[ETA]MrSpadge

Member
  • Content Count

    799
  • Joined

  • Last visited

  • Days Won

    7

Everything posted by [ETA]MrSpadge

  1. "New SMR" drives makes it sound like SMR would replace PMR. I'd rather expect that PMR will at some point be replaced by heat assisted recording, and SMR by heat assisted SMR. The performance hit of SMR is so severe that it can't replace regular drives anytime soon. They'll coexist until HDDs are only needed for archives and something else replaces the regular high capacity drives. MrS
  2. [ETA]MrSpadge

    Crucial BX100 SSD Review Discussion

    I agree, kalmquist. I'd be a lot more interested in a comparison of 120 / 250 GB drives with the "usual budget suspects" and newcomers: BX100, MX100, MX200, 850 EVO and ARC100. That's where the different write performance of the drives is most likely to result in real-world performance differences. MrS
  3. [ETA]MrSpadge

    best portable 2+ TB USB 3.0 drive for off site backup

    A 2-drive enclosure could also get you more speed with Raid 0. This obviously increases the failure risk, but that won't matter much if it's only a backup anyway. Apart from that, a 7.2k rpm 3.5" drive in an external enclosure is pretty much the only half-practical thing which could give you more speed. It's not as portable as a 2.5" drive, but if you leave its power supply at home it's not that cumbersome either. MrS
  4. The Z97 allows SRT. If you want to use it, the SATA controller must be set to RAID in the BIOS/UEFI. If you have already installed Win, this will make it unable to boot (yes, even 8.1). By changing a registry value prior to switching to RAID you can get around this problem. It's pretty much the same as changing an installed Win between AHCI and IDE mode, which triggers the same boot failure. Google finds instructions for this Afterwards you must have some unpartitioned space on the SSD to active SRT in the "Intel Rapid Storage Technology". I have not yet tried this, as I'm always using the full 60 GB of my SSD as cache, but I remember seeing the option to leave some space on the SSD unallocated. There is a risk that the software insists on wiping the SSD of all partitions when you set up SRT without using the full space for it. Just be careful here, I really don't know if this works or not. Well, you could try it with some of your current HDDs before deciding on how fast your new one needs to be. MrS
  5. Sounds good, Chris. For large games you definitely want a 7.2k rpm HDD, but you don't need to spend big in e.g. a WD Black or Seagate Enterprise, because the performance difference between "fast 7.2k rpm " and "normal 7.2k rpm" are not that big. Over here I see the "HGST Deskstar IDK 4TB" for 153€ and the "Toshiba MD04ACA 4TB" for 118€. All others are more expensive, e.g. the WD Black is at 200€. The price difference would be enough to make me choose the Toshiba. BTW: which mainboard do you have? If gaming performance really matters to you, you could use a part (20 - 60 GB) of the 256 GB SSD as cache for the large HDD, if you can use Intel SRT. I'm using a full 60 GB SSD as cache for my 3 TB Seagate 7.2k rpm and am quite pleased with the result. It's obviously not full SSD performance, but way better than just the HDD. MrS
  6. The Ultrastar 7K6000 should be very expensive once it's available. It's the Desktar you're looking for. And if you're going for the 7K4000 make sure to get the 4 TB one, the 4 GB version is a scam Do you also have an SSD? For desktop use I'd get a 128 or 256 GB Crucial MX100 together with a large HDD. the Seagate Hybrid with SSD cache would also be a nice option, but the 4 TB model uses 5.9k rpm, which defeats the purpose of making a fast system drive. And don't read too much into the blackblaze study. It's valid for what it is, but they state themselves that they're operating the drives in an unusual environment with very high vibration from neighbouring drives. Neither are the drives meant to be used like this (they are cheap desktop models without vibration sensors and compensation - the manufacturers know this is an issue when running many of them and will happily sell you the more expensive unit with the appropriate hardware) nor is this environment representative of usage in a real PC. Every year there's a return rate report from a big french hardware shop. Their results are very different, there's pretty much nothing special to see. This probably means the Hitachis are more robust in some way.. but does it matter? Like if you state a CPU has an expected lifetime of 100 years in environment A, but only 50 years in the hotter room B. Sure, B is worse.. but does it matter? MrS
  7. As far as I understand NVMe mainly reduces latency, which matters the most under high queue depths. Since regular desktop work doesn't see such loads, there won't be a significant difference. It's certainly nice to have, though, and one could argue that even such small CPU time savings add up world-wide to save power. MrS
  8. [ETA]MrSpadge

    Where did my HDD capacity go?

    Windows Explorer may not count files that it can not access, since it's running under your account with your access rights. We're talking about 187 GB, though, which would be a lot of system files. Did you already perform a disk cleanup, including system files? MrS
  9. ... and this is not even using 3D NAND, so they've got another ace up their sleeve! MrS
  10. [ETA]MrSpadge

    Toshiba Announces 3TB 2.5-Inch HDD Discussion

    Agreed, impressive progress! With 4 platters it's probably 12.5 or 15 mm high and doesn't fit standard 2.5" bays, but I won't complain about someone pushing the edge of storage further MrS
  11. [ETA]MrSpadge

    Crucial Mx100 or M550 which to go for

    I say go for the 256 GB MX100. The extra space will come in handy at some point; it always does. Sure, there are even faster SSDs - but I don't think you're rebooting your server often enough for a few seconds to be worth the added cost. Better save the money now and upgrade to a PCIe (M.2) based drive later on, when they're readily available. BTW: if you're really concerned with boot time you should love Win 8.1, with regular boot and/or "suspend to disk". On my laptop the latter is about 4 times as fast with a typical RAM usage of 1 of 4 GB. The price difference between for a premium SSD compared to the MX100 can be as large as the price of a full Win licence... MrS
  12. [ETA]MrSpadge

    Samsung SSD 850 EVO SSD Review Discussion

    Considering the MX100 is quite a bit cheaper than either EVO I'd still recommend the MX100 for regular dekstop usage. The reason is that while benchmarks show a performance difference, most real world usage scenarios don't hammer the SSD like that. And if they do, the MX100 is still fast. For heavier I/O load feel free to use faster drives.. but then you might want to skip the EVO and go straight for the top performers. MrS
  13. ... nor for anyone else. Actually it's amazing that the Samsung is able to pack entire 2 TB into such a small package at all! Regarding SATA 2/3: it's backwards compatible and for slow HDDs like this one even SATA 1 is not a bottleneck. MrS
  14. Check all internal cable connections to the mainboard since the machine was moved. I'd also try to test them on a different PC to eleminate all possible issues with the controller etc. If it can not even serve an mp3 the drive seems to be in really bad trouble and may fail soon. Make sure you have decent backups. MrS
  15. I wouldn't worry about a single write error. But always backup your important data, no mater on which medium they're stored. MrS
  16. [ETA]MrSpadge

    Weird Seagate 1TB problem...

    Do you still have a different PCB attached? That might cause strange behaviour if something's not right. MrS
  17. That drive tray seems fine. It takes 2.5" HDDs, which currently do not provide any more than 2 TB of disk space. It also can't take drives with 15 mm height, so Toshiba, WD and Seagate Enterprise are out of the question. This leaves the Samsung Spinpoint M9T as your only choice. It's only 9.5 mm high and runs at 5.4k rpm. It doesn't matter which drive you put into which place, as long as you handle your GRUB well. Your "old" Toshiba drive with 7.2k rpm may be faster as system drive than the M9T, but both are slow regular HDDs anyway. MrS
  18. The 4 TB uses 5 x 800 GB platters, hence the noise (in combination with 7.2k rpm). This also means they can not scale straight to 5 and 6 TB. Generally WD still seems to avoid 1+ TB platters at 7.2k rpm where ever possible. MrS
  19. [ETA]MrSpadge

    Opinion on the WD 6TB Green Drive?

    You could use the drives separately and spread the files manually when you write them, but use the Win 7 libraries to lump them all together when accessing them. MrS
  20. [ETA]MrSpadge

    New internal drives for data storage

    Yes, the reasons I mentioned. Thinner = better compatibility 3 platters instead of 4 = lower noise & power consumption, higher sequential transfer speed I don't know what's inside the STDR1000202, but I think it's unlikely to be a fundamentally different drive. Maybe it's just different case design. MrS
  21. And without port multiplier: it still doesn't matter. because each drive gets its individual channel. MrS
  22. [ETA]MrSpadge

    New internal drives for data storage

    USB powered means you're limited to 2.5" drives. As far as I know there's only the Seagate / Samsung Spinpoint M9T offering 2 TB (with 3x667 platters) in a single package of standard 9.5 mm height. Others my offer enclosures with this same drive. Edit: Toshiba and WD also have 2 TB drives by now, but both are non-standard 15 mm high. I would say the Spinpoint is better because it could also be used within a laptop (as 2nd drive), should you wish to do so later on. MrS
  23. [ETA]MrSpadge

    New internal drives for data storage

    What I think Brian means is that your sample size will be small. So even if there are small differences in drive reliability, it will be down to bad luck if you get a failure. You could argue that the failure rates of the Seagate drives reported there are "significantly higher" rather than "slightly higher". however, they're testing in an extreme environment with many discs producing vibrations. As far as I know that's the wekaest spot of that study. It mainly shows how well the drives fare under high-vibration environments, which does not have to be linked to how well they perform elsewhere. MrS
  24. Disclaimer: I don't know how what I'm going to suggest works with Android or Linux, but under Win it would solve all your problems (as far as I know). It's actually relatively simple and similar to the "Dropbox with Viivo" approach you've already looked at. I'm bringing it up because it works very well for me and might be better than those. I'm using "Cloudfogger" to encrypt my files before they go into the cloud. It works with different platforms, is bound to an account (the key won't get lost unless you loose the login), doesn't interfere with local copies (if the cloud allows them) and is transparent on the local machine. A drawback: logging in online you only see encrypted files (just as the provider and NSA do). I'm using it mostly together with OneDrive on my PC and laptop and am happy that I don't have to bother with "exotic" OS's. MrS
  25. Currently there is no pure HDD significantly faster than your WD Black. Slightly higher benchmarks are possible, but I don't think it would be noticeable. If you're looking for space and performance I'd first suggest caching for the HDD, e.g. via Intel SRT. Regarding the SSD: I expect some tempting PCIe M.2 drives to arrive sometimes in the next few months. Not sure how pressing your upgrade is. What I noticed during the recent years is that once you plug an SSD into a system, even just a 120 GB Samsung 840 on SATA2, modest CPUs go to full load on 1-2 cores as soon as anything disk-intensive is started. I'm not entirely sure if the CPU is waiting for the disk, or if it's otherwise. Win 8.1 shows just a few 10% disk utilization, whereas with HDDs it always goes straight to 100%. So I think single threaded CPU performance, and to a lesser extent multi threaded performance, is starting to become a significant factor for I/O performance. MrS