You are not logged in.
I use Bumblebee and have ran a few SIMPLE tests. Every time I run them, using "primusrun" performs slower than "optirun". In my bumblebee.conf, the Bridge mode is set to "auto". Is that causing the issue? From what I've read, primus is supposed to be faster.
Example of my latest "quick test":
glxspheres:
Nothing: 58 f/s
Optirun: 128 f/s
Primusrun: 62 f/s
Hardware: Acer Aspire laptop, i7, 8 GB RAM, nVidia GT 730M
Does anyone else have lower performance from Primus that from nVidia?
Matt
"It is very difficult to educate the educated."
Offline
It isn't lower performance. Primus uses vsync in it's rendering. It makes sense because the intel driver also using vsync as default behaviour.
Offline
Do not ever use glxgears as a performanc benchmark. Use something proper like the heaven benchmark.
Offline
On my laptop ,the optirun glxspheres is just 60f/s
Offline
I use Bumblebee and have ran a few SIMPLE tests. Every time I run them, using "primusrun" performs slower than "optirun". In my bumblebee.conf, the Bridge mode is set to "auto". Is that causing the issue? From what I've read, primus is supposed to be faster.
Example of my latest "quick test":
glxspheres: Nothing: 58 f/s Optirun: 128 f/s Primusrun: 62 f/s
Hardware: Acer Aspire laptop, i7, 8 GB RAM, nVidia GT 730M
Does anyone else have lower performance from Primus that from nVidia?
Does the bumblebee work well on your laptop everytime? Does the bumblebee appear the kernel panic sometimes?
Like this :
http://imgsrc.baidu.com/forum/w%3D580%3 … 1beec0.jpg
Last edited by jiangkero (2013-04-15 09:46:04)
Offline
It works well every time. Never seen a Kernel Panic either.
@Svenstaro: 1) I know it's not a "benchmark", but it IS a testing tool. 2) I used glxpsheres not glxgears. 3) The Wiki on Bumblebee mentions using glxspheres for testing performance with and without the nVidia Optimus.
Matt
"It is very difficult to educate the educated."
Offline
It works well every time. Never seen a Kernel Panic either.
@Svenstaro: 1) I know it's not a "benchmark", but it IS a testing tool. 2) I used glxpsheres not glxgears. 3) The Wiki on Bumblebee mentions using glxspheres for testing performance with and without the nVidia Optimus.
The problem with glxspheres is that most graphics card can draw hundreds of frames per second which will put optirun is a better light since it doesn't use vsync by default.
In my experience primus is faster than optirun across the board, by about 15 to 25 fps.
Offline
OK. So things like using Blender should see the increase, is that more correct?
Matt
"It is very difficult to educate the educated."
Offline
I have the same things with glxspheres.
I try Oblivion with wine with primusrun it's very slow, but fluently with opitrun.
Also try some game on steam for linux like Serious Sam 3, I have the same trouble.
Offline
From the FAQ in the README provided with Primus: (link)
Q: Performance does not exceed 60 fps, I was getting more with optirun/VirtualGL.
A: This is the effect of vblank synchronisation. For benchmarking, you can use [vblank_mode=0 primusrun ...], but in practice this will probably only waste power, as your LCD panel does not display more than 60 frames per second anyway.
A proper explanation of how vsync works and how you should use it: http://hardforum.com/showthread.php?t=928593
Come on people on the Arch Linux bbs, you should spend more time reading the documentation (RTFM) instead just spraying useless noise all over the forums. (We're not Ubuntu)
Offline
Easy, Kaustic! Let's not get on a soap box around your peers, huh?!? Check the forum, I help more than I ask. That's called give and take.
Matt
"It is very difficult to educate the educated."
Offline
Easy, Kaustic! Let's not get on a soap box around your peers, huh?!? Check the forum, I help more than I ask. That's called give and take.
I'm not talking about you specifically. It is a general problem which occurs on many forums where people fail to invest any time in researching the answer beforehand and contributing nothing but noise. Arch is overall pretty good at avoiding this combined with excellent moderation.
Edit: jasonwryan: A Taxonomy of Help Vampires is worth a read when you get a chance.
Last edited by Earnestly (2013-04-20 11:53:10)
Offline
My apologies, Kaustic. I inferred incorrectly.
Thanks for the clarification. And I'll take a look at that link later this weekend.
Matt
"It is very difficult to educate the educated."
Offline