Monero Mining Performance

Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.

MEMOOO

New Member
Jul 30, 2017
2
0
1
34
Thank u for reply my question!

1. Yes every processor
2. Two processor.
3. Which HashRate of gives Intel Xeon x5650 XMR MONERO

Thanks you for replaying ..

1- So "2x Intel Xeon E5-2670 V1 = 829H/s" mean "two of this Processor = 829H/s" .. ?!
3- i see many servers have many " Intel Xeon x565o inside " with nice price so i want to know hashrate for this CPU .. ?!
 
Last edited:

Hanatos

New Member
Jul 30, 2017
3
1
1
30
Two processor's. I need hasrate for two processors.
2. Maybe :)

Thanks you for replaying ..

1- So "2x Intel Xeon E5-2670 V1 = 829H/s" mean "two of this Processor = 829H/s" .. ?!
3- i see many servers have many " Intel Xeon x565o inside " with nice price so i want to know hashrate for this CPU .. ?!
 
  • Like
Reactions: MEMOOO

spfoo

Member
May 23, 2017
102
16
18
Mostly cache size and frequency driven with little on IPC. It makes sense from that perspective but still lower than I would want to see.
Just makes me wonder how AMD and Intel are going to make money in the future. Upgrading CPUs doesn't seem very interesting as there's not much to be gained. Well on the other hand we are only looking at mining performance.
 

Patrick

Administrator
Staff member
Dec 21, 2010
12,519
5,825
113
Xeon Bronze 3104 (6 cores / 6 threads), L2 + L3 cache 14.25MB

H/s by number of threads used:

6 = 187H/s
7 = 183H/s
8 = 176H/s

I am starting to think, at least on the low end, that the exclusive cache means that you can use less effective cache space for XMR mining.
 

spfoo

Member
May 23, 2017
102
16
18
12-core AMD 6234 for example. I got 16 of those with heatsinks on Ebay for about 270 USD. Just the heatsinks for these are typically 30 USD a piece.
Here is a dual AMD Opteron 6234 in action. It's a 12-core with 12+16MB L2/L3. Released in 2011. It's not optimized for threads yet since the hs variance is odd. Anything from 18-28 threads get 700+. Need more and longer testing. Using my tuned version of Wolf's miner.



 

madLyfe

Member
Jul 18, 2017
48
4
8
44
Here is a dual AMD Opteron 6234 in action. It's a 12-core with 12+16MB L2/L3. Released in 2011. It's not optimized for threads yet since the hs variance is odd. Anything from 18-28 threads get 700+. Need more and longer testing. Using my tuned version of Wolf's miner.



are you using a SM board? i have the same readout for lm-sensors and im not sure how to read it. im new to nix in the last couple days.

here is what mine looks like:
putty_2017-08-01_10-37-25.png
 
Last edited:

spfoo

Member
May 23, 2017
102
16
18
are you using a SM board? i have the same readout for lm-sensors and im not sure how to read it. im new to nix in the last couple days.

here is what mine looks like:
View attachment 6274
Yes that's a SM board. I take it the watts are per socket/CPU and temps are per die. Those Opterons have 2 dies per CPU. Not sure how accurate the watts are. On my UPS I get 290W consumption for the MB in full tilt.
 
  • Like
Reactions: madLyfe

spfoo

Member
May 23, 2017
102
16
18
Yes that's a SM board. I take it the watts are per socket/CPU and temps are per die. Those Opterons have 2 dies per CPU. Not sure how accurate the watts are. On my UPS I get 290W consumption for the MB in full tilt.
By the way I had 2 of those boards originally. One broke down. I guess they are pretty much end-of-life now after having served some data center 24/7 for 5+ years. Probably same goes for the CPUs. Fun toys though.

I've noticed that Intel CPUs seem to be generally much better in power consumption compared to AMD. That explains the difference in price.
 

madLyfe

Member
Jul 18, 2017
48
4
8
44
By the way I had 2 of those boards originally. One broke down. I guess they are pretty much end-of-life now after having served some data center 24/7 for 5+ years. Probably same goes for the CPUs. Fun toys though.

I've noticed that Intel CPUs seem to be generally much better in power consumption compared to AMD. That explains the difference in price.
what xeons would you rather run over opterons?
 

spfoo

Member
May 23, 2017
102
16
18
what xeons would you rather run over opterons?
Just looking at Patrick's listing on page 1 the E5-26xx seems to be very good in mining and power consumption seems to be around 10% lower than 62xx Opterons. That can make the difference between profitable/non-profitable. I'm just basing this on Intel TDP-numbers and what I know of the Opteron power consumption.

I haven't played with the newer 63xx Opterons, but power wise they seem to be the same as the 62xx-series. 63xx seems hard to find as well. Either it's not being upgraded yet or Intel took the market completely after the 62xx Opterons.
 
  • Like
Reactions: madLyfe

Patrick

Administrator
Staff member
Dec 21, 2010
12,519
5,825
113
Xeon Gold 6132 (14 cores / 28 threads), L2 + L3 cache 33.25MB

H/s by number of threads used:

12 = 648H/s
13 = 669H/s
14 = 691H/s
15 = 670H/s
16 = 640H/s

Not bad, not great. EPYC will be much faster in this range.
 
Last edited:

Patrick

Administrator
Staff member
Dec 21, 2010
12,519
5,825
113
Xeon Silver 4112 (4 cores / 8 threads), L2 + L3 cache 12.25MB - note this is a part optimized for low core counts

H/s by number of threads used:

4 = 199H/s
5 = 215H/s
6 = 195H/s

At this point, I am so accustomed to much higher core count systems that 4C/ 8T seems really low.
 
Last edited:

Nero24

Member
Jul 26, 2017
32
5
8
48
www.planet3dnw.de
Threadripper 1950X ist around 1,180 H/s with xmr-stak-cpu under Windows 10 x64 Pro. But keep in mind, that Threadripper is a HEDP CPU, tuned for max. performance, not for max efficiency. A Threadripper 1950X system consumes more power 215 w running 16 Monero threads) than two Ryzen 7 1700 systems (2x 85 W = 170 W) and is much more expensive than two Ryzen 7 1700. So perhaps it's more useful to buy two Ryzen 7 1700 systems than one Ryzen Threadripper 1950X for Monero :)
 

amalurk

Active Member
Dec 16, 2016
313
117
43
102
It would be interesting to compare h/s per watt as measured at the outlet with otherwise minimal system config. It is probably unrealistic to get that for all these but you really should start adding at least the watts while mining for any new ones so at least some rough comparisons can be made.