Monday, December 19, 2005
Microsoft Denies Changes in Vista Graphics
Mary Jo Foley says that Microsoft Denies Changes in Vista Graphics.
My last post was in response to the TechWorld article that spawned the Mary Jo Microsoft "report".
I never read the article as if MS was changing Vista suddenly to pull the graphics subsystem out of the kernel. Instead I read it as if they ALREADY had to improve stability compared to Windows XP, that it was designed that way to begin with. I wonder how many people read it like I did vs. how many MS thinks read it like?
Microsoft reponse channeled through Mary Jo says the TechWorld report is wrong, but the MS response is a bit misleading. I think it goes out of its way to suggest that more is business as usually in Vista with regards to graphics as compared to XP than is in motion. The Windows Longhorn Display Driver Model - Details and Requirements PowerPoint I linked to off of Microsoft's site makes it pretty clear that there is significant change to DirectX and the driver model in Vista when compared to XP. DirectX 10 is the native version in Vista and requires new hardware, so I think this should be read as for the "advanced mode" cards I blogged about in my last post. Those DX10 APIs are not backward compatible with any previous version of DirectX. Previous versions of DirectX are "implemented" on Vista.
Here is a nice summary post with links to the info that came out of PDC05. Download the slide decks that this article is linked to. These decks where used during the PDC and re-iterate much of the information in the slide deck I linked to above, but with some interesting new details. The Vista Desktop will use Direct3D from DirectX9! This is on slide 16 in the PRS311_Balaz.ppt deck. So this is the "existing DirectX APIs", but its really Direc3D from DirectX9 Extension. This is what "implemented" means because they took the DX9 stack and it runs side by side with the DX10 stack, but they made D3D9 work with the new driver model, with the division of driver labor in user mode and kernel mode.
What's really interesting is that MS thinks this is serious enough to issue a press release through Mary Jo. Graphics in Vista are not business as usual compared to XP, and this is what got a lot of people excited on Thursday, including yours truly. The level of change compared to XP and the way Vista has been designed to work has escaped a lot of people's attention, and I think once people started to realize this, they are either scared of MS getting this right (check) or scared what it means to there existing hardware and games. I have been saying since the PDC03 that Longhorn was going to be so huge a change you are going to want new hardware to support it, and this information only reinforces that view.
Thursday, December 15, 2005
Microsoft removes the graphics subsystem from kernel in Windows Vista and the relationship to OS X Tigers Quartz 2D Extreme
Bink.nu has linked to a story on Techworld.com that states Microsoft is going to remove the graphics subsystem from the Windows kernel and move it back into user mode according to Microsoft infrastructure architect Giovanni Marchetti.
I guess I must have missed this news, but it's huge. Here are some links I got from Google:
Vista graphics drivers to be more stable than XP drivers, ATI says - This article actually speaks to the driver model going to user mode
Graphics Hardware and Drivers for Windows Vista - Really no insight, but this is the top hit for Longhorn Display Driver Model.
Windows Longhorn Display Driver Model - Details and Requirements - PowerPoint - Look at slides 7 and 8 for the comparison between XP and Vista. This has a lot of information about exactly what MS is trying to achieve and what the benefits are.
Don't get confused though, this isn't a reversal of moving the graphics subsystem into the kernel in Windows NT 4.0. That was done, if I remember correctly, to increase performance of GDI over Windows NT 3.51 since the number of context switches to get graphics work done was so costly that running heavy video apps or games was not practical. This is the now infamous move of Win32K.sys. Windows XP maintained this model, offloading a bit of work through GDI+ to the graphics card, but this was just a band-aid, GDI was showing its age and doesn't work modern graphics hardware and the way applications are being used. Don't misunderstand, I am not a kernel engineer, but the trial and tribulations of this architecture have always been interesting to me. I also have read all the Windows Internals books, nee Inside Windows NT in the Helen Custer days, except the latest. Here is the book on Amazon to that (thanks Mark R. again).
Vista doesn't revert to the previous NT 3.51 style model. The graphics subsystem in Vista is a complete brand new. Look at slide 9 in the PPT file, GDI commands are now rendered in SOFTWARE. The Vista graphics subsystem attempts to treat the GPU and video memory just as it treats the CPU and RAM. Multiple applications can access the GPU simultaneously and video memory is virtualized just as RAM is. These functions are contained in the kernel and called by DXGkrnl. DXGkrnl still calls a IHV piece of code, but the way I read this the footprint is significantly reduced and IHV code is less likely to be a cause for hangs. Most of the IHV work is done by the UMD, and I think that means user mode driver. Take a look at slides 18-20
Here is why you should wait until Vista and probably 2nd generation video cards are out to support the OS. Until the hardware is built for it, GPU scheduling is done in software instead of hardware. That means slow and will hit the CPU. In this slide deck, you really want hardware that supports the "Advanced" model.
The Techworld.com article states this brings Windows up to par in terms of graphics subsystems with Mac OS X and Linux. Ars Technica has a very good explanation of the OS X graphics subsystem, called Quartz, in 3 pages of information created when Mac OS X 10.4 Tiger was launched in April:
Mac OS X 10.4 Tiger - Quartz
Mac OS X 10.4 Tiger - Quartz 2D Extreme
Mac OS X 10.4 Tiger - QuickTime 7
Apple's Mac OS X Quartz Feature
The only problem with moving all the work onto the GPU in Tiger with Quartz 2D Extreme is that it's disabled. The code is in there, but it's not turned on. Reasons for this have never been explained, the only information that has come out is in the About 10.4.3 Software Update, which states: Disables Quartz 2D Extreme—Quartz 2D Extreme is not a supported feature in Tiger, and re-enabling it may lead to video redraw issues or kernel panics. So I think the graphics subsystem in Vista is going to surpass the capabilities in Tiger, at least until a post 10.4.3 software update either enables Quartz 2D Extreme or OS X 10.5 comes out. The strange thing is you can still enable Quartz 2D Extreme, but I thin the 10.4.3 update note is to discourage those on the bleeding edge that they are doing A Bad Thing(tm). I would love to know what problems where encountered that Apple decided to disable the functionality in Tiger.
One of the things I am wondering about is the Vista "Advanced" mode video cards. Is Apple going to use these? If so, are they going to change OS X to utilize the hardware based scheduling features? Even though Apple has been using hardware accelerated compositing and VRAM virtualization in OS X for years now, is Apple going to have to change their implemenatation to get closer to the way Vista is handling graphics operations to use the new video cards? What is the patent situation with the capabilities of these new cards? Is MS licensing these patents to ATI and Nvidia? Are Nvidia and ATI effectively just manufacturing cards for MS to spec? MS and Apple cross license patents, I wonder if MS is using any Apple patents to do the Vista graphics subsystem, or vice-versa if Apple is using MS patents for Quartz. Games on Mac OS X perform slower on similar GPU hardware than games running on Windows XP, see World of WarCraft as an example. I wonder how game performance will be now that the entire UI and all applications are sharing the GPU as a virtualized resource. Do games perform on Vista on par with OS X, are they as fast as XP, faster than XP? I haven't really seen anyone talk about this yet, even the XP to Vista comparison. Does MS have a special "games-mode" hack in the graphics subsystem to make it all work faster?
The Vista 1.0 graphics subsystem scares me from a backward compatability and Vista native app stability perspective. My money's on at least 2 Service Packs before the Vista graphics subsystem is solid. Of course I could be wrong and Vista is taking so long to get it right, but MS has done nothing to prove it won't take 3 versions to get this right. I don't think MS has been very forthcoming about the "advanced" model video cards, perhaps to prevent an Osbourne effect for this holiday season on PCs or graphics card upgrades. Just a theory, but I am glad I am not in the market for either upgrade until the "advanced" approved video cards are out.
Monday, December 12, 2005
Yet Another Way to Subvert Windows Security
Once again Mark Russinovich of Sysinternals.com has dove deeply into Windows internals and found something disturbing. Certain Group Policy settings can be circumvented by Limited Users.
Go read Mark's article, I'll wait. Back? I was designing AD forests on Windows 2000 back in my sys admin days, always thought Group Policies were one of those killer features for admins. Deploy settings to whole batches of machines, lock down users to what apps they could run, and how they could run them. Frankly, I am surprised that it has taken nearly 6 years since Active Directory and Group Policies were publicly released for the Software Restriction Policies (SRP) to get cracked. This always seemed like a juicy target.
There probably isn't a lot of immediate exposure to corporate networks. As with most digital protection mechanisms, the bar doesn't have to be to high to stop the normal user. But there will be those intermediate to advanced users that have been restricted using application blacklists instead of whitelist that will want to subvert the SRP blacklist through a proxy application. If any media that normal users might read, PC Magazine comes to mind, start to pick up on this, it may be a big deal. Of course, this all depends on how many corporations are using SRP. When I was deploying AD back in the day, I only betaed SRP, never implemented, nobody wanted to do ongoing management of it, no one wanted to become the application police.
Monday, November 21, 2005
Why I am not getting an Xbox 360 (or renewing my Xbox Live subscription)
Microsoft's great hardware, software and services hope, the Xbox 360, as just about everyone remotely interested knows launches tomorrow. I will not be getting one for a number of reasons.
Sure, Wow's Impact has something to do with it, but it's not the whole story. I have gone back to the Xbox recently to give Star Wars Battlefront II a run through and I am totally enjoying it. The most startling aspect, load times are quick compared to my summer of PSP loading hell, and I don't think I can deal with the PSP load times at all anymore, since completing my Madden 2006 season (a total non-event, not even a trophy ceremony) I haven't touched the PSP.
Back to the 360, here are my reasons for not wanting this box:
- Too Expensive. MS knows the $400 price point is too much, so they offer the $300 version, which you can't even save to or play previous Xbox titles because you have no storage. The cost/benefit it heavily in favor of the $400 version, and that is too much.
- Not backward compatible enough. I know, 215 titles, but no Battlefront I or II, the game I am playing, and something smells funny
- No Must Have Game. I am really disappointed that all the games are just re-treads of previous versions or sports titles, and the only thing being upgraded is the graphics. Maybe I am getting old on this, but I expect the graphics upgrade, there is no surprise, the gameplay hasn't changed, so I am disappointed.
- Need a TV upgrade SDTV has a fixed resolution, the 360 graphics can only get so good. The Xbox Classic already looks pretty good, so to really see the 360 outclass the Xbox, I need an HDTV, bringing the price for gaming nirvana into the thousands. Ah, no thanks.
- None of my friends are actually playing Xbox Live. This is the main reason I am not renewing my Live subscription. Everytime I boot the Xbox to watch a movie or play some Battlefront II, no one I know is playing anything. Most of them are hard-core into WoW or won't play on Live because they are hardcore mouse and keyboard control freaks. I like the Xbox controller, but I appreciate the mouse and keyboard precision.
- I am tired of subscriptions. Here is another reason I am not renewing Live, I am really done with the whole subscription model. I feel this need to try and use whatever I am subscribing to as much as possible so that I feel like if I run the breakeven analysis, I am paying a fair price. I know, Live is only $4.16583333 a month, but if I don't use it that month, I get nothing for something, and that doesn't sit well with me. A subscription is always constantly on the mental To-Do list somewhere, and I just don't need anymore of those.
- I am waiting for all next-gen consoles to ship. Sorry MS, but I saw this story before and believed the hype of the last great white hope, the Sega Dreamcast. Unlike the 360 though, the Dreamcast actually had extremelly compelling games that I wanted to play, like NFL 2K, which redefined video game football. Sorry, Madden 2006 Another Console Edition, is not on my buying list. Plus, you never know if Sony can pull off greatness in the PS3 or if Nintendo redefines console caming with Revolution.
- I am holding out for all the next Apple gear. This one is much much lower, cause I could make new Apple gear happen if I wanted to in the 2006 and get a 360, but the coming Intel Mac's are really affecting my buying decisions.
- I am not repeating the Sony PSP. Same kind of games were out at launch (seasonal sports titles adjusted) for the PSP as 360, and then nothing forever. Call me crazy, but I suspect this is the best MS and partners could do for now and we won't see anything truly new until say March or maybe next fall (Halo 360?). Sure, some stuff will trickle out that will be decent, but I haven't heard of a true system seller yet (Gears of War?).
Tuesday, November 08, 2005
My first SQL Server 2005 Experience
I got SQL Server 2005 installed yesterday in a relatively painless experience. Told the installer to dump everything it had onto my system in a named instance so I could run SQL Server 2000 side-by-side, it all seemed to work.
I launched the SQL Server Management Studio and connected to a remote SQL Server 2000 database. In the Summary tab, I double-clicked down to Databases, and then tried to double-click into a table, when I received the following error:
Once I get this error, the whole ListView is busted and doesn't work. I have to relaunch the application to get this working. Maybe I have some crazy config, but this would seem like one of the top cases that must be tested against, and I really can't believe this is a shipping product.
Thursday, November 03, 2005
GotDotNet is so frustrating and How many developer sites does MS need?
I don't go to GotDotNet because I frequently find that when I do, something just breaks and I can't be bothered. I went today to the FxCop Team Page to see if they had a tool to parse the FxCop logs to create a report in Excel. I want't to get the trend of FxCop violations over several months of automated reports. An additional wrinkle is that my FxCop reports have been merged into my CC.NET log files.
So I go to the FxCop Team Page, then click on link for FxCop area within GotDotNet User Samples and I get the following error:
This is in IE6 too, not sure that matters. If I refresh a couple times I can see the forum, but it's totally unreliable. How many developer sites does Microsoft need (GotDotNet, MSDN, Channel 9, www.asp.net, www.windowsforms.net, I know I am missing some)?
I am tempted to build the FxCop Excel Report Generator myself, I built an FxCop parser at my last company, it was easy using xsd.exe to generate classes from the schema files, but if it's already out there, I am not going to bother.
Wednesday, November 02, 2005
Sony PSP Media Manager
Shoutout to Paul Thurrott for bringing the Sony PSP Media Manager to my attention.
You can buy, that's right buy, the Sony PSP Media Manager here
I can't believe Sony is charging $19.95 for the download, $29.95 for the boxed version with a special introductory offer of $24.95! This is terrible, Sony should be giving this away with the PSP, it should just be a free download. Completely evil.
Paul also posted a link to the software release notes. Notice that it requires Quicktime 7 and it's a .NET application, both of which I find interesting. And that's right, the final strike against this software, before I have even seen it live, is that it doesn't work on OS X. That's it I guess, I have been waiting for the FREE Sony software for the PSP to do what iTunes does for the iPod, obvious now it's not coming.
Scott Hanselman has a good post on manually getting TV content onto your PSP with Windows tools, and I just saw he updated his post with a link to the Media Manager software. After reading Hanselman's post, it seems Handbrake for Mac and Linux can do the video encoding work, you would just have to get the videos onto the PSP manually, which still of course sucks because of Sony's folder and file naming scheme. Maybe a weekend project to try, though video has to be small, I am not spending the money on the 1 GB memory stick until I can get a test done. I am still not paying for software to complete the scenario of getting media on the PSP, no wonder Sony has lost the MP3 player race to Apple, they just don't understand completing the user experience.
Possible file corruption for NTFS volumes on Windows Server 2003 Service Pack 1
I subscribe to the Windows Server 2003 Knowledge Base feed to see what issues have been discovered, if my current environment might be impacted, and if a fix is available. This has already saved me hours of troubleshooting as I discover a KB article for something I know we have seen in our environment.
Sometimes you get something scary, like this:
Potential file corruption problem on NTFS volumes during extensive stress tests in Windows Server 2003 Service Pack 1
Microsoft explains in the KB article that the scenario that causes corruption is very rare in real world scenarios. Still this should scare anyone in a large environmnent using file services, because Microsoft does not say the probabilty is zero for the unlikely cases, just that it's low. Low! What does that mean? 1%, 5%?
If you think you meet the likely criteria, you can either not install Windows Server 2003 SP1 (which is where the problem lies, but I think most organizations have moved forward with deployment already since the "bake in" period is well over) or, uh, MS let you know when then have a hotfix.
I wonder if a root cause of the late discovery of this issue, relative to release date for SP1, is the maintenance team building the service packs. I am sure they are all very bright people, but let's face it, the service pack team at Microsoft are not the starters, they're not the A-Team, they're the second string. Microsoft calls it the Windows Sustained Engineering group at least as of 2003 in one reference I found. It is normally this group that produces hotfixes and service packs, but for extraordinary times, like for Windows XP SP2 the main Windows development team is brought back into the fold. Think of this normally as two tracks, the main Windows development team, and the Sustained Engineering team for hotfixes and service packs. The people that wrote Windows are not usually the ones that fix Windows. This isn't to say there is no interaction, I don't know of course, but it is a different team without as much experience on the code when the work is done. Service Packs are also not looked at as major Windows versions (XP SP2 excluded), they're just maintenance releases. I would put money on them being tested less rigourously, not by the test team, but by the outside people the project that really uncover the kruft of the code. Windows Server 2003 Service Pack 1 was based on XP SP2, but XP SP2 certainly wouldn't be stress tested for NTFS corruption. The Windows XP KB feed has no mention of a potential NTFS corruption issue.
I am not saying lack of testing or less experienced developers caused this issue, but I am saying NTFS stress tests didn't reveal any potential corruption in Windows Server 2003 RTM and its there in Service Pack 1.
Tuesday, November 01, 2005
MSDN Subscription Transition Update #4
Ugh, now this is just annoying. I received another response from MSDN this morning:
Hello Dave, This is -Name Removed To Protect the Innocent- and I am taking ownership of this case. I apologize for the issue that you encountered in the MSDN North America Visual Studio Team Transition Preference Login page. The site is now working correctly. You should now be able transition your MSDN Universal Subscription to a Visual Studio Team Subscription through https://microsoft.na.subservices.com/msdn/default.asp?hidToPage=VST without difficulties. If you need further assistance in transitioning your subscription, please call MSDN Subscription Services at 800-759-5474, 5:30 a.m. to 5:30 p.m., Monday – Friday, except holidays. Did I complete completely address your issue? Please let me know by replying to this e-mail. Please do not hesitate to write back should you have other concerns. Thank you, -Name Removed To Protect the Innocent- Microsoft Online Customer Service Representative*Sigh* The site is still down as of this post.
Sony DRMed CD's Install Virus/RootKit/Malware
Mark Russinovich with his SysInternals site and tools has to be one of the smartest people on Windows (any version) working today. No one has done more outside MS to make Windows systems maintanable. If you don't already go download his Process Explorer, FileMon, RegMon, Autoruns, and RootKitRevealer tools. You will need them one day and be glad you have them on your system. Some go so far as to have the latest version of all these tools on a USB stick in case shit happens
Mark has outdone himself with an amazing piece of detective work that reveals that Sony installs DRM software on Windows with their copy-protected CDs that is for all intents and purposes spyware/malware/virus because of the techniques used to install and evade detection. I think a boycott of Sony is in order, this is just unacceptable, looks like the PSP is going on the auction block.
I don't like DRM software at all, but of all the schemes out there, seems like Apple with FairPlay on their music and video files strikes an acceptable balance between consumer and content provider. This is one of the core reasons Apple has been so successful with iTunes, the DRM is largely transparent.
Subscribe to:
Posts (Atom)