More Vista nonsense

[Fushigi]

Overwrite has the same net effect as a delete/recreate - replace an old file with a new one - but the process is different. An overwrite would not be a delete/copy; it would look more like allocate/open a file, write the new contents, truncate to force EOF in case the new file is shorter, close.


BTW I should have commented more about the registry for Case 2. Registries are fine, but there should really be separate system & application registries and applications (anything that's not a Windows component) should never be allowed to update the system registry. A single registry to rule them all is a bad idea on Microsoft's part. It is far too easy for a rogue application to mess up system aspects of the registry under the current design.

Something similar should have been done with DLLs; applications that call already installed Windows DLLs are fine, but if they supply their own, even if they are normal MS DLLs that simply aren't installed already on the machine, they should only place them in the application's directory. "DLL Hell" could have easily been avoided or at least radically reduced.

In essence with proper design there would have been very few times when non-Windows software actually needs to write to the Windows directory structure.

 

[sechs]

I agree, that's a pretty ridiculous assertion.

 

[Drakantus]

I've been very happy with Vista 64. Then again, I don't use much old software, I don't use hardware with poor driver support, and I enjoy tweaking settings and such to get the best possible performance.

Vista's stability seems to be the exact same as 2000/XP to me: it's about 99.9% stable with good drivers and hardware, but if you install a bad driver or memory or whatever you are going to have problems.

The OS uses more memory, yes. Just like XP uses more than 2000, and 2000 uses more than windows 98, and Windows 95 uses more than windows 3.1, which uses more than DOS. I don't let it bother me because of two reasons. One, without a 64bit OS I am going to be hard-limited around 3GB RAM, so with my current hardware I would be losing 1GB of RAM if I went back to XP or 2000. Two, memory is cheap. If you are going to buy Vista for $200-$300, would it really kill you to spend an extra $40 to add 2GB of ram to your system?

Also, I like the fact that Vista is far more usable without activation than XP. If I am testing an old PC, the last thing I want to do is lock in an XP install by activating before I know the hardware is good. But I can't even install the windows updates without activating, which really limits the testing I can do. Vista works fine (for 30 days) without activation- you can still install all the windows updates, and the computer will pass genuine advantage checking.

 

[Bozo]

Microsofts home page is proclaiming " Vista Ultimate; new lower prices"

They just don't get it...

http://www.microsoft.com/en/us/default.aspx

Bozo :joker:

 

[ddrueding]

Home Premium @ $129 and Ultimate @ $219? They are getting desperate...

 

[Fushigi]

Those are upgrade editions, not retail. So they'll work but you'll have to do the upgrade trick to use it on a clean install. IIRC I paid $235 for Ultimate Upgrade last May. Not buying direct you can get Ultimate Upgrade for $189 and Retail for $250.

 

[Clocker]

Or if you have a friend of a friend from M$ you can get Vista Ultimate (Full) for $40.

 

[Fushigi]

Nice. I can get the full version of Office Premium for $19.95 + about $2 shipping via my employer & the MS Home Use program, but OS versions aren't on the list.

 

[Adcadet]

From my previous University I got both Office 2007 and Windows Vista Ultimate for $10 a piece. I've found Office 2007 to be mixed, although for poster creating I've found the Excel-PowerPoint integration to be wonderful. I just installed Vista on a spare HD, and found it to be nothing special so far.

 

[Tea]

Vista's so bad no one steals it: http://www.theinquirer.net/gb/inquirer/news/2008/06/10/vista-bad-steals

 

[Bozo]

It just gets worse. Apparently, Microsoft can't see past it's nose:

http://news.cnet.com/8301-13924_3-9996317-64.html?part=rss&subj=news&tag=2547-1_3-0-20

Bozo :joker:

 

[Fushigi]

Yeah. Sure. Whatever. They're BSing because they're late to the game. Otherwise Sandisk would have provided details as to why Vista isn't 'optimized' for SSDs. Sure, Vista hits the disk more than XP, but any drive - spinning or solid state - should be able to handle that just fine.

 

[ddrueding]

Seconded. I could understand that a standard disk needs a certain access pattern to optimize it's cache, but an SSDs primary advantage is the near-zero access time to anything on the disk! This means that no optimization should be necessary!

 

[Handruin]

I don't want it to sound like I'm jumping on the band wagon, and I dislike vista as much as the next noble IT guy, but I agree with the prior two posts...vista shouldn't be the reason for delaying their product. I also can't understand why they wrote that the very low end market was fine...who puts an SSD in a very low end machine???

In the very low-end of the market, however, this is not an issue. "In very low-end, ultra low-cost PCs, existing controllers can get the job done for 8-, 16-, and 32-gigabyte storage because these are relatively unsophisticated and demanding requirements," he said.

It's like saying a porsche carerra GT's V10 got the job done inside a Ford Pinto.

 

[ddrueding]

I suppose they mean for the low speed SSDs in cheap computers (like the Eee and such).

 

[Bozo]

Quote from the Tech Report

"The CEO didn't detail the exact cause for the problem, although judging by his statement, he might be referring to Vista's I/O prioritization scheme. According to this Microsoft paper, the scheme favors storage responsiveness over throughput in an attempt to compensate for high mechanical hard drive seek times."

Bozo :joker:

 

[Clocker]

Yeah. Sure. Whatever. They're BSing because they're late to the game. Otherwise Sandisk would have provided details as to why Vista isn't 'optimized' for SSDs. Sure, Vista hits the disk more than XP, but any drive - spinning or solid state - should be able to handle that just fine.

I agree. Sure, Vista has some issues. But I think the popularity of 'Vista Bashing' is leading Sandisk to try to blame M$ for their own screw-up.

 

[Fushigi]

I read the paper. Two things come to mind:
1. For storage hardware vendors, it boils down to 'put SATA NCQ in it'. Everything else is done at a driver or higher level.
2. The details apply to Server 2008 as well. Considering SSDs are also being considered for data center usage (lower heat/power, can actually afford the cost, etc.) it is just a little odd that Server08 wasn't mentioned.

It is also funny how OCZ, Samsung, and the others make no mention of issue with Vista/Server08 & their products.

I have no issues with Sandisk in general. I've no agenda or axe to grind. But without actual evidence of an issue with the OSes, I'll stand by my post that they're blowing smoke on this one.

 

[LiamC]

My 5c worth

A post I read on Tech Report from a Linux user was interesting (quoted below). It also gels with comments from Intel posted here a while back about the need to develop a different controller for SSD's. I'm guessing that what they meant was a different software stack and hardware controller to work with the different access patterns of SSDs.

The upshot is that I also call BS on this report from SanDisk. Yes there appear to be issues, but these issues aren't show stoppers. Other SSD manufacturers can run large flash disks under Vista. As such, the SanDisk position is a straw man. SanDisk are covering up something.



"...User: stdPikachu

Not if my tests are anything to go by. I've got a PG database running off a Samsung flash SSD and the no-op scheduler gives about a 20% performance boost over deadline or cfq for our workload.

No-op should make a difference because, much like with the vista scheduler, t'other schedulers were written with the mechanical latencies of spinning platters in mind. Since no-op acts on a first-come-first served basis and SSD have no random seek penalty (you additionally lose the CPU overhead of re-ordering IO requests) it's generally the fastest option (indeed, not seen a situation where no-op wasn't the fastest for solid state yet).

Personally, I don't think tweaking SSD firmware around the limitations of a single OS (which may change in the future anyhow) is the way forward. There needs to be a method by which operating systems can go "Oooh! An SSD!" and react accordingly. I imagine there are manual steps you can take in Vista much like going for no-op in Linux. "...


http://www.techreport.com/discussions.x/15167

 

[mubs]

stdPikachu ... Pikachu is a Pokemon. Used to be my daughter's favorite. Odd handle for a tech guy.

 

[Bozo]

Samsung and MS

http://reseller.co.nz/reseller.nsf/inews/759FFDABB045D634CC25749E000C758C

Bozo :joker:

 

[Clocker]

I switched back to XP from Vista after several months of use. I had forgotten how fast my PC could be. I have 4GB RAM installed but Vista x64 is still a pig compared to XP. It was nice to give it a try though. I'll keep my Vista Ultimate since I only paid $40 for it but I will probably never use it unless Windows 7 is an even bigger pig.

 

[Mercutio]

I will 100% assure you that Windows 7 will also be a pig, unless Microsoft decides to modularize the whole thing, which I 100% assure you they won't do.

The other day I put together two quad core machines on 1TB Samsung drives with 4GB RAM each. One has Vista Business and the other XP Pro. The XP Pro box boots in about 25 seconds. The Vista machine takes not quite two minutes.

And hey, I can make it even more dramatic if I load XPLite rather than full XP.

 

[Bozo]

I like this, Microsoft's creative counting.

http://apcmag.com/xp_still_killing_vista_in_sales_volume_hp.htm


Bozo :joker:

 

[Bozo]

Vista security broke??

http://www.neowin.net/news/main/08/...ty-rendered-completely-useless-by-new-exploit

Bozo :joker:

 

[Fushigi]

Sounds like potentially all MS OSes are broken. What other MS OSes include DEP besides Vista & XP? Need to know more, like are alternative browsers safe, can this be defeated by disabling JS, AX, etc.

Also, they mention 'loading' into any address space. They don't mention 'executing' that address space. Technicality, yes, but if they don't load the code into the right place, it won't execute properly. And the code still needs to be called; otherwise it's just random junk occupying RAM.

 

[ddrueding]

In order to load something into memory and execute it, the system needs to already be compromised. I considered those features to be semi-useless anyway. Sure, it allows rights escalation, but there have always been a hundred ways to do that on Windows.

 

[Gilbo]

Vista security broke??

http://www.neowin.net/news/main/08/...ty-rendered-completely-useless-by-new-exploit

Bozo :joker:

Sounds like potentially all MS OSes are broken. What other MS OSes include DEP besides Vista & XP? Need to know more, like are alternative browsers safe, can this be defeated by disabling JS, AX, etc.

Also, they mention 'loading' into any address space. They don't mention 'executing' that address space. Technicality, yes, but if they don't load the code into the right place, it won't execute properly. And the code still needs to be called; otherwise it's just random junk occupying RAM.

The devastating aspect to the technique they claim to have implemented is that it allows them to guarantee insertion into the correct memory location despite Vista's memory randomization algorithms. I.e. part of developing this exploit was defeating Vista's memory randomization safeguards. Being able to do this guarantees that code will be executed if the hacker so desires.

I read some time ago (more than a year) that the OpenBSD developers were highly displeased with Vista's address space randomization. They had some very technical criticisms I didn't understand and I can't find the link. It's possible the algorithm wasn't secure, and this exploit plays on known deficiencies, combining them with other clever tricks.

In order to load something into memory and execute it, the system needs to already be compromised. I considered those features to be semi-useless anyway. Sure, it allows rights escalation, but there have always been a hundred ways to do that on Windows.

Unfortunately, their demonstration code apparently demonstrates a complete pathway through both IE 7 & Firefox (version unreported) all the way through Vista's safeguards into executable memory. If it is as they describe it, it is a completely usable vulnerability with pretty much maximum exposure. I.e. you visit a website with exploit code (or with an exploited cross-site scripting vulnerability) -> you're fucked.

 

[ddrueding]

Time to install NoScript again...

 

[Gilbo]

Well, they haven't done their presentation at Blackhat yet. There's no proof-of-concept code available. People could be exaggerating things because they like the attention.

If it's a genuine exploit though, one that lives up to the details they've provided, it's going to be a serious issue... Wouldn't be the first time.

 

[timwhit]

Time to install NoScript again...

I had NoScript installed for about a week. I found it highly annoying. I visit such a wide range of websites that I was constantly having to allow scripts to run. Plus, there are very few websites that will work right without scripts enabled these days.

 

[Gilbo]

]Ars Technica has details on the range of attacks the methods the researchers have developed allow.

Essentially, you need badly behaved software to implement the work arounds. These work arounds allow you to load run executable code into memory and run it (they don't detail any new UAC or privilege elevation issues, so you would need to combine this with other vulnerabilities to, say, install a root kit).

The bad part is that IE 7, Firefox, the Flash Plugin, and Java are all examples of badly behaved software with respect to these vulnerabilities.


Ars is rather anti-sensationalist on this, a viewpoint which I can understand (considering web outlets propensity for screaming that the sky is falling), but it seems to me that they end up downplaying the vulnerabilities' significance. 99% of all machines have this software installed, and "use .NET" is not going to be a solution for the foreseeable future. The fact that Vista can no longer stop buffer overflows is definitely a big deal.

 

[Mercutio]

The thing I take away from all this is that Vista offers no more protection than XP did.
And for the most part, we're OK with the level of security that XP has. A lot of people end up turning off (or turning down, if they know how) UAC anyway.

 

[Gilbo]

The thing I take away from all this is that Vista offers no more protection than XP did.

Exactly. It's just a bit of a shame because the buffer overrun protection was supposed to be one of the big reasons Vista's kernel was better. In was actually something that would have been really useful... if it had worked.

 

[CougTek]

I'm retrying Windows Vista tonight. I don't know long it'll last (no more than 120 days for sure - 30 days + 3 rearm). I do it because I have to. Last I gave Vista a try on one of my systems was more than a year ago and I get more and more tech support calls from unfortunate people stuck with Vista.

This time, I try the 64bit version on a much more powerful system than I tried last time. It is my main test system : the one I fold on. The one before was a P4 3.2GHz with a gig of RAM and a Radeon X1600 Pro. It scored 3.4 on the performance index. This one scores 5.8 and only because of the hard drive, otherwise it would be 5.9. It is after reading this article that I decided to retry M$' latest consumer operating system. The simultaneous streaming videos got me intrigued. So here I am, typing on Vista, despite all my despise against Windows 6. I'm almost pissed because so far, it doesn't feel sluggish. Of course, I tweaked it a little, but I haven't spent a lot of time doing so. It is...usable, I guess. At least on a box that cost 1200$ without a monitor.

More later...


P.S. Expect pictures of torn Antec enclosure by this time next week...

 

[Fushigi]

My work laptop only scores a 3.2 due to the graphics; gaming graphics are 3.8. Processor/RAM/HD are 5.4/5.1/5.2. GPU is a 128MB nVidia NVS135M, CPU is a T7800, 4GB RAM, 120GB 7200RPM Seagate.

I would hazard a guess that the GB of RAM was the limiting factor on the old machine; slower GPUs wouldn't mangle system performance that much except for graphics-intensive apps. Vista loves RAM and I wouldn't recommend less than 1.5-2GB for the average user.

Also, I noted a rise in performance ratings with SP1. I'm assuming it is due to performance improvements although it is possible MS just messed with the rating system.

 

[ddrueding]

I've noticed some performance differences in systems that all scored 5.9 (the maximum). It's a shame they don't have the scoring go higher...

 

[Bozo]

Also, I noted a rise in performance ratings with SP1. I'm assuming it is due to performance improvements although it is possible MS just messed with the rating system.

I believe MS just screwed with the numbers. During the beta test numerous testers complained of low scores although they were running top of the line hardware.

Bozo :joker:

 

[Mercutio]

After much experimentation, I've found a fun new Vista pet peeve:

Take a 17" or 19" widescreen flat panel display. Brand name, generic, whatever (I like HannsG displays for cheapies these days). Most likely, it will have a screen resolution of 1440x900.

Attach that display to an XP computer. Any XP computer. It can even be pre-SP1.

XP will recognize the display and properly configure itself at 1440x900, as long as it has a driver and hardware that supports that resolution (i.e. not a 4MB S3 Trio)

Take that same monitor and plug it into a Vista computer. It can by exactly the same as the XP machine.

There is about a 20% chance that Vista will recognize the new monitor and configure itself at 1440x900. The rest of the time, it will drop down to 800x600 until the user loads a driver. If it doesn't do it the very first time, it never will, even if you force Windows Updates or screw around with resolution settings.

This, frankly, is total bullshit.

I'll even go one step further and say that Linux - at least SuSE and Ubuntu in GNOME - will both correctly display the proper resolution if they have the right display drivers installed, but Vista doesn't.

 

[Fushigi]

I had to reinstall my wife's PC again from scratch. XP Pro, C2Q6600, 2GB DDR2, Radeon 3870, Acer 20" LCD via DVI, 500GB Samsung, etc. Despite a very vanilla install, it always winds up becoming a BSoD box over time. Both she and I are getting rather annoyed. This time I've left off all of the motherboard drivers except for the audio, which doesn't work without it.

So I format & install it again. XP SP1a (my OEM media) and the subsequent SP3 - the first update applied - failed to auto-detect or allow manual changing to the LCD's native resolution of 1600x1050. It ran at something like 800x600.

Loading Catalyst fixed it (automatically). Now, what bugs me is that Windows had to have seen the VESA modes to know it would do 800x600, but it didn't see/use the monitor's actual resolution. WTF?

On a related but sort-of non-MS front, another bothersome thing is Catalyst Control Center requires .NET 2.0 as a pre-req but doesn't say anything about it until the install is almost done. It installs and then says 'hey, I ain't gonna work without it.' Of course the driver itself works, just not CCC, but it fails to mention that tidbit.