9 Jul
2006
9 Jul
'06
12:43 p.m.
Hi @ all,
I have sympathy for a new notebook with AMD Turion 64 X2 dual-core
cpu. It would open some new possibilities, like virtualization
(pacifica, xen etc.), pure AMD64 Debian arch, other 32-bit OS' and -
and that is the question - good realtime-audio performance?
Do the realtime patches of Ingo Molnar work with SMP-Kernels for
example? And the LSM-modules?
Or is the realtime-preemption in Linux just optimized for single cpu
systems?
My main interest would be to use the laptop sometimes as
"Linux-sound-module" triggered by an external midi-keyboard. Profit
virtual-instruments (virtual-synths, VSTis) much from having two CPUs
and therefore enhanced multithreading? Or is it contra-productive
eventually?
If using a SMP-Kernel would produce hassles - would it be a safe
workaround to compile a kernel *without* SMP support and then driving
one Turion core would be sufficient?
Please forgive these many questions, but I just won't trust the
marketing blindly for the benefits of dual-cores without asking some
experienced users.
Thanks for your time
Oliver
P.S. What about the laptop-fan? Can I expect, that it starts and stops
frequently when the Turion computes "sound"? (that would be annoying)
9 Jul
9 Jul
12:58 p.m.
Hallo,
oz hat gesagt: // oz wrote:
I have sympathy for a new notebook with AMD Turion 64
X2 dual-core
cpu. It would open some new possibilities, like virtualization
(pacifica, xen etc.), pure AMD64 Debian arch, other 32-bit OS' and -
and that is the question - good realtime-audio performance?
I would say be careful with AMD laptops and Linux. Intel machines tend
to be much better supported because there are open source drivers for
the other parts inside these machines like WLAN, gfx card (if it's
Intel) etc. while AMD laptops often use NVidia chipsets and gfx cards
or similar stuff by ATi for which you need to install binary drivers
sometimes, that may mess up other things like power management in
strange ways. 64bit isn't that important for audio anyways. Getting a
laptop to run flawlessly is hard enough, don't make it harder on
yourself by buying hardware without open source drivers.
Ciao
--
Frank Barknecht _ ______footils.org_ __goto10.org__
10 Jul
10 Jul
7:55 p.m.
On Sun Jul 09, 2006 at 04:57:43PM +0200, Frank Barknecht wrote:
Hallo,
oz hat gesagt: // oz wrote:
my ATI IXP based AMD64 works flawlessly on linux. even 3d accel is supported with the open
drivers after hacking a line into drm_pciids.txt
I have sympathy for a new notebook with AMD
Turion 64 X2 dual-core
cpu. It would open some new possibilities, like virtualization
(pacifica, xen etc.), pure AMD64 Debian arch, other 32-bit OS' and -
and that is the question - good realtime-audio performance?
I would say be careful with AMD laptops and Linux. Intel machines tend
to be much better supported because there are open source drivers for
the other parts inside these machines like WLAN, gfx card (if it's
Intel) etc. while AMD laptops often use NVidia chipsets and gfx cards
or similar stuff by ATi for which you need to install binary drivers
sometimes, that may mess up other things like power management in
strange ways. 64bit isn't that important for audio anyways. Getting a
laptop to run flawlessly is hard enough, don't make it harder on
yourself by buying hardware without open source drivers.
Ciao
--
Frank Barknecht _ ______footils.org_ __goto10.org__
8:25 p.m.
On Sun, 9 Jul 2006 16:57:43 +0200
Frank Barknecht <fbar(a)footils.org> wrote:
Hallo,
oz hat gesagt: // oz wrote:
Thanks. I agree in general, but it's a dilemma - feed opportunistic
companies like ati or feed quasi-monopols like intel. :-/
So the only thing, that's sure at the moment, is that I choose a AMD
Dual-Core for my next computer. And one of its OS' (besides other
multi-boots) will be a dedicated 32-bit linux for audio purposes.
Because dual-core Turions are very recent, there seems to be not much
experience with latencies and realtime-kernels at the moment. We'll
see. In reality, I only found one (o-n-e) laptop at the moment, that
could work for me (Asus A6T, nvidia/realtek based). At the same time I
know, its WLAN-card (Airgo) is not supported. The problem you
mentioned. A deal-breaker would be, if dual-cores don't work well
with realtime-audio.
Gruss,
Oliver
I have sympathy for a new notebook with AMD
Turion 64 X2 dual-core
cpu. It would open some new possibilities, like virtualization
(pacifica, xen etc.), pure AMD64 Debian arch, other 32-bit OS' and -
and that is the question - good realtime-audio performance?
I would say be careful with AMD laptops and Linux. Intel machines tend
to be much better supported because there are open source drivers for
the other parts inside these machines like WLAN, gfx card (if it's
Intel) etc. while AMD laptops often use NVidia chipsets and gfx cards
or similar stuff by ATi for which you need to install binary drivers
sometimes, that may mess up other things like power management in
strange ways. 64bit isn't that important for audio anyways. Getting a
laptop to run flawlessly is hard enough, don't make it harder on
yourself by buying hardware without open source drivers. 
2:10 p.m.
oz said:
Do the realtime patches of Ingo Molnar work with
SMP-Kernels for
example? And the LSM-modules?
Pretty sure Ubuntu ships an AMD64 kernel with PREEMPT and SMP enabled. I
have personally used the realtime-lsm module on an Debian as-shipped SMP
kernel, though not running on an SMP machine. Most (all?) of Debian's
(x86) kernels (and presumably Ubuntu's) ship with SMP enabled by default,
because it still works on uniprocesser machines without overhead. I talk
about Ubuntu and Debian because that's what I use.
If you can wait a week or two, I am getting a dual Opteron X2 (4 core
64-bit) machine this week and my intention is to do audio work on it. I
can tell you how it goes. I hope it goes well, I've spent a lot of money!
It's not a laptop. That will bring other complications, I'm sure - take
note of Frank Barknecht's comments on drivers. I have a laptop with Intel
Centrino chipset and it works well, though it is not fast enough for my
audio work (softsynths and rosegarden, moving sequences about, quickly, in
realtime). The graphics card is not Intel, it's ATI, and I'm having to use
their proprietary driver to get what I want out of it. The graphics in the
Opteron box will be nVidia though, so there's no escape as yet...
--
Michael Pacey
9:08 p.m.
On Mon, 10 Jul 2006 17:04:20 +0100 (BST)
"Michael Pacey" <michael(a)wd21.co.uk> wrote:
oz said:
Aha, thanks, that are strong hints, that the realtime patches
compile cleanly against SMP-Kernels (that's good!). But it says not much
about the achieved realtime capabilities.
The situation with the Debian-Kernels seems not to be as clear to me.
Recently I installed a Debian/stable on a dual Opteron and the default
Kernel supported only one cpu (cat /proc/cpuinfo). I had to
post-install a dedicated SMP-Kernel (2.6.15x-SMP). It seems, Debian has
always two versions (SMP and not) of a Kernel. Eventually the
new debian-installer from etch does a better autodetection
and installs a SMP-kernel by default.
Do the realtime patches of Ingo Molnar work with
SMP-Kernels for
example? And the LSM-modules?
Pretty sure Ubuntu ships an AMD64 kernel with PREEMPT and SMP enabled. I
have personally used the realtime-lsm module on an Debian as-shipped SMP
kernel, though not running on an SMP machine. Most (all?) of Debian's
(x86) kernels (and presumably Ubuntu's) ship with SMP enabled by default,
because it still works on uniprocesser machines without overhead. I talk
about Ubuntu and Debian because that's what I use. If you can wait a week or two, I am getting a dual
Opteron X2 (4 core
64-bit) machine this week and my intention is to do audio work on it. I
can tell you how it goes. I hope it goes well, I've spent a lot of money!
Yeah, that would be great to read about it here in the forum!
And I have to wait anyway, because my favorite notebook is not
available the next few weeks.
It's not a laptop. That will bring other
complications, I'msure - take
note of Frank Barknecht's comments on drivers. I have a laptop with Intel
Centrino chipset and it works well, though it is not fast enough for my
audio work (softsynths and rosegarden, moving sequences about, quickly, in
realtime). The graphics card is not Intel, it's ATI, and I'm having to use
their proprietary driver to get what I want out of it. The graphics in the
Opteron box will be nVidia though, so there's no escape as yet...
Ok, but that was a mono-processor (right?). Best wishes for your
quadro-core system! When it will be really fast, chances are good,
that a dual-core in a laptop can be sufficient for me.
Thanks for your comments,
Oliver
9:18 p.m.
On Tue, 2006-07-11 at 01:08 +0200, oz wrote:
Aha, thanks, that are strong hints, that the realtime
patches
compile cleanly against SMP-Kernels (that's good!). But it says not
much about the achieved realtime capabilities.
As one would expect realtime performance is no better or worse than 32
bit i386. You need a recent JACK version as dual core AMDs have a
hardware bug that screws up JACK's timing.
Lee
9:38 p.m.
Le 11 Jul à 01:08, oz ecrivait:
Hi.
I have exactly the same hardware, and I'm particularly happy with
it. It works really great, including with Ingo Molnar's realtime patch
(2.6.16 kernel). The only difference is that I use it with 32 bits gentoo,
not 64 bits.
By the way, does one can expect real improvements with the 2.6.17
kernel, especially regarding the new scheduler for dual core CPU ?
Thanks in advance for any advice,
Y.
If you can
wait a week or two, I am getting a dual Opteron X2 (4 core
64-bit) machine this week and my intention is to do audio work on it. I
can tell you how it goes. I hope it goes well, I've spent a lot of money!
Yeah, that would be great to read about it here in the forum!
And I have to wait anyway, because my favorite notebook is not
available the next few weeks.
6764
days inactive
6765
days old
linux-audio-user@lists.linuxaudio.org
7 comments
6 participants
participants (6)
-
carmen -
Frank Barknecht -
Lee Revell -
Michael Pacey -
oz -
yves_p＠nnx.com