TP201 Project3: write benchmark comparison between kvs and sled

[xuyifangreeneyes]

General Question

I am writing TP201 Project3. When doing write benchmark, I find that sled's set operation is much slower than kvs.

I also run the Project3 code in this repo and find similar phenomenon.

Write benchmark for sled takes so much time that it is not yet finished in the snapshot.

Is it normal that sled's set operation is much slower than kvs or I made mistakes somewhere.

Tips

Expect for asking questions on Github, you can also discuss it in the channel
#wg-talent-plan-courses. of
tidbcommunity
slack workspace.

[TheLudlows]

Maybe the sled's flush is too expensive

[dreamerlzl]

Same here. In my old Intel mac laptop, it takes > 1 hour for the 100 samples, while only 3-5 secs for the referenced solution and my own solution.

[Tom910]

The same issue. In my case helps to upgrade the version of criterion to 0.5.1 version. A new version of the library better adapt for slow sources