Saiweb

Following on from Linux – Generating file manifests and then checking them I was always getting the same questions …

How long left on the manifest Buzz ?
How long left on the verification Buzz ?

And I HATE having to turn around an say … I don’t know …

The problem with the usual command line method is that it give no indication of progress, and by extension no indication that it was infact running and not ‘hung’ …

As such I have now added the ‘manifest’ command set to the Sysadmin toolset

The manifest command take two data types, the first is a folder path from which to build the file manifest from, the manifest itself is also compatible with the “md5sum –check” function.

The second is the path to the manifest itself, in this case the manifest command will verify each file against it’s entry in the manifest:

At each point the command give you an indication of it’s current status, however this does come at a small cost, the script has no concept of the size of you console and as such will always render out the same number of character meaning if you console is not wide enough it will not render correctly, in the videos I have the console on a high resolution monitor as can be seen each video itself is 900 pixels wide.

This process is CPU intensive (20-45% on one core of a intel core2duo 2.8GHZ) and uses around 140KB of memory.Tweet

Tags: cli, indication, indicator, Linux, manifest, md5, progress, python, sysadmin

oooooooo SHINY …. I mean erm … yeh … *whistles*

I’ll be taking a look at CUDA over the next few days, first task being to rig it up into OSX and Eclipse CDT, CUDA is Nvidias current push at programming for the GPU, from what I can see it is a vast improvement over the old Cg days (nVidia’s other attempt at a GPU SDK).

This seems farm more popular and has greater support, with the newest nVidia drivers and card being pushed out ‘CUDA’ ready … and when your app therefor can be used on any system running a compatible GPU I’m pretty sure cross platform is not going to be an issue, and then there’s the performance side of things with GPU’s parallel architecture massively out performing CPU’s.

If I have the time I will be getting a could of cuda shell projects out the door, I have a very basic concept for one allready but you’ll just have to wait and see

UPDATE: Despite a quick playing during my lunch I can’t get CUDA example compiling …

Tweet

Tags: cuda, eclipse, md5, nvidia