Microsoft unveils the final/product name for "Windows 7"

[Albuquerque]

Why does VHD make for a "major version" change? Windows PE didn't exist until XP made it's debut and cannot be created from Windows 2000 media, but it's the same "major version" as Windows 2000. The functionality that was needed was a kernel-level change of course, and the resultant benefits are undeniable for optimized workstation and server implementation and build tasks.

And while Windows PE may not be the same functionality as VHD, the technology modification involved was a pretty significant change based on what your standards seem to dictate. Nevertheless, it was a "tweak" then, and it's still a "tweak" now.

Howabout this:
Since nobody on this forum has a view into the source code that drives the Windows kernel, none of us know what a "major" upgrade this or any other Windows update will truly be. That doesn't nullify my opinion, and by that logic, it doesn't nullify your opinion either.

So, in essence, I'm agreeing with Richard. Outside of Microsoft's numbering scheme, it's pretty much an opinion as to what makes a "major" OS upgrade versus a "minor" one.

 

[chavvdarrr]

Outside of Microsoft's numbering scheme, it's pretty much an opinion as to what makes a "major" OS upgrade versus a "minor" one.

Except transitions like 3.1 -> 95 or NT4 -> NT5 or 98->NT5, or XP -> Vista when there are so many changes that noone in his mind will say the change is minor

How anyone can compare changes introduced with the transitions above with changes known so far in win7 is above me.

 

[Albuquerque]

The XP -> Vista, while still major in my book, was still "less of a major change" than the others you outlined. IMO of course

And yeah, Win7 just doesn't really look that major to me. Yay, virtual image boot -- ok, so it's making Windows PE more functional like a real OS, not unlike what Bart and all those freebie plug-ins have been doing for years. Some of the new threading semantics are also pretty sweet, but not significantly different in architectural requirements than what was done with Windows 2000 -> 2003, and those were both the same "major" kernel underneath.

 

[Richard]

I mean seriously, do you guys see "scales to 256 cores" and the only thought that crosses your minds is "meh, I don't have 256 cores anyway"? How about: hey, so if it scales like this I bet it's not just a constant value changed in the code; perhaps that's because removing kernel locks boosted multiplier when doubling cores? How does that matter to me? Obviously Intel is not releasing i7 this year...

As I said, this (and the increased physical ram limit) is a good feature to have but it just doesn't factor into my "is Win7 worth it" equation. So no, I don't simply dismiss it out of hand, I just put it into perspective. Let's stretch this argument into obsurdity: if MS presented a new Win7 feature that allowed quantum processors to use 50% less energy that would be a HUGE feature that would benefit exactly zero persons.

I have a similar thought process when evaluating 3DMark for instance: while I enjoy reading about and discussing the details of the newest version, because the benchmark doesn't run at playable framerates I don't "trust" it as a purchase decider for me. I.e. what do I care that the lighting used by 3Dmark 2009 simulates light-transport ultra-realistically if that means any game using that would run at single-digit fps?

BTW: what do you mean by "user-replaceable"?

You can't replace the NT kernel with your own, while you can replace explorer.exe with your own shell. The % of people that do that is very small but my point is that you can't put a change in these at the same level because one is essencial, the other isn't.

 

[Sxotty]

Take a look at what I underlined, if they're being careful to maintain compat they're going to take longer to do it. Rewriting Windows memory manager is not something you try every day (or year), that's why I mentioned the short turn-around time. Also, simply making it faster does not make it a major architectural change, considering we're talking about nanoseconds. Rewrite the boot process to go from 45 seconds to 10 or less, now that would be something.

IT takes 15 seconds to get to my desktop in XP

That is starting once the screen wakes up for POST. With a SSD.

 

[RudeCurve]

IT takes 15 seconds to get to my desktop in XP

That is starting once the screen wakes up for POST. With a SSD.

Booting time varies a lot from PC to PC even when running the same OS. Just to give an example, I just built a PC for a relative and it boots up extremely quickly compared to my PC, yet my PC has faster hardware and more memory. We're talking about a 2GHz Pentium dualcore with 2GB of RAM vs a 2.5GHz C2D with 4GB of RAM. I don't know if single channel or dual channel makes a difference in boot time but my system is single channel and it takes about 15 seconds to boot to the logon screen while the other PC takes half the time.

 

[Blazkowicz]

IT takes 15 seconds to get to my desktop in XP

That is starting once the screen wakes up for POST. With a SSD.

about the same for me, with the lowest end desktop CPU available (sempron LE 1100), and a few years old HDD.

 

[suryad]

IT takes 15 seconds to get to my desktop in XP

That is starting once the screen wakes up for POST. With a SSD.

Ah I am jealous. I want SSDs in raid 0. It takes about 30 seconds for my rig to boot up I think.

 

[Dominik D]

As I said, this (and the increased physical ram limit) is a good feature to have but it just doesn't factor into my "is Win7 worth it" equation. So no, I don't simply dismiss it out of hand, I just put it into perspective. Let's stretch this argument into obsurdity: if MS presented a new Win7 feature that allowed quantum processors to use 50% less energy that would be a HUGE feature that would benefit exactly zero persons.

Once again I ask you to differentiate major release from dev POV and major release from your usability POV. You've made it perfectly clear that this release has very few features that you find interesting. And that's fine by me. That's not the point though. This discussion goes like this:
- This is fine burger!
- But I don't eat meat, it's crap.
This is, once again, disappointing way of discussing stuff, especially on this board. And that's why I think it's unfortunately EOT for me.

You can't replace the NT kernel with your own, while you can replace explorer.exe with your own shell.

This is where you're wrong as well: explorer.exe is NOT shell. It's a tiny part of shell.

 

[Albuquerque]

I see you didn't respond to me ConayR, so what about my responses says to you that I'm wrong?

 

[Dominik D]

I missed your response, sorry. I'll make it brief.

First of: I don't really understand what you're trying to say. Windows PE is minor compared to VHD support in terms of dev effort. I don't understand why Windows PE is relevant in any way to Windows 7. Can you please be a little bit more clear on that? Windows PE has nothing to do with VHD boot. WinPE is a stripped down, bare metal Windows. If you want to boot from VHD you need support in bootloader. That has nothing to do with WinPE.

As for the source code: some years ago I'd take the stance "this is awesome technical forum and I'm sure there are Windows devs here; they just won't disclose themselves". Today - I don't know, I can only hope there are developers here.

 

[Richard]

Once again I ask you to differentiate major release from dev POV and major release from your usability POV. You've made it perfectly clear that this release has very few features that you find interesting. And that's fine by me. That's not the point though. This discussion goes like this:
- This is fine burger!
- But I don't eat meat, it's crap.
This is, once again, disappointing way of discussing stuff, especially on this board. And that's why I think it's unfortunately EOT for me.

You're oversimplifying my opinion way too much. Point to where I said this (or other) feature is crap. I specifically said these are good features to have. They just have zero impact on me and on 99% of the people buying Win7. If you don't believe me, listen to Teh Man talk about it.

He specifically mentions how getting rid of the system-wide dispatcher lock (there are still locks), supporting up to 256 cores, etc. are really (very) forward-looking features that won't benefit consumers (of Win7) until years later.

If you want to have a theoretical discussion on what is a major/minor release regardless of whether it has a major/minor impact on the people using Win7 then, as I mentioned before, you'll have to find someone else because that kind of discussion for me has very little interest. In fact, Microsoft themselves share that opinion.

When we started planning the release, the first thing some might think we have to decide is if Windows 7 (client) would be a “major release” or not. I put that in quotes because it turns out this isn’t really something you decide nor is it something with a single answer. The magnitude of a release is as much about your perspective on the features as it is about the features themselves.

This is where you're wrong as well: explorer.exe is NOT shell. It's a tiny part of shell.

That's not what the argument was however, the argument was that changes in explorer.exe (that's why I've been specifically mentioning the filename in my previous posts) no matter how big, are ultimately lower in architectural complexity when the file can be replaced with a user alternative; whereas kernel changes, now matter how small, always have farther-reaching implications.

 

[Sxotty]

Booting time varies a lot from PC to PC even when running the same OS. Just to give an example, I just built a PC for a relative and it boots up extremely quickly compared to my PC, yet my PC has faster hardware and more memory. We're talking about a 2GHz Pentium dualcore with 2GB of RAM vs a 2.5GHz C2D with 4GB of RAM. I don't know if single channel or dual channel makes a difference in boot time but my system is single channel and it takes about 15 seconds to boot to the logon screen while the other PC takes half the time.

If you exclude the motherboard crap you would think it should be similar. My high end rig (high for me not youguys) booted a lot slower (with a raptor) simply b/c the motherboard takes forever to allow windows to start booting.

In my car once the windows screen shows up with the blue blocks the screen flashes and it is at the desktop within 2 seconds. It takes the same amount of time as a resolution change basically so I don't really know how long it is. The POST process there is much quicker as well, though there is still stuff afterward that takes awhile (to get up to around 15 seconds).

From that perspective the motherboard is the most important thing in the time to desktop. It actually sucks b/c I had planned to upgrade my car PC from my current desktop when I upgraded that, but the motherboard is so slow it would be completely intolerable. I wonder if there is a listing of quick booting motherboards. The car one is a cheapo Jetway +sempron under-volted and the desktop was a more fancy pants Asus+core 2 duo 2.67Ghz.

 

[Albuquerque]

I missed your response, sorry. I'll make it brief.

First of: I don't really understand what you're trying to say. Windows PE is minor compared to VHD support in terms of dev effort. I don't understand why Windows PE is relevant in any way to Windows 7. Can you please be a little bit more clear on that? Windows PE has nothing to do with VHD boot. WinPE is a stripped down, bare metal Windows. If you want to boot from VHD you need support in bootloader. That has nothing to do with WinPE.

As for the source code: some years ago I'd take the stance "this is awesome technical forum and I'm sure there are Windows devs here; they just won't disclose themselves". Today - I don't know, I can only hope there are developers here.

Obviously you don't know much about Windows PE then.

Windows PE in XP/2003 boots from an SDI file, which is (surprise, surprise!) a hard-drive-in-an-image file. SDI isn't the same as VHD in terms of file instancing and doesn't allow multiple partitions in a single file, but booting from a disk image file isn't significantly different now than it was then. This wasn't available in Windows 2000 by any stretch of the imagination, but tada, there it is in XP and later.

Better yet, Windows PE in the Vista platform boots from a WIM file, which provides pretty much the same data storage platform as VHD minus sector-level emulation. Nevertheless, both the SDI and the WIM boot image process required bootloader support, so again, why is VHD so "radically different" this time around? You've still yet to make a convincing argument of this.

 

[Humus]

As an example, Vista's explorer is missing the "go back a directory rung" button from XP which many people used.

That's not a missing feature since the exact same functionality (and much more, in fact) is reachable in one click as well.

O RLY?

There are plenty of cases where you need more than one click to go up one directory. The most frustrating cases are the ones like in the screenshot. Then there's the case where you've just copied the address from the address bar, so it's in text mode. Then you need to click elsewhere first. And then there's the case where the window is small and the name of the folder is long enough so that the previous directory doesn't fit. Finally, the exact point where you need to click is not constant, which often makes this way of navigating slower for the very common one directory up case, not to mention that I still after having used Vista for about two years now I still can't seem to get it right all the times. Often I click the top most folder, when I need to click the second. This keeps confusing me to this day.

 

[Randell]

QFT - yep the single most annoying thing about Vista IMHO.

 

[aaaaa00]

This is another very cool talk about the new kernel features coming out in Win7/Windows2008R2 Server (they share the same kernel), especially NUMA support and > 64 CPU support.

http://channel9.msdn.com/pdc2008/ES20/

"Windows Server 2008 R2 is supposed to be a 'minor' release. But being on the kernel team and working on this for a number of years, I would say that for us -- the foundation of the server -- this is the biggest release we've had since Windows 2000." --Arie van der Hoeven, Senior Program Manager, Windows Kernel Group​

This is really exciting stuff to me: even though this stuff is server oriented now, what is "big iron" today will be on your desk someday -- and probably sooner than you think.

People are talking about client machines (not just servers) with dozens and perhaps hundreds of cores. Look at where things like Larabee are rumored to be going. Or even just look at Nehalem -- a quad socket Nehalem can already have up to 64 logical cores (4 sockets x 8 cores x 2 threads).

IMHO, the more you dig into it, the more obvious it becomes that this stuff is NOT "minor tweaks" -- it looks like Microsoft is going bigtime into parallel computing. Changes to the OS to efficiently support hundreds of CPUs (they tested up to 256 because that was the biggest single machine they could get their hands on), user mode thread scheduling, the VS 2010 parallel runtime, DX11 compute shaders...

(Here's an interesting PDC talk on DX11, including compute shaders: http://channel9.msdn.com/pdc2008/PC05/)

To me it seems like they're maneuvering lots of pieces together, with the end goal of providing an OS, runtime, and great tools for developers to solve parallel compute problems.

 

[hoom]

Do I get my animated desktop fire yet?

I've always wanted a cool fire effect generated live & running along the bottom of the screen :smile:

 

[Richard]

Do I get my animated desktop fire yet?

I've always wanted a cool fire effect generated live & running along the bottom of the screen :smile:

Isn't that what DreamScene is for?

O RLY?

Where is the obligatory owl pic?

EDIT: for people testing the PDC build, it seems some functionality of the superbar and desktop is hidden and can be unlocked. If you're interested head over to Rafael's place; the guy never gives up.

 

[Humus]

Where is the obligatory owl pic?

There you go!