# Voxl-streamer x264enc with gpu

Source: https://forum.modalai.com/topic/1014/voxl-streamer-x264enc-with-gpu
Category: FAQs (https://forum.modalai.com/category/4/faqs)
Posted: 2022-06-13 14:48:20 UTC by Philemon Benner
Replies: 6 · Views: 1474

## Philemon Benner · 2022-06-13 14:48:20 UTC

Hi,
i'm using voxl-streamer and i'm wondering if it's possible that the x264enc in voxl-streamer can use the apq8096 gpu for acceleration of the encoding. I had a look at gstreamer and x264 subproject has an opencl feature that can be enabled, but i haven't found any examples on x264 with opencl. Also is there a source for docker opencl prebuilt package ? I wanna build it with a different compiler.

## Reply by Eric Katzfey (ModalAI staff) · 2022-06-13 17:58:32 UTC

@Philemon-Benner x264enc runs on CPU only. But there is omxh264enc on apq8096 that uses Qualcomm HW acceleration.

## Reply by Philemon Benner · 2022-06-14 08:34:59 UTC (in reply to Eric Katzfey)

@Eric-Katzfey Thanks for the answer i will have a look at it.

## Reply by Philemon Benner · 2022-06-14 10:24:42 UTC

@Eric-Katzfey is there a way to compile omx core with another compiler ? If so could you refer me to the site where i can build it from source, i haven't found anything yet. 
I'm getting the following exec format error: 
```
0:00:10.059630626  4097   0x7f88666070 ERROR                    omx gstomx.c:177:gst_omx_core_acquire: Failed to load module '/usr/lib/libmm-omxcore.so': Error loading shared library /usr/lib/libmm-omxcore.so: Exec format error
```

## Reply by Chad Sweet (ModalAI staff) · 2022-06-14 15:34:13 UTC

I believe the omxcore on APQ8096 is 32-bit. That error implies your host application does not match the omxcore ABI. You might want to try 32-bit compilation

## Reply by Philemon Benner · 2022-06-14 16:04:27 UTC (in reply to Chad Sweet)

@Chad-Sweet yeah i'm actually using musl toolchain and compiling in 64-Bit. I wanna build omx core with musl from source.

## Reply by Eric Katzfey (ModalAI staff) · 2022-06-14 16:17:33 UTC (in reply to Philemon Benner)

@Philemon-Benner We build this as part of our Yocto system image build. It is 32 bit. We have not tried to build this outside of the Yocto build.