It seems you say that the SYSVAD can pass the HLK certification in some
conditions that make me think you never really made the test, isn’t it ? Did
you really perform the test by yourself ?
Last year we finally got these 3 test failure (nothing else):
- Audio Codec – AC3 Test – Certification – Desktop
- Audio Codec – Audio Logo Test – Glitch – Desktop
- Audio Codec – Class Driver Audio Logo Test – Certification – Desktop
BTW: The Communications Audio Fidelity Test was declared OK …
You may consider making the HLK test on a fresh Windows 10 installation, and on
a first SYSVAD installation.
(because the last problem we found is that the HLK test does not give the same
result after driver update).
Could you send a screenshot of your HLK test after first pass please ?
Regards
Vincent Burel
De : wdmaudiodev-bounce@xxxxxxxxxxxxx [mailto:wdmaudiodev-bounce@xxxxxxxxxxxxx] ;
De la part de Gary Daniels (Redacted sender "Gary.Daniels" for DMARC)
Envoyé : vendredi 24 mai 2019 19:36
À : wdmaudiodev@xxxxxxxxxxxxx
Objet : [wdmaudiodev] Re: Audio Stack Validation / SYSVAD and HLK Certification
validation
If the installation instructions in the sysvad readme.md file are followed,
then all Audio Hardware Logo Kit test cases in the compat playlist (the list
required for certification) will pass on the SYSVAD sample driver.
If these instructions are not followed and the driver is installed via "add
legacy hardware," the hardware loopback test in the HardwareAudioProcessing
(HAP) test suite will fail. This is the only expected failure.
* When is hardware loopback needed?
* Hardware loopback is a recommended feature for drivers which implement
signal processing modes, so that the most accurate representation of what is
rendered to the speaker can be provided for loopback clients to perform Echo
Cancellation.
* AudioDG will provide a software loopback when hardware loopback is not
supported, however the signal processing would be missing from the loopback
stream.
* Hardware loopback is required for drivers which implement multiple
signal processing modes.
* Because the loopback must be a mix of multiple concurrent streams
running at different modes, AudioDG cannot provide a software loopback.
* Why does this test case fail?
* The test compares the loopback audio to a reference sound played and
fails if the loopback goes out of tolerance.
* The SYSVAD sample driver provides a sine tone for the simulated
loopback audio. Because this simulated audio is unrelated to the reference
sound the test will fail.
* Why does SYSVAD implement hardware loopback, why not provide a more
accurate simulated loopback?
* Being an example driver, SYSVAD implements multiple modes and hardware
loopback to show how they are presented to the OS, the associated properties,
structure definitions, connections, format negotiation, etc. It is hugely
beneficial to have a sample which shows how to do these things.
* The vast majority of drivers implement signal processing and mixing in
hardware. Creating throw-away simulated signal processing or mixing is outside
of the scope of the sample. Driver developers know how to implement signal
processing and mix audio. There is limited benefit to simulating this and doing
so would add additional complexity.
* Why does the test behavior change?
* It simply makes no sense to perform loopback testing on the SYSVAD
sample driver. However, any "real" driver will have to pass this test. With
that goal in mind, the test was modified to skip if it can identify the SYSVAD
driver.
Running tests beyond the compat playlist, failures may be encountered with the
SYSVAD driver. Many of the tests outside of the compat playlist are simply not
applicable to the SYSVAD driver. The Communications Audio Fidelity Test, for
example, requires special hardware and does validation for round trip latencies
for Skype certification. It requires a real audio driver with a real microphone
and a real speaker. It will fail if run with SYSVAD. The Voice Activation
Manager tests will run and pass on the SYSVAD driver using simulated keyword
programming and detection, however this test is really only interesting for
developers creating a hardware keyword spotter. Judgement is required. SYSVAD
is, after all, a development example and not a product.
Another test which is not in the compat playlist and may encounter issues is
the audio glitch tests. The tests may fail due to glitches introduced by Driver
Verifier, by default the HLK enables Driver Verifier for all testing. This is
not unique to SYSVAD. It is recommended that these test be run first with
Driver Verifier enabled so that Driver Verifier can flag any problems with the
driver. If the tests pass on the first run then no additional work is needed.
Obviously, any issues detected by Driver Verifier should be fixed. If the tests
fail due to detected glitches, the test should be run a second time with Driver
Verifier disabled for pass/fail determination.
-Gary
Sent from Outlook <http://aka.ms/weboutlook>
_____
From: wdmaudiodev-bounce@xxxxxxxxxxxxx <wdmaudiodev-bounce@xxxxxxxxxxxxx> on
behalf of Vincent Burel (VB-Audio) <vincent.burel@xxxxxxxxxxxx>
Sent: Saturday, May 18, 2019 1:19 AM
To: wdmaudiodev@xxxxxxxxxxxxx
Subject: [wdmaudiodev] Audio Stack Validation / SYSVAD and HLK Certification
validation
I can change the title of this thread, But can you reply correctly to my
question :
Do you (or Gary Daniels or Pete Brown or Anyone at Microsoft) know if your
company will validate the SYSVAD audio driver and HLK Certification Process one
day?
In other words, can you say me if the HLK certification process works 100% with
the current SYSVAD audio driver example ?
In other words, can we build our audio driver for WIN10 or not ?
Thanks by advance!
Regards
Vincent Burel
De : wdmaudiodev-bounce@xxxxxxxxxxxxx [mailto:wdmaudiodev-bounce@xxxxxxxxxxxxx] ;
De la part de Matthew van Eerde (Redacted sender "Matthew.van.Eerde" for DMARC)
Envoyé : vendredi 17 mai 2019 17:10
À : wdmaudiodev@xxxxxxxxxxxxx
Objet : [wdmaudiodev] Re: Increasing Timer Resolution results in erratic
Scheduling/DPC Latencies on Windows 10
Um… what does this have to do with increasing timer resolution?
If you want to talk about HLK failures, please start a new thread.
If you’re running into a particular failure, please share the .hlkx.
_____
From: wdmaudiodev-bounce@xxxxxxxxxxxxx <wdmaudiodev-bounce@xxxxxxxxxxxxx> on
behalf of Vincent Burel (VB-Audio) <vincent.burel@xxxxxxxxxxxx>
Sent: Friday, May 17, 2019 6:24:08 AM
To: wdmaudiodev@xxxxxxxxxxxxx
Subject: [wdmaudiodev] Re: Increasing Timer Resolution results in erratic
Scheduling/DPC Latencies on Windows 10
Thanks Michael for your experience!
Matthew!? Gary !? Pete !? Anyone at Microsoft will validate the SYSVAD audio
driver and HLK Certification Process one day or what ?
Regards
Vincent Burel
De : wdmaudiodev-bounce@xxxxxxxxxxxxx [mailto:wdmaudiodev-bounce@xxxxxxxxxxxxx] ;
De la part de Michael Johansen
Envoyé : mardi 14 mai 2019 16:17
À : wdmaudiodev@xxxxxxxxxxxxx
Objet : [wdmaudiodev] Re: Increasing Timer Resolution results in erratic
Scheduling/DPC Latencies on Windows 10
Struggling on my own trying to pass audio HLK tests.
Many DF tests fail. When I open the log report it says 0 failed 100% passed. I
don't get it....
KS_Position test sometimes fail and sometimes not.
Others just fail with errors that don't give me any hint on what is really
happening and why it fails.
Br
Michael
Den tir. 14. maj 2019 kl. 15.38 skrev Vincent Burel (VB-Audio)
<vincent.burel@xxxxxxxxxxxx>:
Another problem? Not reassuring.
BTW: Since December, I’m still waiting for Microsoft statement about audio
driver for WIN10.
Can you (or someone who knows) confirm that HLK certification process work 100%
with SYSVAD audio driver example now ?
I will maybe have a time slot in June/July to rebuild a SYSVAD example and test
it in HLK Certification process…
Regards
Vincent Burel