| 09:06 | jnettlet[m]> | Ke: I don't think there is anything there that is USB specific. Some PCIe fixes, and then SFP / networking |
| 12:54 | jnettlet[m]> | any QT5 users here? I am wondering what a good performance test for the GPU is. |
| 17:29 | jnettlet[m]> | for those interested. I know I was talking topi about it a bit ago. Currently our full HB2 carriers suspend to ram power draw is 38mA at 12V so about 468mW |
| 17:47 | jnettlet[m]> | but for those interested in real world etnaviv vs galcore performance. Using the Qt5_CinematicExperience as a benchmark, the etnaviv driver struggles with 30fps at 720p, the binary driver is able to do a solid 60fps...this is in the main screen with everything enabled. So still quite a bit of ground to make up. |
| 17:52 | Ke> | I don't think many people want to use the open source drivers for gaming |
| 17:52 | Ke> | but who knows |
| 17:52 | Ke> | I would be more interested to know, whether you can render 4K video |
| 17:53 | Ke> | or 1080p video |
| 17:57 | jnettlet[m]> | On the MX6 it can do 1080p not 4k |
| 17:58 | Ke> | it's possibly not just about the encoder but also about rescale ability |
| 17:58 | jnettlet[m]> | I haven't tested the MX8m yet. It can do 4K video fine. The GPU is much faster.... Still same bugs I have had to adapt and patch. |
| 17:59 | Ke> | do you mean that actually rendering 4K to a 4K display was possible due to strict limitations of the spec or just too slow |
| 17:59 | Ke> | s/possible/impossible/ |
| 18:00 | jnettlet[m]> | The MX6 doesn't have a clock fast enough to display 4k even 4k30 |
| 18:00 | Ke> | display encoder? |
| 18:00 | Ke> | or gpu |
| 18:02 | jnettlet[m]> | Yes... Basically everything. In theory the GPU can handle 4k sized textures... But the memory controller would be a huge bottleneck. The VPU decoder can't handle 4k video. Even with the new clock changes that allow it to run at 396Mhz |