|
DrDork posted:If £20 is important to you, and you were living with a Q6600 for this long, the performance difference probably isn't something to worry about. I'm happy to pay the extra £20 as long as there's an actual, noticeable benefit, but if an SN550 or whatever is so fast already that i won't be able to notice the difference then i'm happy with the cheaper option.
|
#
?
Jan 12, 2020 00:41
|
|
|
|
| # ? Apr 23, 2024 23:16 |
|
|
Lungboy posted:I'm happy to pay the extra £20 as long as there's an actual, noticeable benefit, but if an SN550 or whatever is so fast already that i won't be able to notice the difference then i'm happy with the cheaper option. The SN550 looks like a solid budget NVMe drive - it doesn't quite have the performance of Phison e12 drives, but it's awfully close. It doesn't have a DRAM cache, but that doesn't hurt it too much and it's TLC so write speeds are still acceptable when it goes through it's 13GB SLC cache (only really applicable for very large writes, like drive cloning). Endurance and warranty are good. You're unlikely to notice a difference between it and faster NVMe drives, and honestly, for most applications you would't even notice a difference between an NVMe and a SATA SSD. The cheapest SATA drive I'd recommend would only save you £5, though, so I'd probably stick with the SN550.
|
|
#
?
Jan 12, 2020 01:05
|
|
|
What budget/used NVME should I get for a Z97 system?
|
|
#
?
Jan 12, 2020 02:52
|
|
|
Shaocaholica posted:What budget/used NVME should I get for a Z97 system? How big and what budget
|
|
#
?
Jan 12, 2020 04:40
|
|
|
FRINGE posted:How big and what budget 512gb ~$130. I just don't want to have something faster than what a Z97 can handle.
|
|
#
?
Jan 12, 2020 05:17
|
|
|
Shaocaholica posted:512gb ~$130. I just don't want to have something faster than what a Z97 can handle. Don't get something slower just because the Z97 will keep you to two PCIe 3 lanes. Eventually you'll want to drop it into a new board. That being said, don't get a PCIe 4.0 drive unless you're planning with relative certainty to upgrade to Ryzen 2 or 3 CPU in the short to mid-term future. Hell, even then it's not worth it.
|
|
#
?
Jan 12, 2020 05:58
|
|
|
Shaocaholica posted:512gb ~$130. I just don't want to have something faster than what a Z97 can handle. No reason for budget or used at with 130 as the target. A new expensive-end one (samsung 970) is in that budget: https://www.amazon.com/dp/B07M7Q21N7/ Or more reasonably: https://www.amazon.com/Sabrent-SB-ROCKET-512-Rocket-Internal-Performance/dp/B07KGMBCKD/
|
|
#
?
Jan 12, 2020 07:22
|
|
|
So I bought a Crucial MX500 as my first SSD, and I'm happy so far. I searched for these phrases first, but didn't find an applicable answer, should I enable either or both of these in their Storage Executive app:
|
|
#
?
Jan 12, 2020 09:21
|
|
|
Every SSD from major brands does at least 10% overprovisioning when it comes out of the box. Theoretically there is lesser write performance degradation when the drive is close to full if you manually reserve more usable capacity for overprovisioning, but such will never matter for 99.99% of the consumers in the real world so leave it alone.
Palladium fucked around with this message at 09:44 on Jan 12, 2020 |
|
#
?
Jan 12, 2020 09:41
|
|
|
Yeah, overprovisioning was a thing years ago when 128GB was a "reasonable sized drive" and drives would ship with little or no OP to save on production costs. Now that sizes are so large and adding a few extra GB for OP, it's generally not worth adjusting from defaults.
|
|
#
?
Jan 12, 2020 15:34
|
|
|
doctorfrog posted:So I bought a Crucial MX500 as my first SSD, and I'm happy so far. I searched for these phrases first, but didn't find an applicable answer, should I enable either or both of these in their Storage Executive app: Momentum Cache is a program that caches writes in main memory, which can make IO faster but also means more danger of unfinished important writes if you get a system crash or power loss. As with Samsung Rapid, not worth it. Over provisioning isn't needed, but keep in mind that flash drives don't like to get lots of writes when they're full. If you notice that your drive is getting full do a windows drive cleanup or uninstall a game. Palladium posted:Every SSD from major brands does at least 10% overprovisioning when it comes out of the box. I don't think that's true. The MX500 for example has 512 total GB of nand, and 500 capacity. 12 GB extra space is plenty for most purposes, but it isn't 10%. The thing about overprovisioning is that the percentage isn't a meaningful measurement. If you have a full drive and write 25GB of data, you probably want 25 to 50GB overprovisioned to avoid write amplification. Doesn't matter what the size of the drive is. If you only write less than 1GB at a time* you can get away with a lot less. So for average people the spare space that they round down is plenty. *ignoring degenerate cases like a drive that's entirely full of 6-7KB files and wants to delete and then write 1GB of more 6-7KB files
|
|
#
?
Jan 12, 2020 15:42
|
|
|
Klyith posted:I don't think that's true. The MX500 for example has 512 total GB of nand, and 500 capacity. 12 GB extra space is plenty for most purposes, but it isn't 10%. Don't forget that space reported by the computer is in gibibytes while size on the box is in gigabytes. So that 500 GB will only be ~488 GB available space. That makes ~24 GB of overprovisioned space, still just around 5%, but double what you were thinking.
|
|
#
?
Jan 12, 2020 15:59
|
|
|
Geemer posted:Don't forget that space reported by the computer is in gibibytes while size on the box is in gigabytes. So that 500 GB will only be ~488 GB available space. That makes ~24 GB of overprovisioned space, still just around 5%, but double what you were thinking. No, the flash packages are also made in multiples-of-1000 gigabits.
|
|
#
?
Jan 12, 2020 16:15
|
|
|
Klyith posted:No, the flash packages are also made in multiples-of-1000 gigabits. Oh huh. Never mind me, then.
|
|
#
?
Jan 12, 2020 18:26
|
|
|
.
some dillweed fucked around with this message at 06:37 on Dec 22, 2023 |
|
#
?
Jan 13, 2020 03:23
|
|
|
I need some help with an 860 EVO SATA 1TB SSD I just got, I hope this is the right place to post. I tried cloning an old Intel 120gb SATA SSD to this new drive using Samsung's Data Migration tool, but after disconnecting the old drive, I was unable to boot to the new one (disk read error in BIOS). I tinkered around in the BIOS but couldn't figure out anything so I reinstalled the old drive and tried cloning again. Now I still can't boot to the new drive and when booted to the old drive, the new one only shows one partition with a 31mb (not gb, mb) partition and nothing else. I can't try cloning to it again and I don't know how to even get it back to square one. Did I brick it somehow?
|
|
#
?
Jan 13, 2020 03:58
|
|
|
Grog posted:Just to be a pedantic, correcting rear end in a top hat, the MX500 uses 256 Gb dies, and seems to use two of those per package (512 Gb), not multiples of 1000. From the Anandtech review of the 1 TB MX500: Nope, because each of those 256 Gbit dies has 256,000,000,000 bits on it, not 274,877,906,944. That's what I meant by multiples of 1000. So if you were to mod the firmware of that MX500 1TB to make the entire flash capacity usable and have zero overprovisioning, you would have: 256 * 2 dies/package * 16 packages * 1000^3 = 8192000000000 bits /8 = 1024000000000 bytes which in Windows would then be: 1024*10^9 / 1024^3 = 953.67 GiB For more definitive evidence, in a Samsung press release about 64L 3D VNAND of pretty much the same process type, they say "85.3 billion cells", which when multiplied by 3 for TLC is approximately 256 gigabits, not gibibits. Storage manufacturers have been using Gb forever to mean decimal numbers of bits.
|
|
#
?
Jan 13, 2020 04:47
|
|
|
willroc7 posted:I need some help with an 860 EVO SATA 1TB SSD I just got, I hope this is the right place to post. I tried cloning an old Intel 120gb SATA SSD to this new drive using Samsung's Data Migration tool, but after disconnecting the old drive, I was unable to boot to the new one (disk read error in BIOS). I tinkered around in the BIOS but couldn't figure out anything so I reinstalled the old drive and tried cloning again. Now I still can't boot to the new drive and when booted to the old drive, the new one only shows one partition with a 31mb (not gb, mb) partition and nothing else. I can't try cloning to it again and I don't know how to even get it back to square one. Did I brick it somehow? First off wipe the new drive with diskpart (careful to select the right drive). Then try to format the whole drive with a regular ntfs partition and copy a few gigs of files to it to make sure it's writable correctly & you don't have a DOA. After that, try macrium reflect instead of the samsung tool.
|
|
#
?
Jan 13, 2020 04:56
|
|
|
Klyith posted:First off wipe the new drive with diskpart (careful to select the right drive). Then try to format the whole drive with a regular ntfs partition and copy a few gigs of files to it to make sure it's writable correctly & you don't have a DOA. I tried this and diskpart only shows the 31mb volume so that's call I can clean/format/etc.
|
|
#
?
Jan 13, 2020 05:17
|
|
|
willroc7 posted:I tried this and diskpart only shows the 31mb volume so that's call I can clean/format/etc. Can you screenshot? Does it looks like this? code:
|
|
#
?
Jan 13, 2020 05:23
|
|
|
Yes but before I tried to clone it, it showed up correctly as 1TB.
|
|
#
?
Jan 13, 2020 05:34
|
|
|
Go download something like Paragon Partition Manager and load that up and see what it sees on your drive. If it only sees that 31MB partition, yeah, something's dicked up with your drive. More likely, the clone software made that 31MB partition the primary partition for some reason and you should see it and one or more other partitions. Delete them all, make one single large NTFS partition, and use Macrium Reflect or some other non-Samsung program to do the transfer.
|
|
#
?
Jan 13, 2020 06:29
|
|
|
Apparently Sabrent has been quietly switching the Phisons for Kingston without letting anyone know.
|
|
#
?
Jan 13, 2020 06:33
|
|
|
.
some dillweed fucked around with this message at 06:36 on Dec 22, 2023 |
|
#
?
Jan 13, 2020 09:11
|
|
|
DrDork posted:Go download something like Paragon Partition Manager and load that up and see what it sees on your drive. If it only sees that 31MB partition, yeah, something's dicked up with your drive. More likely, the clone software made that 31MB partition the primary partition for some reason and you should see it and one or more other partitions. Delete them all, make one single large NTFS partition, and use Macrium Reflect or some other non-Samsung program to do the transfer. So I found other people have had this problem. They used "Atola Technology HDD Restore Capacity Tool" to restore the capacity. Unfortunately this software was shuttered in 2017 and the versions I can find to download do not work on this windows 10 machine. Apparently this has something to do with a BIOS bug in the older Gigabyte board and the host protected area on the drive? Does anyone know if an existing tool I can use that will help me fix this? Thanks guys... Sources: https://superuser.com/questions/58753/windows-7-detects-my-1-tb-disk-as-a-disk-with-only-31-mb-space https://community.wd.com/t/wd10-eads-problem-from-1tb-to-31mb/14105
|
|
#
?
Jan 13, 2020 15:06
|
|
|
That's a weird issue to be having these days. I'd still give Paragon Partition Manager a shot and see what it sees on the drive. PassMark also has a tool that claims to be able to adjust the HPA on a drive: https://www.passmark.com/products/diskcheckup/
|
|
#
?
Jan 13, 2020 15:47
|
|
|
DrDork posted:That's a weird issue to be having these days. I'd still give Paragon Partition Manager a shot and see what it sees on the drive. I'll try passmark but paragon and every other partition manager I've tried reports just the 31mb partition and nothing else.
|
|
#
?
Jan 13, 2020 16:00
|
|
|
willroc7 posted:I'll try passmark but paragon and every other partition manager I've tried reports just the 31mb partition and nothing else. Reporting the 31mb partition isn't meaningful, you need to completely re-initialize the drive. Just deleting the partition doesn't work because what's messed up is lower than that. Can you carefully follow the diskpart instructions I posted before, the clean operation should completely reset the drive including re-initializing the area that's been corrupted. You're not using it to erase and format the partition. After that, try samsung magician which has a factory reset operation. Grog posted:Okay, I don't even know what is actually right anymore, but are you sure that's not just Samsung marketing giving one number instead of another to prevent people from expecting the full binary/gibibyte amount of cells available? As in, 256 Gbit would still be 32 GB in the memory sense, or 32 GiB (32*1024^3 bytes). I don't work for a drive manufacturer so I got no real evidence.  I don't think that samsung would lie on press releases, but they do say "over" 85 billion so it could be 91 billion cells. But as a more general thing I wouldn't expect any manufacturer to be more generous than they needed to be. >10% extra capacity seems like more than the drive needs for reliability and replacing worn out cells?Grog posted:e: Same deal from articles on Anandtech and elsewhere, and easy to find things like the description of NAND capacity on Wikipedia. They all say NAND works in strict binary multiples like you would expect from any form of memory, and the rest is marketing/advertising and conversion from the advertised decimal amounts to binary in the operating system. NAND definitely doesn't need strict power of 2 capacity, 86Gbit and 384Gb dies have existed. Not evidence for whether those examples are actually giga- or gibi- size numbers though. But think about it this way: if NAND needed to be a binary multiple size it would be for addressing reasons. (That's why ram is strict power of two.) But that addressing is to the actual cells, and each TLC cell has 3 bits. So the final capacity should be a power of 2 times 3, and we'd see TLC dies with 192 Gbit and such. OTOH with all these drives having SLC cache, they'd need a fair amount of extra capacity for that to fit! So maybe you're right about the die size being 1024^3 Gbits, it seems like there wouldn't be enough room for SLC without that. But that also means the overprovision is less than the strict math you did earlier, by 2x the SLC cache size (or minimum size in the case of drives like the MX500 that are variable). Fascinating! I don't know what's going on in there! But for anyone else following this, I think we both agree that however much extra space is in your SSD, it's enough that you don't need to manually overprovision.
|
|
#
?
Jan 13, 2020 17:26
|
|
|
Thanks for the thorough answers on my SSD questions, thread!
|
|
#
?
Jan 13, 2020 17:47
|
|
|
I’ve seen at least one (old) article suggesting that the dynamic slc cache covers the overprovisioned area as well as currently unused space. If that were true, it seems possible that the minimum cache size would shrink over time as the overprovision shrinks, rather than allocating some fixed cache space that’s unavailable as an overprovision? Seems like the sort of thing that someone might have tested on old drives with dynamic caches, but my brief googling hasn’t turned anything up.
|
|
#
?
Jan 13, 2020 17:57
|
|
|
DrDork posted:That's a weird issue to be having these days. I'd still give Paragon Partition Manager a shot and see what it sees on the drive. So I ran PassMark and it did show like 930gb reserved. It let me adjust it and now the reserved space says 1mb instead, but still the drive only shows up at 31mb physical. I tried clean in diskpart again and still no joy. Klyith posted:Reporting the 31mb partition isn't meaningful, you need to completely re-initialize the drive. Just deleting the partition doesn't work because what's messed up is lower than that. Can you carefully follow the diskpart instructions I posted before, the clean operation should completely reset the drive including re-initializing the area that's been corrupted. You're not using it to erase and format the partition. The factory reset in samsung magician doesn't appear to work. When I try to select the erase data step it says this drive is not supported. It does still see the disk as an 860 EVO 1tb.
|
|
#
?
Jan 13, 2020 19:15
|
|
|
willroc7 posted:So I ran PassMark and it did show like 930gb reserved. It let me adjust it and now the reserved space says 1mb instead, but still the drive only shows up at 31mb physical. I tried clean in diskpart again and still no joy. So this is on the GA-Z77x-UD3H mobo you posted about like 2 years ago? Are you on the most up to date F20e bios? And did you do a nvme BIOS mod like you were talking about earlier in the thread? Because I would revert any weird stuff like that. At this point I can only imagine that you can't reset because it wasn't a one-time error during the clone, but an ongoing problem. Attempts to clean / reset the drive are just producing the same results. Also, that mobo I guess has both intel and marvell sata ports so be sure you're using the ones controlled by the intel chipset. Does the BIOS seem to recognize / show capacity of the drive? doctorfrog posted:Thanks for turning a simple question into a nitpicky and ultimately fruitless debate!  
|
|
#
?
Jan 13, 2020 20:07
|
|
|
Klyith posted:So this is on the GA-Z77x-UD3H mobo you posted about like 2 years ago? Are you on the most up to date F20e bios? And did you do a nvme BIOS mod like you were talking about earlier in the thread? Because I would revert any weird stuff like that. At this point I can only imagine that you can't reset because it wasn't a one-time error during the clone, but an ongoing problem. Attempts to clean / reset the drive are just producing the same results. Actually this is an even older board, a GA-965P-DS3. And I'm sure the BIOS are not up to date.
|
|
#
?
Jan 13, 2020 20:24
|
|
|
|
|
#
?
Jan 13, 2020 20:45
|
|
|
willroc7 posted:Actually this is an even older board, a GA-965P-DS3. And I'm sure the BIOS are not up to date. good god wtf um, looks like on that you want to set the ICH8R southbridge into RAID mode -- that will enable AHCI and the only options are legacy IDE and RAID. AHCI may give you better results talking to the SSD. Or maybe try the purple ports which hook to the add-in controller? if that succeeds in reading the SSD it still probably won't boot and you'll have to reinstall windows. but honestly with that drive being hosed up and you not having a PC modern enough to unfuck it, sending it back is looking like a decent option. I don't think you've permanently bricked the drive but I think the fix is to reset it in a PC that's not ancient. why are you putting a 1TB ssd on a 15 year old PC?
|
|
#
?
Jan 13, 2020 23:10
|
|
|
Haha. I had a GA-965P-DS3 with a Core2Duo E6600 in it. Looking at a picture of the board and the box just brought memories back. Memories that seem in the fairly distant past: can't believe you're still using one. Does it run a modern OS alright?
|
|
#
?
Jan 13, 2020 23:48
|
|
|
I'm still running an MSI P965 Platinum with Windows 10 on it and it's fine given how old it is.
|
|
#
?
Jan 14, 2020 00:31
|
|
|
I guess we're in an era of many spare clock cycles if you just want a PC to browse and play media with.
|
|
#
?
Jan 14, 2020 01:10
|
|
|
Klyith posted:good god wtf It's my mom's hand me down system she got from me. It runs fine and has windows 10, but it has a 120gb Intel SSD and she's just out of room. I thought it would be easy to clone it to a new drive but, alas, I was very wrong. I have access to several other more modern PC's if that'll help me. I connected it briefly to my coffee lake desktop but it wouldn't even recognize it as a samsung SSD. It just said unrecognized device under device manager so I gave up on that.
|
|
#
?
Jan 14, 2020 02:29
|
|
|
|
| # ? Apr 23, 2024 23:16 |
|
|
Klyith posted:I don't work for a drive manufacturer so I got no real evidence. It’s not lying, more like glossing over a shitload of technical details most people won’t understand and giving them the number they will. This is an example, that‘s definitely a load-bearing “over”. NAND flash pages always have extra storage capacity that’s not even overprovision, just more bits for storing error correcting codes. Everyday ECC DRAM is single error correct, double error detect, and needs 1.125x storage to do that (72/64). The BCH (or similar) ECC codes used in SSDs have have to detect and correct more than 2 bit errors. They should gain some efficiency from the larger block size but I wouldn’t be surprised if the overhead ends up being similar. Fake edit, SUPER NERD JUNK BELOW: Looked up an ancient Micron planar MLC datasheet as an example (they want you to sign your soul away to see data for modern parts), here’s the details: Page size x8: 4320 bytes (4096 + 224 bytes) Block size: 256 pages (1024K + 56K bytes) Plane size: 2 planes x 1024 blocks/plane Device size: 16Gib: 2048 blocks, 32Gib: 4096 blocks (Page is the minimum unit of write. Block is the minimum unit of erase. Planes are logical flash chips which cohabit the same die or package, and exist to provide parallelism, eg one plane can perform a block erase while another does a page read.) The 224 bytes extra per page are what’s intended to be used for error correction. If you multiply things out, that means our nominally 16Gib part actually stores 16.875 Gib (binary gigabits) or ~18.12 Gb in decimal. To return to the earlier topic, I think that before TLC it was standard to have powers-of-2 counts of blocks per plane and pages per block, for the same reasons why DRAM chips have power of 2 X and Y dimensions (treating the data bus width of the DRAM chip as a third Z dimension). I wouldn’t be terribly surprised if plane internals other than the page size are still powers of 2 today, it may just be the total number of planes per die that’s not a power of 2.
|
|
#
?
Jan 14, 2020 03:42
|
|
