[BUG] SNR value is smaller than acceptable value in TestPtpVolumeLoopback

[dkrzysiak]

Describe the bug SNR value is smaller than acceptable value in TestPtpVolumeLoopback Issue found in: [WHL] 02_03_TestPtpVolumeLoopback48000Hz24b32b8ch

Topology Description: Simultaneous playback and capture on external loopback (same SSP) with signal verification. Playback pipeline contains volume component.

Topology:
           +--------------------------------+
    +----+ | +---+  +---+  +---+  +-------+ |
    |Host|-->|Buf|->|Vol|->|Buf|->|SSP Dai|---+
    +----+ | +---+  +---+  +---+  +-------+ | |
           +--------------------------------+ |
                                              |
           +------------------+               |
    +----+ | +---+  +-------+ |               |
    |Host|<--|Buf|<-|SSP Dai|<----------------+
    +----+ | +---+  +-------+ |
           +------------------+

To Reproduce 02_03_TestPtpVolumeLoopback48000Hz24b32b8ch 1/50

Reproduction Rate 1/50

Environment

1. Branch name and commit hash of the 2 repositories: sof (firmware/topology) and linux (kernel driver).
   • SOF BRANCH: main HASH: https://github.com/thesofproject/sof/commit/6c4f3ea42f0d04e9a421dd8325cdbc9be31056bf
2. Name of the platform(s) on which the bug is observed.
   • Platform: WHL
3. Test environment: BRANCH: master HASH: b4886bebbe49454850d59f1a49a0460e590db71c

Screenshots or console output

logs.zip

[lgirdwood]

@singalsu looks at first glance like the gain sign could be flipping ? but I dont see how this could be per sample, possibly DMA/SSP ?

[singalsu]

@dkrzysiak Is there by chance a similar test without volume. Does it pass always?

[lgirdwood]

@dkrzysiak btw, is this with Zephyr or xtos ?

[dkrzysiak]

@singalsu there isn't. Volume has a specific topology, where SSP Dai is part of pipe_plb. @lgirdwood xtos

[wszypelt]

unsupported platform on TGL issue does not occur Tested FW hash: 4943aaa8907149cf6c02e9046e2ef5d40ba72d74