1 |
As regulars are aware, I /was/ running an old Radeon 92xx series card, |
2 |
r200 series chip. My system was /relatively/ good, even if it's half a |
3 |
decade old now, because it's a dual socket Opteron, which I had upgraded |
4 |
to top-of-the-line dual-core Opteron 290s (2.8 GHz), with plenty of |
5 |
memory (8 gigs, tho it's now six as a stick went bad on me and I've not |
6 |
replaced it yet), and running four SATA drives in md/kernel RAID. |
7 |
|
8 |
Well, a few weeks ago I switched the system partitions from RAID-6 to |
9 |
RAID-1. In many tasks the RAID-1 is actually faster than the RAID-6 was, |
10 |
tho part of that might be that the new partitions aren't fragmented, |
11 |
yet. While I was at it, I rid myself of the LVM2 layer I was running |
12 |
most of the non-rootfs system on. No real issues with it here, but it |
13 |
was a bit of a hassle since I couldn't put the rootfs on it directly, and |
14 |
I have seen some horror stories I didn't like, tho whether they're |
15 |
accurate on the current LVM2 I don't know. But anyway, I decided that |
16 |
layer was more hassle than it was worth, and experience with the new |
17 |
layout so far says I was right. |
18 |
|
19 |
But that just lays the groundwork for the REAL upgrade. I FINALLY got |
20 |
the video card upgrade that I'd been needing for awhile, thus bringing it |
21 |
more inline with the rest of the system. It's a Radeon hd4650, rv730 |
22 |
chip, gig video RAM (tho I have a feeling I'm not using anything near |
23 |
that), dual DVI output (I'm not sure if both are dual-link tho, might be |
24 |
one dual-link and one single-link), AGP bus as that's what my system is |
25 |
-- five years old, remember, I have PCI-X but not PCI-E. |
26 |
|
27 |
Of course, the xorg native xf86-video-ati driver (and xf86-video- |
28 |
radeonhd, tho that seems to be falling behind now, unless you have HDMI |
29 |
you want to support or something) only have 2D for anything r600 or newer |
30 |
in their released drivers, thru the 6.12 series (with 6.12.4 being the |
31 |
latest, and a possible 6.12.5 coming up). There's not even a beta |
32 |
tarball out for the 6.13 series yet, so if one wants OpenGL support, |
33 |
really the whole point to the upgrade, one has to run the "live" driver, |
34 |
straight from git or available in the x11 overlay as the traditional live |
35 |
version 9999. |
36 |
|
37 |
So that's what I grabbed. I already had the latest non-live xorg |
38 |
components installed from the tree and x11 overlay, so I was fortunate |
39 |
and didn't need any further live packages, only xf86-video-ati-9999. |
40 |
|
41 |
Meanwhile, I basically gave up on the kernel bug I was git-bisecting, as |
42 |
I couldn't duplicate it on the (then still unaccelerated) new radeon |
43 |
hardware, tho I saved a bisect-log in case it comes back with the new |
44 |
hardware after I enable acceleration, git-pulled, did a git-checkout of |
45 |
v2.6.32 (Linus git tree), did the usual oldconfig, then a menuconfig and |
46 |
changed my config around a bit, enabling KMS, etc. |
47 |
|
48 |
Did a reboot into the new kernel and played around at the radeondrmfb |
49 |
enhanced CLI for awhile, tweaking a couple things there, then started X/ |
50 |
kde4 and started tweaking things for the new hardware, there. |
51 |
|
52 |
After editing xorg.conf and restarting X a few times, playing with |
53 |
glxgears, etc, I started trying out the newly available kde4 OpenGL eye |
54 |
candy options. =:^) As I run dual 22" 1920x1200 LCDs, stacked for |
55 |
1920x2400, and the old card couldn't handle OpenGL at resolutions above |
56 |
2048 either direction, I hadn't had the OpenGL effects available to play |
57 |
with on the old card. What a change the new card made! =:^) |
58 |
|
59 |
So now I'm running kde 4.3.4 with OpenGL effects. It's nice. I've |
60 |
actually had the "snow on the desktop" effect turned on as I worked, for |
61 |
several hours now, tweaked a bit to add more "flakes" but reduce the size |
62 |
to make them a bit more realistic, and with the "behind windows" option |
63 |
turned off, so they float in front of the windows. Much like watching |
64 |
real snow fall outside the window while you're nice and warm inside, it's |
65 |
quite a calming effect. |
66 |
|
67 |
OTOH, there's still enough glitches to see why it's not released yet, and |
68 |
I did have one crash. Also, font anti-aliasing /really/ looks bad now, |
69 |
it's /gotta/ be a bug somewhere I'm sure, so I turned off font anti- |
70 |
aliasing entirely. MUCH better! With that, it's working well enough to |
71 |
be usable if a few visual glitches, mostly background repaints turning |
72 |
bits of the plasma panels and desktop weird colors at times, which goes |
73 |
away with desktop switches, etc, but also a semi-regular flashing of bits |
74 |
of one particular corner of the desktop, and artifacts appearing on |
75 |
scrollbars and the like occasionally. But it's good enough I've no |
76 |
intention of going back, even if the driver code is unoptimized at |
77 |
present and the snow makes new launches rather less than responsive! But |
78 |
I can always turn the snow bit off, if I want, and have a reasonably |
79 |
responsive system with the other effects still. |
80 |
|
81 |
So now I suppose I'm experiencing kde4 as it was meant to be seen, fully |
82 |
accelerated opengl effects, cube desktop switching, snow on the desktop, |
83 |
wobbly windows (which unlike many, I think I'll keep tho I turned down |
84 |
the effect power a notch, and can turn it down another if I want), cover- |
85 |
switch for alt-tabbing, etc. Very nice, even with the glitches. It'll |
86 |
be even nicer when the radeon r600 opengl driver and kernel KMS matures a |
87 |
bit. Unfortunately, even in 4.3.4, kde4 itself is still buggy enough I'd |
88 |
consider it beta, tho late beta now. The first kde 4.4 beta is out now, |
89 |
and since 4.3 still feels like beta anyway, I'll probably upgrade before |
90 |
the scheduled February release date, tho it'll probably be beta2 or rc1 |
91 |
before I get to it. I expect kde 4.4 to be what I'd call release |
92 |
candidate quality, the critical bugs gone and no show-stoppers, but still |
93 |
not quite there, and 4.5, in August, to finally hit what I'd call good |
94 |
release quality suitable for an ordinary user. After that, it'll all be |
95 |
frosting on the cake, especially now that I have a decent video card and |
96 |
can enjoy it as it was designed to be enjoyed. =:^) |
97 |
|
98 |
-- |
99 |
Duncan - List replies preferred. No HTML msgs. |
100 |
"Every nonfree program has a lord, a master -- |
101 |
and if you use the program, he is your master." Richard Stallman |