It's time to get serious about gaming... presenting : GameOS A ...

ruralrompΛογισμικό & κατασκευή λογ/κού

2 Δεκ 2013 (πριν από 3 χρόνια και 9 μήνες)

95 εμφανίσεις


It's time to get serious about gaming... presenting :


Game
OS


A custom built OS specifically for hard
-
core gamers and LAN gaming events

Purpose
built on high performance Intel

& Gentoo Linux technology using the ICC complier


############################
###############################


Vista 32bit Ultimate specs:


nVidia graphics driver: 7.15.11.7824
-

date: 10/7/2008
-

(Digitally signed:
Microsoft

windows hardware
compatibility publisher) downloaded directly from nVidia site and installed on fresh system
.


Vista sp1 applied with latest patches


No virus software installed


spy sweeper on gaming mode


Killer Nic in gaming mode
-

driver 4.1.0.1 (Digitally signed: bigfoot networks, inc.)
-

date: 9/26/2008


Creative Sound Blaster X
-
Fi Fatality sound card
-

dr
iver: 6.0.1.1361 (Digitally signed: creative labs, inc.)
-

date:
7/15/2008


Chipset: Intel 875P Chipset
-

driver:
Microsoft

6.0.6001.18000 (Digitally signed:
Microsoft

windows)
-

date
6/21/2006


Boot drive
-

x2 WD Raptor X 10k rpm HDD Raid 0 hardware array

-

Adaptec

integrated Sata HostRAID
-

driver: 4.0.36.2 (Not digitally signed)
-

date: 4/10/2005


CPU: Intel P4 3.4Ghz Extreme Edition (Northwood core)
-

800Mhz FSB
-

2MB L3 cache
-

HT enabled in BIOS
and Vista


Pagefile drive
-

x1 WD 500Gig ATA/133 Ultra D
MA mode 5


Filesystem: NTFS filesystems for all drives
-

default cluster and RAID 0 stripping settings


MSINFO32 file: http://www.tyler.mcadams.com/msinfo


Dxdiag file: http://www.tyler.mcadams.com/DxDiag


Services file: http://www.tyler.mcadams.com/servic
es


Services file: http://www.tyler.mcadams.com/AutoRuns.arn


Services configuration: default +/
-
: TCP/IP NetBIOS (disabled), Server (disabled), Terminal Services
(disabled), nVidia nTune service (enabled), Norton Ghost service (enabled)


Msconfig: (No sta
rtup executables disabled)
-

Windows defender, nVidia Compatible Windows display driver,
Version 178.24, nVidia Media Center Library, CTXfiHlp Application, Logitech Setpoint, Norton ghost,
Spysweeper, Micro
soft Windows Operating system, A
shampoo Unistaller

Watcher, Spybot
-

Search & Destroy,
nVidia nTune, Killer NIC, Logitech Setpoint(2)


Super Micro P4SCT Mobo BIOS version: R 1.2b
-

4SCT6126.zip
-

Release Date: 6/14/06
-

http://www.supermicro.com/support/resources/results.cfm


Motherboard: Super Micro P4SC
T mobo:
http://www.supermicro.com/products/motherboard/P4/875/P4SCT.cfm


Memory configuration: 4gigs Dual channel x4 1 gig DDR PC3200 Dimms per channel (default bios ram
timings
)


Bios configuration: Intel e1000 nic card disabled, Serial and parallel ports

disabled, factory ram timings, HT
enabled


System sanity checks: Ram test, Spinrite 6 disk check, SMART check, Vista checkdisk, vista sfc /scannow,
Vista fresh install backup, Norton ghost image backup, system restore ON and rstore point made before tests

-

all tests good
-

no system or NTFS filesystem errors


ACPI OS settings: Performance


Cooling: air cooled system ("typical" configuration)


Monitor: SGI F190 18" Flatscreen monitor
-

1280 x 1024 pixels @ 60Hz 32bit
-
color (System dual monitor
capable
-

co
nnected via VGA
-

system powered down, second monitor removed and powered back up before
tests
-

keeps video card from
splitting

up video ram and other
resources
)
http://techpubs.sgi.com/library/tpl/cgi
-
bin/summary.cgi?coll=hdwr&db=bks&docnumber=860
-
0401
-
0
01


nTune settings: No system tuning first round benchmarks
-

no GPU or video memory overclocking first
round testing


###########################################################


Gentoo Linux engine specs:


Install method:
Minimal 2008 install ISO
-

http
download install
-

stage 3 package


Customizations:

stage 3 base system
recompiled

with aggressive make flags and various USE flags for
desktop support: http://www.tyler.mcadams.com/make.conf


Kernel version:

2.6.25
-
gentto
-
r7 (gentoo
-
sources) compiled wit
h GCC version 4.1.2 (first round tests) config
file: http://www.tyler.mcadams.com/config.txt


X11
-

version 7.4
-

compiled with same use flags as base system and new GCC version: 4.3.2
-

xorg config file:
http://www.tyler.mcadams.com/xorg.conf


GUI:

failsa
fe mode used for launching games without a gui manager overhead (base system gui KDE 3.5.9
complied with GCC 4.3.2)


Filesystems:

ext3 (default format settings) /boot partition (at /dev/hda1), XFS root filesystem (
formatted

with: mkfs.xfs
-
l internal,size=
128m /dev/hda2/), 512MB swap partition


Intel compiler cflags:
-
O3
-
xW
-
ipo
-
gcc


Base system packages recompiled with ICC:

dev
-
lang/python, dev
-
db/sqlite, dev
-
libs/libxml2, app
-
arch/tar, app
-
arch/bzip2, app
-
arch/gzip, app
-
arch/unzip, app
-
arch/cpio, sys
-
po
wer/powertop, media
-
libs/x264
-
svn, media
-
sound/lame, media
-
libs/faac,
media
-
libs/giflib, media
-
libs/speex, media
-
libs/libtheora, media
-
libs/xvid, media
-
libs/libogg, media
-
libs/libvorbis, dev
-
libs/lzo, net
-
irc/xchat, media
-
libs/freetype, dev
-
libs/glib, x11
-
libs/gtk+, media
-
libs/libpng,
dev
-
libs/libgpg
-
error, media
-
gfx/gimp


Default services running:

alsasound, auditd, bootmisc, checkfs, checkroot, clamd, clock, consolefont, cpufrequtils, dbus, hald,
hostname, iptables, keymaps, local, localmount, modules, n
et.eth2, net.lo, netmount, ntp
-
client, rmnologin,
spindown, syslog
-
ng, urandom, vixie
-
cron, xdm


System sanity checks:

Ram test, Spinrite 6 disk check, SMART check, xfs_check, xfsdump root filesystem
backup before tests, revdep
-
rebuild


nVidia driver versi
on:

(module) 178.24 32bit
-

date: 11/12/2008


Killer Nic driver version:

(module) version 1.2
-

date: 08/01/2008


Creative Sound Blaster X
-
Fi Fatality sound card driver version:

(module) XFiDrv_Linux_Public_US_1.00
-

date: 11/06/2008


Boot drive
: Hitachi 7
200rpm ATA/133
-

8mb cache
-

UDMA mode 5


Pagefile

drive: same drive


Virus software:

ClamAV



###########################################################
###########


Overview:


As you might notice off the bat, these hardware and software specs show that W
indows has a definite
advantage
-

System throughput is much faster with a hardware Raid 0 array, bigger cache size, and a
separate drive and micro controller for the page file. Also Windows has the advantage of being the mainstay
gaming platform for years
-

the drivers for the video card, sound card and nic card are all much more
mature
-

the Linux system is using beta drivers for both the sound card and the nic card. (The nVidia Linux
driver has made massive gains in the last few years but still lacks in s
ome areas that Windows does not.)


With all these inherent advantages you would think that Windows would make the ultimate gaming platform,
but as our results show, you will notice that Linux actually has the advantage
-

and the advantage comes
from it's
openness. We can pop the hood and tweak certain things that we can not in Windows (Like
recompiling the kernel)
these

tweaks give us the ability to take Linux and build it from the ground up for one
purpose
-

to make the ultimate gaming system.


Next
-

Wha
t are the advantages?

What are the advantages?


1.)

Customization
: Since we can choose to strip away parts of the system's infrastructure we do not
need and / or disable parts that are "in the way" we can produce a system that is purpose configured
for gaming
. For instance
-

we can strip away the entire gui and start a game from the command line.
This technique alone yields close to a 200fps performance gain on a system that is still running the
gui underneath it.


2.)

Increased performance on the same hardware:
L
inux has an advantage in the way it can utilize the
hardware it runs on
-

when Windows is installed on a system it is installed in a very generic manner.
This means that the software itself is not taking full advantage of what the hardware can give it.
Sin
ce we have access to every single piece of code that makes up our gaming system's infrastructure
we can customize that code for our specific purpose with our USE flags and then optimize our code
for the actual hardware with cflags.


3.)

Cost:

-

Fast systems ev
en with older hardware: It is well known that Linux performs better than
Windows on older hardware (as these test results will indicate) That means you can use an older
system with a tweaked Linux OS where otherwise you would have to buy a completely new W
indows
system to play the latest games. That means you'll have to fork over more money
-

where as with
Linux it's free
-

doesn't sound fair does it?


4.)

Stability:

With Windows there is always something running in the background zapping away
recourses from yo
ur memory, cpu and gpu. With a properly tuned Linux system we can stop anything
that may be running in the background (or have nothing in the background running to start with!)
Not only do these hidden processes take up cpu time, they can cause stability i
ssues when you are
trying to run your games
-

especially so for overclockers.


5.)

High performance through the use of custom compilation:

Since once again we have the raw code
like clay in our hands we have the ability to further optimize our code with the In
tel Linux C++
compiler. This code optimization can yield around 30% increases in performance when compared to
the default GCC complier and even more gains when compared to a generic code base. Not only can
we compile the base seem with this method but the
actual Linux kernel itself! This is the biggest
advantage that Linux has when compared to Windows.


6.)

Security:

Tired of being hacked by cheaters? Linux offers security where Windows can't. Windows is
infamous for it's security holes.. Why would you want to
game on it and loose because you were
cheated? Linux on the other hand offers a secure infrastructure developed by the NSA


SE Linux.
Never get another virus, never get cheated again.












Next
-

The test cases:



The test cases:


We will start out
with generic systems that have default characteristics, test them and then we will move to
more and more aggressive configurations of both Windows and Linux in an attempt to squeeze out as much
performance as we can from each system.


The benchmark games:
Quake 3 and Doom 3


Why Quake 3?


1.)

It's a benchmark standard
-

used for years to test gaming and graphics performance
-

plenty of P4 /
Q3 benchmarks online to validate how empirical the results are


2.)

It's OpenGL
-

we are comparing apples to apples


3.)

We are te
sting with an Intel P4 arch
-

Quake 3 is known to perform best on the P4 ar
ch when
compared to similar
chips
-

this particular P4 we are testing with was

the flagship P4 EE 3.4ghz
chip w/ 2MB L3 cache, so it should br
ing out the be
st in both OSes. The
Inte
l

core in current chips
have very similar cores but have been further optimized so these

test results

will one scale greater

but predictably

in newer chips.

Further, since the

code we are modifying is in the kernel
,

and thus
executed in the cpu
,

this
bench
mark

gives a great measure of how optimized code can increase
system performance levels
.


4.)

We are testing with nVidia graphics
-

nVidia is known to be the leader in OpenGL performance
-

we
are testing with the fastest nVidia APG x8 hardware available: Gefor
ce 7 series 7950 GT w/ 512MB
RAM
.


5.)

We are tes
t
ing with old school hardware
-

Supermicro P4SCT mobo
-

with slower old school
hardware differences in performance will be easier to see and measure, plus tweaks to the system
will yield more illu
strative perfor
mance gains (since the system is single core in both the cpu and
gpu performance scaling can be more easily measured)


6.)

It’s

an actual game and not a synthetic benchmark


that means real world
results
.


Why Doom 3?


1.)

See: Why Quake 3