Minimum no of hrs" box during installation feasible?

[morpheous_2812]

Full subject heading : Will a "Minimum numbers of hours" dropdown box during installation of V7 client be feasible?

I am currently running an uniprocessor slot, using the latest 7.1.38 client

Working on a 7013 (0,102,58) WU

Estimated TPF is 24 mins 57 secs

Estimated time of completion is 1.68 days, according to the information shown in the client control.

Some basic information about this system:

Intel Pentium E6700 3.2GHz CPU
Motherboard P43T-C51
Memory 2048MB

Above details obtained from CPU-Z

However, this system is not running 24/7, but only about 6 to 8 hours a day, sometimes more, sometimes less.

I did some calculations, and I am sure that with some adjustments, I should be able to complete this WU before the preferred deadline of 8 days.

An idea came to my mind while I was doing my calculations, and i wanted to share on the forum, as I am not sure whether it will be too much work, or too late to include into the final version of the V7 client.

Will it be possible to include a drop down box, that indicates to the person installing, how many hours he/she will be folding on their computer? There has to be a minimum amount of hours set by the client, maybe 6 hours a day?

The main reason for this suggestion, is that I realize that the deadlines overall for the various projects seem to have become shorter, and there may be folders, who leave their clients running in the background and may not notice that their systems are now unable to complete their WUs within the given deadline.

Of course, I realize that people may not follow exactly to the settings they have indicated during the installation, however, it does let the new folders understand that although it is 100% voluntary, failure to meet the deadlines will slow the progress of science significantly.

Another benefit is that once this information is sent to the servers during installation, based on the system configuration, they will be able to quickly analyze and therefore not send WUs with short deadlines that will not be able to be completed by the folder in time, hence reducing the amount of expired WUs.

The above suggestion is of not much use to experienced folders, as they most likely know of how to calculate TPF, or run highly customized rigs which have no difficulties in meeting the deadlines ahead of schedule.

My proposed idea is more for new or inexperienced folders, who are getting to know folding@home for the first time, or want to contribute but not on a 24/7 basis.

It is my understanding that the purpose of having an unified V7 client, is not only to combine all 3 clients, uniprocessor, SMP and GPU, into one easy to use client, but also to make it easy for new contributors to folding @home, to make it simple for them to start folding right away.

Nonetheless, I understand that implementation of this feature might not be feasible, either due to other important issues that need to be solved before the V7 client is ready for the public, or the time and effort needed to add in this feature into the client outweighs the potential benefits.

If the above feature is not practical, due to time or resources constraints, I can understand. The purpose of this post is just a sharing of my thoughts on how the V7 client can be made more easier for new folders.

When I joined several years ago as a new folder, I started on the V5 client, only used uniprocessor folding and did not dare to use any beta clients at all. Now after learning a lot about what folding@home is all about, I have contributed more time and resources to folding and whenever I have the opportunity, I will try my best to let more people know what folding@home is all about.

[codysluder]

I don't think it's possible to do what you're asking.

The benchmarking process is used to set both the points per WU and the deadlines. The servers don't have WUs that can be completed in less hours per day than a single number which depends only on the speed of your computer.

Suppose the completion times on the benchmark machine for two projects are 1 day and 2 days. The deadline1 will be half of deadline2. If your machine is twice as fast as the benchmark machine, you can complete the first WU in 12 hours and the second in 24 so you have to run at least 24 hours by dealine2 which is exactly the same percentage of each day as 12 hours by deadline1.

[Jesse_V]

Well, he does have a good point IMO. If v7 installs SMP+GPU by default, then some people may not know about deadlines and end up hurting things more than they think. There's a lot of subtleties in F@h (deadlines are one) and it'd be nice to make sure people have some idea of the important ones. The brilliance of the uniprocessor is that 1) its set-and-forget, and 2) it has long deadlines. These two make it applicable to almost everyone, from the guys who fold only during the daytime to people who fold 24/7. V7 basically accomplishes #1, but #2 doesn't apply to either SMP or the GPU, which the combination is now the default if there's the right hardware. I'd like to see a better message warning them about the deadlines, and that if they aren't going to run this 24/7 to please choose the uniprocessor.

I recently proposed a "laptop or desktop?" prompt in the installation for setting the pause on battery feature, and that was shot down because everyone wanted to make sure the installation stayed super simple. So I doubt they will add some kind of "how many hours a day do you plan on running F@h?" thing. Just a message like I said would be better.

[7im]

Hang on there Jesse. Your laptop prompt suggestion was NOT shot down. It was a good idea that was adapted to fit in to the existing design goals. And your suggestion was made in to a feature request, and will get put in to the client, just in a different way than you suggested. Please don't try to paint a dark cloud for morpheous_2812's suggestion when there isn't a dark cloud.

Not all ideas can be added exactly as conceived. And fah cannot make a perfect program. There will always be some situations that cannot be programmed around. For example, your concern about SMP+GPU as being the default. It's a valid concern, but your concern doesn't make it the wrong choice. The project has already weighed and reweighed that feature and that setting over and over. They have reasons that is the default setting. It may not work well for everyone, and is not expected to work well for *everyone* nor could you fix it to work well for *everyone* but it is the best option for *most everyone*

And uniprocessor is applicable for most everyone as you say, but SMP is preferred by most people these days. So applicable is not the same as preferable. Programming for the lowest common denominator is not a good goal to set.


Back to the morpheous_2812's suggestion. Good idea, but difficult to estimate, and harder to commicate accurately. Fah could run a benchmark (something it already plans to do in the future) and then show how many hours per day that your specific hardware would need to fold to meet the deadline.

But then just finishing a work unit just minutes before the deadline is not the goal. Some might prefer the estimate to be shown with time to spare so they can upload the next day. Or with time to spare for their slow internet connection. It all gets very complex very quickly.

For example, the computer may be on 8 hours a day, and can fold @ 100% for 6 hours, and fold @ 50% for 2 hours while you game. How does fah know that you game for 2 hours a day, and adjust the "hours per day" estimate to match? It's all most impossible to do it accurately.

However, when the ETA information on main screen of FAHControl is fixed to give more accurate information, it won't be hard for you to make your own rough estimates, based on your own knowledge of how you run your computers.

[Napoleon]

Hmm, maybe the installer could open the Best Practices FAQ by default once it's finished? It is brief, already there and let's one know about the basic considerations. Would also gently nudge new users towards other FAQ pages, in case they are eager to learn more about FAH than just install-and-forget.

[7im]

Include a ReadMe file with the client? I'm not sure how well any prompting like that actually works. You can't PUSH knowledge on to people. You can only make it easier to find when they want it, and easier to comprehend.

For example, does anyone know there are links to the fah home page and to this forum in the About screen for FAHControl? Maybe those links need to be on the main screen somewhere? (I don't think so, but maybe something like that only better) Like a Help button on the Toolbar next to the About button. With links to the home page, DL page, FAQs, install guides, forum, etc.

[Napoleon]

I'm clueless when it comes to (G)UI design, considering that in the past I've favoured v6 console clients for their ease-of-use and simplicity. Client v7 convinced me to jump ship.

Anyhow, less clutter is always better. I'm in favour of providing links. People choose whether to follow up or not. 7im, you're certainly right about pushing things on people, it alienates them right quick. Still, I'd think that linking to Best Practices FAQ in its current form doesn't cross the line, at least it didn't do that for me. Good stuff to know from the get-go, no fluff.

[7im]

Agreed. Wherever the help is made available, Best Practices should be on the list.

[morpheous_2812]

Thanks for all the feedback and comments.

After reading all the responses, it seems that implementation may be more trouble than its worth, if its even possible to implement it at all.

The reason I thought of the idea in the first place, was because the previous time I used the uniprocessor client, which was quite some time ago, it was really install and forget.

It was possible to do it at that time, because the deadlines was quite long, about more than 30 days on average for most WUs, that used the uniprocessor. For the 7013 WU, the preferred deadline is set at 8 days. It is possible that this WU may not be completed by a folder, who installs and forgets about the client, yet only folds several hours a day.

Another question that I have been wondering about, does the WUs get "tougher" to fold, as old projects are completed and new projects arrives to take over?

One of the most important reason why I like the folding client, is that it runs in the background and does not affect the Usability of other programs. Therefore when a person who is doing more intensive work, such as autocad or video editing, they will not notice any slowdown or lag, as the folding client will release the resources needed for other work, then take it back when the user is no longer doing any intensive work.

However, in my opinion, its also the ease of use that has the potential to cause unforeseen problems, especially to the folders who are running the uniprocessor clients, as it is set and forget. Since it runs in the background, some folders may even forget that they are running folding@home.

According to the Folding@home Frequently Asked Questions (FAQ),

What are the minimum system requirements?

All computers can contribute to Folding@home. However, if the computer is too slow (e.g. wasn't bought in the last 3-4 years or so), the computer might not be fast enough to make the deadlines of typical work units. A Pentium 3 450 MHz or newer equivalent computer (with SSE) is able to complete work units before they expire.

Someone who wants to know more about what folding@home is about, may visit Stanford's website, see the minimum system requirements, decide that their computer meets the requirements, download the client and start folding WUs.

However, if the WUs get "tougher" to fold as old projects are completed and new projects arrive to replace them, it might be a good idea to review the minimum requirements needed for folding.

Of course, this might mean a drop in the number of folders, as people realize that their old computers are not able to complete the current WUs, unless they spend more time to fold the WUs.

Another scenario could be that there may be folders who install version 6 of the client a year ago and noticed that they meet the deadlines easily. If they are install and forget folders, it is possible that they are totally unaware that their system are unable to meet the current deadlines now, if they spend the same amount of time folding as one year ago.

I understand some of the opinions that I have expressed are hypothetical scenarios, which may or may not be valid. Nonetheless, if V7 is to be the only client in future, since it is an unified client, all folders will be affected, whether they are casual folders who install and forget, or experienced folders who know exactly how the client works.

I do not consider myself a very experienced folder, but just sharing my thoughts on the problems that might occur. I stopped folding on an older laptop, even though it could meet the deadlines, as the margins for error were very small and I had to monitor it constantly, which became tiring over time.

Finally, I realize that any changes or features that are added into the client must be thought of carefully, so as not to alienate any particular group of folders, or unintentionally causes negative impact. That is why I mention my idea is just a suggestion, if it does not improve folding as a whole, then it should be dropped, so that valuable time and resources can be dedicated to the appropriate issues.