Okay, ran a read benchmark with 1,000 1GB samples - so a terabyte read. T1 and T2 sensors sat at 40 degrees and 65 degrees respectively, plus or minus a degree, until about the halfway mark when it for some reason went to 52 and 74 degrees respectively. There was a brief drop near the start, and a slightly more sustained drop at the halfway point - but as it picked back up again afterwards, I'm seeing those more as something else wanting to use the disk for a bit rather than a throttle operation (but that's just me guessing based on the available evidence.) Benchmark done: SMART log at the very end of the run: Code: Smart Log for NVME device:nvme0n1 namespace-id:ffffffff critical_warning : 0 temperature : 40 C available_spare : 100% available_spare_threshold : 10% percentage_used : 0% data_units_read : 3,173,584 data_units_written : 3,834,557 host_read_commands : 10,611,605 host_write_commands : 26,910,489 controller_busy_time : 170 power_cycles : 197 power_on_hours : 176 unsafe_shutdowns : 5 media_errors : 0 num_err_log_entries : 1 Warning Temperature Time : 0 Critical Composite Temperature Time : 0 Temperature Sensor 1 : 40 C Temperature Sensor 2 : 65 C Thermal Management T1 Trans Count : 0 Thermal Management T2 Trans Count : 0 Thermal Management T1 Total Time : 0 Thermal Management T2 Total Time : 0 No throttle operations recorded.
It would appear that my Crucial doesn't record throttling. Crucials own disk manager thing doesn't have anything, CrystalDiskMark has a few "VendorSpecific" sections but it doesn't say any more than that. Crucial software reports the highest temp as 81 degrees. Doesn't reports any problems, drive health still 100% so can't be doing much harm. I'll probably stick the heatsink on when I next pull the cpu hsf now that I've got it. I'll not worry about it too much now,
Yes it cant hurt, although some seem intent on arguing against being prudent. I would not want any component at 80 plus degrees, whether it is truly affecting performance or not. A read of systematic tests of nvm-e drives suggest it can affect performance - which I trust more than a guy with one drive and one computer at his desk without a validated testing suite and a control etc.
Unless you put it on the NAND, where it will increase power draw and reduce operational lifetime. SSD heatsinks are like RAMsinks: shiny marketing devices with no effect on performance outside of extreme circumstances. They are an effective device in converting your cash into a lump of metal, but not into performance.
Your link to another thread, and the article therein, did not speak about increased power draw and reduction in operational lifetime? The article on a 950 Pro stated the exact opposite, in-fact. I am also unsure if a 60GB sequential write is an extreme circumstance, either. Edit - I saw your above posts about the nand. Genuinely interested in finding out more about it.
You're very welcome. Care to link to these "systematic tests of nvm-e [sic] drives," which I'm assuming somebody was paid for rather than doing out of the goodness of their heart?
It wasn't meant as a personal criticism. I think we all appreciate your input, not just on the forum but your articles too. Here is one specific to the 970 evo which they saw throttle at 79c, 5 degrees short of your peak during the read benchmark you graciously showed us. They state that the copper label and nikel coating is much improved, and that only heavy use cases need consider active cooling (and maybe that speaks to my ocd that I bought the heatsink before the drive). Nevertheless, and I return to the OP, as we are talking about one of the (best) most recent m.2 drives. It sounds like the OP is unsure how his is actually operating.
I'm only being a sarcy bee-eye-tee-sea-aitch, don't mind me. Ta! So, it looks as though I'm *probably* OK as-is, as I'm unlikely to be throwing more than a terabyte around at a time (given it's a terabyte-sized drive.) Write will probably get it hotter, mind you, but like 'eck am I going to chew through the NAND write lifespan to test the theory out! Ubuntu Live DVD and the Gnome Disks Benchmark would at least give us a comparative with mine, if the SMART log isn't being helpful and tracking throttle operations.
Well this thread has been informative for me too. It motivated me to look in to the issue more, and there is a consensus that airflow is as good, if not better, than a heatsink. So I apologise to yuusou in post #7 for trying to contradict him, airflow is a key helper in the storage cooling stakes!