Intel runs all of their products through test following (or exceeding) the JEDEC standards for SSD devices. These are fairly involved tests that involve operating the devices at the high and low end of their temperature ranges, subjecting the drives to vibration, etc. More involved tests perform more complex operations like writing cold and reading hot, and vice versa. Our briefings took place at the Folsom campus so Intel could walk us through the spaces where such certification tests are actually completed.
Here is a temperature controlled test cabinet. These cabinets each contain over a hundred SSDs and keep them all busy while temperatures are kept at the high or low ends of the design spec (i.e. 5c, 85c, etc).
This is not the only cabinet full of test samples, either:
There are additional temperature controlled cabinets that contain actual client systems, each driving four SSDs with specific test workloads that are more complex than what the simpler batch-testing cabinets are capable of. On one side of this cabinet sits 120 mini-ITX client systems:
…and on the other side we see 480 SSDs densely crammed in and under test:
…and once again, there is not just one of these cabinets:
Smaller batches of these samples are taken near their end of usable flash life and then stored in a worst-case-temperature environment for several months:
After months of 'baking', those samples are then evaluated for their ability to retain data for extended periods – even with their flash cells at their EOL, completely worn state.
All in all, this is a significant level of testing. I was even shown a large area full of serialied boxes of stored samples – a portion of *every* production run of *every* model of SSD Intel has *ever* produced. This enables Intel to respond quickly to reports from the field. If there is a suspected issue with a particular batch of product, they can simply pull those specific samples from storage and attempt to reproduce the reported issue on that very same batch.