That means if the application is poorly written and stops handling frame window commands at any point you can't even minimize the window until it gets done processing. Minimize, kill and move should pretty much never stop working for any given window, even if the application is displaying a goddamn modal dialog box (Another pet peeve of mine and Microsoft seems to encourage programming by modal dialog.)
You have no idea what you are talking about. I think you are confusing the 'single input queue of OS/2' with Windows, which since Win3.x has always had a multi-input queue.
Vista also has several changes that address this even futher. For example the composer can even redraw unresponsive applications without any I/O lock.
Anyone that has used Windows with an NT base like 2k/XP/Vista knows that 99% of the time you can still 'Close and sometimes Minimize/Move' a crashed application; and in Vista it is 100% of the time on all of the above.
Meanwhile OSX and E17 demonstrate that you can put a glitzy interface on an OS that's quite suitable for server purposes
You are kidding right? Have you ever even seen performance numbers comparing Windows 2003 server to OSX Server? Have you even seen deployments of remote RDP users on a Windows 2003 server with all the themes and UI glitz of XP active?
The scary thing is that Longhorn even takes this to the next level, letting remote users run the 3D Aero interface remotely, fully accelerated locally because the Vista/Longhorn composer is pusing Vector and 3D information over RDP. Lets see you run a 3D application on any other Server OS or even Desktop OS 4,000 miles away with hardware acceleration and with a 3D UI with all the glitz. And this is something Vista does today, and Longhorn Beta will do later this year. I have seen our techs easily using glass and accelerated 3D applications from a Vista or Longhorn server session on a 56K connection, which is past impressive to being a bit scary.
I'm pretty sure the only way that Microsoft could design an OS that didn't suck would be to tear the whole thing down and start from scratch, though
And maybe if you knew what you were talking about you would understand the NT kernel of Windows is considered to be one of the best OS foundations, even from critics in the OSS world, it is the Win32 subsystem that takes a beating and MS could very easily replace it at any point.
But then again, if you had any clue you wouldn't have made the irresponsible and inaccurate statements in your post.
Next time do a google or even ask the 10 year old computer nerd that lives next door before trying to add information on something you know nothing about.
Why would Windows be able to provide multi-core GPU performance for old games presumably not written specifically to take advantage of it, but the other OSes would be unable to do the same for their apps once they had rewritten their graphics subsystems?
Well technically it is possible, but if you think you are going to get Linus or Apple to completely rewrite their video subsystem from the kernel and driver model up and add in a low level GPU scheduler any time soon, you have far greater expectations than I do.
Another thing to note about multi-GPU usage in Vista in contrast is just looking at older methods that currently are used by GPU companies. Even the hybrid dual chip cards are using a form of SLI, something that is quite dated.
And instead of OSes bridging this lack in the industry, we are using hardware hacks. If it was fairly easy to do what Vista is doing we would have already seen someone out there doing it in the OSS world, as there is a lot of innovation from some super bright people.
Vista was a monster in just the video subsystem and WDDM changes, let alone all the other changes that delayed its release. Not only did MS have to invent the way the new driver model works, add in GPU scheduling, and GPU virtualization, they also had to get ATI and NVidia to completely rewrite the long standing Windows consolidated drivers from scratch all the while developing a new DirectX architecture for new hardware and creating a new graphics foundation for developers that bridges desktop application and DirectX development.
MS also did all of this in a way that left the door open for legacy drivers and legacy modes of operations, not something that is easy to do considering how differently Vista operates when a WDDM driver is loaded compared to an XP or non WDDM driver is loaded. (Just the fact that it can handle old kernel level video drivers and the new user level WDDM drivers is kind of a trick.)
So with that said, it is VERY possible that other OSes will catch up, but until people in the non-MS world actually realize that Vista is actually doing some things right and WE need to pay attention so that the OSS world doesn't fall behind, the OSS world WILL fall behind.
Ignorance is not bliss in this instance, it is a nail in the coffin.
The chief complaint of the original article and of my post is that not only does the latest version of windows break this familiarity
I can agree with this, but in an industry that is driven by advancement of technology it is foolish to assume that the methods of the past will always be the most logical or functional.
Certain breaks just have to happen, or we would have to be teaching grandma how to edit config files and use chmod in addition to making sure she sets her ftp before downloading the newest version of text based solitaire.
I am both old and new school, as my career goes back a lot of years, but I have also spent a lot of time in the industry working with advancements in usability.
One old example that I use when teaching 'old school' people is that if you use WindowsXP like it is file manager in Windows 3.x or just a fancy command shell, you are missing the bigger picture of moving your thought process to a new generation that is actually easier if you take a moment to undestand why and how it works differently. And once you realize this shift in thought you can be far more productive than you ever could in older OSes and their UI concepts.
MS also seems to forget that a lot of the original adoption of Windows 3.x was from pirated copies getting PC users into a GUI and away from single tasking DOS apps.
A very subjective review with no hard facts about Vista... And featured on SlashDot, how could this be?
#1) What drivers were used? The optimized ones from NVidia or ATI? Vista has a new Video subsystem with a new driver model, and NVidia and ATI have had to write their drivers from scratch, something that maturity of the XP and other OS drivers just don't have.
#2) Was Aero left on to get the speed improvements? Turning off Aero reduces Vista video performance to XP levels, and turns off many of the accelerated features.
#3) Usability is addressed, but based on what grounds? MS spends millions on usability testing, are we all to be so stupid to conclude that their research in this area is not somewhat valid? Are they taking new users, old file manager type users, Mac users, or what? Facts please.
#4) File copy performance? Again based on what circumstances? Our internal tests show Vista can shove mass amounts of files in many settings several times faster than XP, also without exhausting the system RAM or cache as XP and prior NT bases would. I would like to see how these numbers were obtained.
#5) Menu lag? Again, was Aero turned off, how could they be showing numbers that are in direct contrast to our testing? If Aero is enabled, the UI is not only more responsive, but things like Menus and Windows opening are significantly faster than XP and especially OSX.
#6) Mouse precision? This has to be a joke right? The Windows Input model allows for extremely high resolution devices, and is SOLELY based on the input device used. If you pick up a high resolution mouse that obtains 10x the precision that a low end mouse provides in Vista it is very measurable and based upon the device. If you select another input device like a Wacom Tablet, your input resolution can be adjusted based on the device to scale in factors to several 1000 times the variances they use as examples in the article.
This can easily be demonstrated by a simple example, Ink Input in Vista is extremely high resolution, and captures at an extremely high rate.
Are they using a generic mouse and just hooking it up to the systems to get these numbers?
The mouse precision is the biggest joke of the article...
Just and FYI, for you and the parent post above you.
The Styles in XP, were a performance factor on systems with less than 80mb of RAM and around a P200 processor. On any system with more RAM or a faster processor, the XP Styles show NO measurable performance differences.
Also with regard to Vista, don't make the same mistake of loading vista and turning off the Aero interface. As the Aero interface in Vista turns on the Composer which not only allows more native drawing to be accelerated, but also accelerated applications based on the simple fact they don't have to redraw independantly.
On Vista, Aero speeds up the interface and application drawing on scale of 10x over XP, the Vista Basic interface, or using a non WDDM video driver.
The Aero in Vista is more about implementing a new video subsystem than adding 'pretty' animations to the desktop. The new video subsystem adds many new features to existing applications by accelerating GDI/WPF drawing as well as using a really smart composer that is controlled by a GPU scheduler so applications can no longer lock or fully control the GPU, something no other OS can do.
Aero was an overhaul of the interface designed to sell copies due to the "wow" factor.
Actually it is just a pretty side effect from the overhaul of the dated graphics subsystem in XP.
Vista not only caught up with graphics standards used in other OSes like OSX, but leapfrogged everything out there with the WDDM in Vista.
Only in Vista can you run multiple 3D applications and the OS implements a GPU scheduler ensuring applications don't lock access to the GPU or make other applications have to wait for another 3D application to yield.
Only in Vista can the Composer write directly to the Frame buffer from both System and GPU RAM, meaning GPU RAM constraints no longer exist. This also adds performance in that the UI doesn't have to be double buffered like OSX does to achieve the 'tear free' drawing.
We can bitch about Vista a lot, but the Aero/Glass is not desgined as a 'wow', in fact it has very little 'wow' in terms of visual appeal. However what it does do is usher in a new age of video technologies that can be used by applications and GPU architectures.
MS learned a lot from the XBox 360 development process, and the WDDM in Vista with the new subsystem is one shining example of moving video technology to the next generation, even if current applications only get a fraction of the usage or boost from it.
So when ATI and NVidia start releasing multi-core GPUs, and old Windows games automatically scale across them because Vista handles this, don't be surprised when the rest of the industry cries foul because other OSes don't have the ability to take advantage of the multicore GPUs without application being written specifically for them.
Even with single core GPU technologies of today, Vista already has an edge in being able to pre-emptively multi-task 3D applications and applications that use the GPU for non-visual computing. Again, something that just doesn't exist on other OSes.
So forget the myth that Vista created the visual effects for 'wow', they are just a nice side effect showing the basics of what is possible in Vista.
Hackers will always find a way around whatever scheme MS or any other company devises.
Very true, but in the 90s, companies pirating mass quanities of MS software was a major concern. Not only did it kill MS revenue, but it screwed consumers as they thought the copies they were buying were legitimate.
From working in the OEM world at that time, even our trusted vendors would often have workers that would offer 'discounted' OEM copies of MS software from another company all the time. It was so bad that our company had to put in place buying policies to ensure only geniune copies were ever purchased. And even with that policy in place, we had several hundred fake copies of MS software slip through OEM distributors that themselves didn't realize the copies were fake.
This is where all this crap started, but in the process MS lost sight of the goal of ensuring consumers didn't get screwed.
People in MS honestly think the Activation and Keys and WGA were good things to help the consumers to ensure they didn't get worthless copies.
MS just needs a real wakeup call that there are alternatives to dealing with mass production piracy that DO NOT harm or even bother the consumers.
Digital distribution could actually be a real solution as the control of licensing and purchasing could potentially be more easily controlled than relying on Keys and WGA to ensure copies weren't obtained from shady companies.
I don't agree with MS on this, although I do have understanding of where and why all this protection crap started.
That's ridiculous! I roll out TS and Citrix professionally for corporate clients as part of my job and they're great technologies in the right situation, but they do not come cheap.
Well since you have rolled out TS or Citrix (they aren't the same you know), you must be the expert here.
I shall in the future refer all our company clients to you. When GM or EDS calls, be sure to tell them how you once deployed a citrix and TS project professionally, that will ensure you get the contract. Geesh...
Why does it seem that everyone here is either a newb or an Apple fanboi anymore?
There are no restrictions in the OS X license saying you CANT run it in a VM. Apple just says you can only run it on apple branded hardware
Wow, I'm glad you cleared this up. Would you send us all a URL for where we can get an Apple approved VM?
Apple won't let people install their OS on anything without an Apple logo and you see this as being better than MS licensing restrictions for Home versions of Vista in a VM?
Do people frequently use the term insane when describing you?
Vista has a brand-new TCP/IP stack that was shown during beta testing to be susceptible to numerous antiquated attacks including the LAND attack which was fixed in win2k and XP. I suspect THAT error, which was a buffer overflow problem when we saw it the first time, has been fixed; but it only indicates that the new stack in Vista is probably full of more gaping holes than a pornstar convention.
Thank you for proving that you are either a shill or an ignoranus.
Ok, Vista does have a new stack that merges IPv4 and IPv6, but as for it being full of holes, would you care to cite something other than the LAND attack crap that was fixed early on in the beta process?
Since you haven't apparently heard:
"The second attack, called "Land," sends a TCP SYN packetto the target's address using the same source and destination address and ports. This can cause a target toreply to itself. The effect this attack has on a Vista target isto cause the IPv4 stack to become unresponsive for a fewseconds. This vulnerability has been addressed by build 5270."
So even pre build 5270 when it was fixed in Vista, all it did was create a moment of unresponsive behavior on the IPv4 side of the stack. I'm glad you picked a such a major 'hole' in Vista to point out. Geesh
Are you really this stupid, or just trying to get a response? If you need attention, just post something like "MS Sucks" or "I want to have Steve Job's baby" next time.
Of all the windows variants to pick, Win2k is the least secure in the NT product line.
If they wanted realtime, embedded, or a more hardened version of Windows anything from the post SP2 of XP or the Windows 2003 fork would have been a much better choice.
Let's at least hope the Win2k was based on the time this was approved, and migration plans are already in place to move to a more secure and stable version.
In comparison to Win2k - XP, Win2003, and Vista are light years ahead when just comparing stability, let alone security.
XP has been poked and beat on for 6 years and security updates are now at a level below good OSS OSes, Vista will have a honeymoon, but in theory is building upon the mistakes and fixes from XP, besides the fact that Vista is built on the Win2003 fork of NT, which has even more extensive security and stability.
I'm not saying Windows is the best choice here, but of all the versions, Win2k is at the bottom of the list as it was a major revamp of NT 4.0 and the transitional bugs and security didn't start getting ironed out until WinXP and especially the Win2003 MS security revamp era where some of the XP SP2 changes came from as well.
(For everyone out there that thinks Win2k is faster, more stable, or more secure than XP, Vista, or Win2003 you are deceiving yourself on many levels.)
Microsoft's TS licensing for a decent number of users is extortionate. A VMware player package on some central network share would be free and far more supportable, being totally copyable, deletable & restorable as flat files.
There is a lot you don't consider when stating something like this.
Let's just take one concept and let you think about it. What is better shoving VM Images over your corporate network or RDP data streams? Now consider this in an office environment where people work from other locations or outside the office.
So just to run this 'one' application, you want to shove a feg gigs across the network, and this is easier and cheaper how?
In an office with 5 people this could be debateable, but in an office with 20-1000 users, using VMs to support one or a few applications would be insane.
Thinking like yours is actually keeping companies from non-windows deployments on the desktop, as they have some crappy old windows based application people need, and they don't consider a light easy way to get that application to the desktop without having Windows running locally.
Once again I as a consumer have to ask. What is MS doing that no one else is doing?
Windows is a 'licensed use' 'closed source' OS. That up front should tell anyone what they need to know about it.
So I have to ask, why is there such outcry that you can't install certain Vista versions in a VM for production or daily use? The last time I checked you can't install OSX in a VM NO MATTER WHAT according to the Apple license.
So every user complaining about this policy from MS, should also write a letter to Apple demanding they let OSX run in VMs legally as well.
At least MS fully licenses the non Home versions to work in VMs, and still allows developers to test home versions in VMs.
So if this really angers you, then you have choices. First you should write Apple and all other Closed source OS companies that don't allow their OSes to run in VMs.
Your next choice is simple, don't like it, don't freaking use it, there are plenty alternatives.
If companies have a software product THEY NEED that only runs on Windows it would be FAR CHEAPER and easier to install a cheap Windows server and let users run that application via terminal services. Also a lot easier to deploy and support than mass amounts of VMs scattered throughout the offices.
As for developers, most developers can get free or trial copies of any windows version for testing, and you can get by the 'license' if you need to test your product on Home Basic even in a VM.
MS is also working with Xen and doing virtualization as a lot of OSS and technical people would want, yet because this puts VMWare at a disadvantage, they get to cry wolf and try to create some PR out of how they get hurt.
If VMWare wants to cry about this, then fine let them cry. But if they want to succeed then they need to create a product that is simply BETTER than MS's VM or anything out there. That is the only way they will succeed, especially considering they have the entire *nix VM Host market as MS doesn't even try to make a non Windows Host version of their VM software.
So get over it VMWare and just do what you do best.
If this was REALLY about OS licensing to run under VMs, then they would also be talking about OSX and tons of other OSes that do not allow usage in VMs; instead they are focusing only on two versions of MS Vista.
This should have been the first clue to everyone that VMWares motives are not as pure or consumer minded as they want people to believe.
until the courts decide that software patents are stupid and then microsoft will have nothing left at all.
Actually this would be a good thing for everyone. MS doesn't survive based on patents, and smaller software companies could work without fear of patent hookers.
they'll reveal one patent violation (assuming there is at least 1)
Sadly, there are many obvious features that technically could be damaged with patent claims.
From NTFS all the way to UI concepts that came out of the Word team that is used throughout GUIs on all OSes. (Things that came out of the Word team are things like drag and drop text, select and modify ie. highlight words then change their font, even the squiggles for misspelled words used in OSS software was a MS concept with protection.)
Microsoft has tons of patent and copyright leverage that we already know about, let alone internal patented concepts of the core OSes that could have been done at MS first and patented or patentable.
If the world shoves MS into a corner and they take an active/offensive role, they could cripple EVERY user level OS in existence, and they also have the money to do this.
Luckily so far MS patents have been leveraged for protection against suits or countersuit leverage. If they were as evil as some believe, MS could be making far more money from patent suits and licensing than just developing and selling software like they currently do.
So technically Balmer is right; however, Gates or someone needs to tone him down. Actually, MS could make a large PR move by just kicking Balmer to the curb and throwing his chair out with him.
Texture caching is the responsibility of the driver. Back in the days of 3DFX and their Glide API, the programmer was responsible for texture cache management, not so with Direct3D and OpenGL.
You say I have no idea what I'm talking about, then you go on to describe the EXACT THING I was talking about.
The WDDM is the new DRIVER MODEL in Vista, and Vista uses new tools opened through the driver to implement a GPU Task and GPU RAM scheduler.
And the whole point of the WDDM and Vista is that NO APPLICATION has 100% control of the GPU at any time, as GPU architectures assumed in the past. This is lower than DirectX, OpenGL, etc and DOES happen at the Vista and Vista driver level.
This is why I can run a DirectX app and OpenGL app with full hardware acceleration in Transparent windows on a Vista desktop with an animated video playing behind it. Sure the composer is creating the final image, but the GPU is being used by not only the composer, but the two applications, and both applications assume they have full scheduling and control of the GPU. But, they no longer do, Vista does give the GPU pre-emptive tasking abilities.
Vista also adds native multi-processor GPU support, so old applications can scale across GPU cores.
This will even be more important with DirectX10, and ONE reason it requires Vista, as DirectX10 is built based on these concepts and brings to the table GPUs being used for physics, etc without having to worry about an application locking the GPU to render a scene.
Everyone just needs to go read the MS/ATI/NVidia docs on this.
I wish I had more time this morning to respond to everything, I might try to follow up if I get more time later. Just a few quick points.
You're not making a distinction here between open formats and proprietary formats. (Where an "open" format is something which I consider to be unpatented, fully-documented, and (as I'll get to), truly portable).
Ok, first the sources I cited that created outrage original are NOT OSS. JAVA, PDF, etc...
OpenXML is the only exception and I should have pointed out the distinction.
However, your own definition of Open doesn't even hold in the OSS world. It is a great ideal, it just does not exist.
If you look at the licensing for many technologies in the OSS world, they are flawed, everyone from MS to Linus have criticized many of these for various reasons.
If MS were to use MANY of the OSS world technologies, they would open themselves up to being required to provide the full source of everything they do. This is NOT their business model, although there are some internal advocates inside MS that are pushing OSS into the MS world.
As you will note more OSS is coming out of MS in a free to use license with NO restrictions, even more open than GPL. But in the same note, less is being used internally as they do not want to jeopardize Windows or other development efforts that they do not want to fully disclose because they stuck a GPL piece of code in it.
Keeping these separate in a company that is closed source is hard, and MS is trying more than people realize. Apple did this the easy way, they released back only the core modifications, and then closed the project. Since NT is based on some very unique concepts, disclosing the code completely would be a large financial risk.
Give OSS a few more years, and the OSS advocates in MS to help in defining the lines, so that MS can use 'standard' OSS technologies without opening up everything they do. Eventually, it might push MS to keep opening up as they go along. Maybe not GPL, but at least released software and concepts with a fully free license that has no restrictions to even put an acknowledgement to MS in the code people use.
(Yes WMA is patent-encumbered, look at what happened to VirtualDub with ASF).
Ok, this was about them reverse engineering, not about them using WMA. There are 1000s of tools out there that use WMA that ARE NOT LICENSED and are NOT IN ANY JEOPARDY from MS. This was not about MS further closing or locking WMA. Go reverse engineer and use the SAME code you obtained and release it from any closed source project, see what happens.
I could make an equal counter argument about DIVX as they not only used code from MS MPEG4 to create the original versions, but DIVX to this day is STILL based on the MS implementation of early MPEG4.
Unfortunately, what you'll find if you read the 6000-page format specification of OpenXML* is that it is little more than a confusingly-twisted XMLization of the proprietary Office format - designed exclusively for the Office suite and virtually impossible for any other suite to support fully (in fact the spec explicitly recommends against non-Office suites implementing certain features).
The restrictions on use are not as bad as you make them seem to be. If I wanted to make a notepad applicaiton and only implement the OpenXML ability to read and store (Font, Bold, Italic), I could easily do this, even though my implementation would be FAR from what Word 2007 does with the same document.
The other thing I see you do, that others have done drives me crazy.
OpenDoc was slammed by a lot of techs because it ONLY delivered OpenOffice features and badly, and left NO ROOM for storing anything that OpenOffice couldn't do.
And the same People ran to defend OpenDoc based on these grounds are the first to scream that OpenXML is based on Office 2007 document technologies.
Double Standard...
However, to understand OpenXML, take a bigger look at the picture.
First people were mad that MS had abandoned 'industry' standards by not shipping an MP3 ripper in WMP in Windows.
Then when they do put MP3 ripping in, this is what happens?
Why was MS the only company sued, when there are thousands of companies using MP3 technology, many that didn't even pay for the license like MS did?
So when people complain about MS not using a 'standard' and instead using their own technology like WMA, maybe they should remember crap like this happens all the time.
I don't blame them for using their own toys like OpenXML or anything else, every time they have tried to work with non-MS technology they get bit for it. And anyone here could cite things like PDF,Flash,Java and numerous other technologies that do nothing but force MS to lock their doors tighter and only use their own technology.
Software patents and patent lawsuits are BAD. I don't care if it is MS, Apple, IBM or Joe Blow. And this case really smells, considering MS even paid for the licensing and tried to do it through proper channels and didn't even change or alter the technology, but instead just licensed it to give their customers choice when ripping a freaking CD.
So is anyone else ready to tell all things associated with MP3 and variants to shove it up their ass?
Just a note, Vista completely changes how and what RAM is swapped out. So no longer will a file copy or a game exhast other active processes.
Vista now prioritizes RAM based on the level and history of usage, instead of shoving everything to VM equally as NT did in the past. NT was good in that everything was considered an equal, but when it came to memory management and caching and how or what got shoved to VM, it wasn't efficient, as the app you were using the most could get shoved to the HD if you opened something wanting lots of RAM.
NT in the older days would also would shove apps to VM based on whether they were minimized, a very bad idea when users would minimize their word processor to look up something, and then have the HD grind when they restored the wordprocessor to write something.
The Vista changes are beyond just caching and superfetch, there are changes in the whole memory management architecture.
If you haven't, check out the link I posted in the parent post, it is a brief insight into what changed and why and also contrasts it to XP and previous NT versions. A good quick read for people that find this stuff interesting.
The article also gives some good ideas and insight into ways other OSes could be more efficiently handling RAM and caching as well, which I hope gets picked up in the OSS world, and doesn't just stay a hidden Vista concept because people hate MS more than they want to learn.
Quality will not decide this format war - the PS3 will.
Well, possibly, but considering the number of people that are buying the USB2.0 XBox 360 HD-DVD player just to hook up to their computer will probably have as much if not more of an impact.
Why buy a PS3, when you don't even have to buy an XBox or PS3?
Just go buy an XBox 360 HD-DVD($200US), hook it up, and if the movies are VC1 encoded, they will even play with WMP11, if not install an Mpeg4 codec, and you have both possible compression formats at a price point cheaper than OEM internal drives by a factor of 2X the savings.
And since they are just USB2.0, you should be able to get them to work with any OS, although XP and Vista have native drivers for it.
Dude. You are making a dunce out of yourself in public. In the context of GPU operation, if GPU memory is "virtualised" that means it is mapped into the OS memory areas (since it must be maintained in a persistent state from the perspective of non-directx 10 apps), so that the various applications which attempt to get at the "real" GPU memory can be held off until their "scheduled" time comes up at which time their individual idea of the GPU memory contents is mapped in.
Did you even read the links?
Vista maps the GPU and system RAM used for 'video' together, so by using the new WDDM model and a AGP or PCI/e bus, there is no need for Vista to shove the System used RAM for video back into the GPU RAM space in order to let the GPU use it or draw with it. It can stay in System RAM, and the GPU sees it as native GPU RAM.
I think this is the bigger point you are missing.
Which means that when you use Aero, the OS uses the GPU fully plus a considerable chunk of OS memory with all those currently "un-scheduled" GPU states.
What is actually happening is that Vista sacrifices memory for virtualization, thus expanding the system memory usage significantly when compared to a pre-Vista situation.
99.9% of desktop users that have 128MB on their Video card will not see this effect, the chance that you have enough active Windows open to eat into system RAM is not as likely as you might think.
Now with games, this WILL happen; however, in gaming what is more beneficial to you in terms of performance? Loading the game textures off the Hard Drive continuously, or letting the system pretend you have more GPU RAM so the textures are 'virtualized' in system RAM?
If you know anything about gaming, yanking crap off the hard drive compared to being able to use another small chunk of system RAM to hold them is going to yield a far better experience. And this is also superior to just caching them, as the Game is not having to load/request the textures continously. Instead the Game sees the textures as already loaded in GPU space and can use them as if they were sitting on the Video card itself.
Vista will move textures to the GPU RAM if they are in high performance use. So some of the initial calls to a texture will be virtualized and used from System RAM, but if the texture is frequently used and another texture is not being used, Vista will flip them out so the higher priority texture will then be sitting in GPU RAM, which is faster.
This is also why you can throw your texture quality to the roof in Vista and not see a massive performance loss, as the Game is not continually having to grab/load textures to create a scene, instead the game/application sees them as sitting in GPU RAM whether they are in the real GPU RAM or in System RAM in 'virtualized' GPU RAM space.
agree with you in the sense that all this hoo hah about Vista's caching is utterly devoid of point.
However, you should bear in mind that Linux and FreeBSD already have (and have had for a long time!) caching performance equal to or greater than that of Vista.
As far as Unix-based OS's are concerned, this has been standard. In fact, I think I'm correct in saying that XP implemented it in some minor fashion when it was released (quite a while back by the way).
Most OSes have lazy and even look ahead type of caching, Vista's is doing things a little bit different, as it looks at user usage patterns, application usage patterns, and even boot and driver usage patterns, etc.
There is a lot more going on in Vista than just 'caching', Superfetch is more about optimal caching with several new ways of handling what and when to cache data. There is also a reason ReadyBoost in Vista works, even though Flash Sticks are slower than a HD for continuous R/W.
In the OSS world, there are things we can learn and use from this, instead of dismissing Vista with the common, oh we have had these features for years, when in fact we haven't.
Even NT 3.1 had both lazy and limited look ahead caching, so yes you are right that XP had implemented different variations of caching even back in 2001; they were new to XP. However some of the caching policies in XP were both ahead of Linux and BSD and some of the caching was behind how things are done in Linux and BSD. File copying with no priority (behind), active prefetch monitoring (ahead), etc.
This article not only explains a bit about why Vista is different, but also contrasts the differences by using Win2k and XP as examples of how common caching was handled in the past. MS was well aware of the 'same priority for all' and how this didn't scale well for the user experience.
Take a look at the guts of what it is doing and why it is different than even what you find in Linux or BSD.
Except for how Vista's memory sweet spot is FOUR TIMES that of any contemporary desktop OS. Do you really consider that being in the same ballpark?
You are missing the point... Vista's sweet spot is 512mb to 1GB as well.
The difference with Vista, is that as you keep throwing RAM at it, it uses the extra RAM in its 'smart' caching system to take advantage of the 'free' RAM to increase the performance of the computer. So even if Vista and all your running applications only require 500mb or RAM, Vista will keep getting faster with extra RAM that the OS and applications don't need.
So Vista's sweet spot is where it surpasses XP (512MB to 1GB), but you can keep throwing RAM in Vista and Vista will continue to get faster as it will 'intelligently' cache more data from the Hard Drive in anticipation of it being used, hence less disk access because the data is already in the RAM cache.
If you go by the 'concept' that Vista's sweet spot is 4GB because that is as much RAM as you can throw in the computer, then on Vista 64bit, if we throw 128GB in the machine, you could also call that its sweet spot, as Vista would be using all that extra RAM to pre-cache even more of your applications and data. Although there would eventually hit a limit based on the amount of applications and data actually on your Hard Drives.
I posted this link a few times, but in case peeps have missed it, go read about this from the horses mouth and not some bad review.
Slashdot Mirror
That means if the application is poorly written and stops handling frame window commands at any point you can't even minimize the window until it gets done processing. Minimize, kill and move should pretty much never stop working for any given window, even if the application is displaying a goddamn modal dialog box (Another pet peeve of mine and Microsoft seems to encourage programming by modal dialog.)
You have no idea what you are talking about. I think you are confusing the 'single input queue of OS/2' with Windows, which since Win3.x has always had a multi-input queue.
Vista also has several changes that address this even futher. For example the composer can even redraw unresponsive applications without any I/O lock.
Anyone that has used Windows with an NT base like 2k/XP/Vista knows that 99% of the time you can still 'Close and sometimes Minimize/Move' a crashed application; and in Vista it is 100% of the time on all of the above.
Meanwhile OSX and E17 demonstrate that you can put a glitzy interface on an OS that's quite suitable for server purposes
You are kidding right? Have you ever even seen performance numbers comparing Windows 2003 server to OSX Server? Have you even seen deployments of remote RDP users on a Windows 2003 server with all the themes and UI glitz of XP active?
The scary thing is that Longhorn even takes this to the next level, letting remote users run the 3D Aero interface remotely, fully accelerated locally because the Vista/Longhorn composer is pusing Vector and 3D information over RDP. Lets see you run a 3D application on any other Server OS or even Desktop OS 4,000 miles away with hardware acceleration and with a 3D UI with all the glitz. And this is something Vista does today, and Longhorn Beta will do later this year. I have seen our techs easily using glass and accelerated 3D applications from a Vista or Longhorn server session on a 56K connection, which is past impressive to being a bit scary.
I'm pretty sure the only way that Microsoft could design an OS that didn't suck would be to tear the whole thing down and start from scratch, though
And maybe if you knew what you were talking about you would understand the NT kernel of Windows is considered to be one of the best OS foundations, even from critics in the OSS world, it is the Win32 subsystem that takes a beating and MS could very easily replace it at any point.
But then again, if you had any clue you wouldn't have made the irresponsible and inaccurate statements in your post.
Next time do a google or even ask the 10 year old computer nerd that lives next door before trying to add information on something you know nothing about.
Why would Windows be able to provide multi-core GPU performance for old games presumably not written specifically to take advantage of it, but the other OSes would be unable to do the same for their apps once they had rewritten their graphics subsystems?
Well technically it is possible, but if you think you are going to get Linus or Apple to completely rewrite their video subsystem from the kernel and driver model up and add in a low level GPU scheduler any time soon, you have far greater expectations than I do.
Another thing to note about multi-GPU usage in Vista in contrast is just looking at older methods that currently are used by GPU companies. Even the hybrid dual chip cards are using a form of SLI, something that is quite dated.
And instead of OSes bridging this lack in the industry, we are using hardware hacks. If it was fairly easy to do what Vista is doing we would have already seen someone out there doing it in the OSS world, as there is a lot of innovation from some super bright people.
Vista was a monster in just the video subsystem and WDDM changes, let alone all the other changes that delayed its release. Not only did MS have to invent the way the new driver model works, add in GPU scheduling, and GPU virtualization, they also had to get ATI and NVidia to completely rewrite the long standing Windows consolidated drivers from scratch all the while developing a new DirectX architecture for new hardware and creating a new graphics foundation for developers that bridges desktop application and DirectX development.
MS also did all of this in a way that left the door open for legacy drivers and legacy modes of operations, not something that is easy to do considering how differently Vista operates when a WDDM driver is loaded compared to an XP or non WDDM driver is loaded. (Just the fact that it can handle old kernel level video drivers and the new user level WDDM drivers is kind of a trick.)
So with that said, it is VERY possible that other OSes will catch up, but until people in the non-MS world actually realize that Vista is actually doing some things right and WE need to pay attention so that the OSS world doesn't fall behind, the OSS world WILL fall behind.
Ignorance is not bliss in this instance, it is a nail in the coffin.
The chief complaint of the original article and of my post is that not only does the latest version of windows break this familiarity
I can agree with this, but in an industry that is driven by advancement of technology it is foolish to assume that the methods of the past will always be the most logical or functional.
Certain breaks just have to happen, or we would have to be teaching grandma how to edit config files and use chmod in addition to making sure she sets her ftp before downloading the newest version of text based solitaire.
I am both old and new school, as my career goes back a lot of years, but I have also spent a lot of time in the industry working with advancements in usability.
One old example that I use when teaching 'old school' people is that if you use WindowsXP like it is file manager in Windows 3.x or just a fancy command shell, you are missing the bigger picture of moving your thought process to a new generation that is actually easier if you take a moment to undestand why and how it works differently. And once you realize this shift in thought you can be far more productive than you ever could in older OSes and their UI concepts.
If pirating "killed" MS revenue, come one
I did say 'kill' not killed. lol
MS also seems to forget that a lot of the original adoption of Windows 3.x was from pirated copies getting PC users into a GUI and away from single tasking DOS apps.
A very subjective review with no hard facts about Vista... And featured on SlashDot, how could this be?
#1) What drivers were used? The optimized ones from NVidia or ATI? Vista has a new Video subsystem with a new driver model, and NVidia and ATI have had to write their drivers from scratch, something that maturity of the XP and other OS drivers just don't have.
#2) Was Aero left on to get the speed improvements? Turning off Aero reduces Vista video performance to XP levels, and turns off many of the accelerated features.
#3) Usability is addressed, but based on what grounds? MS spends millions on usability testing, are we all to be so stupid to conclude that their research in this area is not somewhat valid? Are they taking new users, old file manager type users, Mac users, or what? Facts please.
#4) File copy performance? Again based on what circumstances? Our internal tests show Vista can shove mass amounts of files in many settings several times faster than XP, also without exhausting the system RAM or cache as XP and prior NT bases would. I would like to see how these numbers were obtained.
#5) Menu lag? Again, was Aero turned off, how could they be showing numbers that are in direct contrast to our testing? If Aero is enabled, the UI is not only more responsive, but things like Menus and Windows opening are significantly faster than XP and especially OSX.
#6) Mouse precision? This has to be a joke right? The Windows Input model allows for extremely high resolution devices, and is SOLELY based on the input device used. If you pick up a high resolution mouse that obtains 10x the precision that a low end mouse provides in Vista it is very measurable and based upon the device. If you select another input device like a Wacom Tablet, your input resolution can be adjusted based on the device to scale in factors to several 1000 times the variances they use as examples in the article.
This can easily be demonstrated by a simple example, Ink Input in Vista is extremely high resolution, and captures at an extremely high rate.
Are they using a generic mouse and just hooking it up to the systems to get these numbers?
The mouse precision is the biggest joke of the article...
Just and FYI, for you and the parent post above you.
The Styles in XP, were a performance factor on systems with less than 80mb of RAM and around a P200 processor. On any system with more RAM or a faster processor, the XP Styles show NO measurable performance differences.
Also with regard to Vista, don't make the same mistake of loading vista and turning off the Aero interface. As the Aero interface in Vista turns on the Composer which not only allows more native drawing to be accelerated, but also accelerated applications based on the simple fact they don't have to redraw independantly.
On Vista, Aero speeds up the interface and application drawing on scale of 10x over XP, the Vista Basic interface, or using a non WDDM video driver.
The Aero in Vista is more about implementing a new video subsystem than adding 'pretty' animations to the desktop. The new video subsystem adds many new features to existing applications by accelerating GDI/WPF drawing as well as using a really smart composer that is controlled by a GPU scheduler so applications can no longer lock or fully control the GPU, something no other OS can do.
Aero was an overhaul of the interface designed to sell copies due to the "wow" factor.
Actually it is just a pretty side effect from the overhaul of the dated graphics subsystem in XP.
Vista not only caught up with graphics standards used in other OSes like OSX, but leapfrogged everything out there with the WDDM in Vista.
Only in Vista can you run multiple 3D applications and the OS implements a GPU scheduler ensuring applications don't lock access to the GPU or make other applications have to wait for another 3D application to yield.
Only in Vista can the Composer write directly to the Frame buffer from both System and GPU RAM, meaning GPU RAM constraints no longer exist. This also adds performance in that the UI doesn't have to be double buffered like OSX does to achieve the 'tear free' drawing.
We can bitch about Vista a lot, but the Aero/Glass is not desgined as a 'wow', in fact it has very little 'wow' in terms of visual appeal. However what it does do is usher in a new age of video technologies that can be used by applications and GPU architectures.
MS learned a lot from the XBox 360 development process, and the WDDM in Vista with the new subsystem is one shining example of moving video technology to the next generation, even if current applications only get a fraction of the usage or boost from it.
So when ATI and NVidia start releasing multi-core GPUs, and old Windows games automatically scale across them because Vista handles this, don't be surprised when the rest of the industry cries foul because other OSes don't have the ability to take advantage of the multicore GPUs without application being written specifically for them.
Even with single core GPU technologies of today, Vista already has an edge in being able to pre-emptively multi-task 3D applications and applications that use the GPU for non-visual computing. Again, something that just doesn't exist on other OSes.
So forget the myth that Vista created the visual effects for 'wow', they are just a nice side effect showing the basics of what is possible in Vista.
Hackers will always find a way around whatever scheme MS or any other company devises.
Very true, but in the 90s, companies pirating mass quanities of MS software was a major concern. Not only did it kill MS revenue, but it screwed consumers as they thought the copies they were buying were legitimate.
From working in the OEM world at that time, even our trusted vendors would often have workers that would offer 'discounted' OEM copies of MS software from another company all the time. It was so bad that our company had to put in place buying policies to ensure only geniune copies were ever purchased. And even with that policy in place, we had several hundred fake copies of MS software slip through OEM distributors that themselves didn't realize the copies were fake.
This is where all this crap started, but in the process MS lost sight of the goal of ensuring consumers didn't get screwed.
People in MS honestly think the Activation and Keys and WGA were good things to help the consumers to ensure they didn't get worthless copies.
MS just needs a real wakeup call that there are alternatives to dealing with mass production piracy that DO NOT harm or even bother the consumers.
Digital distribution could actually be a real solution as the control of licensing and purchasing could potentially be more easily controlled than relying on Keys and WGA to ensure copies weren't obtained from shady companies.
I don't agree with MS on this, although I do have understanding of where and why all this protection crap started.
It seems even MS is willing to admit a major screw up and is willing to give consumers the benefit of the doubt once again.
Do you think people inside MS are starting to finally stop listening to Ballmer and his business/money only mentality?
It could be quite earth shaking for MS to start caring about consumers more than $$ again.
That's ridiculous! I roll out TS and Citrix professionally for corporate clients as part of my job and they're great technologies in the right situation, but they do not come cheap.
Well since you have rolled out TS or Citrix (they aren't the same you know), you must be the expert here.
I shall in the future refer all our company clients to you. When GM or EDS calls, be sure to tell them how you once deployed a citrix and TS project professionally, that will ensure you get the contract. Geesh...
Why does it seem that everyone here is either a newb or an Apple fanboi anymore?
There are no restrictions in the OS X license saying you CANT run it in a VM. Apple just says you can only run it on apple branded hardware
Wow, I'm glad you cleared this up. Would you send us all a URL for where we can get an Apple approved VM?
Apple won't let people install their OS on anything without an Apple logo and you see this as being better than MS licensing restrictions for Home versions of Vista in a VM?
Do people frequently use the term insane when describing you?
Vista has a brand-new TCP/IP stack that was shown during beta testing to be susceptible to numerous antiquated attacks including the LAND attack which was fixed in win2k and XP. I suspect THAT error, which was a buffer overflow problem when we saw it the first time, has been fixed; but it only indicates that the new stack in Vista is probably full of more gaping holes than a pornstar convention.
Thank you for proving that you are either a shill or an ignoranus.
Ok, Vista does have a new stack that merges IPv4 and IPv6, but as for it being full of holes, would you care to cite something other than the LAND attack crap that was fixed early on in the beta process?
Since you haven't apparently heard:
"The second attack, called "Land," sends a TCP SYN packetto the target's address using the same source and destination address and ports. This can cause a target toreply to itself. The effect this attack has on a Vista target isto cause the IPv4 stack to become unresponsive for a fewseconds. This vulnerability has been addressed by build 5270."
So even pre build 5270 when it was fixed in Vista, all it did was create a moment of unresponsive behavior on the IPv4 side of the stack. I'm glad you picked a such a major 'hole' in Vista to point out. Geesh
Are you really this stupid, or just trying to get a response? If you need attention, just post something like "MS Sucks" or "I want to have Steve Job's baby" next time.
Of all the windows variants to pick, Win2k is the least secure in the NT product line.
If they wanted realtime, embedded, or a more hardened version of Windows anything from the post SP2 of XP or the Windows 2003 fork would have been a much better choice.
Let's at least hope the Win2k was based on the time this was approved, and migration plans are already in place to move to a more secure and stable version.
In comparison to Win2k - XP, Win2003, and Vista are light years ahead when just comparing stability, let alone security.
XP has been poked and beat on for 6 years and security updates are now at a level below good OSS OSes, Vista will have a honeymoon, but in theory is building upon the mistakes and fixes from XP, besides the fact that Vista is built on the Win2003 fork of NT, which has even more extensive security and stability.
I'm not saying Windows is the best choice here, but of all the versions, Win2k is at the bottom of the list as it was a major revamp of NT 4.0 and the transitional bugs and security didn't start getting ironed out until WinXP and especially the Win2003 MS security revamp era where some of the XP SP2 changes came from as well.
(For everyone out there that thinks Win2k is faster, more stable, or more secure than XP, Vista, or Win2003 you are deceiving yourself on many levels.)
Microsoft's TS licensing for a decent number of users is extortionate. A VMware player package on some central network share would be free and far more supportable, being totally copyable, deletable & restorable as flat files.
There is a lot you don't consider when stating something like this.
Let's just take one concept and let you think about it. What is better shoving VM Images over your corporate network or RDP data streams? Now consider this in an office environment where people work from other locations or outside the office.
So just to run this 'one' application, you want to shove a feg gigs across the network, and this is easier and cheaper how?
In an office with 5 people this could be debateable, but in an office with 20-1000 users, using VMs to support one or a few applications would be insane.
Thinking like yours is actually keeping companies from non-windows deployments on the desktop, as they have some crappy old windows based application people need, and they don't consider a light easy way to get that application to the desktop without having Windows running locally.
Once again I as a consumer have to ask. What is MS doing that no one else is doing?
Windows is a 'licensed use' 'closed source' OS. That up front should tell anyone what they need to know about it.
So I have to ask, why is there such outcry that you can't install certain Vista versions in a VM for production or daily use? The last time I checked you can't install OSX in a VM NO MATTER WHAT according to the Apple license.
So every user complaining about this policy from MS, should also write a letter to Apple demanding they let OSX run in VMs legally as well.
At least MS fully licenses the non Home versions to work in VMs, and still allows developers to test home versions in VMs.
So if this really angers you, then you have choices. First you should write Apple and all other Closed source OS companies that don't allow their OSes to run in VMs.
Your next choice is simple, don't like it, don't freaking use it, there are plenty alternatives.
If companies have a software product THEY NEED that only runs on Windows it would be FAR CHEAPER and easier to install a cheap Windows server and let users run that application via terminal services. Also a lot easier to deploy and support than mass amounts of VMs scattered throughout the offices.
As for developers, most developers can get free or trial copies of any windows version for testing, and you can get by the 'license' if you need to test your product on Home Basic even in a VM.
MS is also working with Xen and doing virtualization as a lot of OSS and technical people would want, yet because this puts VMWare at a disadvantage, they get to cry wolf and try to create some PR out of how they get hurt.
If VMWare wants to cry about this, then fine let them cry. But if they want to succeed then they need to create a product that is simply BETTER than MS's VM or anything out there. That is the only way they will succeed, especially considering they have the entire *nix VM Host market as MS doesn't even try to make a non Windows Host version of their VM software.
So get over it VMWare and just do what you do best.
If this was REALLY about OS licensing to run under VMs, then they would also be talking about OSX and tons of other OSes that do not allow usage in VMs; instead they are focusing only on two versions of MS Vista.
This should have been the first clue to everyone that VMWares motives are not as pure or consumer minded as they want people to believe.
until the courts decide that software patents are stupid and then microsoft will have nothing left at all.
Actually this would be a good thing for everyone. MS doesn't survive based on patents, and smaller software companies could work without fear of patent hookers.
they'll reveal one patent violation (assuming there is at least 1)
Sadly, there are many obvious features that technically could be damaged with patent claims.
From NTFS all the way to UI concepts that came out of the Word team that is used throughout GUIs on all OSes. (Things that came out of the Word team are things like drag and drop text, select and modify ie. highlight words then change their font, even the squiggles for misspelled words used in OSS software was a MS concept with protection.)
Microsoft has tons of patent and copyright leverage that we already know about, let alone internal patented concepts of the core OSes that could have been done at MS first and patented or patentable.
If the world shoves MS into a corner and they take an active/offensive role, they could cripple EVERY user level OS in existence, and they also have the money to do this.
Luckily so far MS patents have been leveraged for protection against suits or countersuit leverage. If they were as evil as some believe, MS could be making far more money from patent suits and licensing than just developing and selling software like they currently do.
So technically Balmer is right; however, Gates or someone needs to tone him down. Actually, MS could make a large PR move by just kicking Balmer to the curb and throwing his chair out with him.
Texture caching is the responsibility of the driver. Back in the days of 3DFX and their Glide API, the programmer was responsible for texture cache management, not so with Direct3D and OpenGL.
You say I have no idea what I'm talking about, then you go on to describe the EXACT THING I was talking about.
The WDDM is the new DRIVER MODEL in Vista, and Vista uses new tools opened through the driver to implement a GPU Task and GPU RAM scheduler.
And the whole point of the WDDM and Vista is that NO APPLICATION has 100% control of the GPU at any time, as GPU architectures assumed in the past. This is lower than DirectX, OpenGL, etc and DOES happen at the Vista and Vista driver level.
This is why I can run a DirectX app and OpenGL app with full hardware acceleration in Transparent windows on a Vista desktop with an animated video playing behind it. Sure the composer is creating the final image, but the GPU is being used by not only the composer, but the two applications, and both applications assume they have full scheduling and control of the GPU. But, they no longer do, Vista does give the GPU pre-emptive tasking abilities.
Vista also adds native multi-processor GPU support, so old applications can scale across GPU cores.
This will even be more important with DirectX10, and ONE reason it requires Vista, as DirectX10 is built based on these concepts and brings to the table GPUs being used for physics, etc without having to worry about an application locking the GPU to render a scene.
Everyone just needs to go read the MS/ATI/NVidia docs on this.
I wish I had more time this morning to respond to everything, I might try to follow up if I get more time later. Just a few quick points.
You're not making a distinction here between open formats and proprietary formats. (Where an "open" format is something which I consider to be unpatented, fully-documented, and (as I'll get to), truly portable).
Ok, first the sources I cited that created outrage original are NOT OSS. JAVA, PDF, etc...
OpenXML is the only exception and I should have pointed out the distinction.
However, your own definition of Open doesn't even hold in the OSS world. It is a great ideal, it just does not exist.
If you look at the licensing for many technologies in the OSS world, they are flawed, everyone from MS to Linus have criticized many of these for various reasons.
If MS were to use MANY of the OSS world technologies, they would open themselves up to being required to provide the full source of everything they do. This is NOT their business model, although there are some internal advocates inside MS that are pushing OSS into the MS world.
As you will note more OSS is coming out of MS in a free to use license with NO restrictions, even more open than GPL. But in the same note, less is being used internally as they do not want to jeopardize Windows or other development efforts that they do not want to fully disclose because they stuck a GPL piece of code in it.
Keeping these separate in a company that is closed source is hard, and MS is trying more than people realize. Apple did this the easy way, they released back only the core modifications, and then closed the project. Since NT is based on some very unique concepts, disclosing the code completely would be a large financial risk.
Give OSS a few more years, and the OSS advocates in MS to help in defining the lines, so that MS can use 'standard' OSS technologies without opening up everything they do. Eventually, it might push MS to keep opening up as they go along. Maybe not GPL, but at least released software and concepts with a fully free license that has no restrictions to even put an acknowledgement to MS in the code people use.
(Yes WMA is patent-encumbered, look at what happened to VirtualDub with ASF).
Ok, this was about them reverse engineering, not about them using WMA. There are 1000s of tools out there that use WMA that ARE NOT LICENSED and are NOT IN ANY JEOPARDY from MS. This was not about MS further closing or locking WMA. Go reverse engineer and use the SAME code you obtained and release it from any closed source project, see what happens.
I could make an equal counter argument about DIVX as they not only used code from MS MPEG4 to create the original versions, but DIVX to this day is STILL based on the MS implementation of early MPEG4.
Unfortunately, what you'll find if you read the 6000-page format specification of OpenXML* is that it is little more than a confusingly-twisted XMLization of the proprietary Office format - designed exclusively for the Office suite and virtually impossible for any other suite to support fully (in fact the spec explicitly recommends against non-Office suites implementing certain features).
The restrictions on use are not as bad as you make them seem to be. If I wanted to make a notepad applicaiton and only implement the OpenXML ability to read and store (Font, Bold, Italic), I could easily do this, even though my implementation would be FAR from what Word 2007 does with the same document.
The other thing I see you do, that others have done drives me crazy.
OpenDoc was slammed by a lot of techs because it ONLY delivered OpenOffice features and badly, and left NO ROOM for storing anything that OpenOffice couldn't do.
And the same People ran to defend OpenDoc based on these grounds are the first to scream that OpenXML is based on Office 2007 document technologies.
Double Standard...
However, to understand OpenXML, take a bigger look at the picture.
First people were mad that MS had abandoned 'industry' standards by not shipping an MP3 ripper in WMP in Windows.
Then when they do put MP3 ripping in, this is what happens?
Why was MS the only company sued, when there are thousands of companies using MP3 technology, many that didn't even pay for the license like MS did?
So when people complain about MS not using a 'standard' and instead using their own technology like WMA, maybe they should remember crap like this happens all the time.
I don't blame them for using their own toys like OpenXML or anything else, every time they have tried to work with non-MS technology they get bit for it. And anyone here could cite things like PDF,Flash,Java and numerous other technologies that do nothing but force MS to lock their doors tighter and only use their own technology.
Software patents and patent lawsuits are BAD. I don't care if it is MS, Apple, IBM or Joe Blow. And this case really smells, considering MS even paid for the licensing and tried to do it through proper channels and didn't even change or alter the technology, but instead just licensed it to give their customers choice when ripping a freaking CD.
So is anyone else ready to tell all things associated with MP3 and variants to shove it up their ass?
Just a note, Vista completely changes how and what RAM is swapped out. So no longer will a file copy or a game exhast other active processes.
Vista now prioritizes RAM based on the level and history of usage, instead of shoving everything to VM equally as NT did in the past. NT was good in that everything was considered an equal, but when it came to memory management and caching and how or what got shoved to VM, it wasn't efficient, as the app you were using the most could get shoved to the HD if you opened something wanting lots of RAM.
NT in the older days would also would shove apps to VM based on whether they were minimized, a very bad idea when users would minimize their word processor to look up something, and then have the HD grind when they restored the wordprocessor to write something.
The Vista changes are beyond just caching and superfetch, there are changes in the whole memory management architecture.
If you haven't, check out the link I posted in the parent post, it is a brief insight into what changed and why and also contrasts it to XP and previous NT versions. A good quick read for people that find this stuff interesting.
The article also gives some good ideas and insight into ways other OSes could be more efficiently handling RAM and caching as well, which I hope gets picked up in the OSS world, and doesn't just stay a hidden Vista concept because people hate MS more than they want to learn.
Take Care.
Quality will not decide this format war - the PS3 will.
Well, possibly, but considering the number of people that are buying the USB2.0 XBox 360 HD-DVD player just to hook up to their computer will probably have as much if not more of an impact.
Why buy a PS3, when you don't even have to buy an XBox or PS3?
Just go buy an XBox 360 HD-DVD($200US), hook it up, and if the movies are VC1 encoded, they will even play with WMP11, if not install an Mpeg4 codec, and you have both possible compression formats at a price point cheaper than OEM internal drives by a factor of 2X the savings.
And since they are just USB2.0, you should be able to get them to work with any OS, although XP and Vista have native drivers for it.
Dude. You are making a dunce out of yourself in public. In the context of GPU operation, if GPU memory is "virtualised" that means it is mapped into the OS memory areas (since it must be maintained in a persistent state from the perspective of non-directx 10 apps), so that the various applications which attempt to get at the "real" GPU memory can be held off until their "scheduled" time comes up at which time their individual idea of the GPU memory contents is mapped in.
Did you even read the links?
Vista maps the GPU and system RAM used for 'video' together, so by using the new WDDM model and a AGP or PCI/e bus, there is no need for Vista to shove the System used RAM for video back into the GPU RAM space in order to let the GPU use it or draw with it. It can stay in System RAM, and the GPU sees it as native GPU RAM.
I think this is the bigger point you are missing.
Which means that when you use Aero, the OS uses the GPU fully plus a considerable chunk of OS memory with all those currently "un-scheduled" GPU states.
What is actually happening is that Vista sacrifices memory for virtualization, thus expanding the system memory usage significantly when compared to a pre-Vista situation.
99.9% of desktop users that have 128MB on their Video card will not see this effect, the chance that you have enough active Windows open to eat into system RAM is not as likely as you might think.
Now with games, this WILL happen; however, in gaming what is more beneficial to you in terms of performance? Loading the game textures off the Hard Drive continuously, or letting the system pretend you have more GPU RAM so the textures are 'virtualized' in system RAM?
If you know anything about gaming, yanking crap off the hard drive compared to being able to use another small chunk of system RAM to hold them is going to yield a far better experience. And this is also superior to just caching them, as the Game is not having to load/request the textures continously. Instead the Game sees the textures as already loaded in GPU space and can use them as if they were sitting on the Video card itself.
Vista will move textures to the GPU RAM if they are in high performance use. So some of the initial calls to a texture will be virtualized and used from System RAM, but if the texture is frequently used and another texture is not being used, Vista will flip them out so the higher priority texture will then be sitting in GPU RAM, which is faster.
This is also why you can throw your texture quality to the roof in Vista and not see a massive performance loss, as the Game is not continually having to grab/load textures to create a scene, instead the game/application sees them as sitting in GPU RAM whether they are in the real GPU RAM or in System RAM in 'virtualized' GPU RAM space.
Virtualization is a Good thing...
agree with you in the sense that all this hoo hah about Vista's caching is utterly devoid of point.
s /2007/03/VistaKernel/
However, you should bear in mind that Linux and FreeBSD already have (and have had for a long time!) caching performance equal to or greater than that of Vista.
As far as Unix-based OS's are concerned, this has been standard. In fact, I think I'm correct in saying that XP implemented it in some minor fashion when it was released (quite a while back by the way).
Most OSes have lazy and even look ahead type of caching, Vista's is doing things a little bit different, as it looks at user usage patterns, application usage patterns, and even boot and driver usage patterns, etc.
There is a lot more going on in Vista than just 'caching', Superfetch is more about optimal caching with several new ways of handling what and when to cache data. There is also a reason ReadyBoost in Vista works, even though Flash Sticks are slower than a HD for continuous R/W.
In the OSS world, there are things we can learn and use from this, instead of dismissing Vista with the common, oh we have had these features for years, when in fact we haven't.
Even NT 3.1 had both lazy and limited look ahead caching, so yes you are right that XP had implemented different variations of caching even back in 2001; they were new to XP. However some of the caching policies in XP were both ahead of Linux and BSD and some of the caching was behind how things are done in Linux and BSD. File copying with no priority (behind), active prefetch monitoring (ahead), etc.
This article not only explains a bit about why Vista is different, but also contrasts the differences by using Win2k and XP as examples of how common caching was handled in the past. MS was well aware of the 'same priority for all' and how this didn't scale well for the user experience.
Take a look at the guts of what it is doing and why it is different than even what you find in Linux or BSD.
http://www.microsoft.com/technet/technetmag/issue
Except for how Vista's memory sweet spot is FOUR TIMES that of any contemporary desktop OS. Do you really consider that being in the same ballpark?
s /2007/03/VistaKernel/
You are missing the point... Vista's sweet spot is 512mb to 1GB as well.
The difference with Vista, is that as you keep throwing RAM at it, it uses the extra RAM in its 'smart' caching system to take advantage of the 'free' RAM to increase the performance of the computer. So even if Vista and all your running applications only require 500mb or RAM, Vista will keep getting faster with extra RAM that the OS and applications don't need.
So Vista's sweet spot is where it surpasses XP (512MB to 1GB), but you can keep throwing RAM in Vista and Vista will continue to get faster as it will 'intelligently' cache more data from the Hard Drive in anticipation of it being used, hence less disk access because the data is already in the RAM cache.
If you go by the 'concept' that Vista's sweet spot is 4GB because that is as much RAM as you can throw in the computer, then on Vista 64bit, if we throw 128GB in the machine, you could also call that its sweet spot, as Vista would be using all that extra RAM to pre-cache even more of your applications and data. Although there would eventually hit a limit based on the amount of applications and data actually on your Hard Drives.
I posted this link a few times, but in case peeps have missed it, go read about this from the horses mouth and not some bad review.
http://www.microsoft.com/technet/technetmag/issue
Take Care,
TheNetAvenger