-
sech1
-
selsta
all 8 cores have the same hashrate?
-
sech1
yes, this is weird
-
sech1
xmrig is trying to figure it out now (it's his M1)
-
sech1
2300 h/s @ 13.1 watts (CPU package only)
-
selsta
how does that compare to Ryzen from efficiency side?
-
sech1
Better than Ryzen. The best Ryzen 3900X rigs are able to do something like 12 kh/s, 100W at the wall which would put CPU at 75W, so ~160 h/s/W (CPU package)
-
sech1
M1 does ~175 h/s/W
-
sech1
16- and 32-core M2 next year will be RandomX monsters
-
sech1
well, maybe not monsters, but I expect 3900X levels with lower power consumption
-
nioc
My 3900X gets 15,200 at I think 80w for cpu
-
sech1
nioc check "CPU Package" line in HWInfo
-
nioc
I will wen cat gets off lap :)
-
sech1
CPU package is usually 20-25W more than just CPU
-
nioc
Ok
-
sech1
It's I/O die that adds these watts
-
nioc
sech1 cpu + soc = 93W
-
nioc
package = 105W
-
sech1
Not bad. So M1 ~ Ryzen in efficiency
-
nioc
not at most efficient cpu frequency
-
nioc
4.0 GHz
-
sech1
When I did experiments with Ryzen, the most efficient was 0.9V on CPU and highest stable frequency at that voltage - usually ~3.7 GHz
-
sech1
and running DDR4 at 2400 MHz with lower voltage on SoC also helps
-
nioc
the only comparison I have is 4.1 GHz was 10 watts more at the wall
-
hyc
8 threads uniform could just be the OS scheduler shuffling things around. does that ench run with pinned threads?
-
hyc
eh wait, that USer Agent says Intel Mac OS X
-
sech1
It also says L2: 32.0 MB for some reason. I guess it's just hwloc glitches with this system
-
hyc
gmaxwell is apparently accusing us of profiting with insider knowledge off the transition to RandomX. what a crock.
-
hyc
also claims a bunch of 7742s hit the market after the randomx switch that were certainly mining
-
hyc
big deal. server chips are much more expensive than desktop chips, would have been a waste of money
-
hyc
as for insider knowledge, none of the RandomX work was hidden from the public so wtf is he talking about
-
hyc
<gmaxwell> oh no, I was saying that I'm confident that people exploited the change for profit. Whats impossible to know is if the people pushing the change (e.g. Howard Chu) were a part of that.
-
sech1
gmaxwell who?
-
moneromooo
Well, that miners will try to make as much profit of mining is kinda the point anyway.
-
hyc
gmaxwell is a BTC dev
-
hyc
he invented CT
-
sech1
Why would he care even?
-
sech1
Did he have 1000 Vega farm on Monero?
-
hyc
has to keep up the myth that ASIC-resistance is futile
-
hyc
otherwise all of BTC devs look like fools. which they obviously are.
-
asymptotically
you people are disgusting! wanting miners to get block rewards instead of shady asic manufacturers
-
sech1
hyc is apparently accusing BTC devs of being fools!
-
hyc
:D
-
sech1
anyway, what insider knowledge is he talking about? RandomX was public since December 2018
-
hyc
yeah that makes no sense at all
-
sech1
-
sech1
Even November 1st, 2018 :D
-
hyc
and we had announced our CPU-centric goals back in March 2018
-
hyc
the prototype randprog and randomjs repos were all public too, all the ongoing work process and discussions were public. so wtf.
-
selsta
he does not seem well informed