
|
[haiku-appserver]
||
[Date Prev]
[05-2006 Date Index]
[Date Next]
||
[Thread Prev]
[05-2006 Thread Index]
[Thread Next]
[haiku-appserver] Re: BWindowScreen broken
- From: "Rudolf" <drivers.be-hold@xxxxxxxxxxxx>
- To: haiku-appserver@xxxxxxxxxxxxx
- Date: Tue, 02 May 2006 22:08:33 +0200 CEST
Hi,
> // TODO: hardcoded to vesa for now. Fix me
> string = "graphics/vesa";
hehe, that makes sense :)
I just ran the benchmark util trice: exact same speeds over and over.
MTRR is in use, and AGP as well (FW).
Differences between R5/dano and haiku I noted:
-R5 clears the screen to black immediately after start of tha app,
haiku does not (app delays for 3 seconds so I can see this clearly:
rubbish on Haiku. Not very important I think though: cosmetic)
-more of a fault: Haiku forgets to restore the original screen
settings....
Apart from that this one app runs perfectly :)
-------
Bench results: (laptop at 600Mhz, mains powered):
(waiting.. ah there it is)
Dano:
write 64 648Mb/sec
32bit 638Mb/sec
8-bit 479Mb/sec
Read 64 bit 7.75Mb/sec
32, 4.07
8, 1.01
---
Haiku (tested 3 times, all about the same (diffs about one percent or
less)
write 64, 629
32, 629
8, 466
read 64, 7.64
32, 4.02
8, 0.99Mb/sec
So, All up and running nicely!
Aparantly Haiku loads the bus a bit more than Haiku over here though.
Off course, I realize that these accesses are outside of the
app_Server: they are assembler routines doing their work to minimize
the CPU limit factor.
Stephan's benchmark from bebits will be intesting to run for
app_server results :)
I guess I will be running pageflipper next up ;-)
Bye!
Rudolf.
> return B_OK;
> }
>
> I'll fixed it now if you don't mind, although I think BWindowScreen
> is
> broken anyway - we'll see.
>
> Bye,
> Axel.
>
>
>
|

|