2.6.28 kernel and above

[Peter_Pan123]

Hello,

do somebody know why is the kernel 2.6.28 and above not usefull? i tried a lot of configurations and everytime i got with hl1 and hl2 problems. At Hl1 i got with pb3/tick2000 > feeling like speedhack on the server. At Hl2 i got only 920 FPS with lags. What do i wrong?

best reguards

[DiSTANT]

Peter_Pan123 Wrote:Hello,

do somebody know why is the kernel 2.6.28 and above not usefull? i tried a lot of configurations and everytime i got with hl1 and hl2 problems. At Hl1 i got with pb3/tick2000 > feeling like speedhack on the server. At Hl2 i got only 920 FPS with lags. What do i wrong?

best reguards

You should post your configs so we can look.

[Peter_Pan123]

http://www.nopaste.com/p/aekUlF4

hope you could help me :-(

[janne]

I've tried alot of kernels above that and I havent had any problems with HL2.
Currently I'm on 2.6.29-rc7.

[Peter_Pan123]

mh how many fps do you have? is your system a quad core too?
could you give me your config to try it on my system?

[BehaartesEtwas]

pb3 and ticrates>2000 oftenly increase the game speed, just dont use that combination.

i had no big problems but "only" an increased irq load, so i switched back to 2.6.26 (with RT patches)

[Peter_Pan123]

ok but i dont know which kernel i should use
i want to know why is that a problem to use a kernel above 2.6.28? the feeling ist terrible.-..

[pengy]

For hl1 use pingboost 3 and sys_ticrate 1001 works for me and i got http://www.legion-gaming.com/fps.png

[Amo85]

pengy

But whats your cpu usage?

[pengy]

Only 4% max on a 3.2 Ghz Quad core

[BehaartesEtwas]

cpu usage is typically not measured correctly (I know 3-4 ways to measure it, all completely different results in the same moment)

[pengy]

Stats in console, top in ssh, VNC and view resource monitor... thats all i know how, all of them show way different things... i think the resouce monitor is the most accurate though

[BehaartesEtwas]

"vmstat 1" gives you some useful information. only starting with the the second line it seems to be accurate. but I wouldn't look too much at the cpu usage. more interesting are the interrupt rates ("in") and the context switch rates ("cs"), which might be very different among different kernels. if you change your kernel and still achieve the same fps but get lower "cs" and "in", it will be an improvement (less cpu load).

[Peter_Pan123]

which counts are ok ?

[BehaartesEtwas]

That's what I get on my server (note, the first line differs much from the rest, probably bullshit). I don't know it those values are good, but it works for me...

Code:

procs -----------memory---------- ---swap-- -----io---- -system-- ----cpu----
r  b   swpd   free   buff  cache   si   so    bi    bo   in   cs us sy id wa
0  0 312092 127380   3780 375720    1    1   445   223    2    3 21  5 70  4
0  0 312092 127372   3780 375724    0    0     0     0   41 26065  5  8 88  0
0  0 312092 124516   3780 375728    0    0     0     0   85 26255  4  4 92  0
0  1 312092 104848   3780 384908    0    0  9264  1475  154 26431 14  6 56 25
0  0 312092 104432   3788 385120    0    0   176    76   93 26509 10  2 79 11
0  0 312092 114792   3788 385184    0    0     0    27   39 26176  1  1 99  0
0  0 312092 114748   3788 385184    0    0     0    22   41 26139  3  1 97  0
0  0 312092 114864   3788 385184    0    0     0     0   54 26228  0  1 100  0
0  0 312092 114864   3788 385184    0    0     0     0   22 26064  2  1 98  0
0  0 312092 114864   3788 385184    0    0     0   125   48 26167  2  2 97  0
0  0 312092 114864   3788 385188    0    0     0     8   35 26192  1  0 100  0
0  0 312092 115140   3788 385188    0    0     0     0   57 26145  3  2 96  0