|
makere posted:You can buy a bag of M2 screws, with the standoffs.
|
#
?
Jul 17, 2023 11:55
|
|
|
|
| # ? May 14, 2024 02:53 |
|
|
Does it have any negative effects on NTFS, if you expand the partition after cloning it to a bigger drive? With what not initially having been laid out to drive size (like MFT layout and such)? --edit: Yikes, Macrium under regular Windows is already showing two additional partitions that Disk Management is hiding. One is behind the system partition. This is gonna be fun. —edit: Macrium aborts at 40% with a CRC error 
Combat Pretzel fucked around with this message at 13:12 on Jul 17, 2023 |
|
#
?
Jul 17, 2023 12:01
|
|
|
Combat Pretzel posted:Does it have any negative effects on NTFS, if you expand the partition after cloning it to a bigger drive? With what not initially having been laid out to drive size (like MFT layout and such)? No, in any non-degenerate case. The resize operation should handle things like moving files in/out of the MFT reserved zone. And especially no on a SSD, where concepts like "contiguous" are pretty meaningless. Run an optimize after you resize. Combat Pretzel posted:--edit: Yikes, Macrium under regular Windows is already showing two additional partitions that Disk Management is hiding. One is behind the system partition. This is gonna be fun. If they are very small 16-128mb partitions, they're a semi-useless thing that microsoft puts in for unexplained reasons on GPT drives. Possibly something for dynamic disks / storage spaces? I dunno. If you have plain-jane partitions and storage, you can delete them with no effect.
|
|
#
?
Jul 17, 2023 13:28
|
|
|
Thanks for the info. That was certainly odd. The first two attempts of an "intelligent sector copy" in Macrium RE failed with a read CRC. Then I tried a forensic copy, that also failed. Then I ran CHKDSK /R /F on the source drive, it found nothing wrong, then I attempted another intelligent sector copy and then it finished properly.  I hate computers.
|
|
#
?
Jul 17, 2023 13:48
|
|
|
kliras posted:are they standardized on old crusty boards like asus c7h? i've heard some horror stories, and could never find any good places to buy from here in europe Most boards by that era were standardized, but the Crosshair VII Hero has an M2 heat sink on the top slot, so it may not have a standard mounting depth. Honestly it doesn’t matter much as long as the drive is held down, I’ve seen plenty of people in server land who just kapton tape them down lol. Also on that board, you really should only use the top m2 slot if you can. Using the lower m2 will bifurcate the lanes off of the first PCIe slot, forcing the GPU to 8x.
|
|
#
?
Jul 17, 2023 15:50
|
|
|
kliras posted:yeah that slot is basically dead to me. hell if i'm going to wrestle my d15 every time i want to manage something there they should bring this back for the primary CPU-linked m.2 slot 
|
|
#
?
Jul 17, 2023 15:55
|
|
|
Combat Pretzel posted:That was certainly odd. The first two attempts of an "intelligent sector copy" in Macrium RE failed with a read CRC. Then I tried a forensic copy, that also failed. Then I ran CHKDSK /R /F on the source drive, it found nothing wrong, then I attempted another intelligent sector copy and then it finished properly. Do you have a bunch of smart errors on the drive now? Particularly uncorrectable errors. If so, it could be that macrium hits a read error and says "data's not trustworthy, abort" while CHKDSK says "oh well, take whatever garbage the drive gave me and mark it repaired, move on with life". And so when Macrium ran successfully, those bad sectors had been filled in by chkdsk. In that scenario your image now has some errors in it, which is... I dunno, probably fine? What are the chances they hit something vital or data you care about. Your call to either move on with life, or clean install / restore backups. Alternately if you have no smart errors on the source drive, it's probably not worth worrying about... but maybe means you want to check sata cables for bend / pinch, or re-seat a m.2.
|
|
#
?
Jul 17, 2023 16:28
|
|
|
Cygni posted:
shame pcpartpicker can't specific which nvme slots you use on your build, would make it neater for pcie lane management
|
|
#
?
Jul 17, 2023 16:40
|
|
|
I guess I'm glad I upgraded the system drive. That said, I still need to install smartctl and see what's up. This is what Crystal DiskInfo says: Available Spare is 0x64, which is 100 in decimal. Which I presume is a percentage. It doesn't list uncorrectable errors or anything. That said, I put the source drive in a different slot, maybe there's issues there. According to the manual, it reads like it's hooked to the CPU, but by location on the mainboard, I'd say it's on a chipset. There's no usable block diagram of the X670E TUF from Asus, so I have no idea. --edit: The LBA in the error log that smartctl lists is always 0. So who knows what's loving up. I can't find a list of these CmdId's to decipher. Combat Pretzel fucked around with this message at 16:58 on Jul 17, 2023 |
|
#
?
Jul 17, 2023 16:42
|
|
|
Combat Pretzel posted:I guess I'm glad I upgraded the system drive. That said, I still need to install smartctl and see what's up. This is what Crystal DiskInfo says: Media and Data Integrity errors are the bad ones, similar deal as uncorrectable read errors for smart. NVMe spec posted:Media and Data Integrity Errors: Contains the number of occurrences where the controller Uh, I don't know what I'd do in your shoes. 3 is ... a small number. But you'd really prefer that number to be zero! I suppose now would be an easy time for a RMA, since you just got a new drive. Or you could let it ride and check crystaldisk every few months to see if the number is going up. Combat Pretzel posted:--edit: Error log entries are 100% ignorable if you aren't an engineer working for the company that made the thing. NVMe drives can dump their internal error log, but it will mostly be trivial unimportant poo poo like corrected errors or every time the PCIe bus has a hiccup. (LBA 0 probably indicates errors with the bus or other non-flash stuff.) ~1000 isn't unusual at all. You'd have to dump the full log to look for where the bad errors were, get real deep in the weeds.
|
|
#
?
Jul 17, 2023 17:26
|
|
|
Klyith posted:Uh, I don't know what I'd do in your shoes. 3 is ... a small number. But you'd really prefer that number to be zero! However I'm still confused as to why CHKDSK didn't find anything. When it checked the user data, it took its sweet time, compared to free space checking. Alas, if there was actual data corruption, it's too late, anyway. --edit: Samsung says the 970 EVO has 5 years of warranty. Hmmm. Combat Pretzel fucked around with this message at 17:37 on Jul 17, 2023 |
|
#
?
Jul 17, 2023 17:35
|
|
|
my entire computer literally fell over, and only crystaldiskinfo was able to tell me my mechanical drives were f'ed - after it was pretty much too late to do anything about it i'm an extremely annoying cdi evangelist now
|
|
#
?
Jul 17, 2023 19:52
|
|
|
Klyith posted:Also, considering the failure rates you have in every thread in SH/SC, I'd suggest a different storage medium: In fairness, that was our first handheld productivity device, and it's a shame we don't learn how to use them in school.
|
|
#
?
Jul 17, 2023 20:02
|
|
|
kliras posted:yeah that slot is basically dead to me. hell if i'm going to wrestle my d15 every time i want to manage something there https://www.amazon.com/dp/B08S7NV2B7 I bought this when I needed to change out a m2 drive and instead of being an exercise in frustration it only took a few minutes. It took longer to unplug everything and unseat my video card than it did to switch the NVME drive.
|
|
#
?
Jul 17, 2023 20:36
|
|
|
TOOT BOOT posted:https://www.amazon.com/dp/B08S7NV2B7 but my experimental ambitions were also just dashed after looking at my manual and realizing that this old piece of crap motherboard apparently has friggin pci 2.0 slots starting to think we should just go full usb and colour code the slots based on version, maybe even multiplier ...
|
|
#
?
Jul 17, 2023 21:22
|
|
|
Combat Pretzel posted:However I'm still confused as to why CHKDSK didn't find anything. When it checked the user data, it took its sweet time, compared to free space checking. Alas, if there was actual data corruption, it's too late, anyway. I don't think CHKDSK can truly 'check user data' by itself. Looked at its documentation and nothing hints at that. All it can really do is issue a read and let the disk (whether HDD or SSD) report if it read the data successfully. In HDDs, a weakly bad sector can sometimes be successfully read by trying over and over again until one attempt works just well enough to drop the detected bit errors below the fully correctable threshold. It's not implausible that this could happen with SSDs too. In both, the drive's controller should notice "hey, I just pulled off a difficult read" and move the data elsewhere. With HDDs this was done with a few reserved spare sectors and a mapping table to identify which sectors have been spared out, and where. With SSDs, thanks to wear leveling every LBA is always remapped, so it's even easier to move the data to a better location. Also, in both, if you overwrite a troublesome sector the drive's firmware should take that as an opportunity to retire that physical sector and remap it (HDD) or put it on a "never use again" list (SSD).
|
|
#
?
Jul 17, 2023 21:42
|
|
|
Any quick benchmarks I can do before I replace my SN550 with an SN850X (both 2TB)? Mostly for fun. I'll be cloning the drive.
|
|
#
?
Jul 19, 2023 03:58
|
|
|
Rinkles posted:Any quick benchmarks I can do before I replace my SN550 with an SN850X (both 2TB)? Mostly for fun. I'll be cloning the drive. CrystalDiskMark (thread title)
|
|
#
?
Jul 19, 2023 04:16
|
|
|
Klyith posted:If they are very small 16-128mb partitions, they're a semi-useless thing that microsoft puts in for unexplained reasons on GPT drives. Possibly something for dynamic disks / storage spaces? I dunno. GPT secure boot requires partitions with a specific structure and contents to boot from, and these partitions are where windows ( and Linux ) put that media. If you are using bitlocker, this is also where the programs that can read TPM and automatically decrypt your drive live. Please do not delete them.
|
|
#
?
Jul 20, 2023 07:23
|
|
|
EoRaptor posted:GPT secure boot requires partitions with a specific structure and contents to boot from, and these partitions are where windows ( and Linux ) put that media. If you are using bitlocker, this is also where the programs that can read TPM and automatically decrypt your drive live. Different thing, I'm talking about a Microsoft Reserved Partition. You're talking about the EFI System Partition. (Disk management will show you the efi partition, but won't let you do anything with it. It hides the MSR completely.) MS says: quote:Microsoft Reserved Partition (MSR) So I don't think it's needed for bitlocker or secure boot, but MS isn't 100% clear. OTOH perhaps I shouldn't tell people it's ok to delete this unimportant thing that's also possible to mistake for the very important thing. 
|
|
#
?
Jul 20, 2023 13:48
|
|
|
How good are the Crucial P3 and P3+? I remember trying the P2 and finding it slow, and the P5 was my go-to. My brother in Canada is getting what appears to be a S.M.A.R.T. warning on boot, so I want to recommend a replacement he can order on amazon.ca, but the P5 at 256GB isn't popping up in the results.
|
|
#
?
Jul 24, 2023 19:10
|
|
|
TITTIEKISSER69 posted:How good are the Crucial P3 and P3+? I remember trying the P2 and finding it slow, and the P5 was my go-to. My brother in Canada is getting what appears to be a S.M.A.R.T. warning on boot, so I want to recommend a replacement he can order on amazon.ca, but the P5 at 256GB isn't popping up in the results. P3 & P3+ are QLC drives, so I would not recommend them at 256GB size. Small drive capacity makes the bad parts of QLC much worse. Looking at pcpartpicker in canada, there are WD Blue 570s for cheap at non-amazon places but amazon.ca is weirdly high. You could get a Teamgroup MP33, or jump all the way up to a WD SN770 for $40. (256GB is, in general, not going to be a fast drive even if you pick from one of the higher-end models. There's not enough flash to spread parallel reads & writes across. Like, the performance of the 250GB WD SN770 is about a third that of 1TB-2TB version. Also 256 is now starting to fall below the size of even a single flash module, so you're paying more $/GB.)
|
|
#
?
Jul 24, 2023 19:37
|
|
|
Trip report on upgrading the lab computers: both were a great success. Ended up going with 500 GB MX500s. One of the computers took a whopping hour and a half to clone over the ~70 GB of data (yes it was defragmented), it's seen the greatest boost from the procedure. The other took 15 minutes for roughly the same amount of data and it's database no longer swamps the I/O. Thanks for the advice.
|
|
#
?
Jul 24, 2023 20:26
|
|
|
Klyith posted:P3 & P3+ are QLC drives, so I would not recommend them at 256GB size. Small drive capacity makes the bad parts of QLC much worse. Thanks very much, it's an old laptop so I don't want to make it any slower. He is also considering a new laptop.
|
|
#
?
Jul 24, 2023 20:32
|
|
|
TITTIEKISSER69 posted:Thanks very much, it's an old laptop so I don't want to make it any slower. He is also considering a new laptop. Oh. We're talking about "fast" and "slow" by the standard of current PCIe gen4 drives which are very very fast. In a laptop that's more than 2 or 3 years old you're probably limited to gen3 anyways. Pretty much any SSD is fast enough to make general desktop use as snappy as the other hardware can manage. (Also, how old are we talking about here? Have you confirmed that the current drive is NVMe / the laptop can support NVMe? Some early laptops with m.2 can only do SATA drives.)
|
|
#
?
Jul 24, 2023 20:52
|
|
|
I'm pretty sure I put a P5 in there but not 100%. But it's an old Dell Latitude e7470 that I snagged from the junk pile at work (with permission).
|
|
#
?
Jul 24, 2023 22:43
|
|
|
What’s the buzz on Fanxiang? I see pretty competitively priced but I always forget what I want in TLC/QLC/SLC (this one has a “dynamic SLC cache”). Amazon reviews are good but mostly from people putting them in PS5s. I got a couple for some Raspberry Pi projects on a whim, but I’m looking to upgrade my wife’s desktop’s primary drive (gaming and light photo editing) so now I figure I should get a second opinion.
|
|
#
?
Jul 28, 2023 17:00
|
|
|
Subjunctive posted:What’s the buzz on Fanxiang? I see pretty competitively priced but I always forget what I want in TLC/QLC/SLC (this one has a “dynamic SLC cache”). Amazon reviews are good but mostly from people putting them in PS5s. "Dynamic SLC cache" = the drive can use a flexible amount of NAND as pseudo-SLC, so when the drive is less full you have more fast pSLC space. At first this was a big feature of QLC drives, because avoiding the QLC write penalty was such a big deal. But now lots of consumer drives have it, including TLC ones. It's a pretty meaningless feature to put in the bullet list in 2023, but at least it indicates their controller is something relatively not-garbage. You want TLC over QLC if the prices are anywhere close, but if a QLC drive is much cheaper then it depends on your application. If your use isn't write heavy -- such as gaming and photo editing -- the QLC downsides will really not be noticeable. OTOH if you are specifically looking for a 4TB drive the SN850X is $40 cheaper and not chinesium. Or TBQH for the use-case a Crucial P3 or P3 Plus would be fine, they're QLC but a huge drive means you have huge SLC cache. VVV edit: I have no data / knowledge of fanxiang -- they seem legit enough to be not a terrible idea. If I was looking at a 4TB drive I'm not sure I'd pick them though. It's a big enough investment that I'd be more worried about a known solid warranty. Looking at canada prices, I'd probably do the P3 plus. Klyith fucked around with this message at 18:40 on Jul 28, 2023 |
|
#
?
Jul 28, 2023 17:51
|
|
|
No, that’s a US price and I’m in Canada where the SN850X is $50 more for 4TB (and entirely out of stock on Amazon), but if Fanxiang is an unknown quantity then I’ll pay the premium. Thanks!
|
|
#
?
Jul 28, 2023 17:57
|
|
|
Starting to wonder if my board is cursed or if I am. This is a RMA replacement 970 EVO (was brand new when I first booted it up, only slightly used now) giving the same errors that killed the original drive. It's still under warranty but I am strongly just considering replacing it with something else and then keeping the RMA return as a spare. Got a bad block error log entry and found Media and Data Integrity Errors listed at 1. From experience this is going to get worse and I'm already going to need to use clonezilla set to bypass errors to swap out to the new drive even with just the one error. I am stuck on PCI-E 3x for now. I was considering replacing this one with a WD SN850X assuming I won't lose enough performance to notice anything. Plan to keep it for use on a new board later or otherwise I'd just pick up another 3.0 SSD. future ghost fucked around with this message at 00:47 on Jul 31, 2023 |
|
#
?
Jul 31, 2023 00:43
|
|
|
future ghost posted:Got a bad block error log entry and found Media and Data Integrity Errors listed at 1. From experience this is going to get worse and I'm already going to need to use clonezilla set to bypass errors to swap out to the new drive even with just the one error. Media and Data Integrity errors are not good, but I don't think you can know for sure that a single one is a sign of a dying drive. If the drive is bad that'll keep going up, but it could also hang around at 1 for a really long time. Don't read tea leaves to predict failure, have backups. future ghost posted:I am stuck on PCI-E 3x for now. I was considering replacing this one with a WD SN850X assuming I won't lose enough performance to notice anything. Plan to keep it for use on a new board later or otherwise I'd just pick up another 3.0 SSD. A PCIe gen4 drive like the SN850X will perform just fine in a PCIe gen3 motherboard / slot. Like it'll hit the bus max bandwidth and be limited by that, so it'll only be as fast as the fastest PCIe gen3 drive. But not slower. Possible exception might be some high-performance dramless drives like a SN770, maybe gen4 bandwidth to main memory is important to their great results on 4k-random type stuff that's normally difficult for a dramless drive? I dunno, but wouldn't spend any time worrying about it. 4k random isn't a big deal for desktop use.
|
|
#
?
Jul 31, 2023 13:27
|
|
|
That's the one SMART error I do not trust on NVME disks. The previous 970 EVO also started on just one error too and ended at 1240 when I replaced it. Not going to bother running a full sector scan until after I get this one cloned, but the last one looked like a christmas tree when I sent it in for RMA. Just really hoping it's not an issue with this board, given that this is two in a row with the same problem now. Maybe I'll have better luck with a different drive vendor. Good to know it'll at least run fine even with less than max performance. Edit: Error count increased after cloning it to another disk. future ghost fucked around with this message at 20:00 on Jul 31, 2023 |
|
#
?
Jul 31, 2023 14:24
|
|
|
So anyone going to be at FMS….anyone?????
|
|
#
?
Jul 31, 2023 23:55
|
|
|
WhyteRyce posted:So anyone going to be at FMS….anyone????? I was in SJ during the last one and was bummed I couldn’t make it. I wanna see more EDSFF drives
|
|
#
?
Aug 1, 2023 02:04
|
|
|
priznat posted:
I like my e3.s thicc
|
|
#
?
Aug 1, 2023 02:48
|
|
|
WhyteRyce posted:I like my e3.s thicc Give me a x16 you cowards
|
|
#
?
Aug 1, 2023 03:07
|
|
|
Klyith posted:The tea leaves predict all drives will fail, have backups.
|
|
#
?
Aug 1, 2023 03:34
|
|
|
I am playing catch-up on this thread, but thinking about doing some all flash storage around the 6500 ION (home use) purely as a spinning drive replacement -- I would switch my spinning drives to just be a backup that turns on once a week, syncs and then turns back off. I also have a pair of used PM1735 that I was thinking about RAID 1ing as a basic homelab VM store / mini-SAN -- seems like the most reasonable configuration to me that would remain performance + single fault tolerant. What's the state of the art for doing a JBOD like setup w/ NVMe drives on ZFS or btrfs? I think an EPYC 7443P while having gently caress loads of lanes will be incredibly power hungry... vs getting a 7600, under-volting it and then getting a 6-bay Icy Dock and feeding it with the 24 lanes from the CPU (x4/x4/x4/x4 PEG, x4/x4 NVMe). Ideally -- 4x 6500 ION JBOD / RAID 0 (maybe... since its being backed up anyways?), and then 2x PM1735 in RAID 1 would fit onto the lanes on a 7600. I can put a 10 Gb adapter off the chipset since that has a x4 link to the CPU.
|
|
#
?
Aug 8, 2023 19:12
|
|
|
You’re looking at like $2k plus for a 30TB ION but they are beautiful things The p5316 absolutely needs a massive price slash but that doesn’t look like it’s happened based off a quick google. But I assume enterprise parts vendors just set the price once and never bother to change WhyteRyce fucked around with this message at 19:27 on Aug 8, 2023 |
|
#
?
Aug 8, 2023 19:23
|
|
|
|
| # ? May 14, 2024 02:53 |
|
|
I forget if I posted this already, a friend/ex coworker sent me this https://m.youtube.com/watch?t=254&v=IFwFDZDQWAA&feature=youtu.be Watching him puzzle out how a pcie switch works is amusing to me (I worked on that exact one)
|
|
#
?
Aug 8, 2023 19:36
|
|