Poll: Future Of The SMx Project
profdlp
The Holy City Of Westlake, Ohio
This project began in January of 2002. As hard as it is to believe, in just over a month we will mark the fifth anniversary of our Team 93 combined effort to boost the Folding@Home research program. During that time, some of the computers which were considered viable contributors at the time have slipped into near obsolescence. I doubt that very many of you who are reading this still use the same computer you had five years ago.
The question before us now is whether we should press on by adding new rigs, put our efforts into upgrading the older rigs in our lineup, or a combination of both.
This is a NON-BINDING poll. The whole idea right now is to take the pulse of the team, not to decide a particular strategy at this point in time. Voting has been made public, since quite frankly we are going to give more weight to the opinions of those who have been the most active in supporting the project and the team in the past.
If you'd like to be involved in helping us begin actively working toward getting our SMx Project back in high gear, we would really appreciate your input. Any parts you have which you would consider donating to the project would also be much appreciated. Just send me a PM and I'll be happy fill you in. Our critical needs are Motherboards, Memory, and CPUs.
The question before us now is whether we should press on by adding new rigs, put our efforts into upgrading the older rigs in our lineup, or a combination of both.
This is a NON-BINDING poll. The whole idea right now is to take the pulse of the team, not to decide a particular strategy at this point in time. Voting has been made public, since quite frankly we are going to give more weight to the opinions of those who have been the most active in supporting the project and the team in the past.
If you'd like to be involved in helping us begin actively working toward getting our SMx Project back in high gear, we would really appreciate your input. Any parts you have which you would consider donating to the project would also be much appreciated. Just send me a PM and I'll be happy fill you in. Our critical needs are Motherboards, Memory, and CPUs.
0
Comments
I think if any upgrade path is chosen we eliminate the systems that use parts that are no longer available at retail. I think that is about 40% of out SMx rigs. No matter how we do it, team 93 will gain in points and all will be good
I've noticed we've had less people applying to be hosts, so let's use this time build up interest and get cost-effective folding points
Well said!
Perhaps the very old units could be canibalized to give some of the other older ones a bit more productive life, saving on cost put towards new rigs.
Prof, how accurate is the SMX rigs list? How did the last round of upgrades go?
I only ask because SM6 is listed as a 2.5ghz. celery. (It's not :bigggrin: )
Gargoyle summed up my feeling "...cost effective folding points".
That's a great point. We don't want to replace an entire rig if we can help it, and we have a number of older rigs which have motherboards capable of a decent CPU upgrade. A simple pop-and-drop of the processor would bring them up to current standards without any further ado. Any of you folks that have an old Athlon XP in the 2000+ to 2500+ range that you aren't using and that you don't need could singlehandedly help us turn a poky old computer into a fairly robust Folding machine.
There are a few computers in our lineup which would need a new MB to support a better CPU, but I'd guess that some of our members have a still-useful old board laying around that will do nothing but gather dust until the day they do their Spring Cleaning and toss it in the trash.
In my opinion it would be a money pit to 'upgrade' older machines. Because about the only worthwhile upgrade that you are gonna do to them is to practically build a new system. So why not just leave them alone and build newer machines. Granted some money could be saved by sending the new hardware to the existing hosts as a result of not having to buy another case I want to see the SMx project continue to expand and fulfill its mission rather than drift into 'Maintenance Mode".
I know that building these machines takes money and/or parts, but what would be cool is if someone could code a bar graph of some type that would keep track of donations, and how far away we are from building the next rig. That may give people more encouragement to donate if they could see exactly how much of a need there is, and that they are indeed making a difference. and maybe split the bargraph into 6 or 7 parts, depending on what goes in one (mobo, proc, mem, NIC, CD/DVD,...) that way if one of those parts gets donated, it could be set to 'paid for' on the graph. I'm aware that the idea needs refining, but I believe that it could work.
Otherwise, new motherboard/cpu combos would probably be best directed to a new rig. SM11 is about as good of a Socket A system as it can be, and SM24 is the best P3 system it can be. Some of the other rigs have room for drop-in upgrades, though.
I don't see GPU folding be prudent to direct funds into, yet. I'm all in favor of the SMx committe buying me a PS3, however
Sorry to say, but I see hosting the bigger issue. I don't host. I am running two machines at home on dial-up now. Of course I have three of my own machines in my office at work using their power and internet connection.
Do we have people that can/will host more machines?
As far as the old old rigs, I think it would be smarter to just cut our losses and either let them keep folding if it is economical/efficient from an electrical usage point of view, or strip them of all useful parts (such as case..etc.) and then do something with the other parts, trash them or donate them to something that could still use them.
Personally, I'm entirely in favor of liquidating our old equipment stocks and setting a minimum standard and upgrading all rigs that fall below that level. It's a waste of the host's time and electricity to run an un-efficient folding machine, and I think we owe it to them to make sure their resources are well used. In fact, I'm entirely in favor of doing an annual upgrade... maybe make January "upgrade the project" month where anyone below the new standards gets a significant upgrade. (To be clear, I mean upgrade anyone below the level to a high level such that they will not need an upgrade for several years, not continually dragging a large number of machines up one notch).
I definitely think we can start rolling out some new rigs come spring, but for now I'd like to see us maximizing our current resources
Well said Mr. Keebs
What I mean is that lets say we have a P3 rig that is as maxed out as its gonna get, about the only thing left to do is to drop a new mobo/cpu in. Which I'm not against. As we can't let the rigs lay to waste. But at the same time I want to see New rigs roll out. I can understand the amount of dedication, time, and money it takes to get the rigs rolled out, but I think we should have a goal to shoot for. Maybe a rig every 2-3 months?
It is more of a short term option though. The rigs will become obsolete faster than building new ones with newer gen hardware.
However a valid point has been raised.... There have been far less volunteers of late and I think that's because the SMX project was producing rigs at such a rate of knots at one stage, that almost every member of good standing, who qualified for a rig and was willing to host one, got one!
We will always need quality people to host and if supply outstrips demand that may not always happen.
And then of course there's the age old business principle...It's always easier and more cost effective to service existing clients than look for new ones. To an extent I think that applies here.
Exactly my point.
I would have said that it was 100% accurate to the best of my knowledge, but my knowledge increased somewhat with the second part of that post.
Send me an updated parts list and I'll get on it.
We attempted to have Hosts contact us privately with a request for an upgrade based on what their machine could best use. (i.e. "My board has a Duron 1600, but could handle as much as an XP 2800+")
It was not exactly what you might call a rousing success.
We're working on a plan where we can publicly match donors with Hosts. What would you all think of a thread here in the Team Short-Media Forum where Hosts could post their needs (no, not those kind of needs, you pervs...), hopefully leading to volunteers from the team stepping up with a parts donation?
95% of the components in our 36 SMx rigs are from donated parts. The exceptions have been when we've had a computer all ready to go save for one crucial part. An example would be something like SM24 where we needed a pair of matched CPU's for the dual-socket MB.
The SMx Fund covers things like buying cases (we provide new ones, which look a lot nicer in the Hosts home), shipping costs, and emergency parts replacement.
Your idea is a good one, though. We'll look into having a place where we can list rigs under construction, highlighting the parts needed to finish them off.
I'm all for that. We're not interested in going through all the rigmarole involved to bump a 1700+ up to a 1900+. If we can make it a 2500+, then now you're talking.
I'm not disagreeing; I just don't know. But, what if a $150 graphics card turned a lame machine into a real powerhouse? Would there be support for spending some dough on one card to use as a guinea pig?
Interest has been in steady decline for the last year. We recently opened a thread on the subject and a quick count shows that we had three people currently qualified who expressed interest, two of whom already Host a rig. (Which is not an obstacle in and of itself.) We also had some people who I think will be fine candidates who are waiting on their 18th birthday or need a little more time with the team to be considered veteran members.
We're still global and will ship anywhere we can find a qualified Host. At the same time, it would be dandy if a group of our teammates in a particular region got together to simplify things by building a rig locally. We would also save a bundle on shipping.
One idea is to sell the old parts on eBay, then plow the money back into the SMx Fund. What would you all think of that?
For the first year I was building them, we averaged a new rig every single month. We also reissued several rigs due to Host turnover and replaced a few others which had gone AWOL. As has been pointed out, maintaining a steady stream of qualified Hosts has become a big problem.
We have no intention of turning the SMx Project into a closed club of three dozen Hosts who get annual upgrades. As long as we have new faces who really want to Host one of our rigs and are qualified, we want to keep rolling out new computers. Several teams have emulated our project and a few have gone on to have more computers than we do at the moment. That doesn't set well with me.
In simple terms, if we have enough people willing to become a Host and enough parts, we can upgrade old rigs and build new ones quite rapidly.
Agreed. My only point was to supplement the shortfalls with the main parts bin, but I guess I'm pointing out the obvious
Last time I checked, there were no AGP cards authorized for use by F@H. Sure, they can be hacked into working, but personally I think we should wait to spend money until something is officially supported and is guaranteed not to send back junk data. Plus, these more powerful GPUs may require PSU upgrades in some rigs. Many rigs have bargain-basement PSUs.
I'm still hesitant to upgrade mobo/cpu at the same time, and offer the stipulations in post #9 again. What do people think about those?
Remember, the current mobo/cpus are producing points already. Unless the motherboard prevents the SMx unit from producing points at our new minimum standards, it would probably be best to keep it as is and make the new mobo/cpu combo an additional rig. Mainly because 2 rigs are better than 1. The side issues of a mobo upgrade are also important: if we upgrade to a new platform, we might need to get a new PSU and RAM as well, which could increase the upgrade cost by 50 to 100%. At which point, let's just finish the job with a case and call it a new rig.
I'm completely in favor of dropping in better CPUs or RAM, but am hesitant about motherboards. Definitley a case-by-case sort of thing where we'd want to compare current ppd and potential.
I especially like the recycling the parts that are of no further use to us into ebay. I would also like to see exactly what the option are on the GPU folding, and what kind of return in ppd we can expect from one for the money.
and Garg shares my sentiments about upgrading older rigs to a new platform, why not just let the older ones keep folding, get a new case and call it a new rig.
I think the motivation for building new rigs is to get more people involved and not at all fiscal. In terms of team money, time, and hosts' electric bills, I think it makes more sense to upgrade.
That said, I do not favor the cessation of new rig creation at all. I simply think we need to take a month or two and direct our efforts at seriously upgrading our fleet.
That idea is on the fast track right now.
Let's see if we can dig up some hard data on the subject. The point made about the increased PSU requirements is a good one, too. We'll need to do some real research to make this a possibility.
What would everyone recommend as a reasonable lower limit on CPU power?
When I moved in the Summer of 2004 I "retired" (to the dumpster) a P200 and a P166 rig. They were doing about 50ppw and, at the time, that almost seemed worth keeping. It was the cost of having them hauled nearly 600 miles that made me decide to scrap them. Since then I've also stopped Folding on a pair of K6-3/400 rigs; it wasn't worth the electricity after the deadlineless WU's dried up. I know there are some people who wouldn't want anything in their lineup lower than a couple GHz.
How about a Poll on the subject? :headbange
If we upgrade old systems so every SMx rig is at least 1.8 GHz... the host would be spending about the same about to power them but getting more out of them.
In general, hosts are doing this for the honor and the joy of seeing the project prosper. Shouldn't they be rewarded every few years with a newer setup? Why should someone put their name in for a 2500+ when the next SMx could be a 3000+? Which is gonna crank out more?
I vote we get every SMx rig to at least 2.0GHz/ 1800+ by 2008 (a bit over a year.) In the mean time, we try to issue a new rig or two (maybe more if there is interest and money).
i think thats a good method of renewing the machines we have out in the enviroment now. say we set the minimum level of WU's at 30 per month, in 6 months we can increase it to 40WU's a month, any thing that falls under 40WU's for X ammount of time can be end of life'd and if we get a good stream of donations in, we can send a processor or some ram to increase the WU's but at least this way you have a base line, a minumum of what we want back as a return for the electricity, shipping, and mantanance on the units. as well as stopping older units form becoming money pits just to get the WU's needed to keep it out of EOL
but i say we come up with a system for EOLing older units.
arrrggg, i need to quit my job, lol
True, I'm just concerned about instances where the only part we're not replacing is the case. However...
This, with adjustments for Armo's point about the SMx's actual production (although it correlates to CPU MHz) is a good idea. Sure, the cutoff point we choose will be arbitrary, but we'll be able to prioritize the rigs that are farthest below that cutoff, so no issue there. That gives us a framework we can really work with.
For reference, here's a table of ppd of other cards: All I have to go on is the one guy in this thread that reported his frame times.