windows 7??

@Blasto: Using RTM. I didn't install all optional updates, only the critical ones and the ones related to my devices.

@thevortex: Windows will never have RT kernels - the nature of the architecture does not allow it.

FWIW, a bit of latency is not a killer and may actually help a multitasking environment along in a typical PC - though it is critical in more computation-critical environments such as math clusters and render farms, which are single-process machines. Which is why even on Linux, the RT kernel is an add-on.

It is also more likely to be stable than a RT kernel in a typical home setup where a operating system has to deal with multiple requests even if you have only one application running.

I understand the general theory of a real time system not being the most stable one. Although in a few days of my testing real time Linux distributions did not behave too badly for me.

But my question is what is there as the audio engine in Windows 7? Is it WASAPI?
 
Edit: The latency has not improved much. I get average 400uS now, and peaks of 700uS. I'd be willing to look at ideas. Did your test hardware have an add-in graphics card?

Posted the specifics to you in a private message. Reply back from your mail id. Will take it offline from there.

Will delete this message once done. :)
 
that's very unfortunate. My impression was that it is exactly what windows is for. For serious work there are other platforms.

Well, Then according to you, 80% of the corporates do not do any serious work.
None of the s/w companies in india/us post profits.

Unless, browzing can somehow mint money.
 
I've been cagey about M$ for a long time. And I didn't know which way this debate would swing. I still think that Win7 has some way to go in terms of overall performance, and maybe device manufacturers should get their wagons rolling a little earlier to support new OS's, for a better overall experience from shiny new systems.

However, blasto's attitude has been a revelation. After looking at my system config, he's extended personal help to me in troubleshooting it, even after the contention and arguments we went through. I don't know what the solution will be or will there be indeed a solution, but the approach is pretty refreshing.

Thanks :)

Hi,

I am happy cranky is having a re look at his measurements:D

On a serious note, about the attitude of BLASTO, this is the beauty of this forum.I have seen lot of forums all over the net where guys abuse each other as there is no tomorrow. I applaud both cranky and BLASTO for stepping back from the heat of the arguments and handling this as two mature men.:clapping:

Long Live HiFi Vision!

cheers,
sri
 
Hi,

I am happy cranky is having a re look at his measurements:D

On a serious note, about the attitude of BLASTO, this is the beauty of this forum.I have seen lot of forums all over the net where guys abuse each other as there is no tomorrow. I applaud both cranky and BLASTO for stepping back from the heat of the arguments and handling this as two mature men.:clapping:

Long Live HiFi Vision!

cheers,
sri

Agreed! Even in Hifivision though this has been an exception rather than a rule.
 
80% corporates? Do you mean people who exchange emails, fill out web forms?
IE/ outlook, antivirus are the most serious app for corporates. Though I am sure they can be ported to linux with little extra effort.

regards

Well, Then according to you, 80% of the corporates do not do any serious work.
None of the s/w companies in india/us post profits.

Unless, browzing can somehow mint money.
 
Hi,
I have seen lot of forums all over the net where guys abuse each other as there is no tomorrow. I applaud both cranky and BLASTO for stepping back from the heat of the arguments and handling this as two mature men.:clapping:

To better explain the scenario for more techy members,

Cranky was looking at 'out of the box' permance in trems of DPC (Deferred Procedure Call) handling. 'Lantency', while is a generic term used to denote any 'delay', apparently has been used in the music/video making world to denote only 'DPC Latency' which Iam not aware of. Or in other words, 'Latency' would not inherently mean 'DPC latency' to 'me' as I deal with a lot of other latencies too.

Now, as for what DPC latency itself is,

The Sytesm/Kernel drivers work at a higher prority than the user processes. To make this happen, the kernel dll are made capable of 'deferring' the procedure call made by the low priority user processes for a short time typically a max of100 uS. Usually this max time is never reached. The kernel dlls return within 20-50uS.

When seen from the user process's (here a audio/video application) perspective, it CAN get its procedure call deferred (delayed) by some amount of time due to kernel dlls deferring its procedure calls. If this delay is high, It might lead to skipping or blank spots in the audio/video output. This delay is call 'DPC latency' and a A/V application can run seamlessly on a system with <250uS.

As you can see, For the whole process to work, each individual kernel dlls should not defer the procedure calls by user processes >100uS. Only drivers which satisfy this criteria are 'certified'.

It so happens that the vendors does not bring out drivers during an OS launch as quick as we might expect. This is the problem we face during all OS releases. The end customers choose to use the driver they are aware of from an earlier version of the OS which might cause various performance issues including DPC latency (and a whole lot of other issues as well). This is termed as a performance issue of the OS itself.

While my machine was capable of producing <125uS DPC latency (attached earlier) for every 1000uS measured, Cranky's system could only archieve even <500uS with the same tool. There is definitely one or more faulty drivers which are responsible for this.

My test in 5 machines around me (which are not as high end as cranky's) yielded me results ranging from 100uS to 200uS (laptop). So an inherent OS issue is ruled out.

I and cranky are in the process of narrowing down on the faulty driver in his system. Will get back with he results. IMO, His machine is capable of attaiing ~120uS with full AERO and <100uS without AERO from the comparison I made with a similar system.
 
On a serious note, about the attitude of BLASTO, this is the beauty of this forum.I have seen lot of forums all over the net where guys abuse each other as there is no tomorrow. I applaud both cranky and BLASTO for stepping back from the heat of the arguments and handling this as two mature men.:clapping:

Agreed! Even in Hifivision though this has been an exception rather than a rule.

I dont know where you guys picked this up. :) I was only talking about something I know rather than something I heard or read about.

I repect cranky. In his field he is a genious and the reputation says the story.

But that does not mean junking measurements made by independant organisations and a whole lot of MS engineers because the tool he used for mesaurements in 'his' system gave bad results.

Also, If the DPC were only dependant on the OS, there will not be a site which will have the DPC measurement tool hosted. Instead they will measure it and post the results in place where they host the tool. :D
 
Blasto,

Thanks for the informative post. This issue with drivers not complying with OS specs has plagued Linux distributions as well. As you correctly point out, this has nothing to do with the underlying OS. If a certain driver is implemented so that it waits in a while() loop doing nothing, then every call to that driver will result in infinite wait time! I assume Win7 (or the Win API) has an interface specification to tell driver programmers how they should implement their drivers?

One thing I do not understand is this statement:
"For the whole process to work, each individual kernel dlls should not defer the procedure calls by user processes >100uS."

If my user process makes (say) 3 kernel calls sequentially and if each is deferred by 100us (max time allowed), my application will still not be able to work seamlessly since I need ~250 uS of delay for my application. So just making each dll have delay less than 100 uS is not enough to guarantee anything. That is why Win is not a real-time system. A RT kernel would guarantee that every such task set is schedulable before runtime. Does what I say make sense?

Thanks,
Ajinkya.
 
this may not seem very relevant in this techy discussion but since there are friendly and knowledgable people around i somehow feel tempted to put a few points across.....
my thinkpad came pre-loaded with Vista which i had to replace with XP, for obvious reasons...in the process i ended paying Microsoft twice over, once for the vista which i don't use and the XP that does tolerably well...lured by all this discussion about 7, i visited the microsoft website and downloaded stuff that checked my system's compatibility for the new OS..everything was fine except that it told me that my notebook would not be able to support aero graphics...
also most of the reviews (like the one that came on washington post site) seem to make much of the fact that 7 is not Vista...
now my question is, now that this new OS is out, what would be the life of XP??
at one time life was simple since you didn't do much with your PC but now with a whole lot of convergence happening what would be the lifetime of the XP????
 
One thing I do not understand is this statement:
"For the whole process to work, each individual kernel dlls should not defer the procedure calls by user processes >100uS."

If my user process makes (say) 3 kernel calls sequentially and if each is deferred by 100us (max time allowed), my application will still not be able to work seamlessly since I need ~250 uS of delay for my application. So just making each dll have delay less than 100 uS is not enough to guarantee anything. That is why Win is not a real-time system. A RT kernel would guarantee that every such task set is schedulable before runtime. Does what I say make sense?

Thanks,
Ajinkya.

You should also take into account the advantages a multi-core processor offers. When a core is handling DPC, other cores are free for the user application. So, in a 4 core machine like cranky, the probability of encountering a delay >250uS is very less unless a driver is clogging up cpu consistantly.

There are situations where the latency can go beyong the 250uS. This is called a DPC spike or DPC burst depending on the intensity. There will be DPC bursts within short period of time which might go beyond the the 250uS. Infact the MAX DPC delay can be as high as 500uS over a period of time. Theoritically it can be anything beyond that. But effective handling or cores should see it does not happen.

Infact if the music application is configured as 'realtime', the probability of it encountering the latency is very less as one core is made available to it neveretheless. As a warnging, tagging an application as realtime can have other bad effects in terms of resource clogging which might not interest people here. :rolleyes:

Windows is not a RT OS. But since the comparison here is between Win XP and Win 7, An user who was happy with XP should be happy with Win 7 is what is the point Iam trying to make.
 
Last edited:
my thinkpad came pre-loaded with Vista which i had to replace with XP, for obvious reasons...in the process i ended paying Microsoft twice over, once for the vista which i don't use and the XP that does tolerably well...
That is unfortunate. There were many laptops with linux/free dos around and a million reviews saying what vista was capable of :)

lured by all this discussion about 7, i visited the microsoft website and downloaded stuff that checked my system's compatibility for the new OS..everything was fine except that it told me that my notebook would not be able to support aero graphics...

If vista supported aero in your machine, Win 7 will be more than happy to. Since you downgraded to XP, the tool might have got confused. :eek:hyeah:
But since as you say even XP does only telerably well, I would like to see your configuration. Upload the output of 'msinfo32.exe' somewhere and PM me the link.

also most of the reviews (like the one that came on washington post site) seem to make much of the fact that 7 is not Vista...
now my question is, now that this new OS is out, what would be the life of XP??
at one time life was simple since you didn't do much with your PC but now with a whole lot of convergence happening what would be the lifetime of the XP????

XP dies by 2012. No support after that.
 
thanks a lot BLASTO.
although i am not going to immediately sing a requiem for XP, hope this is not going to be a triumph of 'style over substance'.
anyway i will miss XP...it was better than Vista...

and when i meant 'tolerably' i guess i wasn't talking about it in a 'system' sense (Intel(R)Core(TM)2Duo CPU [email protected],778MHz, 1.96GB of RAM) but rather as any other 'long suffering' windows user who has to wait 'forever' for the computer to start and then fight through all those security vulnerabilities.....
 
Hi,

When i install windows 7 from an ISO image it gives the error message that DVD/CD device driver is missing and won't go further.I tried by disconnecting the power chord of the DVD writer.But no luck.I have never heard about drivers for DVD drives.By the way i am installing it on a machine which already has XP and i want to dual boot.

Regards
 
Windows7 SP1 available-
Windows 7 and Windows Server 2008 R2 Service Pack 1

Windows 7 and Windows Server 2008 R2 SP1 will help you:
Keep your PCs supported and up-to-date
Get ongoing updates to the Windows 7 platform
Easily deploy cumulative updates at a single time
Meet your users' demands for greater business mobility
Provide a comprehensive set of virtualization innovations
Provide an easier Service Pack deployment model for better IT efficiency
 
For HTPC Users :
Improved HDMI audio device performance:
A small percentage of users have reported issues in which the connection between computers running Windows 7 and HDMI audio devices can be lost after system reboots. Updates have been incorporated into SP1 to ensure that connections between Windows 7 computers and HDMI audio devices are consistently maintained.
 
Windows 7 is an Awesome OS. Infact it's the best I have used so far. I'm not a MAC user, but I'm a Linux user. Windows 7 is hard to beat. Some people buy new PCs and use Windows XP, because they feel 'safe'. But XP is not utilising the full capability of the new hardware. Windows 7 has great UI and is actually very fast, even on my 2 and half year old Core 2 Duo Laptop.
 
That is unfortunate. There were many laptops with linux/free dos around and a million reviews saying what vista was capable of :)



If vista supported aero in your machine, Win 7 will be more than happy to. Since you downgraded to XP, the tool might have got confused. :eek:hyeah:
But since as you say even XP does only telerably well, I would like to see your configuration. Upload the output of 'msinfo32.exe' somewhere and PM me the link.



XP dies by 2012. No support after that.


blasto can definitely help us wid windows....... :)
 
my laptop with celeron almost cried wid vista.. win7 ultimate with aero on works just fine... desktop is cranking it hard with quad core now....


blasto... no updates buddy... its been long we guys met..
 
The Marantz PM7000N offers big, spacious and insightful sound, class-leading clarity and a solid streaming platform in a award winning package.
Back
Top