Damn damn double damn. Time to build a server at home with hexacores in it or something.
We need a wiki page for CPU hash rates on scrypt
Topic: How to mine CPU coins for free - page 2. (Read 21473 times)
We need a wiki page for CPU hash rates on scrypt
Damn damn double damn. Time to build a server at home with hexacores in it or something.
Has anyone done the math to see if mining Litecoin would be profitable on an Amazon EC2 High-CPU Extra Large (8 cores)? On-demand pricing is $0.68/hr and and spot pricing is $0.216/hr.
I've tried it. It says it has 8 cores, it had only 3...no really, just 3 cores. The Cluster Compute Quadruple Extra Large however has 8 real physical cores, not limited, however it's priced at 1.60 per hour.
Calculations for the free EC2, assuming a generous speed of 1.5 khash/s:
Using http://allchains.info/calc.html tells me that you get 3.961 coins per day, or if you use 750 hours, 750*3.961/24 = 123.78125 LTC
At the current exchange rate (~0.01 BTC/LTC) that gives 1.237 BTC, or (3.25 USD/BTC) about 4 dollars. In total.
Obviously, if you get the more expensive EC2, even with 8 cores, it's not worth the money, as you get about 4/750 = 0.0053 dollar per core per hour. The LTC/difficulty ratio would need to increase ~5 times for this to be beneficial, assuming you actually get 8 cores at $0.216/hr.
Using http://allchains.info/calc.html tells me that you get 3.961 coins per day, or if you use 750 hours, 750*3.961/24 = 123.78125 LTC
At the current exchange rate (~0.01 BTC/LTC) that gives 1.237 BTC, or (3.25 USD/BTC) about 4 dollars. In total.
Obviously, if you get the more expensive EC2, even with 8 cores, it's not worth the money, as you get about 4/750 = 0.0053 dollar per core per hour. The LTC/difficulty ratio would need to increase ~5 times for this to be beneficial, assuming you actually get 8 cores at $0.216/hr.
legendary
Activity: 1862
Merit: 1011
Reverse engineer from time to time
Has anyone done the math to see if mining Litecoin would be profitable on an Amazon EC2 High-CPU Extra Large (8 cores)? On-demand pricing is $0.68/hr and and spot pricing is $0.216/hr.
I've tried it. It says it has 8 cores, it had only 3...no really, just 3 cores. The Cluster Compute Quadruple Extra Large however has 8 real physical cores, not limited, however it's priced at 1.60 per hour. You guys do understand that you only get so much processor time/power per month under that free account. After the limit is passed, you will start getting billed.
LOL that was not mentioned in OP...
I've started to use my VPS for mining after reading this thread, its getting 1KH/s on one core, i have 4 available. I am limited by the machines unused cpu power down to a guaranteed share. I think it'll get faster at night.
Has anyone done the math to see if mining Litecoin would be profitable on an Amazon EC2 High-CPU Extra Large (8 cores)? On-demand pricing is $0.68/hr and and spot pricing is $0.216/hr.
You guys do understand that you only get so much processor time/power per month under that free account. After the limit is passed, you will start getting billed.
I tried march=native, it didn't do any better than just using SSE2 for me.
Specifying all the SSEs as well as AVX (I was using a 2600k) made a big difference.
You seem to have misunderstood the thread. It's about Amazon EC2 computing. I.e rented computers, not a personal computer.Specifying all the SSEs as well as AVX (I was using a 2600k) made a big difference.
If march=native doesn't detect high level SSEs and/or AVX, it doesn't detect it.
Doesn't matter what the computer is being used for.
Well, I got some speed bumps to 2.5kh/s all of a sudden. Then it goes back down to 0.4kh/s. If you run a large virtual environment at work like I do you would understand why it does this.
legendary
Activity: 1862
Merit: 1011
Reverse engineer from time to time
I tried march=native, it didn't do any better than just using SSE2 for me.
Specifying all the SSEs as well as AVX (I was using a 2600k) made a big difference.
You seem to have misunderstood the thread. It's about Amazon EC2 computing. I.e rented computers, not a personal computer. Specifying all the SSEs as well as AVX (I was using a 2600k) made a big difference.
I tried march=native, it didn't do any better than just using SSE2 for me.
Specifying all the SSEs as well as AVX (I was using a 2600k) made a big difference.
Specifying all the SSEs as well as AVX (I was using a 2600k) made a big difference.
legendary
Activity: 1862
Merit: 1011
Reverse engineer from time to time
yes, I have edited it at the same time, it should be artfortz.
Thanks for the tip for confgiure. I am getting up to 1.41 khash/s now, almost the same as my own i5:
Do you know what that is exactly? It's load balancing in effect. You for just a few seconds get a bump of that 1.30 or so speed, then get reduced to below 0.05 Thanks for the tip for confgiure. I am getting up to 1.41 khash/s now, almost the same as my own i5:
Quote
[2011-11-02 21:14:38] LONGPOLL detected new block
[2011-11-02 21:14:38] thread 1: 1664 hashes, 0.03 khash/sec
[2011-11-02 21:14:38] thread 0: 525 hashes, 0.03 khash/sec
[2011-11-02 21:15:06] thread 0: 1647 hashes, 0.06 khash/sec
[2011-11-02 21:15:07] thread 1: 1581 hashes, 0.06 khash/sec
[2011-11-02 21:15:08] thread 0: 3696 hashes, 1.41 khash/sec
[2011-11-02 21:15:10] thread 1: 3385 hashes, 1.22 khash/sec
[2011-11-02 21:15:10] thread 0: 2281 hashes, 1.16 khash/sec
[2011-11-02 21:15:10] PROOF OF WORK RESULT: true (yay!!!)
[2011-11-02 21:15:12] thread 1: 2707 hashes, 1.13 khash/sec
[2011-11-02 21:15:12] PROOF OF WORK RESULT: true (yay!!!)
[2011-11-02 21:14:38] thread 1: 1664 hashes, 0.03 khash/sec
[2011-11-02 21:14:38] thread 0: 525 hashes, 0.03 khash/sec
[2011-11-02 21:15:06] thread 0: 1647 hashes, 0.06 khash/sec
[2011-11-02 21:15:07] thread 1: 1581 hashes, 0.06 khash/sec
[2011-11-02 21:15:08] thread 0: 3696 hashes, 1.41 khash/sec
[2011-11-02 21:15:10] thread 1: 3385 hashes, 1.22 khash/sec
[2011-11-02 21:15:10] thread 0: 2281 hashes, 1.16 khash/sec
[2011-11-02 21:15:10] PROOF OF WORK RESULT: true (yay!!!)
[2011-11-02 21:15:12] thread 1: 2707 hashes, 1.13 khash/sec
[2011-11-02 21:15:12] PROOF OF WORK RESULT: true (yay!!!)
Just got mine working and yeah it's only 0.40-0.50 kh/s and obvious it's only a single vcore when you try 2 threads. Oh well, it's a virtual machine shared with thousands of users.
legendary
Activity: 1862
Merit: 1011
Reverse engineer from time to time
You need all the SSEs in there, not just SSE2. It makes a huge difference.
Thats why i do -march=native -mtune=native which enables them all, -msse2 is just a habit. In any case, ALL i got and i mean from ALL the optimization i tried, just 0.50kh/sAnd i also use a 64bit linux, which is supposed to be faster than 32bit OS
Not to mention that there is extreme load balancing in the micro instance. You mine at full speed for a few seconds and then you get reduced to 0.02 or less KH/s
Also, with that exact same miner, i get over 3KH/s per core on my 64bit Virtual machine on my own computer.
You need all the SSEs in there, not just SSE2. It makes a huge difference.
Thats why i do -march=native -mtune=native which enables them all, -msse2 is just a habit. In any case, ALL i got and i mean from ALL the optimization i tried, just 0.50kh/sAnd i also use a 64bit linux, which is supposed to be faster than 32bit OS
Not to mention that there is extreme load balancing in the micro instance. You mine at full speed for a few seconds and then you get reduced to 0.02 or less KH/s
What CPU do you have?
FWIW, I used:
FWIW, I used:
Code:
CFLAGS="-march=core2 -O3 -Wall -msse2 -msse3 -msse4 -msse4.1 -msse4.2"
and I'm getting ~3 k/s legendary
Activity: 1862
Merit: 1011
Reverse engineer from time to time
You need all the SSEs in there, not just SSE2. It makes a huge difference.
Thats why i do -march=native -mtune=native which enables them all, -msse2 is just a habit. In any case, ALL i got and i mean from ALL the optimization i tried, just 0.50kh/sAnd i also use a 64bit linux, which is supposed to be faster than 32bit OS
Not to mention that there is extreme load balancing in the micro instance. You mine at full speed for a few seconds and then you get reduced to 0.02 or less KH/s
You need all the SSEs in there, not just SSE2. It makes a huge difference.
legendary
Activity: 1862
Merit: 1011
Reverse engineer from time to time
I am writing because i've already tried them.
Micro instance has 1 core, or when i compiled, just 0.50kh/s. And i doubt that i can launch 30 or more FREE instances.
I've also tried the CLuster Computer Quadruple extra large which is an 8 core, 16 thread Intel Xeon which does 2.77(MAX) kh/s per core
So no, you do not get 1.5kh/s just 0.50
And if anyone wants, here is something i use to automatically download all dependencies plus the miner, download it, configure it, and compile it
That is what i do when i first SSH to the new EC2 instance. It's fast, and you barely waste time.
Micro instance has 1 core, or when i compiled, just 0.50kh/s. And i doubt that i can launch 30 or more FREE instances.
I've also tried the CLuster Computer Quadruple extra large which is an 8 core, 16 thread Intel Xeon which does 2.77(MAX) kh/s per core
So no, you do not get 1.5kh/s just 0.50
And if anyone wants, here is something i use to automatically download all dependencies plus the miner, download it, configure it, and compile it
Quote
apt-get update;apt-get install -y libcurl4-openssl-dev;apt-get install -y git;apt-get install -y make;apt-get install -y autoconf;git clone https://github.com/Lolcust/Tenebrix-miner.git;mv Tenebrix-miner /opt/;cd /opt/Tenebrix-miner/;aclocal;./autogen.sh;CFLAGS="-O3 -Wall -g -ftree-loop-distribution -fstrict-aliasing -fstrict-overflow -march=native -mtune=native -msse2"
That is what i do when i first SSH to the new EC2 instance. It's fast, and you barely waste time.
Jump to: