Windows vs non realtime kernel Linux

[GhostofJohnToad]

I have been a dabbler in Linux for a number of years mostly Debian and Fedora. My biggest headache was always Jack and realtime. Just getting it to work correctly was a pain. I discovered Planet CCRMA for Fedora and things got better. But each new RT kernel update borked my setup. So, I stayed with XP for all my critical audio recording needs.

I recently got a "new to me" laptop with pretty good specs(dual core 2.4 GHz, discreet video card, 4GB RAM, 250GB-72000RPM HD)and am thinking about going all out and using Linux exclusively for an album I am working on. I have read that on newer systems RT is not that critical anymore. I never had any problems with Windows latency which got me wondering. How does a vanilla kernel fare against a Windows setup?

[brummer]

Hi

There is already a lot of the rt-kernel patches slipped in to the mainline, so you can enable rt-prio to jack without runing a rt-kernel,you can use High-resolution timer, . . .
Many ppl reported that they dont use rt-patched kernels any more, because they didn't feel the need for it.
But, the last quantum of performance you can only get out of your box, with a real-time kernel.

I use always since years a rt-kernel for any task I do, and I experience never a problem with it, okay, sometime a update to a new kernel could fail, but you can simply boot in the older well running kernel, so there is no reason to be angry before a kernel update.

I never use windows, and I never use a non rt-kernel on linux, so I cant say if the performance of a linux box without a rt kernel is better or not then a windows box, but I think it is, as far as I know, windows didn't have a rt-kernel, and the internal system calls in a windows box eats a lot of the free CPU power. . . .


regards brummer

[SR]

What is the typical amount of latency you can expect on Windows?

I think YMMV on Linux with a non-rt kernel because of hardware variations.

[spm_gl]

What is the typical amount of latency you can expect on Windows?

We can get below 1ms (0.6 I think) with an RME HDSP card.

[funkmuscle]

but don't you get next to 0 latency with the RME HDSP card no matter what OS? That's at least how I've seen it advertised. and thorgal can varify that more if he reads this..

My daughter just got a new laptop so her old xp machine will now be mine but I wanted to see what xp could do running it as an effects unit hosting vsts via VSTHOST before installing ArchBang as the effects box...
I never had such bad latency before... After I turned off everything I was told to at a windows audio forum, it improved drastically but still latency.... Led me to believe it must be the soundcard...

[thorgal]

the RME HDSP PCI card can go down to 0.6 ms in s/w/ latency. Much less in h/w latency (monitoring bypassing the whole PC).

[funkmuscle]

the RME HDSP PCI card can go down to 0.6 ms in s/w/ latency. Much less in h/w latency (monitoring bypassing the whole PC).

like I said... also windows needed asio for low latency and to be like jack, it needed unwire or something like that...
I never done audio on windows except using kristin or something like that.. it worked for the only 2 tracks allowed with no latency issues.

[spm_gl]

No doubt, setting up a Windows system for audio is no trivial task. Took me about a week, and now I'm scared to touch it, XP breaks so damn easily. I personally prefer configuring Linux systems, and actually find it easier. There is far more information to be found too.

[schivmeister]

Give and take. They're both on par.

Now, I've rarely seen any studio box (Windows/Mac) "break". When it comes to a professional setting, you usually deploy and forget. Most of them do not have access to the internet, and a lot are running Windows XP SP2 for what, maybe more than half a decade?

[Capoeira]

before divorcing from WIN-XP I used to make plastic-surgory on it, and I got a pretty fast and light-weight XP: http://www.nliteos.com/nlite.html

[brummer]

This thread makes me remind a old thread I once found in the debian forum, . . .

Microsoft has now released the source code of Windows.

Code: Select all

/* Copyright Microsoft Inc., Redmond, USA.
*/

#include "win311.h"
#include "win95.h"
#include "evenmore.h"
#include "oldstuff.h"
#include "billrulz.h"
#include "implants.h"
#include "monopoly.h"

char make_prog_look_big[1600000];

void main(){

display_copyright_message();
display_bill_rules_message();
do_nothing_loop();

if (first_time_installation) {

make_500_megabyte_swapfile();
do_nothing_loop();
totally_screw_up_HPFS_file_system();
search_and_destroy_networked_UNIX();
search_and_destroy_networked_LINUX();
install_XP_on_all_machines_found();
disable_Netscape();
disable_RealPlayer();
disable_Corel_Products();
hang_system();
}

write_something(anything);
display_copyright_message();
display_buymoreXboxshit_message();
do_nothing_loop();
do_some_stuff();

if (still_not_crashed) {

display_copyright_message();
do_nothing_loop();
basically_run_windows_3.1_fool_as_XP();
do_nothing_loop();
do_nothing_loop();
}

if (detect_cache()) disable_cache();

if (fast_cpu()) {

set_wait_states(lots);
set_mouse(speed, very_slow);
set_mouse(action, jumpy);
set_mouse(reaction, sometimes);
set_intelli_mouse(NO_INTELLI);
}

/* printf("Welcome to Windows 3.11"); */
/* printf("Welcome to Windows 95"); */
printf("Welcome to Windows XP");

if (system_ok())
crash(to_dos_prompt);
else
system_memory = open("a:\swp0001.swp", O_CREATE, 0666);

while(SATAN) {

sleep(5);
get_user_input();
sleep(5);
act_on_user_input();
sleep(15)
illegal_operation_performed();
}

create_general_protection_fault();
printf("Smash head on Keyboard\n");
make_user_reboot_machine();
}