Can’t Say I Wasn’t Warned About the Samsung 990 Pro…
I had seen posts that the Samsung 990 PRO NVMes fail or keep vanishing from systems and recommending the WD Black SN850X instead, which is also cheaper, and I definitely trust the WD Black range, but I nevertheless went for a 1 TB Samsung 990 PRO with Heatsink for the new computer‘s primary storage, hoping to not be one of the unlucky ones… But it seems that I am, and an added problem is that, with Windows 10 being installed on it and that key being purchased as it was, completely legally and legitimately on my part but likely in a gray area on the part of the seller and with warnings that it may not work if not used quickly, I may be unable to activate it again if I’ll need to reinstall on a new drive, if I’ll replace this one under warranty.
The thing is that, at this point, the problem seems to just appear once in a while and not have lasting effects, but that’s unlikely to last. I mean, the problem first appeared in the early hours of December 13, when I had been pushing the system, having BOINC run Asteroids@home and MilkyWay@home work units at the same time, which seems to really push the temperatures up, the CPU approaching 70°C, while I also had a fair number of browser tabs open, including some music on YouTube, so when I found myself staring at a BSOD listing a WHEA UNCORRECTABLE ERROR and seeming unable to even gather information, being stuck at 0%, I initially thought that I had pushed something too hard, but after pressing the reset button I found that the NVMe wasn’t seen on the system anymore. However, it reappeared after I pressed the power button, which also resulted in a reset, the system booted back up normally and chkdsk didn’t find any problems, so at that point I just ended up getting in bed at 5:20 AM, concerned but at the same time relieved that things seemed to have sorted themselves out for the moment. And after getting up I put that error’s raw data, which was the only piece of information that was saved, through a hex to text converter and confirmed that it was caused by the NVMe, so before going to bed at night I sent a message to Samsung support about it, the initial reply being that it’s good that I have the latest firmware, since I had mentioned that, as well as the fact that the reported temperatures are low, and that as long as it’s seen again, things should be all right.
Well, that obviously wasn’t the case, since it happened again last night, at a time when I wasn’t at the computer and BOINC was just running MilkyWay@home, which actually keeps the CPU below 50°C, so I got back to my room and found it in BIOS, with the Samsung NVMe once again gone, and everything yet again appearing normal after I pressed the power button. And, after chkdsk once again didn’t find any problems last night, after I got up today I also ran the complete extended self-test on the Samsung NVMe from BIOS, and it passed. And I also followed the advice from a post, turning off the LED from the software and setting the M.2 to GEN4 in BIOS… Though another interesting thing that I noticed in that post is that the user in question also has an ASUS TUF GAMING motherboard, though on an Intel platform. And another discussion also mentions ASUS motherboards at first, though further down others are also listed and people just trash these NVMes.
Still, the possibility that the problem is caused by something else does remain, and in another discussion I saw that a motherboard problem may cause it as well, but my first concerns were about the RAM, since the first BSOD was actually on December 7, when, after finally switching back to the old computer to make sure that everything’s properly backed up and also allow it to finish the last Asteroids@home work units, I switched back to this new one, used Group Policy to turn off Cortana and telemetry, rebooted, and found myself staring at that BSOD, with a page fault in nonpaged area error, apparently from the TV tuner‘s remote control driver, which I thought I had stopped from loading on startup, so that seemed rather weird. But using the memory diagnostic function from Windows didn’t reveal any errors, and memory or processing problems would have likely at least resulted in multiple invalid work units for BOINC projects by now, which doesn’t seem to be the case, so I passed that off as an oddity, since everything was normal after a hard reset and nothing of the sort happened since then… But, on the other hand, it is interesting that all of these problems started after turning off telemetry.
I could try switching positions, but I now have that 4 TB Kingston NV2 as well, which I placed in the second slot, and since I put the motherboard‘s heatsink back on in that place and that one covers slots two and three, and I removed that film from the heat pad for slot two, there may be consequences if I move it again, and the fact that I got the Samsung one with the heatsink means that I’ll have to keep that heatsink off if I’ll put it in either of those two other slots. Not that getting the one with the heatsink instead of the regular version wasn’t a bad choice either way, seeing as the motherboard has its own heatsinks for those slots and those would probably also serve to hold the NVMe in place better, but since I didn’t really know anything about NVMes before getting this one, that never even crossed my mind until I actually had it, so I’d have likely ordered the version with heatsink instead of the regular one even if I’d have gone for the WD Black… Which I really, really should have done. Not that it’d have guaranteed a lack of problems, and if they’re not actually caused by the NVMe it might have made no difference, but my chances would have been much better.
Either way, since I’m writing this, I should probably mention some other things as well, starting with the fact that, on December 16, after ending up giving my hosting provider temporary access to the blog and having it be all for nothing, since they couldn’t sort out the problem that I had reported, I ordered that Kingston NVMe, with dad finally managing to pay by card, so I could also extend the warranty for it to seven years. However, I’m still not seeing it among the products that I can request to be serviced from my account on the store’s site and therefore can’t verify that the extended warranty does apply and I’m worried that the payment problems messed something up, since dad had made a first attempt on the evening of December 15 but then realized that he couldn’t authorize it, then left me the information in the morning and went to the bank to sort things out, so I tried again while he was there but it still didn’t work, probably for different reasons, and it finally worked when he tried yet again after getting back, the nice thing being that the store had reserved the product, which seems to have been their last one, during that time, offering a link that was valid for 24 hours and allowed us to try to pay again without needing to actually order again or risk having someone else buy it. Since the product was delivered, on December 18, it would seem that everything was fine, but the delivery guy didn’t ask for a signature, or even for the PIN that would have confirmed that I was the intended recipient, so I’ll give it a little more time but, if it still doesn’t appear there, I’ll have to ask what’s going on.
The thing is that I’m not using it at all yet. I mean, after vacuuming the hallway and having Liza jump straight on my back when I was close to finishing, trying to cling on and scratching my back and shoulders in several places, I did install it that day and ran the full extended test in BIOS, but that was it… I even forgot to disconnect the case’s power LED when I opened it. But I’m going to have to poke around again anyway, not only in case the problems with the Samsung NVMe will continue, but also because the plan for the Kingston one is still to put Linux on it, at least just in case, and I want to disconnect the Samsung at that point, so Linux won’t detect the presence of another OS on the system when it’s installed and the two will remain completely separate. Either way, this means that, while other partitions will be for Windows or for both, if the first one, where I’ll put the OS, is intended for Linux, I didn’t even initialize it under Windows yet. But, since I had to turn off the computer that day anyway, I took the opportunity to switch to the new UPS again, plugging the data cable into the computer but not installing any version of the software yet, so I can confirm that it’s still sending data, but have no idea whether it won’t appear to stop doing so if and when I will install the software, as it did on the old computer.
Speaking of the old computer, on December 12 I updated and activated Windows 10 on it as well, being somewhat surprised that the second purchased key still worked. But I put the computer in the living room and connected it to dad’s peripherals while I did so, and then, when I switched back to the installed Windows 7, I forgot to make a system image backup before making an account for dad, the last one being made back at the end of September… And I should have really made another, because after making that account I ended up with that conhost.exe crash that makes it unusable that I was mentioning back in 2018 and 2019. So I’ll see what I can do about that, but I didn’t try anything so far, just bringing it back here and leaving it next to the desk, since dad isn’t ready to switch to it yet.
Otherwise, I’ll also mention here that no visits were recorded on December 7, 8, 9 and 21, and no visits for three consecutive days was quite surprising. And I’m also going to say that I got in bed at 5:20 AM on the morning of December 8, and the same thing happened this morning. On the other hand, after having started it on December 2, on December 8 I finished The Dromost Gate, also writing the quick review. And before going to bed in the morning, so in the early hours of December 9, I started The Emerald Gate, which I finished on December 14, though what I scheduled to be posted at 11:59 PM was just the rating for the quick review, the text being added later that night. And then, on December 20, I started reading The Lost Metal, which would be only the 11th book this year, so I have to finish it and one more by the end of the year if I’m to stick to the usual target, which is quite embarrassing in itself. But the fact that yesterday I played Cave Story’s Secret Santa, mainly in order to write the quick review for it as the week’s first post, means that I finished four games this year, making it the first one since 2021 when I meet at least the low goal, 2020 remaining the last year when I can say that I finished five. But I’m still wondering when or if I’ll manage to write the reviews for the “proper” games that I finished this year, meaning Diggles, the first Vampires Dawn and Heretic Kingdoms: The Inquisition. On the other hand, after it was the only one purchased for myself out of what actually were the first games purchased since August of 2020, on December 9 I actually started one more, Emperor: Rise of the Middle Kingdom, albeit just completing some of the tutorial missions so far.