DDR5 SPD Hub temperature occasionally spiking

L

LeDoyen

Well-Known Member
Hi,
I tried searching for this error but.. the forum seems to discard short search terms, so i'm out of luck with "SPD" and "hub" lol

My rig is using Aquasuite (with all its hardware monitoring disabled) and HWinfo for sensor polling to aquasuite.

I noticed that on both ram sticks at the end of the day i have a maximum temperature reached of 63.8°C exactly which is very odd.
The hottest i could get them while playing games like hogwarts legacy was 49 - 50 °C tops.
I only once saw the spike happen live. i opened Steam, and the RAM temperature did 43°C - 63.8°C - 43°C in 3 consecutive readings.
I was wondering if that was a known issue, or if anyone had any idea what may cause it.

I know it hapens everyday, but it seems to happen at random, and pretty rarely.

I don't have any other monitoring software apart from HWinfo. I actually reinstalled Windows 11 last weekend, so it's still a very clean install.
I was seeing the Hyper-V warning message on HWinfo main window so i wondered if that would cause a problem, but there is no way to make it go away except by disabling virtualization in the Bios. Hyper-V has never been installed on my PC. I tried installing/uninstalling it see if it would refresh but no.. still getting the virtualization warning. Anyway, i have virtualization totally disabled and the weird readings keep happening.
 
It might be better to start logging the values and then see if this is a single spike or if you observe a ramp-up.
 
i've seen one occurrence and it was a spike between two normal datalog points. I'm starting datalogging it now.
 
I suppose so. That's how it looks like in the logs :
It's always the same value when it happens.
1687026043294.png

It doesn't bother me now that i know it's some bogus readings, but i worried for a minute that the ram might overheat. It just makes the max value column useless on HWinfo.
 
It has to be a bug. Both of my systems (specs in signature) do exactly the same thing. It's like a momentary glitch that causes the sensor to register 63.8°C. The memory modules are generic green Hynix A-die on one system and TeamGroup Delta 7200 on the other.
 
oh thank god i'm not the only one.
i thought this is ecc kicking in because of an unstable oc and melting my sticks.
but so it's definitely some kind of bug.
we're all experiencing the exact same spike temp.
i just hammered my ddr5 ram oc with 105 iterations of y-cruncher vst.
no errors.
peaks at 55c but for a second or two i see 63.8c.
 
Last edited:
Another reason might be concurrent use of other system monitoring or tweaking tools along with HWiNFO.
 
I am having here same behaviour, but also in the minimal reading - 16C. Its absolutely impossible while I was running gpu stress and memtest simultaneously and room temperature is about 23C.
Besides hwinfo, I have also aquasuite running, but same as the op, with disabled monitoring.

My dimms are Kingston Fury KF560C36BBEK2-32, and I guess they have the same spd hub chip and thus similar behaviour.
 
Same here, G-skill F5-6000J3238G32GX2-TZ5RK on Windows 11 pro.

Every now and then, the reading will spit a 16c or a 47c. No monitoring or RGB software running, fresh install of Windows

Not the end of the world, just somehow annoying. For now, ill add an exception to discard the below 20c and above 60c reading but hopefully this is fixable thru software rather than a hardware bug...

DD5.jpg
 
Last edited:
Ok I did some thinking. Going with the conclusion that it is a sensor issue (hardware), I have a request for you. Not sure if this is even doable or how much work would be involved but I'll put it here.

Would it be possible to add more options to the "invalid if" part of the customize settings section?

Right now we have + X value and - X value, which is a bit limited. I'm thinking of a =X value (actually more than one, maybe 4-5 of them so we can add multiples).

In my case, the 2 invalid value that seem to pop up are 16.0c and 47.8c (see below). Others in this thread are reporting 63.8c and -16.0c

Being able to discard just those specifics would retain the precision of the graphs while altering it as less as possible.

DDR5.jpg

I could easily add a under 20c and over 47c, but the 47.8c is within operating range so I would lose readings completely, which is far worst. Note that 47.8c happened as a spike from a flat graph of 31c, so it is confirmed invalid.

Also, while I'm pitching ideas, do you think you could also add ranges to the "invalid if"? something like a from-to. I do also have a use case (my corsair PSU).

Again, just pitching idea to try to improve the user experience! No clue how hard this would be, or if this is even possible...
 
Yes, this has already been requested. I'm rather thinking about filtering those few values directly for this readout so that users won't need to manually configure them.
For that I will need a bit more precise information. Could you please adjust showing those values with 3 digit numbers (can be done in sensor settings - customize) and then once you catch them, please tell me the exact values.
 
You must log in or register to reply here.
Back
Top