Sleep and Hibernate Fail with L1 caches

[BonzaiDuck]

Here's the configuration and history of this system:

Hardware: i7-6700K, Sabertooth Z170, 32 GB (8x4) Trident 3200 DDR RAM, Win7 and Win10 partitions on 1TB Samsung 960 NVME in dual-boot, L2 for Win10 only on 256GB 960 NVME

Seldom if ever use the Win7 OS these days. Win 10 is "up-to-date". The use of drive labels was applied very carefully to keep any caching for Win7 and Win10 separate. I haven't booted from Win7 for maybe a year or more.

Originally, system was built with half the RAM or 16GB. I made sure that hibernate and sleep mode worked, and I believe I had them working with Primocache -- which had both L1 and L2 cache. L1 -- maybe 2 to 4GB -- for the Win10 NVME boot-volume, and L2 for an SATA SSD 2TB. I think the SSD had both L1 and L2 in Primocache. But I seldom used sleep or hibernate thereafter -- although I should for at least a couple reasons.

Then I doubled the RAM, tested the configuration thoroughly -- "1,000% coverage" with HCI Memtest -- , and can't remember what I did regarding sleep and hibernate.

But now, with any L1 cache for any drive, sleep and hibernate don't work. I disabled hibernate for the time being, because (as I remember) there are issues about the size of hiberfil.sys -- which is in turn determined by a minimum of 50% total RAM. I verify that sleep works properly when there are no L1 caches but with L2 cache enabled for the SATA SSD or other drives. The Windows Page file(s) may be split up between the NVME boot/system volume and the SATA SSD.

I can't remember what it was that I did when first setting up the system with 16GB of the Trident RAM. But I had to fiddle with hiberfile.sys, and the pagefile.sys(s) were set automatically for the two drives.

The sleep and hibernate failures throw an event-ID 41 failure in process of coming out of sleep/hibernate, when L1 is implemented. The error codes point to "software or driver" problems -- not the RAM hardware. Eliminating the L1 caching to show sleep "success" indicates to me this is a problem with Primocache configuration, possibly pagefile and hiberfil configurations.

Can I have any insight into what must be done with Primocache to enable both L1 cache and hibernate/sleep?

[Jaga]

I always wrote off my problems getting a workstation running Primocache to come out of sleep correctly as "motherboard/driver issues". But I've experienced the same as you, around 3-4 years ago. I finally stopped using sleep mode and simply turn off the system now (using a NVMe boot drive with a L1 means really fast startups anyway).

Wish I had notes on my system log from back then, but I don't. I do remember looking into it, and not finding any workaround at the time. Sidenote: I too had/have a 6700K with 32/64 GB of RAM and which has been running W10 since the first stable was released. The only suggestions I can make are attempting to disable any/all motherboard features related to booting and the OS. Perhaps Support has additional thoughts.

[Support]

How about checking the option "Release L1 Cache on Hybrid-Sleep or Hibernation" in the Advanced Level-1 Cache Options dialog in the cache configuration?

[BonzaiDuck]

I always wrote off my problems getting a workstation running Primocache to come out of sleep correctly as "motherboard/driver issues". But I've experienced the same as you, around 3-4 years ago. I finally stopped using sleep mode and simply turn off the system now (using a NVMe boot drive with a L1 means really fast startups anyway).

Wish I had notes on my system log from back then, but I don't. I do remember looking into it, and not finding any workaround at the time. Sidenote: I too had/have a 6700K with 32/64 GB of RAM and which has been running W10 since the first stable was released. The only suggestions I can make are attempting to disable any/all motherboard features related to booting and the OS. Perhaps Support has additional thoughts.

I'll answer the two replies in succession. Initially, when I had sleep-mode working again without any L1 cache, I would test by putting the system in sleep and waking it up in about an hour's time. The other night, I thought to sleep it before going to sleep myself, and attempted to wake the computer the following morning -- probably more than 8 hours later. I obtained the same failed response with critical error code 41 and bug-check 0x000000a0. From the utility BlueScreenView, it shows invariably this always involves the driver storport.sys.

I'll draw criticism I'm sure, but this i7-6700K system was overclocked when I built it in 2017. The CPU voltage is below the typical "auto" setting for the CPU as default on Z170 motherboards -- well within that limit. And I made sure (at that time) that the LinX results had minimum variation so there was no error correction occurring with the overclock under stress. I've never bothered -- nor desired -- to reset the system to stock speed.

Except for this sleep and hibernation problem, the system is rock-solid stable with no errors, BSODs or troubles of any kind -- with gaming and all sorts of multitasking. I wouldn't have discovered anything about it until I tried to hibernate and sleep before I first started this thread. The PSU is a Seasonic Platinum unit. I can see that I now have to give the system a shakedown -- test the OC again, perhaps set it back to stock, and test the RAM. The RAM is at its stock settings -- always has been.

So I'll have to run some tests under changed configurations to eliminate possibility of hardware error, even if there is no other indication of any such type of error without using sleep or hibernate.

I'll probably want to update the storport.sys driver.

The problem with memory configurations this large makes hibernation difficult to configure anyway. That may be behind your thinking to abjure any further use of sleep states -- and I could do that. You're right, of course: these systems are fast enough with L1 caching and the NVME drives so that you wouldn't need to do it, but for the desire to leave your spreadsheets and documents etc. in their working state so you can access them again after waking the machine. Something you can't do by shutting down and then booting up.

For Support: I remember setting that feature you mentioned, but I'll have to re-create my L1 cache to test it, and apparently I have some other source of this trouble as I described above, since the recent sleep failure occurs with only an L2 cache in operation.

[Support]

@BonzaiDuck, if possible you may stop caching and check if this issue will happen again or not.