Rock64Pro problems: "couldn't allocate randomx cache using large pages"

[ghost]

Hello. I have a RockPro64 that I have monerod and p2pool running on at the same time. I am mining to the Rock64Pro with a Ryzen 3950X. The Rock64Pro board has 4GB of RAM. Monero daemon and p2pool combined use up 3.5GB of RAM. When I start p2pool, I get the yellow warning message "couldn't allocate randomx cache using large pages". I set the large pages to 3072, which seems to work, but the error still comes up. The board regularly freezes.

Questions:

1. Is the "couldn't allocate randomx cache using large pages" warning impacting performance at all?
2. How might I fix the problem?
3. Does this board have enough RAM to run p2pool alongside Monero daemon, or should I get a second board to divide up the work?

Thank you in advance for your help

[steba000]

I'm using a Raspberry pi4 8Gb with monerod and p2pool. xmrig is running in other systems. I setted-up vm.nr_hugepages=525 and this value is right in my configuration. Maybe I can reduce this value to 515-516.

# grep -i huge /proc/meminfo
AnonHugePages:         0 kB
ShmemHugePages:        0 kB
FileHugePages:         0 kB
HugePages_Total:     525
HugePages_Free:       10
HugePages_Rsvd:        0
HugePages_Surp:        0
Hugepagesize:       2048 kB
Hugetlb:         1075200 kB

[ghost]

How do you think I should set huge pages with Rock64Pro that has 4GB of RAM?

[SChernykh]

If you are RAM limited, you should consider --light-mode or --no-randomx option for p2pool.

[steba000]

How do you think I should set huge pages with Rock64Pro that has 4GB of RAM?

English is not my first language and I don't know if you are kidding me or not... anyway... 5252048 kb=1075200 kB ---> 1Gb 3072 2048 =6291456 kb ---> 6Gb 1Gb < 4Gb < 6Gb Maybe it's still not enough if the kernel uses a lot of memory. I have got also a rockpro64 used for a different project

[ghost]

@SChernykh Will the p2pool be able to handle my Ryzen 3950X @ 20k hashes/sec if it is using --light-mode?

[SChernykh]

hashes/sec doesn't matter, p2pool only checks submitted shares. So any hashrate should work with autodiff or high enough difficulty (30 seconds per share is recommended).

[ghost]

Ok. Does enabling --light-mode hurt my p2pool instance in any way?

[SChernykh]

It will check shares and new p2pool blocks slower (1-2 ms in fast mode vs 10-15 ms in light mode), but it's not critical.

[ghost]

Thank you for your help :)