please add ZADD to the '-r' usage line

I can do that @itamarhaber

@itamarhaber @filipecosta90 hi, I think I should remove zrem test because I can't have a good idea to handle deleting inexistent members, do you think so?

I guess that makes sense... we could benchmark ZMINPOP instead :)

Thanks @itamarhaber I haven't learned redis commands too long time, I didn't know ZPOPMIN just now :). ZPOPMIN can make sense just like SPOP,LPOP.
So I will remove ZREM and add ZOPOMIN.

@ShooterIT @itamarhaber this looks like a good change, I just have one concern, the default benchmark test takes a very long time to run, if we adding more tests should this causing it takes longer time? Redis Benchmark also can run with custom command if we append the test directly in the parameter, such as redis-benchmark -n 100000 -q script load "redis.call('set','foo','bar')" do we leave zadd or zrem as default benchmark to run or leave it to be custom? thanks

@daidaotong good point about time, but I don't see that as a real issue because adding two..eight more tests isn't going to change the run's time by orders of magnitude. I would guess that this wouldn't really affect anyone who's running the benchmark with the defaults... for whatever purpose.

I support the initiative because it seems strange that ZADD isn't present in the default suite - probably an omission rather than intentional. I'm not bent on using ZPOP, but I have nothing against it either.

Semi-related: #7600 - @madolson @yossigo thoughts?

Thanks for your concern @daidaotong I think we should add zrem and zpopmin as defult benchmark. Firstly, zset is the basic data type of redis, we should know its' performace, otherwise, we should test the performace of zadd/zpopmin when we make the size of zset(skiplist actually) bigger/smaller.
For time, I also think it is not a big problem, we may use extra 4s defaultly (qps is 6w if multi-thread is disabled on my machine)

This remind me about #4895, but never mind just go on.

BTW, in HSET we need random field instead of random key, what do you think @itamarhaber ?

thanks @soloestoy I also found this, i think we may want to test performance of hash when we expand it by adding more fields

BTW, in HSET we need random field instead of random key, what do you think @itamarhaber ?

BTW, in HSET we need random field instead of random key, what do you think @itamarhaber ?

Sure, but that's a different "test" altogether - i.e. populating the DB vs populating a nested type.

Yes, but i think 'SET' test already is for populating the DB actually

Hi @itamarhaber I see there is a pending review request from you, could we just merge this PR and then I can make another new one to discuss the HSET problem.

Sure - merging & thanks @ShooterIT :)