Current Server Time: 15:07 (Central Europe)

#apertus IRC Channel Logs

2020/12/13

Timezone: UTC


00:02
BAndiT1983
changed nick to: BAndiT1983|away
01:55
illwieckz
left the channel
01:57
illwieckz
joined the channel
02:17
aombk
left the channel
02:20
aombk
joined the channel
04:37
Bertl_oO
off to bed now ... have a good one everyone!
04:37
Bertl_oO
changed nick to: Bertl_zZ
07:13
BAndiT1983|away
changed nick to: BAndiT1983
08:19
alexML
joined the channel
10:52
BAndiT1983
changed nick to: BAndiT1983|away
11:17
Bertl_zZ
changed nick to: Bertl
11:17
Bertl
morning folks!
11:18
se6ast1an
good day!
11:51
Bertl
off for now .. bbs
11:52
Bertl
changed nick to: Bertl_oO
12:31
comradekingu
left the channel
12:33
Bertl_oO
changed nick to: Bertl
12:33
Bertl
back ...
12:38
comradekingu
joined the channel
13:02
BAndiT1983|away
changed nick to: BAndiT1983
13:16
Guest27507
left the channel
13:16
Guest27507
joined the channel
13:17
MarkVandenBorre[
left the channel
13:18
MarkVandenBorre[
joined the channel
13:18
davidak[m]
left the channel
13:19
davidak[m]
joined the channel
13:55
anuejn
se6ast1an: everything clear now?
14:42
se6ast1an
mostly, so many more questions and ideas already :)
14:43
se6ast1an
for example the mentioned jetson hardware from nvidia
14:43
se6ast1an
I would like to learn more how its used and what it can do
14:43
se6ast1an
its basically a platform for mobile GPU processing I assume?
14:43
se6ast1an
shaders, etc.
14:44
se6ast1an
if yes how does it compare performance wise to dedicated GPUs in PCs
14:53
BAndiT1983
https://medium.com/@aallan/benchmarking-edge-computing-ce3f13942245
14:54
BAndiT1983
don't know if you can compare it to a normal GPU, as V-RAM is one of the biggest factors in calculations
14:55
BAndiT1983
although the specs look impressive
15:06
se6ast1an
do they?
15:06
se6ast1an
"we’ll also run the same models on my Apple MacBook Pro (2016), which has a quad-core 2.9 GHz Intel Core i7"
15:06
se6ast1an
so no GPU acceleration on an old mac book
15:07
se6ast1an
shows better or similar processing sped compared to the embedded solutions
15:07
se6ast1an
*speed
15:07
se6ast1an
speed vs power consumption would have been interesting there
15:11
BAndiT1983
i have the same macbook from the company and it has a GPU, some radeon, don't remember the model
15:23
BAndiT1983
but besides the CPU/GPU debate, there is still proper software development and optimization involved, as intel also has quite many possibilities to let specific software run faster on their CPUs, e.g. embree for rendering
15:24
BAndiT1983
just comparing raw power, won't tell the real state of things
16:24
vup
shared memory can be a big advantage aswell, if you need to share a lot of data between the cpu and gpu, upload and download stuff to the gpu can usually be one of the main bottlenecks, especially when not doing a lot of work on the gpu per byte
16:36
BAndiT1983
true
18:28
se6ast1an
but do I understand correctly that jetson is great because it offers GPU processing for mobile applications for more performance per watt but the same is basically possible in laptops and of course desktop GPUs?
18:29
se6ast1an
the above article talks about AI mainly, but what about "normal" GPU processing?
18:43
Bertl
off for now ... bbl
18:43
Bertl
changed nick to: Bertl_oO
18:47
BAndiT1983
se6ast1an: what is normal gpu processing?
18:47
BAndiT1983
such stuff? -> https://www.fastcompression.com/blog/jetson-image-processing.htm
18:47
se6ast1an
shaders, cuda, etc.
18:48
BAndiT1983
see the article, it's also there
18:49
BAndiT1983
https://docs.nvidia.com/jetson/l4t/index.html#page/Tegra%20Linux%20Driver%20Package%20Development%20Guide/graphics_comp_shader.html
18:49
se6ast1an
right, I see jetson already provides some of the usual suspects for cameras out of the box
18:49
se6ast1an
debayering, etc.
18:49
BAndiT1983
yep, it looks like it was developed with cameras in mind, probably for automation or so, as it supports multiple cameras at once
18:50
se6ast1an
I see