whether such oscillations are normal?

[examplens]

One rig, 6x RX 580 8GB, running on HiveOS and on Sparkpool. GPU clock is 1145/2100 850uV, card temperature is from 46° up to 57°.
All cards working on 30.9MH/s
I mostly monitor over Sparkpool android app, occasionally I see a really big difference in the current mining hash. It's going from 100mh/s up to 300mh/s.
For example:
RealTime 240.00MH/s
-0.23%
24h Avg 184.96MH/s
Reported 185.45MH/s
Reported 24h Avg 185.40MH/s

whether this is normal? Or I need to adjust overclock better?
Here is how the graph looking on the Sparkpool app.

[Ardvark69]

Its pretty normal but that being said you should be able to get 31.5-33 mhs.

[examplens]

Its pretty normal but that being said you should be able to get 31.5-33 mhs.

Well, I am not sure how is smart to push GPU's on maximum.
Card fans were set on 60%, with currently low 2°-10° outside temperatures I am satisfied with their condition. I don't want to burn them because of a few mhash's also it probably reflects power consumption I guess it will be just wasting energy and cards lifetime.

[JayDDee]

The hashrate reported by the pool is artificial, calculated from the number of shares submitted.
Since shares aren't submited at a regular rate the calculated hash rate can vary.

[fmz89]

try on ethermine, and wait for 12hrs, as your statistic is a bit weird even the average hashrate still normal, and check hwinfo for memory error, your memory config hit 2100 is realy high for polaris, better check memory error first then try other pool

[FP91G]

whether this is normal? Or I need to adjust overclock better?
Here is how the graph looking on the Sparkpool app.

You only need to monitor the average hash rate and the reported hash rate.
You have good hash rates, and if your farm is stable, then do not interfere with her mining cryptocurrency.
If during memory overclocking to 2100 there are no stale shares, the memory frequency is selected correctly.
I don't see the point of overclocking anymore