Transparent memory offloading: more memory at a fraction of the cost and power

Re:Transparent memory offloading: more memory at a fraction ... by trhway on Monday June 20, 2022 @06:34PM (#31817062)

>Using a compressed-memory back end, TMO saves 7 percent to12 percent of resident memory across five applications. Multiple applications’ data have poor compressibility, such that offloading to an SSD proves far more effective.

that in general sounds strange. My experience across different application types over more than 2 decades has been very different (not about SSD, i mean the compression in memory have been always a very efficient approach).

> Specifically, machine learning models used for Ads prediction commonly use quantized byte-encoded values that exhibit a compression ratio of 1.3-1.4x.

yep, i haven't used compression for that. That though calls for a question - how many of those models do FB have in memory. Any chance FB instantiates such a model for every user/request? :)

[ Parent | on HN ]

Re:Transparent memory offloading: more memory at a fraction ... by gumby on Monday June 20, 2022 @10:35PM (#31819139)

> …i mean the compression in memory have been always a very efficient approach.

It’s quite possible that a large amount of stuff FB handles is already compressed (e.g. photos, video, and audio) and thus doesn’t offer much opportunity for further compression. Though I wonder how much of that stuff would be memory resident in the back end in the first place.

[ Parent | on HN ]