f_|DSR has quit [Remote host closed the connection]
f_|DSR has joined #linux-sunxi
aggi has joined #linux-sunxi
DuClare has joined #linux-sunxi
DuClare_ has quit [Ping timeout: 480 seconds]
szemzoa has quit [Remote host closed the connection]
szemzoa has joined #linux-sunxi
apritzel has joined #linux-sunxi
szemzoa has quit [Remote host closed the connection]
szemzoa has joined #linux-sunxi
szemzoa has quit [Remote host closed the connection]
szemzoa has joined #linux-sunxi
apritzel has quit [Ping timeout: 480 seconds]
apritzel has joined #linux-sunxi
digetx is now known as Guest16281
digetx has joined #linux-sunxi
Guest16281 has quit [Ping timeout: 480 seconds]
radxanaoki1 has quit []
<spawacz>
Are there significant benefits of using -mcpu=thead-c906 on the D1? Does it speed up some particular workloads by a good ammount?
Schimsalabim has quit [Read error: Connection reset by peer]
Schimsalabim has joined #linux-sunxi
Robot_ has joined #linux-sunxi
aggi has quit [Remote host closed the connection]
aggi has joined #linux-sunxi
<apritzel>
spawacz: is that a solution looking for a problem? Do you have any specific workload you are interested in? Then you could benchmark it, I guess?
<apritzel>
on the ARM and x86 side my experience is those CPU optimisations are overrated, and are not worth the price to pay - code runnable only on a subset of machines
<spawacz>
I don't. I was just curious what these extensions may do
<apritzel>
though the RISC-V experience might be different, with those wild custom CPU extensions
<gamiee>
spawacz : I think to some degree yeah, it can improve performance, but I do not know the statistics about how much.
<gamiee>
also it's better to use T-Head gcc fork for this.
<apritzel>
it's probably more worthwhile to look into libraries that auto-detect features at runtime and chose an optimal implementation
<apritzel>
gamiee: "T-Head gcc fork" sounds very scary ...
<gamiee>
apritzel : yeah, although, upstreaming is going, what isnice