Folding Forum

The stats pages don't have enough information. I can't tell when I returned a certain work unit and I can't tell which of my machines returned the work unit. Every other distributed computing project has comprehensives stats. I know you guys are busy with all the scientific stuff, but we need some better stats. Thanks.

I think the first task on Pande Group's to do list is to get a server that won't take forever to process current stats. If they add some more information now, the current server won't definitely be able to handle it

Yeah, I noticed that it seems like everytime I come to check my stats, the stats update is running. Oh well, I can only hope they come with something new in the near future.

Actaully, does the FAH stat server keep record of every single WU ever sent in, all hundreds of millions of them? (like, the one that the mods can access). If so ,the database would only get larger and more problematic.... would help if more details about WUs are truncated every few months or every year (such as the PCRG, date received, server, etc), leaving on the WU project number and count on the stats page....

Full details are stored for all units as far as I know, although I could be wrong.

geeknik wrote:The stats pages don't have enough information. I can't tell when I returned a certain work unit and I can't tell which of my machines returned the work unit. Every other distributed computing project has comprehensives stats. I know you guys are busy with all the scientific stuff, but we need some better stats. Thanks.

Have you looked at some of the 3rd Party Stats sites linked at the top of this page? EOC? Kakao?

WangFeiHong wrote:Actaully, does the FAH stat server keep record of every single WU ever sent in, all hundreds of millions of them? (like, the one that the mods can access). If so ,the database would only get larger and more problematic.... would help if more details about WUs are truncated every few months or every year (such as the PCRG, date received, server, etc), leaving on the WU project number and count on the stats page....

I seriously doubt it would be more problematic. If Berkeley can keep and store all of that SETI@home information dating back to the late 90s, I'm sure Stanford can do the same with F@H. Hard drives and RAM are cheap.

7im wrote:Have you looked at some of the 3rd Party Stats sites linked at the top of this page? EOC? Kakao?

http://folding.extremeoverclocking.com/search.php - Doesn't find my username or team name.

http://kakaostats.com/usum.php?u=1217442 - all it shows is points. doesn't tell me anything about the WUs I've returned.

lol hard drives are not the problem. It's the sheer size of the database and the slowness even of the latest hard drives compared to RAM which is the problem. As such they are replacing the entire server to get extra RAM capacity... on limited funds that is an issue.

GIving WU info is on our minds and is actually something already architected in. However, in order to maintain privacy, please note that when we do make this information available, we would require a passkey to get the results.

We're in the process of upgrading the server code into production servers and consolidating old servers. Next up is improvements to client and stats.

Until those upgrades are completed, I recommend checking the Tools List in the 3rd Party Applications section of this forum to see if there are any tools to help you track your own WU results.

VijayPande wrote:GIving WU info is on our minds and is actually something already architected in. However, in order to maintain privacy, please note that when we do make this information available, we would require a passkey to get the results.

We're in the process of upgrading the server code into production servers and consolidating old servers. Next up is improvements to client and stats.

Ouch, I better start using it then ( I have one, not sure what it is as I opted for one ages ago and just never really cared ).

Just a question though, why is it a breach of privacy to be able to look up detailed stats from other users? I mean, we already have a cpu count and completed wu count for all donors available to everyone, it's not that big a leap to extend that to all the new info. Just wondering where the privacy comes into play'?

Maybe it is designed to show every piece of information they keep? If so, the external IP of your home network will be visible to the world without using passkeys...

IP's are I admit the only info which creates a possible security risk for the donor.

But other then that, the extended info would only consists of a detailed listing of pcrg's and their submitted results I would guess. Maybe also when they where downloaded, thus giving people a good and easy way to check if there is a problem there.

Afaik, there is only partial info collected about the hardware's capabilites, again not something which I think invades someone's privacy. Most donors list their hw in their forum signature or hw fields, and I think therefore would not object to others seeing their hardware they ran the clients on. ( which is already appearant from the submitted projects currently available ).

Maybe I'm just most curious because I'm not exactly sure what is stored and therefore am unable to see a privacy issue.

On the other hand, it would be a good thing to have an option so donors can select to hide their details from others so people who do object can take that step. But approaching it from the opposite direction I feel isn't needed, and maybe even 'bad' as I think it might influence those who care about ppd allot to gain motivation if their able to seek and match people with simular setups and can try to compete directly with them. Outside of team competition, personal competition now is pretty limited to people who know they got either comparable hw or compareable spending capabilities for purchasing hw, being able to look up donors with a certain hw config would allow greater competition for everyone?

I'm not going to address the privacy issue. Stanford can set that a whatever level they feel is appropriate for the information they will be choosing to display. But I will point out one thing, MtM used the term "most people" which does not include "all people." So while most people don't care what is displayed, some do care, even if that data does not typically cause risk. Those few would choose to stay more private, and I'm rather sure Stanford would prefer to err on the side of caution in that regard. Everyone has a different concept about what should be private.

However, on a functional level, the passkey is also required to keep the searches smaller and accurate (reducing the potential load on the server). If you and several other people share a fah user name, you will get the results from all of those users, which results in a larger amount of data to display, and most would considerate that result inaccurate because it didn't display only your personal WUs. The Passkey makes sure that the search only returns your specific work, and that list is smaller than all the work units turned in by a user name like Anonymous.

Which raises another point. Some people have chosen to donate anonymously, but might want to be able to view their specific work units. A passkey and the new stats would enable them to do that. Or if people were folding under a group name, you could each still use a passkey, and see their own portion of the results from the group.

Privacy was not the only concern.

After the first alinea I want to reply with 'ahh yeah but I added that there should be an option to make it private on each donor's own discrescion'.

But what you added makes allot of sence. Still feel it would in ways also be beneficial to be able to look in the stats for people with simular configurations (client counts and maybe partial hw info as I already admited that atm I don't think any is collected but I think this might change later on), but the points you brought up outweigh that notion by far. Using the paskey as primary key would speed up queries, I'm not familiar with large scale database issues but I do know enough about sql to realise that. Using alternatives as combination of username and team would cause much greater load and would not allow the filtering of information for groups based donors or anonymous one's. You got me convinced.

I do hope it will be considerd to still allow looking up other people's stats if they do not themselfs select an option to make it private or better yet allow it for people who will opt to not make their stats private.