Hi, I bought two identical Ultrastar HC590 26TB Sata drives
They have the same firmware, manufacturing date, parameters, formatting, even batch is the same. Both completely empty and connected to the same SATA port during test. But one is 4-5% percent slower in synthetic tests. Everything else is fine. All SMART parameters are also absolutely identical in both drives.
Is slower drive faulty? Or maybe this is an acceptable speed deviation, due to manufacturing variations, etc.? I’ve never compared the speeds of two identical drives before. Perhaps such speed differences are normal.
If it is normal, then everything is fine, I don’t mind such a speed difference, it’s important that the drive works reliably. But maybe enterprise series drives should have absolutely identical speeds? In this case I would assume that slower drive is faulty and I would replace it.
Weird. It appears that HD Tune is detecting the capacity as 2TiB, which is the maximum capacity that can be reported with 32 bits. It’s not seeing the temperature, either.
HD Tune is more like “real life” read test. Also, it averages the results. In this scenario my HDDs look identical.
But all synthetic benchmarks (like CrystalDiskMark, ATTO, etc.) always shows one of the drives is slower by 4-5%. And I want to find out why it is so. It is very clearly visible in ATTO graph.
BTW, HD Tune it is very old software (from 2008?), I’m not sure if we can fully rely on it.
No, my HDDs are not under RAID controller. They are simply connected to SATA port on my motherboard. During tests I connected them to the same SATA port.
Sometimes users mistakenly select RAID mode for the SATA controller in BIOS. This means that the OS cannot directly access the drive, which in turn could explain why HD Tune is unable to see the temperature.
Can you see the SMART attributes (and temperature) with CrystalDiskInfo?
Regarding bios, everything is configured properly.
Yes, I see SMART attributes (including temperature) in all other programs, except HD Tune – it is simply too old program, and it is not even officially designed to work on Windows 10/11, only on W7. I checked „HD Tune PRO“ – it is commercial, but has a 14 days trial period. It shows everything correctly. I usually use Hard Disk Sentinel. It also shows all SMART attributes correctly with real time updates.
Neither program shows any errors or warnings. All SMART attributes are below allowed thresholds and SMART values are exactly the same for both drives. Also, both drives passed full diagnostics with Kitfox software without errors or warning.
So, the question still remains, why is one drive slower? :).
You do a common mistake, by thinking software diagnostics as them to be a calibration standard. I do not see any meaningful differences at the diagnostic reports.
From the other hand, its in the nature of all HDD, the one with longer operating hours, this to spin a bit faster.
Artificial or human intelligence, non of them can win against old-fashion human technical field experience. Most superior this is Greek. Now our topic starter, he simply has to confirm if the one appearing as faster HDD, is also the one with the longer operating hours.
In any case, the OP is concerned with transfer rate. That can be measured with any number of tools. Spin speed is a different thing. I doubt very much that you have measured the latter.
If you want to learn about HDD motor controllers and how they lock onto the target RPM, find a “SMOOTH” datasheet, eg L7250.
@ Kiriakos-GR is talking about spin speed whereas the OP is talking about transfer rate. The spin speed for both drives is identical, but the transfer rate is variable across all the heads.
Given models of disks should be the same rpm but I am so very cynical. HDtune shows how real disk performance is available which is helpful to see if a disk needs to be erased as the sectors need to be refreshed etc.
CrystalDiskInfo shows some alleged 7200rpm disk as 4800rpm etc. Not that that tool is all that good.
CDI is perfectly good at what it does. In fact, it is much better than HD Tune when displaying SMART attributes. HD Tune’s author still hasn’t worked out than the raw values are best viewed in hexadecimal, and in many cases they are multibyte or multiword values. In some cases HD Tune displays absurd negative decimal values.
In any case, all SMART tools access the same information. They only differ in the way that they interpret it.
CDI, and HD Tune, and every other tool, should access the drive’s parameters via the ATA Identify Device command. This returns 512 bytes of information, including the drive’s reported RPM. Some of WD’s models lie about their real RPM, and some report an RPM of 0 to obfuscate the fact that they are slow (Intellipower). Some tools cannot see the HDD behind a USB-SATA/PATA bridge, and few can see a HDD behind a RAID controller or inside a NAS. Who knows what speed they will report in such cases.
A tool such as Victoria will try to exercise a drive in a way that exposes its real RPM, but it doesn’t always succeed (eg SMR models). Alternatively, one can estimate the RPM via HD Tune’s access time graph. For example, the access time data points are scattered in a yellow band. The width of this band corresponds to the time for one complete revolution of the platters. Therefore, a 7200 RPM drive will have an access time band that is 8.33msec wide. For a 5400 RPM drive, the width is 11msec. Some of WD’s “5400 RPM class” models actually spin at 7200 RPM, but their performance has been crippled in the firmware.
I have Victoria along with HDDscan which I use along with CDI and HDtune. Sometimes I wonder if I should open up visual studio and see what I can crank out.
More recently I found a handy tool called DiskGenius which can migrate Windows to a new storage device. This tool is free unlike many others. https://www.diskgenius.com/
The free version is able to copy windows from SSD to SSD, hard disk to SSd etc free. The paid version is more aimed at VHD users etc. The private dude who wants a bigger system SSD is free and the software seems to be NTFS and EXT3, EXT4 aware so dual boot is not a problem either. I use Linux in a VM which is a safe standard sandbox.
