Well, that's less than $40.00 !!!!!!!!!!!!!!!!!!!!

The first audio tests showed a high DPC latency. Now Cakewalk does a test and Winows 8 performs better than Windows 7. What happened?

Maybe they did some optimizations

Maybe they removed debug code...

I found the DPC latency was much higher in Windows 8. Cakewalk also provided figures saying that 64 bit was 20% faster than 32 bit OS for vsti. Maybe they are cooking the books? Or perhaps giving skewed impression.

And what exactly would be the advantage in that?

Interesting results. I am interested about one thing though. Have been kind of waiting for all the interesting ipad music applications to come to other platforms but it is still not happened. Android has failed to provide low latency so it is unlikely to become music platform in the near future.

But does this information mean that it is even unlikely for Win8 or WinRT come the other multitouch music application platform...? If IPad apps will be ported they would likely become metro applications (with the 100ms and no MIDI)...?

How likely do you think that MIDI and low latency will come to metro in near future...


sigh... I guess I just need to buy ipad or something...., although I hate that it lacks a keyboard.

Youll have to ask Cakewalk why they wanted to publish the figures attall. What do they gain? The results show they are promoting Windows 8.

MS and Cakewalk have had a close relationship for many years. Cakewalk supported only DirectX plugins for a long time. To say they are misleading their customers and potential customers with this information is a bit of a reach though. It may simply be that their application is better optimised for Win8 than some others.

The Cakewalk article makes it appear as if audio performance is a problem of overall throughput and system load that can be measured in "CPU reduction" or number of system calls per seconds. In reality the audio latency problem is about minimizing the amount of bad case scenarios (buffer underruns) which the article does not seem to address. Also while the SONAR Perf utility remains in-house we will just have to take their word for it. Though it would be interesting to know where the reduction in system calls would come from.

As for "high DPC latencies" on Windows 8, this can be attributed to the way kernel timer events are pushed out due to a new feature called dynamic clock interval which reduces power consumption. The system clock does not interrupt at periodic fixed intervals but aperiodically, then when the OS deems appropriate. The "DPC latency" problem is due to the method of measuring which uses a kernel timer, it does not mean that DPCs are actually executing longer. Audio software and drivers generally do not make use of kernel timers.

In any case, the cards have been reshuffled but I have no information that would indicate that Win8 would be a less suitable OS for handling real-time audio.

//Daniel

Thanks, this are good and deeper informations! I read somewhere about changed stuff in the kernel for reduced power consumption.
In the Microsoft forum the developers are not very happy while people using the DPC latency checker tool. Eventually it's true that the DPC tool is not usable for Win8 because of some changed stuff.

Can anybody confirm this?

From their website:

Windows 8 Compatibility: The DPC latency utility runs on Windows 8 but does not show correct values. The output suggests that the Windows 8 kernel performs badly and introduces a constant latency of one millisecond which is not the case in practice. DPCs in the Windows 8 kernel behave identical to Windows 7. The utility produces incorrect results because the implementation of kernel timers has changed in Windows 8 which causes a side effect with the measuring algorithm used by the utility. Thesycon is working on a new version of the DPC latency utility and will make it available on this site as soon as it is finished.

If you have it in the Bios, changing HPET might make it read right ?? ... DPC latency utility in 8

If MS goal for acceptable latency for metro applications is 100ms, doesn't it also imply that we will not see audio applications in winRT tablets or winphone8?

I mean they are almost the same metro applications, arent they? Now that all three os (win8, winRT, winphone8) have the same kernel.

If this is really so I think it is very bad news as I have been hoping to see music applications also in some other platforms than iphone/ipad.


Has anybody still measured the actual latency in metro apps?




Then another question, if metro is unusable for audio and you code with normal win8 API, does it also imply that you cannot use multi touch gestures in your music app?