sech1: no issue clocking 4x16GB dual rank Bdie 3600-16-16-16-36 to 3600-14-14-14-14-38

-28*

31 900 H/s

still feels a little low

1000 h/s per core?

What's CPU clock speed?

stock, so around 3980Mhz

let me check

fixed clock speed give better hashrate even if it's less than what stock shows

yeh I know. With Christmas and family and stuff, haven't hada lot of time to tune it yet

mining something else now so a little too variable to tell, but earlier tests showed it stabilizing around 3980

don't have a cooling solution to really push it, but will try e.g. 4GHz all core and maybe 4.1 and see how that works out

maybe clock down and tighten timings a bit more.

sech1: __try __except won't work with JIT code because the compiler doesn't know how to unwind the stack

have you tried this? docs.microsoft.com/en-us/windows/wi…lingapi-setunhandledexceptionfilter

Unhandled exception filter doesn't work too, I tried it first

or maybe it works, but I didn't have time to figure it out

AddVectoredExceptionHandler might work since it's idependent of the stack frame

tevador I just went "brute force" approach with debugging miner process - main process launches miner as a debugger, catches and fixes crashes.

nicehash doesn't mine into any other public monero pool, right?

cohcho: nicehash mines on whatever pool the buyer points it to

Regarding this "Can anyone test this? tevador/randomx-sniffer #2":

I've launched randomx-benchmark within `systemd-nspawn -D my_test_container` and polled it's registers with gdb few times $: gdb attach `pidof randomx-benchmark` -ex 'p/x (($mxcsr & (3 << 13)) >> 13)'

and got 3 different values 3, 0, 1

I beleive syscall ptrace can do the same as gdb

Is it appropriate test?

yes, that's pretty much what I expected

I've asked about nicehash mining power that wasn't rented, it should be mining to somewhere

there is no "mining power that isn't rented"

AFAIK there are always lowballed orders that would kick in if nobody else buys

it makes sense if you think about it: if the price drops below the mining reward of the coin, you can actually make money by renting hashpower

I expect they have proprietary miner that choose the most profitable coin to mine at any time. If nobody rented this particular user then his hardware is pointed to nicehash internal pool.

And I don't know what is this internal pool.

I never heard about any nicehash internal pool

I remember reading in their FAQ that if there are no buyers, you get nothing

Then i should synchronize my perception of nicehash with reality

in practice there are always buyers

I know for sure that miningrigrentals.com doesn't have any internal pool

cohcho: nicehash.com/support/mining-help/mi…at-happens-when-there-are-no-orders

call me paranoid, ever since i updated to xmrig to 5.4 the pc randomly shuts off after few minutes let idle, didn't happen with 5.3 as far as I can tell.

So both nicehash and mrr are just MITM that count valid shares without any internal pool. Then that 50MH/s from nicehash api can be attached to anything listed at miningpoolstats.stream or counted as unknown. No need to add it as separate pool

yes, exactly

solopool.org - are people seriously that lazy to set up their own node that they are OK to pay a 1.5% fee to solomine?

(0% fee, 32GiB + cpu load + some bandwidth) vs (1.5% fee , nothing)

with 15 MH/s, they pay ~500 USD per month in fees

There is inforamation whether those miners are using nicahahsh/mrr or their own hardware. In 1st case public ip is required

pool provides it too

Thre is info*

There is no info about source of mining power*

pool provides ip too*

you can easily rent some cloud server for much cheaper than the fee

so I'm still betting on laziness

miningpoolstats.stream seems to have stopped showing the number of miners

Their api provides json with poolsminers: data.miningpoolstats.stream/data/monero.js?t=1

still only 26k

+26k miners from nicehash api

22163 now*

good point

do we know how many miners were on nicehash before the fork?

Historical data from their api has only speed for cryptonight: nicehash.com/algorithm/cryptonight

nicehash miners don't really count for decentralization anyways

something around 30-50MH/s for cryptonight, and 3158 miners now

comparing to >20k with 50MH/s for randomxmonero now

The definition of "miner" tends to vary a fair bit. Some of the pools count a miner by address (Sxmr, nodejs-pool based pools), Nicehash does it by per-miner-connection I believe.

so they count the number of rigs and not miners

Yes.

tevador as long as container is not a full-blown VM, sniffer should detect it

RandomX inside a VM can't be detected outside by general-purpose sniffer because it uses separate set of registers.

I even think the sniffer can detect a miner running in a VM with KVM

if sniffer runs on the hypervisor

Same the opposite way. If virus/trojan runs as a supervisor and makes infected OS a VM, it can't be detected from within.

If sniffer is a hypervisor and knows about all VMs, it can examine them of course

Some rootkits use hypervisor mode to hide themselves, it's not a new technique. But it's far from a rookie technique.

we know there are ways to hide from the sniffer, but I think most malware won't bother

total miners 26399

i still get the total miners stat on miningpoolstats