It's been a few weeks, and while I have the new nvme on my desk, I haven't had the time to put it in.

That said, I have put /tmp on tmpfs a while ago, and the average data written had dropped from ~60,000KB/hour to around 5KB/hour. I did find out that Google Chrome likes to write a lot in /tmp ...

I have put /tmp on tmpfs and set the kernel parameter nvme_core.default_ps_max_latency_us=2000 , yet the Data Units Written (divided by the number of hours between 2 snapshots) for the past 24 hours have almost doubled.

Searching the web, I get a lot of hits on kingston ssds having this behavior ...

Anon-E-moose wrote:Do you have nvme-cli installed? It has lots of useful options for nvme investigation.

Yes I do, but I haven't used that before. Any hint as to which commands to look at?

Thank you (and everyone else)!

Just another data point: I discovered the inotifywait command, so I'm running it on /tmp.

In the last 2.5 hours, Google Chrome is the only process that has written anything in /tmp - 740 times, either create, modify, or delete a file there. For most of the time, Chrome is actually just sitting ...

Do you have discard turned off and run fstrim periodically

Unless there's something specific to disable discard, I did not willingly turn it on - the "discard" option is not in my fstab. I actually didn't even know about that until about a month ago when I made the first post of this thread.

As ...

Did you even expect that 700MB writes that one day?

No, all I'm doing is browsing the net, watch a few Youtube videos. Syncthing is for my LAN, not much happening there. The PC is taking backups at night, but sending them to another PC, so that's just reading, not writing. /var/log doesn't seem ...

Filtering out the UFW (Uncomplicated FireWall) that I started using about 2 or 3 months ago (therefore, long after this excessive writing started), I get the output below.

I just found this page for ArchLinux stating there is an issue with that exact same drive with that exact same firmware ...

I have a new NVMe "drive" on my desk, and I will switch in the near-future, but in the meantime, here's some interesting data. I don't think it shows anything (other than it doesn't match). I have run smartctl twice, at 24-hour interval. iotop has been running (in accumulation mode) for that same ...

eccerr0r wrote:One thing that is suspicious is that the read/write ratio is oddly skewed to writes - meaning that it's written and never read back...

/var/log ?

Today's TLC and QLC drives just don't have the endurance anymore but for most uses they are fine. 1PB written however is a LOT, what are you doing to the disk? Using it as a bittorrent dump?

I have machines on 24/7 but they're mostly idle. One of them is PVR and it's accumulated ~ 65TB written ...

2 years ago, I got myself a new computer, this time with a NVMe drive mounted as root drive (/). I also have regular HDD set up in RAID1 for my /home. I did not create a separate partition for /var, so it's on the same NVMe drive as /.

That PC is on 24/7. At some point, I rebooted, and the BIOS ...

Thanks for the tip Goverp. Yeah, I did have the firmware, and while I did (and obviously still do) have the VIDEO_CARDS variable set, it was irrelevant since X was not even installed, much less trying to load. But 5.13 didn't have that issue, and the i915 driver was changed in 5.13, so I guess the ...

About a month ago, I got a new PC. I was unable to get the GPU working (see here ) with gentoo-sources, but it worked with git-sources (5.13-rc4 at the time - I'm still running that kernel right now). Now that 5.13 has been officially released and there is a corresponding gentoo-sources-5.13.0 ...

Oh!?!?!?!

In a last-ditch effort, I moved to git-sources.

[ 0.483594] Linux agpgart interface v0.103
[ 0.484114] i915 0000:00:02.0: vgaarb: deactivate vga console
[ 0.487689] i915 0000:00:02.0: vgaarb: changed VGA decodes: olddecodes=io+mem,decodes=io+mem:owns=io+mem
[ 0.487708] i915 0000:00:02.0 ...

I used to take Asus when I first started building my own PC. 9 years ago I switched to Gigabyte, but I don't recall why - I thought it was just because I liked it better, but Windows-only support might have been it. I'm banning Asus from now on as well.

As expected, Clear Linux has the same behavior on this system. I also tried resetting the BIOS - no luck.

I've managed to install Windows 10, with the graphic drivers, and it works without any problem. I'll still try Clear Linux and resetting the BIOS, but it might look like my other option is to get a separate nVidia GPU, which I wanted to avoid in the first place because support is 3 to 5 years; that ...

Is there a BIOS upgrade perhaps for this CPU?

Forgot to answer that one: yes, there was a BIOS update, which I did apply 2 days ago, hoping it would resolve the issue.

I'll try resetting the BIOS. I also agree that the mobo is suspect. I will need to provide evidence using Windows, 'cause of ...

alamahant wrote:Have you tried with Clear Linux live cd
https://cdn.download.clearlinux.org/rel ... esktop.iso
?

I have not (never heard of that one before), but I'm downloading the iso now!

Yes, the motherboard supports "Intel® Socket LGA1200 for 11th Gen Intel® Core™ Processors & 10th Gen Intel® Core™ , Pentium® Gold and Celeron® Processors".

PCI ID:
00:02.0 VGA compatible controller [0300]: Intel Corporation UHD Graphics 630 [8086:9bc8] (rev 03)

I do find that ID in the kernel ...

The Windows 10 Trial at least goes further than Windows 7. I get to a "full windowing" system, but this is most likely the equivalent of fbdev. At this time I can't get Windows to install because it complains about missing media driver. I've fed it every driver from the motherboard's CD, but the ...

I was missing the microcode, but it did not change anything.

As far as I can tell, the latest versions of the firmware and microcode are already what's coming down from emerge.

Jaglover, yes, this is brand new hardware. Windows 7 failed, but reading on the net it might be "normal", being an old OS ...

Same thing with 5.12.8...

I'll have to dig it out, but I seem to remember seeing "Comet Lake" come up at some point for my GPU.

Right again about the firmware. I've been banging my head on this for almost a week now, and I missed that.

So I've added that filename to the firmware load, as well as loading i915 as a module. I need to enable these:
KVM
VFIO_MDEV
VFIO_MDEV_DEVICE

in order to enable:
CONFIG_DRM_I915_GVT_KVMGT=m ...