they ALL had proper internet connection - so, the cause was NOT our ISP, for once!
At that point, it dawned on me, that something was wrong with my onboard LAN-Controller...
And this is all I tried without success in order to get the LAN-Controller to work again:
1.) Disabling LAN-Controller in the BIOS and then re-enabling it.
2.) Disabling LAN-Controller ( "Intel I219-V (2)" ) in Win7 Device Manager with administrative previlegies and then re-enabling it.
3.) Checking IPv4-Configuration in Adapter Settings (tried auto-retrieve IP address & Gateway, tried manually specify IP adress & Gateway)
4.) Completely* uninstalling Ethernet drivers and then reinstalling them.
5.) Checking LAN-Cable connection both to motherboard and router.
6.) Checking Router settings -> the device name of my computer is not shown, despite the same static IP adress like before and despite network adapter apparently working!
7.) Reverting all OC-relevant settings to their defaults or even below that in the BIOS - default CPU speed, default memory speed & timings, default/auto memory settings, manually fixed and lowest possible voltages for CPU & memory.
8.) CMOS reset and completely re-configuring BIOS from scratch with settings like in 7.)
9.) BIOS Flash/BIOS Update and completely re-configuring BIOS from scratch like in 7.) (-> now, my motherboard has the latest BIOS version
10.) Shutting off PSU *and* shutting off power delivery to PSU -> wait for over 1 hour -> reinstate power.
* = including checking and deleting residual files & folders, additionally uninstalling device driver in Device Manager with administrative previlegies and on top of that deleting the corresponding drivers from Windows Component Store via "pnputil")
I'm out of ideas...
And now, it get's really interresting:
When I boot to Windows 10 instead, the same problem now occurs there too: network adapter apparently working, but no internet connection with the same error popup: "Unidentified Network - No Internet Access" - despite the day before, I could surf on the internet in W10 for hours without the slightest problem! I then tried the same steps (actually, steps 1-4) like in Win7 and that damn thing is still not working!!
It get's even better:
Finally, I pulled out my Linux Live Distribution again, because it has out of the box LAN-Controller functionality without the necessity to manually install a network driver. After booting it up, the same **** occurred but I wasn't surprised anymore: the network adapter was apparently working but my LLD too noticed me that there was no actual internet connectivity - just a day ago, in contrast, it worked like a charm!
I never dreamed or imagined, that the onboard LAN-Controller of a motherboard could be the first victim
of Overclocking - when overclocking, one thinks that the component being overclocked (CPU, RAM, GPU) always "gets it" first and everything else comes secondary to that...
I have very little doubts left but I still ask you here: is my LAN-Controller bricked?