80M to 64 bits ... but not really reserved

[markr]

Originally Posted by petrw1

Yup I was the other you noticed and I too am taking a few hundred every week or so for each machine but it could be a while so feel free to keep grabbing there too if you want.

Thanks, petrw1. To keep things simple, I'll grab from the top end, say 80.8M & 80.9M if that's okay. For the last month or so I was giving the Athlons expiring exponents below 65M but that source is drying up now.

Actually, the user I noticed was "Sid & Andy" which I guess is someone else again.

[petrw1]

Originally Posted by markr

Thanks, petrw1. To keep things simple, I'll grab from the top end, say 80.8M & 80.9M if that's okay. For the last month or so I was giving the Athlons expiring exponents below 65M but that source is drying up now.

Actually, the user I noticed was "Sid & Andy" which I guess is someone else again.

That's me. We started naming our computers after Toy Story Characters last fall so decided to change our user name in the standings to the Toy Story Kids.

Every now and then a few in the 60's or 70's come free.

I'll keep working from the low end as long as I have these old PCs.

[petrw1]

BTW in this 80M range are still 19,441 at 63 bits unassigned.
PLUS I have 1,134 assigned to my 3 old dogs one of which is sporadic.
It will be about 2 weeks until these are done and I am going for more.

[chalsall]

Originally Posted by garo

Yes it is a shame that people with heavy hardware are doing LMH instead of some other more urgent worktype like P-1. LMH should be left for old hardware. I have two Athlons and a P-III 600 on LMH.

To speak to this (as someone with a few cycles doing LMH work)...

It could be argued that the amount of energy required to run a low-end machine to do LMH work is sub-optimal. In the bigger picture, perhaps those old machines should be retired if they're only being kept alive to do LMH. Or, if they're actually doing other things (acting as routers, file servers, etc), perhaps they could be redirected to do DCs when <65 bit LMH work is exhausted.

P-1 work requires a great deal of memory to be done effectively, which cannot always be made available.

I personally use mprime to generate a small, but constant and predictable, amount of web traffic from those machines I am responsible for, and to monitor if suddenly a machine is unexpectedly very busy doing something else...

Please let me know if anyone disagrees with my arguments / position above.

[cheesehead]

Originally Posted by chalsall

It could be argued that the amount of energy required to run a low-end machine to do LMH work is sub-optimal. In the bigger picture, perhaps those old machines should be retired if they're only being kept alive to do LMH. Or, if they're actually doing other things (acting as routers, file servers, etc), perhaps they could be redirected to do DCs when <65 bit LMH work is exhausted.

A logical, but incomplete argument.

Those of us running a low-end machine may also be in the group of those of us having low amounts of investment capital.

You want to give me a more-power-efficient machine, in trade for my nowadays-low-end energy hog? Fine -- I'll swap.

You think it's trivial for me to lay out the cash to purchase a new higher-end system? No -- to some of us, that capital investment cost is not trivial at all.

Sure, I'd recoup some of that investment in energy savings, but how long 'til payoff? (And, if what actually happens is that the new system uses the same energy as the old, to produce its faster operation as a more-energy-efficient system, there's no actual operating cost savings, is there?)

I personally use mprime to generate a small, but constant and predictable, amount of web traffic from those machines I am responsible for, and to monitor if suddenly a machine is unexpectedly very busy doing something else...

Excellent idea.

[petrw1]

Originally Posted by chalsall

It could be argued that the amount of energy required to run a low-end machine to do LMH work is sub-optimal. In the bigger picture, perhaps those old machines should be retired if they're only being kept alive to do LMH. Or, ... perhaps they could be redirected to do DCs when <65 bit LMH work is exhausted.

P-1 work requires a great deal of memory to be done effectively, which cannot always be made available.

Please let me know if anyone disagrees with my arguments / position above.

In my personal observations ...

Yes, I wish I could replace all my low-end machines with Quads ... but there is the $$ issue. But since I can't right now, what do I do:

Since I have it anyway and since power is realtively cheap where I live that is not a major concern.

That being said I find that my PIII 866Ghz and my Duron 1300Ghz are relatively VERY good at LHM, if I keep them to a max of 64 bits. I get 1 Ghz-Day per day with the Duron and about 0.6 with the PIII. This drops to 0.4 and 0.2 above 64 bits. So if GIMPS runs out of sub 65 before I upgrade these machines (I feel pretty safe) I will likely toss them at that time.

I "timed" my PIII at 90 days for a 21M DC ... again about 0.2 points per day; hardly worth the power then.

To compare, my PIV 3.4 doing ECM-F also only contributes 1 point per day and my PIV 2.4 and 2.8 contribute about 1.2-1.3 per day doing DC.

[chalsall]

Originally Posted by cheesehead

You think it's trivial for me to lay out the cash to purchase a new higher-end system? No -- to some of us, that capital investment cost is not trivial at all.

But... With respect, it is not my responsibility to ensure that you can continue to contribute to GIMPS (efficiently or inefficiently).

If you can't contribute to GIMPS efficiently (read: as a side effect of other uses of energy), then perhaps you shouldn't be doing so (from a global energy / carbon emission POV).

No disrespect intended. But my above is considered.

[chalsall]

Originally Posted by petrw1

In my personal observations ...

Thank you for your below / above. I hear you.

My concern is that it appears that several very old machines are being kept "alive" doing nothing but LMH work.

In my opinion, it doesn't matter how cheap power is where you are (and trust me -- we here in Barbados pay more than just about anywhere else) -- at the end of the day it still comes down to the amount of KWHs being consumed doing this work.

Perhaps it is simply time to retire old, inefficient machines doing nothing but LMH work, simply and only because of the tonnes of carbon emitted doing this work.

Thoughts?

[TimSorbet]

I'd guess that carbon emissions aren't usually high on the list of DCers worries. Some crunchers may (whether you think this a bad move or not) keep an old machine alive, inefficient as it is, because they'd rather keep up the cost of the electricity than pony up to get a whole new computer or lose what work it can do.

[garo]

Fair enough. I find LMH useful for keeping track of machine health too. I do that for my parents' machine. At the end of the day, any contribution to GIMPS should be welcomed because something is better than nothing.

[chalsall]

Originally Posted by garo

At the end of the day, any contribution to GIMPS should be welcomed because something is better than nothing.

Please don't misunderstand me...

If an old, slow and inefficient machine is doing something else while it works on GIMPS, then great.

But, iff (if and only if) an old, slow and inefficient machine is being kept alive for no other reason that working for GIMPS, then perhaps, for the greater good, it is not "better than nothing".

With respect. We all make our own decisions, from our own points of view.