Original Link: http://www.anandtech.com/show/2096
Windows Vista Update: RC1/5728 Previewby Ryan Smith on October 3, 2006 4:30 AM EST
- Posted in
At long last the light is at the end of the tunnel. After a several year development period, the longest for any single consumer version of Windows, the end is near for Windows Vista. While it's clearly not ready to be delivered in to the hands of users quite yet, Vista is finally at a point where we can begin talking about what will happen, and not what may.
Although Microsoft uses the Release Candidate nomenclature for Vista builds 5600 and above, including build 5728 we're looking at today, the reality of the situation is that the shipping version of Windows Vista will not be these builds or even a few builds down the line. Given the complexity of an operating system, there are still messy quirks and bona fide bugs in these release candidates, and it's going to be at least another month before we're talking about Microsoft having released a final version, and even then there will be a good amount of post-launch patching to be done as Vista ends up in the hands of the ultimate bug hunters, everyday users.
With that said, this is the first time that we can say without flinching that Vista is in an acceptable state for general use. Compatibility on the x86 version is remarkably improved over what we saw earlier, and in our testing we only managed to come up with a single program - non-commercial at that - that simply wouldn't function correctly under Vista no matter what. Otherwise, everything could be made to work under Vista given enough cajoling, which is an enormous feat given the amount of under-the-hood work the operating system has received compared to Windows XP.
|Click to enlarge|
User Account Controls have not changed much since build 5472, which is not necessarily a bad thing. Like Windows overall, UAC is usable at this point, and not nearly the nuisance it was as of Beta 2. It still has rough spots, and we'll get to that in a bit, but at this point enthusiasts are the only group that will have problems with it.
Hardware support and in-the-box drivers are also coming together, no doubt due to the portability of drivers between Vista and XP in most cases. A quick run-through of our lab turned up only two pieces of hardware that weren't supported under Vista: a Hauppauge TV tuner that had three of the four drivers it needed, and our PhysX card, both of which should have full support soon. All things considered, this will likely be the least-painful Windows transition on the driver front, as vendors have been on top of the few key kernel/driver changes for a while.
At this point we've been using RC1 for nearly a month, and the newer build 5728 for over two weeks, and while we're ready to switch back to XP until Vista is completed due to some video issues, Vista is ready to be taken seriously.
What Lies Ahead
Given that Vista is ready to be taken seriously, there are also a handful of issues that we've encountered so far. These issues are not necessarily showstoppers, but they are fairly significant and will be a problem for enthusiasts. Officially, at only a month left until Microsoft wants to have a version ready to ship, we're not sure if these problems will be addressed. Either way, they're important to mention.
The biggest enthusiast issue is still UAC. Certainly for users who seldom need administrative powers UAC is fine, but the more you need administrative powers the more obvious the problems become. As we mentioned in our build 5472 article, Vista does not have a notion of pre-approved programs. Because marking a program to run with administrative privileges is not itself an administrative task, the actual check comes at every execution. The problem with this becomes readily apparent when using a lot of programs that require administrative powers: every single execution requires authorizing the program to run.
What we would like to see is a way to pre-approve programs to run, using hashing to make sure such a program hasn't secretly been changed, so that selected programs won't require user-authorization every single time they're executed. Apple already does something remotely similar in Mac OS X with their password keychain, so the idea is not unprecedented. This alone would solve one of the biggest nuisances in Vista, and is a much better alternative from a security perspective than disabling UAC outright or setting it to approve all applications requesting administrative privileges.
Another notable issue we encountered cropped up in the same security system, ironically because the security service is doing what we want in this case. Upon attempting to patch Battlefield 2, the patch installer took an abnormally long time to start, and upon some investigation the issue turned out to be that the security service was hashing the patch installer, all 500MB of it. It goes without saying that self-contained executable installers are one of the primary distribution formats for data on the internet, so this isn't a minor issue. The biggest single executable we could find, the installer for the Battlefield 2142 beta, took over two minutes just to hash, and that's not going to make people happy. (Given that the Battlefield 2 patch can take well over 20 minutes to install on a moderate system, however, two minutes isn't the end of the world.)
Although it's clearly easier said than done, if Windows is going to hash all executables it could use some way of figuring out what's an installer package and not hashing the whole thing. Running these kinds of installers is not a daily event, but right now other than a lot of disk activity and some CPU usage by the security service, there's no real notification Vista is attempting to launch the application, and this is going to cause concerns for a lot of people the first time they encounter it. Those that don't understand the specifics of what is happening will almost certainly conclude that Vista is simply slower than XP on some tasks.
The third notable issue is audio for gaming purposes, and while we'll have a lot more on this when the final version of Vista is released, it at least deserves a quick mention right now. As Microsoft has moved most of the Windows audio system into Vista itself and out of hardware and drivers, DirectSound3D is no longer hardware accelerated and EAX effects may never work with it again. There are several exceptions and specific scenarios to talk about here, especially with Creative Labs' soundcards since they're the de-facto vendor of gaming soundcards, but it looks like a lot of older games are going to lose some of their audio abilities. There may also be a greater performance hit due to the amount of processing that is now done solely in software.
Last but not least, let's talk about performance. Here is our test bed, which has been updated from the previous Vista article.
|Vista 5728 Testbed|
|CPU:||AMD Athlon 64 X2 4600+ (2.4GHz/1MB)|
|Motherboard:||Asus A8N-SLI (Socket 939)|
|Chipset:||NVIDIA nForce 4 SLI|
|Chipset Drivers:||NVIDIA nForce 6.86/Vista RC1|
|Hard Disk:||Maxtor DiamondMax Plus 9 120GB|
|Memory:||OCZ PC4800 (512MB x 4)|
|Video Card:||ATI Radeon X1900XTX|
|Video Drivers:||ATI Catalyst 6.9/Vista RC1|
|Desktop Resolution:||1600 x 1200 - 32-bit @ 60Hz|
|OS:||Windows XP Professional SP2|
In some respects, we've been spoiled by having Windows XP for so long. At the ripe old age of 5 years, it's effectively a lightweight operating system for a modern high-end computer. Vista changes all of that, as like most other software it has grown in size to make more use of modern computing power, and no matter what optimizations Microsoft makes, that will be felt on some level. No matter what, most enthusiasts will find Vista's GUI slower than XP's, even with 3D acceleration, and this isn't likely to change with a final release.
Speaking of 3D acceleration, we've seen some improvements out of both ATI and NVIDIA, but there's still some distance to go. Neither can offer XP performance under Vista, and in the case of ATI they still aren't offering OpenGL support under Vista. Where things will be on launch day remain to be seen. We'll have complete Nvidia numbers next month when Vista ships along with Vista x64 numbers, but for now we'll be using our Radeon X1900XTX on just Vista x86.
|General Application Performance|
|Vista 5728||XP SP2|
|Cinebench Multi-CPU Rendering||669||651|
|AutoGK Encoding(XviD 1.2SMP)||13:36||11:59|
|Adobe Photoshop CS2(in seconds)||215.1||204|
Under general performance, Vista is a mixed bag with one interesting result. Encoding under AutoGK with XviD is a little over a minute longer, or about 13% slower. Photoshop CS2 shows a much smaller gap at only 11 seconds, which comes out to only 5% slower. Neither of these results is really poor, but anything over 10% is a pretty stiff hit for just switching operating systems.
Moving on to Cinebench, the tables turn. Although barely enough to consider it outside of the expected experimental error range, the performance boost of just under 3% is the first sign we've seen out of these release candidates that Vista can be faster than XP. By far the most interesting result however is PCMark05 with a 23% performance improvement in favor of Vista, but we're not entirely sure what's going on. Since it's a multitasking heavy benchmark, one possibility is the changes made under the hood for Vista benefit multitasking the most, which may also explain why Cinebench did so well since it too can split its rendering jobs so well. PCMark05 also has some HDD benchmarking activities, so another possible explanation is that Vista has more optimized I/O performance. Multitasking performance in particular is something we will take a closer look at with the shipping version of Vista.
All games were tested at 1600x1200 resolution for the results below. This places a larger burden on the GPU than the CPU, but represents a common resolution for owners of high-end graphics cards. We will conduct more complete testing when the final build of Vista becomes available.
|Vista 5728||XP SP2|
|Half-Life 2: Episode 1 (No AA)||94.4||103.6|
|Half-Life 2: Episode 1 (4x AA)||84.49||86.4|
|Battlefield 2 (No AA)||55.1||77.96|
|Battlefield 2 (4x AA)||52||77.11|
|FEAR (No AA)||52||70|
|FEAR (4x AA)||32||52|
As far as gaming performance goes, the news is universally less pleasant, and sometimes even grim. 3DMark06 comes within 3% of its XP performance, but that's as close as anything gets, and since this is a synthetic benchmark that's about all that needs to be said on the subject. Half Life 2: Episode One shows the best performance out of the real games we tested, only dropping short of 10% of its performance moving to Vista without antialiasing, and even less with antialiasing enabled. Losing performance is never good, but here it doesn't impact playability at all.
Such is not the case for FEAR or Battlefield 2 however. Here the performance drops are all over 25%, the worst being FEAR with antialiasing at 40%. At this point these are large enough drops that they'll certainly impact playability, necessitating cranking down the resolution or settings in order to make up for the drop. As we've said in previous articles, hopefully performance will continue to improve, but the window between now and the launch is getting perilously small, so it seems increasingly likely that Vista gaming performance won't match (or even come close to) XP performance at launch time, at least with ATI's cards. We'll leave the question of why anyone would release a Vista-only game for you to debate.
While we'll hold back on any final opinions on Vista until we have a shipping version, at this point it's becoming increasingly unlikely we'll see any more significant changes between now and when Vista is finished next month. So what we say now we'll likely be saying again.
At this point most of Vista is perfectly fine. Everything works, all functionality is enabled, and driver support is looking good. If anyone had to pick up Vista and use it today, they would be able to do so once they got over the initial shock of just how different it is compared to XP. For doing so, they would be rewarded with a lot of new functionality that XP can't offer, though other operating systems like Mac OS X have offered many of the features for years now.
As the saying goes, it's the little things that matter. Vista isn't perfect - no operating system is - and while we could pick apart things we still don't like about XP 5 years later, Vista's problems are more pressing. Some of the things that Vista does are flat-out quirky and make little sense, and User Account Controls are still going to come across as combative to enthusiasts and other similar user groups. General users will be fine, but as is often the case general users end up adopting some enthusiast practices, and if that includes disarming parts of UAC, it will significantly undermine the security gains of Vista.
Where performance ends up is also a concern. Our general performance numbers are a mixed bag compared to XP, which turns out is a good thing since it would imply that overall Vista is no slower than XP for performance-critical applications. Memory and disk space usage are up, and that's an inevitability of progress, but most recent machines should be no worse off with Vista than XP as far as general usage goes if our numbers reflect the larger whole. This is something we'll investigate much more in depth once we have the final version of Vista.
Unfortunately, gaming performance is still lagging behind, heavily at times, and this is troubling. Gamers will no doubt stay away from Vista if the final version and final video drivers continue to underperform, but there's also the larger issue of how computers are becoming increasingly reliant on the GPU for general tasks, something Microsoft itself is pushing with the new video threading systems for Vista and forthcoming DirectX 10-class video hardware. A large video performance gap could or could not be a problem for non-gamers moving to Vista, but it's too early to tell.
So where does that leave us then? If Microsoft continues to stay on schedule, they will have a shipping version of Vista ready within a month, though retail consumers will be waiting until 2007 to get their hands on it. Whether it will really be ready to replace XP at that time remains up in the air. Early adopters will likely bite the bullet, but many people will probably prefer waiting for the inevitable Service Pack.