Enclosure Can case panels serve as heatsinks without a "block" and "pipes" connecting them to components?

[Reldey]

Hey, that looks kind of nifty! Did you do any before & after testing?

Not as well as I should have. I did a couple mods at once (backplate fan and thermal pad) and didn't test before and after them much. The front bezel does get quite hot under load now though, so I know it is pulling heat away. Before the thermal pad mod and the backplate fan I would hit 85 degrees under load. Now I am around 73-75 max. All my testing was also not super relatable, because I run my case in the horizontal position GPU facing down with a few feet I made.

 

[Eigua]

The controller will do void/null operations to try to heat NAND to that temp. These operation are basically overwriting cells with 1's/0's that arent being used.

I've been searching to find something that corroborates this but I haven't been able to. I have found articles that mention the fact that NAND writes at higher temperatures are easier such as this one, but I haven't found any that state that the controller intentionally heats up the memory by writing to empty cells over and over. That previous article I linked also states that NAND lifespan decreases when stored at (and presumably used at) higher temperatures, so if not actively writing, higher temps would actually reduce the lifespan.

What I have found are many articles that talk about the benefits of cooling such as this one that shows cooling will prevent throttling during read/write operations. There are also a number of SSD's on the market that come with heat sinks, which strikes me as very odd if the inclusion of one would ultimately be reducing the lifespan.

Do you know of some authoritative source that corroborates the information about SSD's controllers writing to empty memory to generate heat? Thanks!

 

[ChainedHope]

Do you know of some authoritative source that corroborates the information about SSD's controllers writing to empty memory to generate heat? Thanks!

I'll pull some docs when Im at work and see which im able to upload for you. There was a series of tests and a few white papers on NAND and their respective controllers. It included the operation I spoke about and how temperature effected the speed of which they occured with lower temps having the operation more frequently. At 0C the operation is performed about 10x more than at 50C. An issue right now is while I can log into my portal, I don't have the licenses on the files so I'm not sure what I can and can't upload and would rather not get into legal issues for "leaking" confidential or trade secret information so I'll need to get a sign-off. There isn't much stopping me from giving a general overview though as long as I dont get too technical (like what I did in my original post).

Something you can try looking into is the sub-zero temp testing, I think there's a publicly available research paper that goes into the technical details of how NAND works as well as some issue when you go under 0C. In there they showed the memory controller forcing refreshes to heat the chips back up. They also have a nice graph of the refresh occurring at a less common interval the hotter it gets towards its nominal temperature. (Google Scholar is your friend if you dont have access to research from an edu or work account)

I also believe JEDEC has some datasheets that go briefly over NAND null ops and conservative preservation. Not all NAND does the null op I was talking about, but the NAND found in M.2 / U.2 / UF (never released, U.2 replaced it) do use it for life conservation and security reasons (Cool NAND = Slow ops = "easier" data mining of encrypted storage). I'll try finding the JEDEC standards and link them.

 

[EdZ]

What I have found are many articles that talk about the benefits of cooling such as this one that shows cooling will prevent throttling during read/write operations.

That's cooling of the controller, not the NAND dies themselves. The throttling is experienced when you perform extremely heavy sustained sequential loads. There's not many real workloads that can sustain multi-gigabytes-per-second reads/writes for the minute (or more, for newer drives like the 960 Evo) beyond copying very large files from one fast NVME SSD to another fast NVME SSD.

 

[Eigua]

That's cooling of the controller, not the NAND dies themselves. The throttling is experienced when you perform extremely heavy sustained sequential loads. There's not many real workloads that can sustain multi-gigabytes-per-second reads/writes for the minute (or more, for newer drives like the 960 Evo) beyond copying very large files from one fast NVME SSD to another fast NVME SSD.

Thanks for pointing that out. That's an important distinction. The SSD heat sinks used in the article and ones you can find for sale cover the controller and memory chips, so they don't make that distinction. Likewise, blowing a fan across an SSD would cool all components involved. I suppose a compromise would be to put a heat sink on the controller chip and not the memory.

As for reaching the thermal throttle limit, I've seen a number of articles that talk about SSD throttling having a negative impact during intense gaming. I'm assuming this is partly due to higher ambient case temperatures. There have also been reports of throttling issues with SSD's mounted on front-side M.2 slots that sit below GPUs during gaming.

 

[Eigua]

I'll pull some docs when Im at work and see which im able to upload for you. There was a series of tests and a few white papers on NAND and their respective controllers. It included the operation I spoke about and how temperature effected the speed of which they occured with lower temps having the operation more frequently. At 0C the operation is performed about 10x more than at 50C. An issue right now is while I can log into my portal, I don't have the licenses on the files so I'm not sure what I can and can't upload and would rather not get into legal issues for "leaking" confidential or trade secret information so I'll need to get a sign-off. There isn't much stopping me from giving a general overview though as long as I dont get too technical (like what I did in my original post).

Something you can try looking into is the sub-zero temp testing, I think there's a publicly available research paper that goes into the technical details of how NAND works as well as some issue when you go under 0C. In there they showed the memory controller forcing refreshes to heat the chips back up. They also have a nice graph of the refresh occurring at a less common interval the hotter it gets towards its nominal temperature. (Google Scholar is your friend if you dont have access to research from an edu or work account)

I also believe JEDEC has some datasheets that go briefly over NAND null ops and conservative preservation. Not all NAND does the null op I was talking about, but the NAND found in M.2 / U.2 / UF (never released, U.2 replaced it) do use it for life conservation and security reasons (Cool NAND = Slow ops = "easier" data mining of encrypted storage). I'll try finding the JEDEC standards and link them.

Thanks for the detailed response! I wasn't aware of your job/background. You have access to much more detailed information that I do and clearly have much deeper knowledge of the subject. I don't really need to see the industry papers and I certainly don't want to get you in trouble, but I was hoping to find at least one article that mentioned that concept since I had never heard that before. I'll use the information you posted here to see if I can find anything. Sometimes it's hard to know exactly the search terms. Thanks again!