yes, that’s true, but I quoted this:
‘‘24H2 grants the whole 200MB for SN770. If SSD firmware cannot handle the >64MB part, it will result write failure, and system file corrupt.’’
if this is true, then SN580 should handle 200mb part (24h2), because:
SN580 HMB size is 200 MB, which is more than the 64 MB that we’re typically seeing on DRAM-less SSDs.
edit: I’m using 23h2 since April. Tried 24h2 (25 days ago) just for few hours and removed it.
Disabled HMB and so far no more BSODs with an SN770 and 24H2. Thanks for tracking that down, now I guess we’ll just be waiting on a firmware update. Until then what does leaving it disabled do, just give a performance reduction? I did see that setting HMBAllocationPolicy to 2 gives a 64MB buffer, should that result in the same performance as having it enabled without crashing?
It doesn’t matter about the rest of my hardware because I have been fighting 24h2 for days. I just purchased a second WD SN580 2TB since windows 24h2 pooched my computer on sunday and I tried everything to get my computer back in running order. I have previously installed 24h2 on a Kingston 2TB and a Intel 256GB NVMe SSDs. They both installed fine. The other WD SN580 was purchased a month ago direct from the factory, it also gave me issues with the upgrade of my 23h2 to 24h2.
So I don’t think it is anything else. Anyone experiencing a blue screen boot loop with 24h2 and you are trying to install it on a western digital drive, be aware, you will probably get a blue screen.
I think it worth a try to set HMBAllocationPolicy to 2.
And I need to mention that probably it is not WD’s problem. The problem maybe at Microsoft side. It may be because Windows cannot handle 200MB HMB. Maybe, Windows treats it as 256MB, and it causes memory exploit. That is also possible.
200MB is a weird number in binary world. I don’t know why WD picks that number.
Setting HMBAllocationPolicy to 0 solved the BSOD but the performance in games is crazy bad now. I will wait a couple of days for a fix or else i throw the ssd in the garbage can and buy another brand
Creating a DWORD32 “HMBAllocationPolicy” and setting it to the value of “2” fixed all issues related to EID 11 and NVMe (SN770) straight up stalling and crashing soon after.
So, yeah… it did alleviate my issues at no performance cost. R/W speeds remain as advertised. Cheers, Snailum!
Also, to comment on the guy that went out here to give their ultimatum for all their problems in the world. Nobody comes here to listen to your whining and calling out ■■■■■■■■.
I changed the HMBAllocationPolicy to 2 to check if it’s better, I don’t know if I should come back to the generic windows driver or if with Micron NVMe driver v2.1.18.0 WHQL the edit registry path it’s the same.
Note that I didn’t have the stornvme error in events since I installed the new driver, but some BSOD still happening
Thanks everyone for the help, you are amazing, hope with this change to value 2 the BSOD stops, I will let you know
I experienced the same issues after moving from Windows 10 to Windows 11 24H2 with this hardware:
SN770 2TB, latest firmware
B550I GAMING EDGE WIFI on BIOS version 7C92v1G
The OS can boot, but after a very short time (often on the login screen, before it even finishes loading) the OS becomes unresponsive and restarts.
The ONLY thing that has been able to alleviate the crashing is setting HMBAllocationPolicy to 0 while using the default NVMe controller driver. Any other combination of the value on this registry key or other NVMe controller drivers (have tried Micron and Phison) will cause the aforemented crashing. The error that is shown in Event Viewer upon each boot (Event 11, stornvme) persists though.
Event 11 disappears with the other drivers, but the crashes persist. Either revert back to the default driver and set the DWORD back to 0, see if that helps (so far it’s worked for me), otherwise I guess Windows 11 24H2 has to be avoided until WD pushes out a firmware update for the drive.
Just updated to 24H2, and experienced crashes on start up and the same controller errors in the event log.
SN770 on firmware 731120WD
Asrock B650M Pro RS
HMBAllocationPolicy to 0 fixes the crashing, but is presumably having an impact on performance without HMB cache.
I tried HMBAllocationPolicy to 2, but that also allocated 200MB for me according to the NVME HMB info tool. I’ve set it to 1 this only allocates 3MB but better than nothing. Can’t find definitive info on the different values but a bit of Googling says “Valid values are 0 for disabled, 1 for minimum, 2 for maximum, and 3 for device usage-based”
I also tried the 64bit Micron NVMe driver v2.1.19.0 WHQL from the level1techs forum, but that just disables HMB according to the NVME HMB info tool.
So we need WD to fix the firmware to ask for 64MB, or MS to patch Windows.
The fact that we are yet to receive an official repsonse or even basic acknowledgement of this issue, makes me doubt we will ever see a firmware update that would patch this stuff outright.
If you can, report this issue directly to Microsoft via the Feedback Hub. Given the severity of this issue and inability for an end-user to use their operating system, I’d wager this would be one of those “high priority” bugs for MS.
In device manager, under Disk drives find your affected drive. Right click Properties → Details tab and look at the parent property it’ll be like ‘PCI\VEN_15B7&DEV_5017&SUBSYS_501715B7&REV_01\4&333c8b5&0&0012’ you need to copy the VEN, DEV and REV parts. For my example it is ‘VEN_15B7&DEV_5017&REV_01’
Back in regedit go to Computer\HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\stornvme\Parameters\Device
Create a new multi string value called ‘HostMemoryBufferBytes’ with the value from 2. followed by a space then the size in bytes you want i.e. ‘VEN_15B7&DEV_5017&REV_01 67108864’