NVME SSD Temperature Questions

[m6tzg6r]

The second drive in that picture is my Crucial 1TB P1 NVME ssd, Drive Temp 5 has a high reading and i'm wondering if its safe.

I contacted Crucial and they said safe temp range is up to 70 celsius, which i think is for Drive Temp or Drive Temp 2. From research the lower temps are maybe the Controller and the higher temp is the Memory Chips, which are supposedly meant to run warm.

In the Crucial software it only shows temp readings for Drive Temp and Drive Temp 2, but no mention of the hotter running Drive Temp 5, so Crucial don't want an average Joe to see the third reading? I did aim a 120mm fan directly at the NVME and average temp came down about 10 degrees.

My temps were obtained whilst playing Doom 2016 which is installed on the NVME.

 

[newtekie1]

From research the lower temps are maybe the Controller and the higher temp is the Memory Chips, which are supposedly meant to run warm.

Not sure where you got this, but under use the controller is always hotter than the NAND chips.

I'd guess Drive Temp 5 is the on die sensor for the controller, or a sensor very close to the center of the controller. Which is why it is hotter, but I wouldn't worry about it at all. It's kind of like the on die sensors in AMD graphics cards. AMD didn't show those temps to users, until just recently, because the average user really doesn't need to see them and would likely freak out to see 100°C+ temps. But the reality is it is perfectly normal to see those temps directly in the GPU die. I'd assume that is why Crucial isn't showing Drive Temp 5 in their software as well. Without knowing what that sensor really is, or where it is located, the temp reading is pretty meaningless, so the average user shouldn't be worried about it.

 

[er557]

my both nvme drives are under a behemoth heatsink of the asus hyper m2 v2 card, and even though I manually switched the fan of the card to off, they run super cool, maybe too cool if that is a thing

 

[newtekie1]

maybe too cool if that is a thing

It actually is.

 

[kapone32]

my both nvme drives are under a behemoth heatsink of the asus hyper m2 v2 card, and even though I manually switched the fan of the card to off, they run super cool, maybe too cool if that is a thing

I have the same M2 card thankfully the temp that it reports is not the temp of the controller but the chips. There is no way the controller would be at 35 C during heavy write cycles as that is what i saw when monitoring the drives.

 

[Vayra86]

I don't generally get nervous about temps, but hot storage... dunno. Going to need some more easing into that I think.

 

[er557]

Yeah, the nvme's controller like to get toasty, as to the memory chips, they like it cooler, And if the manufacturer does not supply heatsinks, I would not worry about it, or just slap those 1$ ebay tape-on parts. my hdd's on the other hand, always were on the hot side but worked reliably many years despite this.

 

[TheLostSwede]

I don't generally get nervous about temps, but hot storage... dunno. Going to need some more easing into that I think.

As I've posted elsewhere, NAND flash has better data retention if it's written to at higher temperatures, as long as it gets to cool down to a more reasonable temperature before power is cut.

This is based on data from Intel. If the data was written at 40 degrees C, with the SSD left to cool down to 30 degrees C, the data retention is 52 weeks. As you can see, the data retention period increases as the temperature during write increases. However, if you turn off your PC straight away, with the NAND at say 50 degrees, then you have a data retention of 9 weeks, if it was written at 50 degrees as well.
Source: https://www.anandtech.com/show/9248/the-truth-about-ssd-data-retention

 

[Vayra86]

Very interesting!

 

[tabascosauz]

Yeah, the nvme's controller like to get toasty, as to the memory chips, they like it cooler, And if the manufacturer does not supply heatsinks, I would not worry about it, or just slap those 1$ ebay tape-on parts. my hdd's on the other hand, always were on the hot side but worked reliably many years despite this.

Other way around. M.2 heatsinks began popping up as a response to the need to keep NVMe controllers cool, as they can put out a surprising amount of heat, and prefer not to run hot, as with most electronics.

NAND, on the other hand, likes it warm when the computer is on. Not a ball-busting 100°C, but if run too cool, bad things start happening.

 

[m6tzg6r]

Thanks for all the replies.

 

[Kirsebaer]

As I've posted elsewhere, NAND flash has better data retention if it's written to at higher temperatures, as long as it gets to cool down to a more reasonable temperature before power is cut.

This is based on data from Intel. If the data was written at 40 degrees C, with the SSD left to cool down to 30 degrees C, the data retention is 52 weeks. As you can see, the data retention period increases as the temperature during write increases. However, if you turn off your PC straight away, with the NAND at say 50 degrees, then you have a data retention of 9 weeks, if it was written at 50 degrees as well.
Source: https://www.anandtech.com/show/9248/the-truth-about-ssd-data-retention

Hi, I need to correct this incorrect information, in case someone reads this while searching for the same information as I. I don't think you've read the article you post, or at least not very well.

The final paragraph: "Remember that the figures presented here are for a drive that has already passed its endurance rating, so for new drives the data retention is considerably higher, typically over ten years for MLC NAND based SSDs. If you buy a drive today and stash it away, the drive itself will become totally obsolete quicker than it will lose its data

 

[chrcoluk]

I read an article which I think the reviewers sourced their info from, the endurance temps are not just for when the power is out, but its for the entire time the data is been stored (even when powered up) as well as when its read so in other words lets say e.g. your NAND is been written to for 10 seconds, but then the next 40 seconds the drive is idle, and then the next 10 seconds after that the data is been read.

For optimal conditions you would want high temperatures for the first 10 seconds, but then low for the next 50, so in short cooler temps are better, unless for some reason your drive is mostly doing writes when powered up which would be highly unusual for a consumer workload.

My 970 EVO is about to be RMA'd and the data it cannot read is data that was put on there before I started cooling it.

The data became unreadable in the space of about 12.5 months. It was written early in the drive's life and never rewritten after.