Thursday, March 28, 2013

Why I am never the first to buy a new technology

There seems to be several different types of people in this world, those that always want the newest of everything, and those that like to see what kinks each product has and wait for them to be either worked out or written off as design flaws.  Maybe its because I was one of the early adopters to a particular smart phone, way back several years ago, lets just say anyone that ever had one, mostly refers to those smart phones as intelligent paper weights, which barely functioned well enough to make calls 90% of the time ( if you want to do something besides make a phone call forget it nearly 100% of the time).

Since then I have learned that, I'd rather not be the first to test out a new version of a software, or buy the newest graphics card before they have been tested and been through several versions of drivers, and so on.  So why am I writing this?

nVidia recently released its newest consumer GPU the Titan, which while it is in theory a massively powerful GPU, I read of quite a few crunchers that bought them basically the second they were released, got them, put them in their computers, only to learn that due to driver issues they either did not work or worked poorly with the projects they really wanted to use them on.  Does this mean that Titan will never be a massive crunching card?  No, it simply means the kinks need to be worked out of the development process and find its way into the tweaking of the drivers before they can really shine.

In a certain sense, the people that like to hold off, like I do, owe thanks to those that want the newest items, as we learn from their mistakes, and when we do buy we have products that work near perfectly thanks to the early adopters.

Tuesday, March 19, 2013

Running On Autopilot

I am happy to announce that I have gotten my systems up to the point that they are running almost entirely on autopilot.  Part of it came with a little training of myself as well, mostly to not mess too often with the computers.  Due to power concerns I am running my least efficient machine only on weekends, and my highest powered machine, which is incredibly quiet runs most of the time.

Every Sunday night I run updates, and each morning before work, each day after work, and then every evening before I go to bed I check to make sure things are running without issues. Its a simple processes that actually takes very little time out of my day, and has been working like a charm.

To do Still:  The main thing I need to do is resetup SSH capabilities into my machines. While not hard I feel as I moved around the ethernet cords in the router, the IP addresses changed so I need to go back in and figure out what they are and record those so I could then have SSH working again, so I can remotely check on my machines.


While that is easier to do remotely, most of the time I want to check on it, I am in my place, and it is as simple as turning on my TV and switching input settings to the appropriate computer.  (If that's not just as easy (if not easier) as going in the terminal and entering commands and passwords, I do not know what is).

Saturday, March 9, 2013

Testing Loyalty

I dedicate all of my GPU crunching time to the project GPU Grid, because I honestly believe in the work that they are doing.  But I hate to say this my loyalty was really tested this past week, possibly made even worse because I was stressed out for another reason and the last thing I needed was to be worrying about how my machine is working on things.

Lets just say a large amount of GPU Grid tasks were released with a major error in their code. I spent over 170,000 seconds crunching these tasks, which as I was distracted I did not notice any problems until they were reported as aborted by my machine some time later.  ( My GPU is on a computer that is dedicated to crunching and not regularly used).  But I was disappointed to see that between that and the one time it froze my computer ( which I did check and manage to fix), that I lost several days of crunching time from this error.

Thankfully I gave it just long enough for the problem to apparently be fixed, but I was really starting to question whether or not GPU grid should continue to receive my support.  But I think I have the biggest soft spot for GPU grid of any project I have yet dealt with as they are people pursuing their doctorates carrying out their research, in a rather similar place in their lives as myself and my brother.  So I can sympathize with them, and want to help them in any way possible.

So to my readers, have any of you ever encountered an error with a project that really had you questioning whether or not it should receive your continued support?  ( I do feel odd posting this so close to a post on Battered Volunteer Syndrome, but I think this is a qualified complaint, and not just a rant.)

Sunday, March 3, 2013

Battered Volunteer Syndrome

I am borrowing the turn of phrase from a poster in the GPU grid forum, because it seems aptly appropriate about what I want to talk about today.  I am amazed at how many people on project forums like to complain about everything that is going wrong at these various projects. Now for one I am known to be a rather laid back person ( passionate yes, but laid back more or less), as such I do not tend to get angy until it is really warranted.

I am not calling out all complaints on forums, as system crashes, and faulty work units should be called out in the right area's so the scientists and programmers can fix the issues.  A bit more on the questionable side seem to come with GPU crunching, in the sense that a very well programmed GPU task will be using everything on the GPU that it possibly can, so when running such a task, it may render your PC unusable from a normal user experience side of things.  While I can see how that can be frustrating, I believe that is why BOINC has options to disable either CPU, GPU or both when you decide to use the PC.

But I have seen some rather outlandish complaints issued against a variety of different projects.  People being completely irate because a project has unexpected down time, often waving around statements like "it would be unacceptable for ATT to go down for this long."  I honestly really need to laugh at that, because guess what you are not paying these projects unless you regularly donate to the projects you support.  As such these projects are almost all on a fairly shoe string budget, and on a shoe string budget one of the first things you cut out is as many redundancies as you can, which means when things crash there's not a backup server just waiting to get put in the game.

Then fewer and far between but almost insanely out of left field are crunchers ( usually with already some substantial amount of credits, so its not like they have only been doing it for a few days), making posts telling the projects that they should be paid for their efforts.  Yes we are doing the science for the projects, but we volunteer to do that work.  Unless the definition has changed since I first learned that word, it means we help with no real expectation for a reward, except maybe emotional, and an atta boy or two.  I fully understand the machinery we buy, and the electricity to run it, not to mention the internet service to keep it connected are all expensive, but if you can not afford the extra bump in electricity, then you likely have bigger financial issues you need to deal with and should not be crunching ( and you definitely should not be buying top of the line machines).

Sorry if this is a wake u call to anyone, but I feel this had to be said.   We are volunteers, our goal should be to help the projects, because we believe in the projects, not to seek some sort of reward.

Friday, March 1, 2013

Computer at Play, While you are Away?

I have my newest computer crunching 24/7 lately, only rarely being restarted for updates, or shut down occasionally for a few hours.  That means it runs when I am at work, sleeping or doing most anything else.  I was more concerned about doing this right after it was assembled feeling it wasn't quite battle tested yet.

Now that its run for many many hours staying incredibly cool, and running with no real issues. So I have no real problem leaving it running while I am away for less than a day.  Now I am away from my place for a few days, and I decided not to shut off the computer while away.  It is leaving me oddly conflicted, but I can not quite explain why.

In all my logic I see no real reason why something should go wrong while away for four days that could not go wrong in a few hours.  In fact I have been so busy lately, I have easily gone more than four days without checking how that computer was doing on multiple occasions lately.  So currently my little bit of solace is to look for new results being returned by the computer letting me know that everything is still running peachy-keen.

So for other crunchers, do you leave your machines running all the time, even while you are away for a few days?

World Community Grid Signature