It has been "years" since I re-installed Windows 7 from scratch on my main Boinc server. I moved the Raid 1 HD's off that system onto my target upgrade system.

So I re-installed Win7 on the old box and brought up SetiBeta for the couple of weeks till I sell the old box. The cpu task time has regularly been above 4+ hours for the 8.05 task.

So imagine my surprise when the beta 8.05 tasks have been running in like 2.5 hours which is MUCH faster than I previously experienced on that hardware.

There are two possibilities here. 1) The SetiBeta data is skewed so I am not getting a representative sample of the data that I would get in production Seti or 2) My operating system is so cruded up inspite of regular registry cleaning that the whole system is enormously slower.

I can spend part of a day backing up all the important data (and maybe even the seti project id data) and do a clean Windows 7 Pro install on the target upgrade machine. But is it really likely to speed up my machine?

Thanks,
Tom

---

A proud member of the OFA (Old Farts Association).

Are you comparing the same Angle Ranges of the workunits you're comparing? Generally, Very Low Angle Range (VLAR) work tends to run longer than Very High Angle Range (VHAR) workunits. The differences can be massive depending on the amount of blanking due to terrerstial noise too.



Also, I'd like to call out the erroneus idea that the Windows registry needs regular cleaning and maintenance. The false idea is that a cluttered registry will cause performance issues and thus requires cleaning. In actuality, Windows only loads the parts of the registry it needs, therefore unread keys don't cause performance issues. Applications work in much the same manner. More often than not it is only going to be applications that clutter up the registry, but if those applications are uninstalled or otherwise not loaded, the extra keys cause no performance loss to the system as they go unread and unused.

It has been my experience that using registry cleaners often cause more problems than they claim to solve. Essentially you are giving your system's "tune-up" to another company/individual and letting them change/delete registry keys without knowing anything about your system or the applications that are installed, so how would they know what the best settings are?


Lastly, the idea that Windows should be re-installed to solve performance issues is a lot like reassmbling your car for a tune-up. It's overkill. Most causes of performance degradation are due to failing hardware, poor maintenance on the file system (CHKDSK, Defrag), and/or inadequate care of the system (proper cooling, dirty environment, etc.). As a hobbiest, I have various Windows systems dating back to Windows for Workgroups and Windows 95/98 and every version since that still run as fast as the day the OS was installed on them.

Now, if doing a wipe-and-load is faster than the time it would take to get the system performing optimally, then by all means go for it. But bad habits usually carry across builds, which means your issues will return eventually. I would strongly recommend ditching the registry cleaners as a start.

There are two possibilities here. 1) The SetiBeta data is skewed so I am not getting a representative sample of the data that I would get in production Seti or 2) My operating system is so cruded up inspite of regular registry cleaning that the whole system is enormously slower.

It finally occurred to me that I needed more data. So I went ahead and joined production Seti whereupon it sent me a bunch of different WU's that regressed to the mean (eg. 4+ hours / wu).

I really didn't want to re-install windows. I don't have the data on separate partitions from the OS so it would be a pain. Now I don''t. :)

Tom

---

A proud member of the OFA (Old Farts Association).