56

I have a laptop. When there is a Windows 10 OS on it, even if there is no data read or written, the USB flash drive will get extremely hot when just plugged into the computer.

However, with the same computer, with Linux installed on it, the situation is totally different. When I write and read data from and to the flash drive intensively, this flash drive just got a little bit warmer.

After I found this phenomenon, I tested various USB flash drives on my desktop, laptop ... with the same Linux and Windows OSes. The situation is similar.

In the Linux environment, I use

dd if=/dev/urandom of=/path/to/my/flash/parent/folder/test.bin bs=1024M count=4

to test a big file write. To test writing a great number of small files I use:

parallel dd if=/dev/urandom of=/path/to/my/flash/parent/test-{}.bin bs=1M count=1 ::: {0001..4096}

In the Windows 10 environment, I just plug the USB flash drive in and do nothing(Not run I/O intensive application on purpose) but wait for 10 minutes. NOTE: The Windows 10 environments on all test computers were freshly installed, and there was no 3rd-party application (I use Education Edition and I disable real-time protection when I start the machine before test. However, I just unset the checkbox in Windows Graphical Setting Manager. I don't know if there need a reboot to make change valid).

Update 1:

It may be due to the USB flash drive itself. Here is the model of flash drive I use in this test.(This is not the advertisement, I have no attitude toward these products)

  1. SanDisk Extreme PRO® USB 3.1 Solid State Flash Drive 128GB

  2. SanDisk 64GB USB 3.0 Flash Drive

  3. DataTraveler 100 G3 16GB

Update2:

There is an answer which consider the effect of the buffer of Linux system.

But for the Linux distro I used, when DD was finished, I typed sync and it took almost zero time to execute. Hence the effect of that is not considerable.

Why? Is this because Windows always does some useless accesses to a flash drive?

In addition when I have to use Windows, is there any practical way to make the USB flash drive cooler?

Improve this question
16
  • afaik, usb port gives 5v out. It would be really cool if we could practically measure the heat of the usb stick in real time.
    – Nick
    Jul 12, 2018 at 12:52
  • 2
    I don't believe this to be a general issue with Windows 10. I have used USB drives on my Desktop, two laptops, and my Wife's desktop; all running Windows 10 (Pro or Home, depending on device). My USB drives do not heat up to any appreciable degree. Is it possible the issue is specifically with Windows 10 Education edition? Or alternatively, that waste heat from the computer is blowing onto the USB stick?
    – Doc
    Jul 12, 2018 at 14:43
  • I've had the reverse happen: the drive got really rather hot when writing a live installation image under Ubuntu, and never worked again (front ports on a desktop, so no thermal conduction issue)
    – Chris H
    Jul 12, 2018 at 14:47
  • May the Windows 10 system be using ReadyBoost?
    – mgarciaisaia
    Jul 12, 2018 at 17:11
  • 1
    You could test the chassis warming the device by using a short extension cable to separate the device from the chassis. I would be quite curious to hear the results.
    – Traveler
    Feb 3, 2019 at 17:40

5 Answers 5

Reset to default
43

On my laptop, the USB 3.0 port is physically very close (~ 2 in) to the fan assembly:

enter image description here

As a result, any USB device plugged in on that side feels approximately the same temperature as the outside of the fan (!).

My laptop is also a little bit slow. Consequently, Windows 10 runs considerably hotter at idle than Ubuntu, and I have noticed that flash drives seem to get hot on Windows as well.

You can test if this is the case for you as well by running a very CPU-intensive program in Ubuntu (say, a 4-core build) and observing whether a plugged-in flash drive gets hot during that time.

Edit: I just now saw Justin's comment. I hope this provides enough information to stand on its own.

Improve this answer
8
  • 8
    Regarding the edit at the end: comments can be purged, answers aren't (except for violating rules), so it is good that you wrote this answer. Moreover, answers in comments are against the rules, so that comment is even more likely to be deleted.
    – user922538
    Jul 12, 2018 at 16:50
  • 1
    That is not possible, for my laptop the fans are in the front near the screen. And for desktop, fans and the port I use are in opposite side.
    – pah8J
    Jul 12, 2018 at 22:45
  • @pah8J Oh, that's really interesting! I see that you have now tested this on a desktop as well. Now I think the reason I have never experienced the same issue is because the usb port got hot for a different reason. :)
    – Nonny Moose
    Jul 13, 2018 at 14:25
  • 6
    This guy gets it. +1. The temperature of the chassis (ground) seeks to equalize itself thermally into everything connected to it, even weak little plastic-encapsulated USB drives.
    – user2497
    Jul 13, 2018 at 20:44
  • @user2497, could you explain it more detailed? Or present this idea as an Answer?
    – pah8J
    Jul 14, 2018 at 6:35
8

Windows does have this thing where you "can use USB sticks to speed up your computer". It does this by using otherwise-unused-stick-memory as a cache, reducing disk I/O. It's possible this extra activity is generating the excess heat.

It's possible to test this by turning off this option, and see if the stick still gets hot

Improve this answer
5
  • 15
    It is never on by default, unless you choose it when prompted after inserting a compatible drive.
    – u1686_grawity
    Jul 12, 2018 at 18:11
  • USB-disks are faster to get small files than traditional non-ssd harddisks due to the disks are rotating which must be waited for on average half a rotation. SSD-disks are just as fast as USB-drives or faster so no advantage there.
    – Thorbjørn Ravn Andersen
    Jul 12, 2018 at 22:20
  • 2
    This feature is called ReadyBoost and is disabled if Windows is installed to a solid-state drive (which we don't know is the case or not for the author of this question).
    – Lance U. Matthews
    Jul 12, 2018 at 23:01
  • 1
    My laptop has its own SSD, but my desktop only have HDD. However, I think it need to enable this feature on a flash drive manually. And the size of Ready Boost is determined by user.
    – pah8J
    Jul 13, 2018 at 0:41
  • @grawity Yeah, and OP says they have tested this with three different flash drives, so I consider it unlikely that they activated this three times without realising the connection to heat.
    – Fiksdal
    Jul 14, 2018 at 20:11
5

There are a few possibilities here.

  • Windows does indeed have more software that automatically scans, integrity checks, power manages, and generally uses USB memory sticks. This can cause the drive to get more activity in Windows than Linux, assuming default installations of each.
  • The Linux drivers may be using the drive as a USB 2.x connected disk, whereas Windows may have a 3.x driver and it may be a 3.x port. This would increase usage speed and heat.
  • Windows is assuming power is required over the USB port for a larger disk and is supplying more, even if unnecessary, through the port. More precisely, the device requests more power from the port than is required. This is unlikely, but can occur if the firmware for the motherboard is permissive and the driver software for the motherboard and/or the devices involved is bugged/not functioning properly. I've seen this happen with older motherboards and some off-brand USB devices. It can also happen if there is a USB extension or dongle involved and the cable is damaged internally, or on a direct connection, the interface board itself is having problems. In theory, Windows 10 will catch this and notify of a power surge on the port, particularly if the cable is damaged, but it does not always successfully deactivate the port when it occurs.
  • Windows Antivirus software may be doing scans of the drive when it isn't in use.
Improve this answer
17
  • 34
    It's also possible that the whole computer is actually getting hotter when running Windows, and the heat is being transfered to the USB drive. (Either through the metal or if the fan blows near it). If you have an external USB hub you could rule this out.
    – Justin
    Jul 12, 2018 at 13:45
  • 84
    Downvoteing for point #3. Windows is assuming power is required over the USB port for a larger disk and is supplying more, even if unnecessary, through the port. This is not how electronics work. Think of it as a rope, the USB drive can pull up to 2A from the port, but the port can not push the rope into the stick.
    – Matt Clark
    Jul 12, 2018 at 14:13
  • 21
    I don't think it warrants a downvote, but I agree with @MattClark. Even if a port is USB-PD compliant, a thumbdrive isn't going to request higher voltages so the motherboard will default to 5V (I'm pretty sure it's a HW implementation, so Windows doesn't even see this other than maybe a flag saying "this port has switched to USB-PD mode" or the like). At 5V, even if the motherboard is capable of supplying ludicrous amounts of power (say up to 20A even though that's not realistic), since the thumbdrive doesn't need that much power, you don't have 20A running through it.
    – Doktor J
    Jul 12, 2018 at 16:50
  • 23
    3rd point is not a technical possibility.
    – user253751
    Jul 13, 2018 at 0:29
  • 19
    Modified answer still is physically impossible. The PC provides +5V on the USB. It may choose to cut off this +5V to a misbehaving device, but then you get 0V. Given a 5V supply, the power drawn depends solely on the USB device.
    – MSalters
    Jul 13, 2018 at 12:20
3

According to OP, the difference in temperature happens on the same laptop, same hardware, so any proximity to heat exhaust is inconsequential.

The difference is likely due to different handling of LPM - link power management. LPM is a more sophisticated version of USB SUSPEND function. USB 3.0 mass storage devices got hot when USB host controller disables LPM (or LPM is not enabled in USB device, this is a mutual thing). Apparently the Windows OS somehow has difficulty in configuring the xHCI controller to run LPM (or has it disabled), while Linux has no problem with that.

To check Windows configuration for LPM, this link might be of some help.

Improve this answer
5
  • Yes, this possibility is considerable. However, If the Windows does not suspend the Pen drive. What component of pen drive is the source of heat? (NAND, Controller or etc.)
    – pah8J
    Jul 16, 2018 at 8:13
  • 1
    @pah8J, in USB 3.0 LPM has three levels of link "suspend", U1, U2, and U3. U3 corresponds to old "USB_SUSPEND", and you are correct, U3 is not used for mass storage devices. However, U1 and U2 provide automatic power savings if properly enabled and meet specifications. If the link is active (U0), flash interface controller can consume 0.5-1 W and will be hot.
    – Ale..chenski
    Jul 16, 2018 at 15:19
  • This make sense in some extent. 1 Watt can rise the temperature significantly due to plastic case of pen drive. However, there may be also other causes. Thanks for this point.
    – pah8J
    Jul 18, 2018 at 3:49
  • @pah8J, to what extent? The easiest way to verify if the link goes in and out of U0 state is to use a device that shows the USB link status, like this one: amazon.com/Status-Speed-Visualizer-Analyzer-Tester/dp/…
    – Ale..chenski
    Jul 18, 2018 at 4:08
  • I believe that the mechanism you mentioned. But a little doubt that 1 W power output will bring the temperature in such high level. It is physics issue now instead of technological one. Maybe for high performance pen drive, the power consumption of controller will be higher. I will collect further information in related area. All in all, you point me a hopeful direction. Thanks sincerely.
    – pah8J
    Jul 18, 2018 at 4:18
2

Another possibility is a difference in how Linux and Windows handle writing to a removable drive. Windows by default will force all writes to happen as soon as possible, in case you remove the drive. Linux, on the other hand, expects you unmount the drive before you remove it, and thus keeps a write cache in memory.

Hence it is possible that your drive is actually getting less use at a given time under Linux, as the writes would be more spread out, and even less writes may be required (if you modify a file that had not been written to the drive yet.)

Improve this answer
2
  • I partly agree with you. Windows will set all removable device as hot-plug. However, for DD command, when finished executing dd if=path/to/liveimage.iso of=/dev/sd1 when I type sync, it executed less than 0.01 sec. Maybe, different distro's situations are different. Finally, I think this aspect may be have effect, but it is not the major cause.
    – pah8J
    Jul 14, 2018 at 3:00
  • This distinction is not as you describe, but at most one of degree. Keep in mind also that the user said no operation was being done - so any usage is automatic background usage, or some unauthorized software behavior to begin with.
    – Chris Stratton
    Jul 14, 2018 at 15:43

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .