<tomeu>
lynxeye: have made lots of progress on the NPU tflite delegate, but I'm stumbled by the GPU hanging every time I try to use the 8th NN core
<tomeu>
the cmd stream and all buffers are bit identical
<tomeu>
do you have any suggestions on how to debug this further?
frieder has joined #etnaviv
<austriancoder>
tomeu: maybe have a look at the registers that get written/read by the kernel driver
<tomeu>
austriancoder: is there any tooling that would make this easier?
<tomeu>
guess I should also decode all the cmdstream that the kernel generates
<austriancoder>
tomeu: the galcore kernel driver has some kind of infra in place that needs to be enabled. Just have a look at e.g. gckOS_WriteRegisterEx(..) and the usage of gcmkHEADER_ARG(..)
<austriancoder>
tomeu: yeah.. having a look at the cmdstreams the kernel generates might be also helpful. Maybe they are also using their debug infra for this.
<tomeu>
I was thinking of some scripts in etna_viv to decode registers, cmdstream, etc
<austriancoder>
for cmdstream there is something: dump_separate_cmdbuf.py
<tomeu>
ok, thanks, will give that a try
lynxeye has joined #etnaviv
Leopold_ has quit [Remote host closed the connection]
Leopold_ has joined #etnaviv
dos1 has quit [Ping timeout: 480 seconds]
dos1 has joined #etnaviv
karolherbst is now known as karolherbst_
karolherbst_ is now known as karolherbst
frieder has quit [Remote host closed the connection]
otavio has quit [Ping timeout: 480 seconds]
otavio has joined #etnaviv
chewitt has quit [Quit: Zzz..]
chewitt has joined #etnaviv
chewitt has quit [Read error: Connection reset by peer]