-
hycsech1: your v2 branch doesn't build on my lichee pi paste.debian.net/hidden/f0921546
-
hycprobably due to v0.7 vector support
-
sech1hyc it needs gcc-14 or newer
-
hycah I only have 13.2. lemme see if I can update
-
hychow about clang?
-
sech1clang should work too. A recent version of course
-
hycI'm using the installed gcc13 to build gcc14 from source. could be a while
-
selstasech1: would the final version also require gcc 14? monero still supports gcc12 on master branch monero-project/monero #9669
-
sech1selsta if you compile for risc-v - yes
-
selstai missed the riscv context
-
selstaall good in this case
-
sech1But risc-v boards generally have newer compilers because they're new
-
m-relay<gingeropolous:monero.social> any reason the benchmark would give me 6788.14, whereas xmrig gives me 9121.8 h/s?
-
m-relay<gingeropolous:monero.social> and i keep trying to justify buying one of these intel core ultra, but buying a dead socket is just..... bluh
-
DataHoarderbenchmark with v2 tests different things, unless you are testing v1
-
m-relay<gingeropolous:monero.social> yeah those are v1 hashrates from the benchmark
-
m-relay<gingeropolous:monero.social> oh might be the wacky affinity for the 1920x
-
DataHoarderhow many numa nodes for that? :D
-
DataHoardercan't remember if the benchmark cares about these and replicates dataset across each numa node
-
m-relay<gingeropolous:monero.social> well the --auto flag has interesting results. v1 = 7445.25 h/s, v2 = 7468.91 h/s.
-
m-relay<gingeropolous:monero.social> lulz nvm. v2 is 6715.3 . turns out -v2 not the same as --v2. who needs syntax checking anyways....
-
sech1Benchmark doesn't do numa
-
sech1You can set affinity, large pages, and enable msr mod via xmrig's script. But not numa