mersenneforum.org  

Go Back   mersenneforum.org > Great Internet Mersenne Prime Search > Hardware > Cloud Computing

Reply
 
Thread Tools
Old 2019-04-22, 23:45   #1
kriesel
 
kriesel's Avatar
 
"TF79LL86GIMPS96gpu17"
Mar 2017
US midwest

10011100111012 Posts
Default seasonal or long term trends

Hi,
I'm wondering if those who use cloud computing frequently have any data about seasonal or long term trends in availability or cost per unit throughput for cloud computing. In home-sited GIMPS engines, whether cpu or gpu primarily, there's a considerable difference in net cost per primality test, depending on whether the waste heat is is a benefit, helping heat the home while needed, or a drawback, increasing usage of central air conditioning. This effect is large enough to swing the run/stop pendulum on some of my older equipment.

If there's less seasonal effect on cloud pricing, it may make sense to switch some throughput from home to cloud for the summer, and back again when wanting the heat.
kriesel is online now   Reply With Quote
Old 2019-04-22, 23:54   #2
chalsall
If I May
 
chalsall's Avatar
 
"Chris Halsall"
Sep 2002
Barbados

7·1,361 Posts
Default

Quote:
Originally Posted by kriesel View Post
If there's less seasonal effect on cloud pricing, it may make sense to switch some throughput from home to cloud for the summer, and back again when wanting the heat.
If you want to actually be fully "green", you would do no compute. Or, at least, unless you needed heat to survive.

Clearly, that's not going to happen. So what is the compromise?
chalsall is online now   Reply With Quote
Old 2019-04-23, 00:36   #3
retina
Undefined
 
retina's Avatar
 
"The unspeakable one"
Jun 2006
My evil lair

29×211 Posts
Default

Quote:
Originally Posted by chalsall View Post
If you want to actually be fully "green", you would do no compute. Or, at least, unless you needed heat to survive.
To extend the "fully green" thing, then the compute heat generation in a cold climate is also a waste since heat pumps (aka A/C) would be more efficient. So in terms of only "being green" there is no justification for running any unnecessary compute. But of course that would make the world a much less interesting place, and there is more to life and survival than simply being as green as possible.
retina is online now   Reply With Quote
Old 2019-04-23, 00:44   #4
kriesel
 
kriesel's Avatar
 
"TF79LL86GIMPS96gpu17"
Mar 2017
US midwest

502110 Posts
Default

Quote:
Originally Posted by chalsall View Post
If you want to actually be fully "green", you would do no compute. Or, at least, unless you needed heat to survive.

Clearly, that's not going to happen. So what is the compromise?
Upper midwest. Heat matters, definitely. I'm pretty green, with my 58mpg hybrid and short trips, and well insulated home, heated other than by GIMPS mostly with wood I cut myself. (On the coldest winter days I might turn the gas furnace on or run the wood stove full throttle or both. Surely you miss those crisp invigorating days down there in hot muggy Barbados.) But that's not what the question was about.

The question is about whether there is a seasonal variation in the economics of cloud compute pricing. Does hot weather, or some combination of seasonal factors, raise the price per unit throughput of cloud computing for GIMPS? Generally there is seasonal pricing of electrical rates. There could also be more energy required for cooling the gear. There could also be seasonal variations in demand for compute.
Quote:
Originally Posted by retina View Post
But of course that would make the world a much less interesting place, and there is more to life and survival than simply being as green as possible.
Supposedly a substantial function of sleep is to reduce energy requirements of an animal, especially predators. It makes the caloric budget easier to balance. Predators sleep a lot. Prey sleep very little, since a lot of time is spent gathering low calorie density low protein density food (grazing or foraging), and letting a predator sneak up on you while napping is dangerous. Omnivores' sleep schedule is intermediate.

Last fiddled with by kriesel on 2019-04-23 at 00:51
kriesel is online now   Reply With Quote
Old 2019-04-23, 00:52   #5
chalsall
If I May
 
chalsall's Avatar
 
"Chris Halsall"
Sep 2002
Barbados

7·1,361 Posts
Default

Quote:
Originally Posted by retina View Post
To extend the "fully green" thing, then the compute heat generation in a cold climate is also a waste since heat pumps (aka A/C) would be more efficient.
LOL. A common mistake. Cold can't be created; heat can be moved, but at cost.

To share, I burnt about 2g of hydro-carbon gas this morning to bring about 100 cc of water to boil (~100 C), which made me two cups of wonderful coffee.
chalsall is online now   Reply With Quote
Old 2019-04-23, 01:10   #6
kriesel
 
kriesel's Avatar
 
"TF79LL86GIMPS96gpu17"
Mar 2017
US midwest

5,021 Posts
Default

Quote:
Originally Posted by chalsall View Post
LOL. A common mistake. Cold can't be created; heat can be moved, but at cost.

To share, I burnt about 2g of hydro-carbon gas this morning to bring about 100 cc of water to boil (~100 C), which made me two cups of wonderful coffee.
You appear to be finding fault with retina for an error you imagine he made. Heat pump is the correct term, and implies doing what you wrote, "heat can be moved, but at cost." Thermal energy is moved from a lower temperature stream to a higher temperature stream by application of work and capital equipment. My central AC is a heat pump configured only to move heat in the opposite direction, providing cooling effect to the interior as in the summer. The real high efficiency heating deal is a small internal combustion engine generator set built to last 20 years, natural gas fueled, that drives a heat pump, and recovers most engine exhaust heat in a heat exchanger for additional heating. I understand the Japanese make use of these. But, the capital cost is high. There are also heat pump based water heaters. For the engine driven heat pump, the engine may produce around 1/3 shaft work, 1/3 hot exhaust, and 1/3 engine cooling requirement from the incoming fuel energy. Assuming a COP around 3 or 4, the heating effect can be about twice the fuel energy. 1/3x 3 or 4 +1/3x0.9 +1/3x1 =~ 5.6/3 or 5.9/3.

The question for this thread is about whether there is a seasonal variation in the economics of cloud compute pricing. Anything useful to contribute?

Last fiddled with by kriesel on 2019-04-23 at 01:15
kriesel is online now   Reply With Quote
Old 2019-04-23, 01:11   #7
retina
Undefined
 
retina's Avatar
 
"The unspeakable one"
Jun 2006
My evil lair

611910 Posts
Default

Quote:
Originally Posted by chalsall View Post
LOL. A common mistake. Cold can't be created; heat can be moved, but at cost.
I don't see where my mistake was. Did I suggest that cold was being created, or that heat pumps were free to operate?
retina is online now   Reply With Quote
Old 2019-04-23, 01:17   #8
kriesel
 
kriesel's Avatar
 
"TF79LL86GIMPS96gpu17"
Mar 2017
US midwest

5,021 Posts
Default

Quote:
Originally Posted by retina View Post
I don't see where my mistake was. Did I suggest that cold was being created, or that heat pumps were free to operate?
Perhaps because the error was Chalsall's?
kriesel is online now   Reply With Quote
Old 2019-04-23, 01:23   #9
chalsall
If I May
 
chalsall's Avatar
 
"Chris Halsall"
Sep 2002
Barbados

7×1,361 Posts
Default

Quote:
Originally Posted by retina View Post
I don't see where my mistake was. Did I suggest that cold was being created, or that heat pumps were free to operate?
Ah, man... I'm sorry if I caused offense.

The laws of thermal dynamics suggest that entropy and enthalpy are intrinsically interconnected.

It then follows that those who use compute, which generates heat, could then use it for other purposes if they then so chose. Or, alternatively, have to deal with it if not taking into consideration.
chalsall is online now   Reply With Quote
Old 2019-04-23, 01:41   #10
retina
Undefined
 
retina's Avatar
 
"The unspeakable one"
Jun 2006
My evil lair

611910 Posts
Default

Quote:
Originally Posted by chalsall View Post
Ah, man... I'm sorry if I caused offense.
I didn't take offence. You would have to try much harder to achieve that.
Quote:
Originally Posted by chalsall View Post
The laws of thermal dynamics suggest that entropy and enthalpy are intrinsically interconnected.

It then follows that those who use compute, which generates heat, could then use it for other purposes if they then so chose. Or, alternatively, have to deal with it if not taking into consideration.
Okay, so the "other use" for the heat is to heat the living space for human comfort. But a heat pump (in good repair) will be more efficient than computing for that purpose. Or perhaps I missed your point entirely and you are suggesting something else? If you were to post more plainly I could understand you better.
retina is online now   Reply With Quote
Old 2019-04-23, 02:06   #11
kriesel
 
kriesel's Avatar
 
"TF79LL86GIMPS96gpu17"
Mar 2017
US midwest

5,021 Posts
Default

Quote:
Originally Posted by chalsall View Post
The laws of thermal dynamics suggest that entropy and enthalpy are intrinsically interconnected.

It then follows that those who use compute, which generates heat, could then use it for other purposes if they then so chose. Or, alternatively, have to deal with it if not taking into consideration.
Dude, quit while you're behind. ;)
Thermodynamics is the word you're looking for.
https://en.wikipedia.org/wiki/Thermodynamics
Thermal Dynamics is a company. http://www.thermaldynamics.com/
We've been through the energy balance and economics for user-owned gear on other threads already.
And based on decades of familiarity with various alternative energy technologies, the economics, and basic principles (Carnot) I'm confident there's not a cost effective way to reclaim available energy from a workstation's warm air exhaust to help power it.

Anything to contribute on-topic, toward the question about seasonal or long term variation in cloud computing cost for GIMIPS?
kriesel is online now   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Nucular Energy: Thoughts, Emerging Trends, Fears only_human Soap Box 121 2020-11-26 00:34
Workload trends SELROC Hardware 23 2019-02-21 22:55
Mprime performance trends SELROC Hardware 0 2019-01-16 09:22
Unfavorable price trends for AWS EC2 spot instances? GP2 Cloud Computing 14 2018-01-11 02:01
Long-term Primenet archive delta_t Data 3 2005-08-25 00:31

All times are UTC. The time now is 17:59.

Fri Apr 16 17:59:11 UTC 2021 up 8 days, 12:40, 0 users, load averages: 2.99, 2.75, 2.89

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.