Dear All,
I am new to Magicoin mining and I recently started mining again after.. I dunno... forever?! (I was mining at the age when bitcoin CPU mining was still on the brink of profitable and GPU bitcoin with an ATI 3800 was good enough to get you good bucks even at absymal bitcoin value. end of 2012 or something?)
Now I am here again... I was mining for fun while writing the thesis back then... doing something similar now I don't have any pretenses of earning huge amount of money, at least not in the short run.
But maybe to learn a thing or two about my rig and about mining and hashing, as back then.
My doubt is regarding Hash/s values. I read some of you and in other threads pulling ~100 kH/s on i5 or i7.
I currently have a dual socket rig, with a Xeon broadwell generation 10 cores, and 128GB of RAM and DDR4 quadruple channel...
Even per socket I would expect to hash a little bit better than an average 300$ i5, but maybe I'm wrong and we are all memory or bus-capped...
But nonetheless I have a dual socket system, so I would expect to hash at least double the amount and go 200kH?
I tried to mine with two workers, performance drastically plummetted... probably could not force it to separate the load and socket cross-talk killed me.
What do you reckon?
xeon processors are good for server tasks. they often have many cores but not such a high speed. i've been disappointed by xeon mining as well. but truth is that a good i5/i7 processor is more worth it than an expensive xeon (unless u got some really high end ones for cheap/free).
Thank you for the answer deztroyr1!
I'm not trying to do a moneybench here. I use the Xeon usually for other stuff of course... I am a scientist and usually I do simulations of atomic nuclei, if you're curious, so let's say that I got them for free. But ATM I don't have the time to lunch other simulations since I'm busy writing/reading/vacationing... so my hardware sits in idle...
I do not do it for the money either... I mainly do it to understand the hashing, the performance/compilation, the blockchain and cryptocurrency economy from the scientific perspective, and how it has evolved in this years so to speak.
So from the purely performance point of view my
two E5-2630v4, 10 core HT 2.2GHz, 3.1GHz turbo, and 25 MB cache each with DDR4 2666GHz should pulverize a
i5-6600 3.30 GHz, turbo 3.90 GHz, with DDR3 1666GHz or similar, and 6MB cache as it happens for physics simulations...
Unless hashing algo is severly single core.
If that's the case, running two workers should give me a factor two improvement on my dual socket (?), and running 40 threads should give me not much improvements, that's not what currently happens.
even if scaling is not linear there is definetely some scaling, while running two workers instead of one decreases the performances.
I can try some tests. Do you have some guidance or guide to link me to?
Show Posts
