1 Reply Latest reply on Aug 21, 2013 3:20 PM by joe_intel

    Meaning of E2h Timed Workload Media Wear

    varao

      I'm continuously sending a high number of random writes with variable block sizes to a 525 (SSDMCEAC030B3) and I'm trying to understand the actual meaning of the Timed Workload Media Wear.

       

      On Linux, while continuously writing to the drive, I had reset the workload timer and after some time started to see the combination <0xE2, 0xE4> change from <65536, 65536> to <100,60> and then increase from then on to where its currently at <4214, 2815>. I guess this means that since the timer rest, the workload has been running for 2815 minutes and the wear percentage is now 4214%?

       

      What I'm trying to understand is the meaning of 0xE2 from the manual:

       

      Timed Workload Media Wear
      E2h
      Measures the wear seen by the SSD (since
      reset of the workload timer, attribute E4h),
      as a percentage of the maximum rated
      cycles.

       

      Is the device really measuring "wear seen"?

       

      It seems to me that the 0xE2 value is just taking Minimum Useful Life/Endurance Rating (2.6 in the manual), say 20GB/day, and considering what percentage the current workload is of 20GB/day. In other words, will the (drive-internal) calculation of 0xE2 change from when the drive is brand new to the point when the drive is (nearly) completely worn out? The reason I'm wondering this is because 0xE2 has been steadily increasing while 0xE9 (Media Wearout Indicator, raw or normalized) has remained constant (100 normalized) for the longest time. There doesn't seem to be any wear occurring, other than perhaps the drive's internal observation of the number p/e cycles, while 0xE2 is linearly progressing together with a constant stream of writes going to the drive.

       

      I'd also like to know or else verify whether the reliability rating in 2.6 of the manual is truly considering "host writes" rather than what I'd intuitively expect to be "NAND writes". It seems to me that the reliability is solely dependent upon NAND writes and, as such, determining the reliability based on host writes seems to be fuzzy, unless there's some unreported constant ratio between host and NAND writes that's supposed to be assumed.