Folding Forum

Wrish wrote:I tried to GPU fold, but I realized it was a huge power draw. The hardware outlay expenses might be lower than for a CPU system, but over the year following, electric costs far outweigh that. Of course, if you pay a very different amount for electricity, the equation could change.

Quite true, as my 'net power cost' for my 19 systems is $36.78 CAD per month (= $35.62 USD per month, at today's exchange rate). This is my 'electricity cost' (= $209.03 CAD per month) minus my 'savings on Natural Gas heating costs' (= $172.25 CAD per month). I live in a relatively cool (some would say cold) climate, so our main cost is Natural Gas for heating in the Winter. Nobody here (at least, nobody that I know or have heard of) has air conditioning for cooling in the Summer, because it is not necessary.

This is also not a 'true' comparison for the 'high-end GPU' systems, as my power cost includes quite a few systems that are designed to be sold to the 'average' user (I am an OEM of custom-built computers), and these systems are drawing power while contributing relatively little to my Folding@Home points scores.

If I limited my calculations to only the 'high-end GPU in a low-end CPU/Motherboard' type of systems, my 'net power cost' may even turn out to be negative.

Best regards,
Art

To hj47 and kiore:

This would really have been much better discussed in PMs, where we are not cluttering up the Forum here.

First of all, I am forwarding to you a PM that I sent to Grandpa_01, who did use a PM to correspond with me on this subject.

This PM should, I hope, answer all of your questions and concerns with this issue.

Also, please see my message to Wrish, that I just Posted here.

Thank You, and Best regards,
Art

I think that we got off topic here fellas.

I'm not interested in points per watt or dedicated system, my machine is part time only, and for gaming primary use, a High end GPU is a necessity for gaming, CPU is also desired, but not anything low end.

My question only was, Intel, or AMD? If more clocks = better output then AMD is by far the better solution $/mhz solution. However if Intel is better and clocks are not as good as "core optimizations" then cost goes out the window.

If some people could tell me their opinion on a part time machine single cpu client on the 1090T and someone else could tell me the same for the i7 980X, that'd be great, but let's not bicker about cost over watts and GPU vs CPU ppd output. This is all about CPU only, GPU is another ball game entirely.

Thanks again for your input! :)

The problem is the part time bit, which makes it hard to calculate. If you only wish to run the classic client then you should just buy whatever is good for you gaming, and in fact the quad you have is pretty good already. You see if you run SMP (multi thread) clients the calculation starts getting more interesting.
Anyway I stick with the advice that the faster the processor and more cores will be more work done faster.

The advantage of Intel's better SSE, plus hyperthreading, has typically put Intel in the lead for the past few years. (fah relies heavily on SSE)

Part time folding (16 hours a day or less, rough estimate) = CPU client. SMP won't make the short deadlines. Install mutliple CPU clients, one for each CPU core.

Get the clients running well, then convert to running them as a service. Out of sight, etc.

7im wrote:The advantage of Intel's better SSE

I thought this was due to Intel purposely slowing their compiler code for CPU's that aren't Intel x86 i.e. any AMD CPU?

I switched from AMD duals to Intel quads a few years ago because the size of the L2 cache was a big factor in SMP speed. This was well before the nVidia GPU client, so SMP was the way to go for max output.

Prior to Folding, I built AMDs for gaming. That's what I started Folding with. You go where the current tech leads you to maximize the activity you wish to perform.

Unless you want to build for running bigadv (i7 or dual cpu), an AMD dual core on a mobo with multiple PCIe slots filled with nVidia 2xx or 4xx gpus would be the way I'd go. Since I already have the hardware, I run SMP on 4 C2Q, bigadv on an i7, and the GPU client on 10 nVidia cards

uncle fuzzy you run pretty much dedicated folding machines, fine for those that will dedicate their machine for that purpose, I however only want one machine. When I get another I will retire my old one to a friend or family member that needs a new PC, it maybe a year or two old, but better than none at all, or something much older. I build my own custom PCs and will support the ones I give away, but do not expect them to run fah for me or anything in return, I help out because I can. If you can afford to build those and support their output, great, I'm happy for you, however because smp isn't something I can run 24/7 and if panda group wants more than 12hrs out of my system per day for smp I won't be able to oblige them. My best guess - I use about 4-6hrs weekdays and 12-16hrs weekend days. The single core versions will complete in about 14 hrs and give me a week to complete, plenty of time, too bad they can't allow more people to use smp, and make it simpler, I hear that v7 helps, but right now it's closed in-house alpha testing, if/when they release a public beta, I'll try it out.

Right now I do run separate single clients, and will continue to do so. I do not wish to make them run as a service, instead I put them in the startup folder, so I can shut them down when I play my games.

hj47 wrote:

7im wrote:The advantage of Intel's better SSE

I thought this was due to Intel purposely slowing their compiler code for CPU's that aren't Intel x86 i.e. any AMD CPU?

This is very interesting, do you suppose that you could tell me how this affects the 1090T vs the 980X thanks.

I started this insane "hobby" by installing the cpu client on my gaming machine- an Athlon XP 3000+. I folded when I wasn't gaming. I continued that way through a couple more builds to an Athlon 64 X2 6400+. If there was an nVidia gpu client at that point, I don't know if I would have switched to Intel. The 6400 was getting "iffy" on completing the newer SMP WUs, so I jumped. I still build to allow gaming, but now do very little of it, so I've put together a little farm that does fold 99% of the time.

I was just answering the question in the title. "Which brand cpu is better?" depends on your daily use and which F@H clients you want to run. For cost, a fast AMD dual or quad is plenty to run cpu clients and feed a gpu client. Hell, a slow cpu is plenty for the cpu client. I still run a few on my laptops once in a while.

"Better" is in the eye of the beholder.

At least 99% of what has been said here has been discussed numerous times on this forum. Everyone has their perspective and "best" depends on what factors you choose to include in your evaluation.

This forum is hardware agnostic. We are not interested in fanboys from either AMD or Intel getting into a shouting match about which is better. (Luckily this topic has not developed into that sort of thing.) In any case, the poll is meaningless because so few people participate and the ultimate answer to the original question: Which is better? is "YES" (both are better). Make your decision based on the primary goals of your machine which was stated as gaming. Install whichever you decide works best for gaming and FAH will use whatever resources you choose to donate.

I'm closing the topic. We are NOT a hardware forum. There are plenty of good hardware forums around. Please continue your discussion on one of them.