From the kernel to the shell, Windows Vista is a very different OS than XPSP2. How so? Here, Charles interviews Architect Nar Ganapathy whose team of highly skilled engineers write the Windows IO system, driver frameworks and related technologies. So, what, exactly, is new in Windows Vista with regard to IO? What does it mean, exactly, to users and developers? Tune in. Learn.
Regarding Windows Vista’s I/O
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSAlert.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
We all know the *only* thing they changed in Vista is the look.
This is possibly good for users if bad drivers no longer bring down windows but are instead disabled.
When people talk about Vista, they often talk about the new visuals, the security improvements, IE7, etc. But then I keep seeing things like this pop up, along with the new audio subsystem, and other little things. Has anybody compiled a list of all these new features that will be in Vista? There’s apparently a lot of new plumbing under the hood, and I wanna know what’s there.
Also, when the guy talks about the revamped IO system that is supposed to prevent (or at least reduce, I guess) applications from hanging when making IO calls. Will this also eliminate the system going into a brain freeze when trying to read from a CD or DVD with bad sectors? If they didn’t fix anything else but this (and the drive-by malware installs), I’d be happy camper
Has anybody compiled a list of all these new features that will be in Vista?
http://en.wikipedia.org/wiki/Features_new_to_Windows_Vista
Not sure how complete it is, though.
Whereas Windows XP only supports loading drivers from floppy disks while the setup environment is initially loading, Vista additionally supports loading drivers from USB and Firewire storage devices, and CDs and DVDs from within the main setup wizard.
And exactly how many years did we have to wait for such a basic functionality ? Geez, how many times I swore over the lack of it during the last decade, given that I had to borrow fdds for windows installs for about seven years now.
Also command prompt during setup would be nice if you could use it for something other than watching a blinking cursor (e.g. ssh-ing, using remote shares, etc.). I have also bad dreams about rootkits using the “features” of the filtering platform to give us major headaches. WMP11’s features really are a new gui and heavy loads of drm. By default it will still not be able to play most popular video formats. In some betas I couldn’t even play some dvd’s (originals), since it give me some access errors without reason. WCS as a self-proclaimed evolution of ICC color profiles is another bit I don’t like. But it’s typical for MS to often choose a path not following standards. I also don’t like when I read lines like “better supported”, “better performance”, “performance optimizations”, since without proof, numbers or some blind trust these words mean nothing.
Thing is, trying betas and wainting the final, I hope what we have seen up to now will improve a bit more. Other than that, we should come back on the topic afer we have the disks in our hands and used them for a few weeks.
“I have also bad dreams about rootkits using the “features” of the filtering platform to give us major headaches. WMP11’s features really are a new gui and heavy loads of drm. By default it will still not be able to play most popular video formats. In some betas I couldn’t even play some dvd’s (originals), since it give me some access errors without reason. WCS as a self-proclaimed evolution of ICC color profiles is another bit I don’t like. But it’s typical for MS to often choose a path not following standards. I also don’t like when I read lines like “better supported”, “better performance”, “performance optimizations”, since without proof, numbers or some blind trust these words mean nothing. “
Spot on. Well said. Well worth repeating.
Having now watched the video.. I do have a more simple answer to their I/O problem..
ASYNC BELOW APP LEVEL, Report into journal the I/O you would otherwise wait for synchronously, would now be waited for… with a reasonable timeout, allow X number of failures before (based on error type) reporting back to the app with an error.
Ideally, you wouldn’t have to go below app level, but backwards compatibility would demand it.
Though, the problem would be far less of a problem if Windows developers could just utilize threads more effectively, though I think that has more to do with Windows API and coding habits more than anything.
I can just create a thread by typeing spawn_thread(),
passing a function and a variable to pass to that function, naming the thread, and giving it a priority. I can then cause that thread to run with resume_thread(thread_id). So, naturally, these problems simply don’t make much sense to me… not to mention my running a real-time capable kernel. Threads cost NIL here, and lots on existing programs and platforms which do not utilize them.
That and many developers simply can’t properly abstract (mentally) the workings of multiple threads.
Oh well, if the solution that MS implemented works for Windows, or not, will only be known in time.
–The loon
The windows kernel architecture is fully asynchronous. The problem that is being solved here is the one which happens when a thread performs synchronous IO which then hangs until the timeout (often on the order of 30 seconds to allow for really slow network connections). Sometimes the driver doesn’t even complete the operation ever. The CancelIO facility allows another thread to stop another thread’s synchronous activities.
Gotta love when one random guy on here thinks he knows better than a whole professional team.
Gotta love when one random guy on here thinks he knows better than a whole professional team.
And professional teams are never wrong? You must be new here.
They can be, yes. I think you’re missing the point though.
Because the last time I tested windows vista build 5728, It took forever to burn a DVD of files located on a workable share on Windows Server 2003. I also mapped the network share to the drive Z: but it hanged there and no CANCELLATION was available to me “the user” to cancel such act, and remember I am talking about a post RC1 build and just one more build is left till RTM is out.
Besides, what the guy said would be nice IF, the developers adhere to the MS guidance in programming, even MS don’t adhere themselves to their own standards, or force developers to adhere to their standards.
Also last time I checked a DVD+RW disk with udf v 1.02 file system, the address bar was growing without displaying any of its contents! While in fedora 5 with full patches and new kernels installed it was able to mount it in just 4 seconds. Why is this?! This is serious as anyone who wish to burn a file larger than 2 GB would have to use udf file system.
And in linux or Mac such a change would go without such propaganda but rather with a silent kernel change that would make life easier, I believe this should be a service pack update to windows XP; thus too little too late.
And in linux or Mac such a change would go without such propaganda but rather with a silent kernel change that would make life easier.
Sure it would. Were you not around when Linux got an O(1) scheduling algorithm?
And in linux or Mac such a change would go without such propaganda but rather with a silent kernel change that would make life easier
You mean they way that Apple adds file versioning and holds a press global press conference, and MS adds it a few weeks earlier but only thinks it’s worth mentioning on a couple of blogs?
You mean they way that Apple adds file versioning and holds a press global press conference, and MS adds it a few weeks earlier but only thinks it’s worth mentioning on a couple of blogs?
Actually a few years earlier, but the point still stands :-).
What will all these neat performance upgrades mean? Absolutely nothing, because when you tack on the extra performance needed to handle the Aero interface, your machine will still run slower than it used to with XPSP2. . .
Aero runs on the GPU, and it isn’t even required that you run it. Current builds of Vista are generally faster than XP.
yeah, right, that’s why the hardware requirements trippled.
If you have the necessary hardware to run Vista, it’s most likely going to run better for you on the same hardware than XP because of how it better utilizes modern hardware.
So it’ll feel faster? That’s good indeed. Well, we’ll see what the performance of Vista will be. I’m quite sure it won’t beat Linux/KDE, tough, as those get noticeably faster every release
How unsurprisingly smug of you. Yeah, Linux and to a much lesser degree KDE are getting noticeably faster with every release, but you should also realize as a fellow Linux user that KDE 3 is a much larger beast than it’s predecessors, much like GNOME 2 and GTK+2. There is no comparison seeing if KDE 2 or 1 is faster than KDE 3, there is no comparison seeing if GTK+1 and GNOME 1 are faster than their current releases. Want to run a quick Linux setup on what most would consider antiquated hardware? Yeah, you can do it, but if you’re expecting the latest major desktop environments to run beautifully, or quickly, you’re pretty deluded. You need a machine just as powerful for XP if not moreso for a snappy Linux desktop experience, I’m not talking XFCE4 or Fluxbox here.
I recently switched from GNOME to KDE as my primary desktop in my Gentoo installation, and play around a lot with XGL and Beryl, but I also have a system that can run those with no real problem because the hardware to handle it. Most distributors recommend 256-512MB of RAM for a speedy desktop experience, and an i686 processor is nice.
P.S. – It’s not that it feels faster, it’s that it is faster. Like I said before, Windows Vista is faster on modern hardware because it actually utilizes it in a more useful fashion, and smartly moves much of the load from system components that no longer have to nor should deal with it. I’m specifically talking about Aero Glass, but the same goes for a lot of the new OS. Maybe you should do yourself a favor and see for yourself if you have the ability to what kind of differences there are in terms of performance.
I have Vista on my second desktop and it certainly is a lot faster than my debian install on the same box. 1.8Ghz, 512mb mem, Radeon 9500.
“What will all these neat performance upgrades mean? Absolutely nothing, because when you tack on the extra performance needed to handle the Aero interface, your machine will still run slower than it used to with XPSP2. . .”
yeah, I love how some of the people on here make the most uneducated statements possible and can’t back them up.
You don’t have to run Aero if you don’t want to. You can be use classic if you want.
The Aero GUI is accelerated by the way of your GPU (Graphics Processing Unit) and in XP it uses your CPU (Central Processing Unit).
Am I old fashioned, or why do I hate video interviews on the internet? Ey, if I’d like to watch TV I’d turn on my TV set… Well, but maybe it’s just that 384 kbit is far too slow for such shit and that Linux’s support for playing realworld videos within browsers still is poor.
Or, realworld video’s support for people who want to be able to use their hard- and software any way they want (without DRM, proprietary codecs etc) is still bad….
Anyway, I hate video interviews as well. Give me an article, or a transcript…
You don’t have to watch it in the browser. The video is downloadable.