I'm adding some benchmark and diagnostic features to Arktika
and the benchmarking took down my vm server
yeah, this ddr2 box can only do 18GB/s to ram
heh
sweet jesus, 142GB/s if I put in jvm ram vs tmpfs
@Fireduck 18GB to 142GB? seriously?
yeah
I wonder if there is something I should change with regards to filechannels
it doesn't seem quite right
the trouble it, tmpfs isn't a shit show like jvm heap
so you can get a lot tigher, like 240gb in tmpfs on a 256gb system
where with java heap, you are lucky to get maybe 180gb on a 256gb system
that seems wrong
how the hell is there that much overhead
that is a separate question that i haven't figured out yet
this is a big enough deal I need to test this in python.
it’s the GC, if it can avoid firing, it will
Oh, that's a good point. Maybe you can set garbage collection to occur differently.
you can, that is a very wide permutation space of JVM tweaking
also, it occurs to me that something might be optimizing and determined that since I didn't actually read the data after fetching it, why bother fetching it?
so that 140GB/s number might be nonsense
still 128GB/s after making it read things and do stuff