Aquaero 5 LT and memtest failures

[ZeroOne]

This is such an odd problem, but reproducible, none the less.

It seems the Aquaero 5 LT is causing massive amounts of memtest errors in a VMware host machine / home server. I realize this is not liquid cooled, but the part in question is from that realm. Its being used to regulate hard drive temps vs. fan speeds much like it does with watercooling and water temp.

I use the Aquaero in a few other systems to regulate water temp with fan speeds and have yet to test, but so far on this server, massive memory errors if this is hooked to the internal USB header while running Memtest86.org's 5.01, single or multi threaded. No issue if the Aquaero is connected to power only.

Specs:
Asus P9D-E/4L
Intel E3-1245v3
Kingston 32GB ECC DDR3 1600
LSI 9300-8i
LSI 9300-8i
Samsung 840 Pro 256GB
WD7500BPKX
Misc WD and Samsung drives on LSI controllers
ESXi 5.5u2
VMs (Win8.1, Win10prev, Ubuntu 14.04.1, CentOS 5, 6)



Has anyone had any kind of memory issues with one of these hooked up??

[Shoggy]

Almost impossible that the aquaero is the real reason here. It is more likely that the used VM has a bug with its HID handling for the USB.

[ZeroOne]

I should add, I didn't intend on blaming the Aquaero as if it's at fault entirely. I have 3 in total and they all work great! Something is up with this setup, however.


In this situation, there is no OS. Booting from Memtest86+ from memtest.org, v.5.01 on USB, the memory tests fail quickly. Within minutes.

In the test, with just motherboard, processor, ram, aquaero hooked via internal usb header (correctly). It fails memtest with 1, 2, 3, 4 sticks, in any slot:

Passing result after 17hrs with no USB connection to Aquaero (but power is supplied for fans):





Take the USB plug and hook it up............







Failing result within 2 minutes, 48 fails so far....





Nothing else hooked up in the system.

[zalbard]

Try a different USB port.

[Shoggy]

Have you noticed that the failed address always changes by 4 steps? Hard to believe that this is a natural random problem and given the fact that these are super small steps this system should not be able to run anything so what happens when you run the regular operating system?

Like before I still think there is something wrong with the HID implementation in the USB driver. Since it runs directly from the stick it must be the operating system that memtest uses. Well, you only have to get another memory tester which is based on a different operating system. Afaik the Windows installation CD also offers a memory test when you boot from it and run the diagnostic menu. Maybe another version of memtest would also solve the problem.

[washu9]

I'd agree with Shoggy. There is a high possibility that it's the USB at fault. Although I'm not sure about blaming it on the driver.
Several things I'd like you to try out.

1. Try a different USB port for the Aquaero if possible.

2. Turn off legacy support for USB in BIOS.

3. If error still comes up, try using a CD drive for the memtest. Unplug anything other than the Aquaero from USB ports.

4. If error still comes up, try securing the Aquaero to a non conductive area. Just put the board on a sheet of cardboard or something to isolate it from the case.

[ZeroOne]

Thanks for the replies! Going to try a different memory test. The previous memtest.org version of 4.20 fails, as does the memtest.com download that is similar. These are so close visually that I'm not sure the underlying issue wouldn't be in all 3.

Will also try the USB changes in the bios and the non-conductive test. It is definitely grounded mounted to case metal right now (though not touching with anything other than the feet / standoffs).