Linux Vs NT performances

Linux Vs NT performances

Post by jed » Sun, 31 Dec 1899 09:00:00



[deletia]

Quote:>>I apppreciate if you reply by also by email. Will be easier to me.
>>Thanks.

>>Gabriele

>Linux itself is compact and fast.  More so than NT.
>Linux/X is less compact and slower than NT.
>X places redirection and stability above speed.
>Graphic drivers are not as good I suspect.

>Let's all be real.
>X really makes Linux proper look bad.
>Sure Linux runs on old machines.  In text mode.
>Most people are bummed out when this is  realized.
>Then when you add a very bloated window manager/environment that matchs SOME
>NT functionality, now your system is slower and bulkier than NT.  MS may

        Then don't run a very bloated window manager. Such a window
        manager is not needed to match the functionality of NT. It      
        isn't that great of a GUI to begin with. Others have replicated
        it before on much more austere hardware than what runs either
        Unix or any version of Windows well.

        Even with poor hardware acceleration, Unix process management
        still trumps NT's. That can make up for quite a bit.

Quote:>have moved GDI into kernel space, but it would have been faster and more
>compact anyway.  X needs a very major rewrite to compact, speed up like 30%,
>or it needs to be replaced.  The saving grace is that Linux/X/KDE/Gnome is
>free.  Linux as a desktop needs a kick in the butt.

>We need X to be as good as Linux proper is.
>Jim

--

  "I was not elected to watch my people suffer and die     |||
   while you discuss this a invasion in committe."        / | \

        In search of sane PPP docs? Try http://penguin.lvcm.com

 
 
 

Linux Vs NT performances

Post by G.S » Sun, 31 Dec 1899 09:00:00


I'm writing this for pure understanding and is not my intention to
have any flame here. My point is about Linux performances. I recently
built a new system totally dedicated to Linux where I'm learning, i
want to say first that I've been in PC as a professional for 20 years.

Well while I'm totally in love with Linux and I like both the product
and the ideology behind it, I don't see all those promised
performances out of it. My other machine is a Win NT machine, a little
better but not too much, I actually feel that is much  faster.

I don't have too many hard data, is more personal feeling but the way
the Linux machine is "answering" to my commands seem slow. (I work
only in X11)

Using Netscape on both machine and having a cable modem I can see how
the machine react to me much better in NT. Am I'm wrong? I'm happy if
soneone can tell it to me.

My Linux machine has the following configuration

K6-2 350 MHZ
MB Tyan 1MB cache
32 MB SDRAM
SCSI 2940 controller
Quantum HD 7200 RPM
S3 968 Graphic Card
floppy, CD
Red Hat 5.2 all by default with the default X enviroment

It is the RAM? may be so, but I heard so many nice things about LINUX
with just a little RAM .

I apppreciate if you reply by also by email. Will be easier to me.
Thanks.

Gabriele

 
 
 

Linux Vs NT performances

Post by Jim Ros » Sun, 31 Dec 1899 09:00:00


-----Original Message-----

Newsgroups: comp.os.linux.advocacy
Date: Monday, March 15, 1999 7:39 PM
Subject: Linux Vs NT performances

>I'm writing this for pure understanding and is not my intention to
>have any flame here. My point is about Linux performances. I recently
>built a new system totally dedicated to Linux where I'm learning, i
>want to say first that I've been in PC as a professional for 20 years.

>Well while I'm totally in love with Linux and I like both the product
>and the ideology behind it, I don't see all those promised
>performances out of it. My other machine is a Win NT machine, a little
>better but not too much, I actually feel that is much  faster.

>I don't have too many hard data, is more personal feeling but the way
>the Linux machine is "answering" to my commands seem slow. (I work
>only in X11)

>Using Netscape on both machine and having a cable modem I can see how
>the machine react to me much better in NT. Am I'm wrong? I'm happy if
>soneone can tell it to me.

>My Linux machine has the following configuration

>K6-2 350 MHZ
>MB Tyan 1MB cache
>32 MB SDRAM
>SCSI 2940 controller
>Quantum HD 7200 RPM
>S3 968 Graphic Card
>floppy, CD
>Red Hat 5.2 all by default with the default X enviroment

>It is the RAM? may be so, but I heard so many nice things about LINUX
>with just a little RAM .

>I apppreciate if you reply by also by email. Will be easier to me.
>Thanks.

>Gabriele

Linux itself is compact and fast.  More so than NT.
Linux/X is less compact and slower than NT.
X places redirection and stability above speed.
Graphic drivers are not as good I suspect.

Let's all be real.
X really makes Linux proper look bad.
Sure Linux runs on old machines.  In text mode.
Most people are bummed out when this is  realized.
Then when you add a very bloated window manager/environment that matchs SOME
NT functionality, now your system is slower and bulkier than NT.  MS may
have moved GDI into kernel space, but it would have been faster and more
compact anyway.  X needs a very major rewrite to compact, speed up like 30%,
or it needs to be replaced.  The saving grace is that Linux/X/KDE/Gnome is
free.  Linux as a desktop needs a kick in the butt.

We need X to be as good as Linux proper is.
Jim

 
 
 

Linux Vs NT performances

Post by Mark Robinso » Sun, 31 Dec 1899 09:00:00



> I'm writing this for pure understanding and is not my intention to
> have any flame here. My point is about Linux performances. I recently
> built a new system totally dedicated to Linux where I'm learning, i
> want to say first that I've been in PC as a professional for 20 years.

> Well while I'm totally in love with Linux and I like both the product
> and the ideology behind it, I don't see all those promised
> performances out of it. My other machine is a Win NT machine, a little
> better but not too much, I actually feel that is much  faster.

> I don't have too many hard data, is more personal feeling but the way
> the Linux machine is "answering" to my commands seem slow. (I work
> only in X11)

> Using Netscape on both machine and having a cable modem I can see how
> the machine react to me much better in NT. Am I'm wrong? I'm happy if
> soneone can tell it to me.

> My Linux machine has the following configuration

> K6-2 350 MHZ
> MB Tyan 1MB cache
> 32 MB SDRAM
> SCSI 2940 controller
> Quantum HD 7200 RPM
> S3 968 Graphic Card
> floppy, CD
> Red Hat 5.2 all by default with the default X enviroment

> It is the RAM? may be so, but I heard so many nice things about LINUX
> with just a little RAM .

> I apppreciate if you reply by also by email. Will be easier to me.
> Thanks.

> Gabriele

How do the machines comapre?  Are the parts name brand or generic?  Are
the drivers name brand or generic?  Has Linux been optimized?  Has NT
been optimized?
 
 
 

Linux Vs NT performances

Post by Anthony O » Sun, 31 Dec 1899 09:00:00


On Tue, 16 Mar 1999 00:39:13 GMT, gsart...@usa.net (G.S.)
wrote:

>I'm writing this for pure understanding and is not my intention to
>have any flame here. My point is about Linux performances. I recently
>built a new system totally dedicated to Linux where I'm learning, i
>want to say first that I've been in PC as a professional for 20 years.

>Well while I'm totally in love with Linux and I like both the product
>and the ideology behind it, I don't see all those promised
>performances out of it. My other machine is a Win NT machine, a little
>better but not too much, I actually feel that is much  faster.

>I don't have too many hard data, is more personal feeling but the way
>the Linux machine is "answering" to my commands seem slow. (I work
>only in X11)

This is (personal theory only) due to the fact that the
mouse lags in X, because it is a task which has to share CPU
like every other task. This makes feedback slow, and hence
the operating system feel slow. Benchmarks don't take this
into account.

As an aside, I ran winamp as a high priority task in Windows
NT 3.51. The mouse lagged. Is this a sign of a quality
windowing system? ;-))

>Using Netscape on both machine and having a cable modem I can see how
>the machine react to me much better in NT. Am I'm wrong? I'm happy if
>soneone can tell it to me.

My personal opinion is that NetscapeN is slightly faster on
Linux than Windows. But I wouldn't like to prove it.

>My Linux machine has the following configuration

>K6-2 350 MHZ

P133

>MB Tyan 1MB cache

Ga586HX - 512Kb cache.

>32 MB SDRAM

64 Mb of which 32 EDO (scrounged),  32 FPM.

>SCSI 2940 controller
>Quantum HD 7200 RPM

Hubba hubba.

Various IDE disks which I scrounged. Main one is quantum
bigfoot.

>S3 968 Graphic Card

I have an S3 ViRGE.

>floppy, CD
>Red Hat 5.2 all by default with the default X enviroment

Slackware something or other - butchered by now. Slackware
"Bride of Frankenstein" is probably the best description.

>It is the RAM? may be so, but I heard so many nice things about LINUX
>with just a little RAM .

Yes, but that's linux alone. Linux runs good on 8meg
excellently on 12.
However, we are talking XFree + NetscapeN here. Which are
hogs.

I ran X (Son of Satan - I don't normally use it) and
NetscapeN,
and did a ps on the running processes (see appendix).

As you can see, I run a lot of crud on this system (and I've
snipped a few lines out). However, if you take the X lines
and the netscape Navigator line

root     14995  2.0  6.4  8080  4048  ?  S    20:56   1:04
/usr/X11R6/bin/Xwrapp
root     14997  0.0  1.5  1764  1004   1 S    20:56   0:00
fvwm95
root     15000  0.0  1.0  1520   668   1 S    20:56   0:00
/usr/X11R6/lib/X11/fv
root     15002  0.0  1.0  1508   692   1 S    20:56   0:00
/usr/X11R6/lib/X11/fv
root     15004  0.0  1.7  2168  1128   1 R N  20:56   0:00
xload -nolabel -bg gr
root     15005  0.0  1.7  2128  1112   1 S    20:56   0:00
xclock -bg #c3c3c3 -g
root     15006  0.0  0.9  1472   600   1 S    20:56   0:00
/usr/X11R6/lib/X11/fv
root     15008  1.8 22.0 19464 13904   1 S    20:56   0:58
netscape

You are looking at over 20Mb for X & NN alone. Being
statically linked is why Netscape is so big compared to the
Windows version, but knowing the reason, doesn't change the
fact.

Checklist for better performance:

1/ Check you are using the acceleration feaures of your card
(XAA is the buzzacronym to look for).

2/ Install Squid as a proxy server on your local machine,
and configure NetscapeN to use it (127.0.0.1:8080 usually).
NetscapeN then runs like a startled cat. I think it is crap
at caching it's own documents.

3/ Buy more memory.

>I apppreciate if you reply by also by email. Will be easier to me.
>Thanks.

>Gabriele

Regards

Anthony

==================================================

Appendix

w:
  9:49pm  up 11:59,  4 users,  load average: 0.91, 0.54,
0.48

free:
             total       used       free     shared
buffers     cached
Mem:         63192      59572       3620      43836
1344      24476
-/+ buffers/cache:      33752      29440
Swap:         8028       1660       6368

ps aux:

USER       PID %CPU %MEM  SIZE   RSS TTY STAT START   TIME
COMMAND
anthony   1140  0.0  1.7  2768  1124  ?  S    10:24   0:08
smbd
bin         74  0.0  0.3   824   228  ?  S    09:50   0:00
/usr/sbin/rpc.portmap
bin         91  0.0  0.7  1096   460  ?  S    09:50   0:00
sh /storage/usr/lib/a
bin         95  0.0  0.4  1096   276  ?  S    09:50   0:00
sh /storage/usr/lib/a
bin         97  0.0  0.4   824   268  ?  S    09:50   0:00
cat -
daemon     396  0.0  0.3   820   224  ?  S    09:55   0:00
/usr/lib/atrun
news       136  0.0  1.7  1596  1088  ?  S    09:50   0:00
/usr/news/bin/innd -p
nobody     104  0.0  0.8  1096   548  ?  S    09:50   0:00
/usr/sbin/httpd
nobody     398  0.0  4.2  3696  2660  ?  S N  09:55   0:06
/usr/local/squid/bin/
nobody     399  0.0  0.5   852   332  ?  S N  09:55   0:00
(dnsserver)
nobody   15031  0.0  0.9  1088   592  ?  S    20:57   0:00
/usr/sbin/httpd
nobody   19442  0.4  1.1  1424   748  ?  S    21:49   0:00
perl /var/lib/httpd/c
nobody   19445  0.4  0.7  1108   504  ?  S    21:49   0:00
sh /usr/local/bin/thu
nobody   19513 27.7  3.2  2988  2056  ?  R    21:49   0:01
/usr/local/bin/jpgtn
nobody   19514  0.0  0.4   880   276  ?  S    21:49   0:00
/usr/local/bin/jpegtr
nobody   19530 30.3  2.6  4044  1672  ?  R    21:49   0:00
/usr/local/bin/jpgtn
nobody   19531  0.3  0.4   880   276  ?  S    21:49   0:00
/usr/local/bin/jpegtr
postgres   142  0.0  0.3  2996   228  ?  S    09:50   0:00
/usr/local/pgsql/bin/
root         1  0.0  0.5   828   356  ?  S    09:49   0:04
init [3]
root         2  0.0  0.0     0     0  ?  SW   09:49   0:00
(kflushd)
root         3  0.0  0.0     0     0  ?  SW   09:49   0:00
(kpiod)
root         4  0.0  0.0     0     0  ?  SW   09:49   0:00
(kswapd)
root        10  0.0  0.2   800   132  ?  S    09:49   0:00
/sbin/update
root        69  0.0  0.5   836   344  ?  S    09:50   0:00
/usr/sbin/syslogd
root        72  0.0  0.5   888   332  ?  S    09:50   0:00
/usr/sbin/klogd
root        76  0.0  0.4   824   272  ?  S    09:50   0:00
/usr/sbin/inetd
root        80  0.0  0.4   844   296  ?  S    09:50   0:00
/usr/sbin/lpd
root        82  0.0  0.5   888   376  ?  S    09:50   0:00
/usr/sbin/rpc.mountd
root        84  0.0  0.5   904   376  ?  S    09:50   0:00
/usr/sbin/rpc.nfsd
root        87  0.0  0.4   856   276  ?  S    09:50   0:00
/usr/sbin/lpd
root        92  0.0  0.4   824   304  ?  S    09:50   0:00
/usr/sbin/crond -l10
root       102  0.0  0.8  1056   516  ?  S    09:50   0:00
/usr/sbin/httpd
root       110  0.0  0.6   868   380  ?  S    09:50   0:00
/usr/local/sbin/nwser
root       111  0.0  1.4  2680   896  ?  S    09:50   0:00
nwbind LARGE c0.a8.2.
root       113  0.0  0.3   832   208  ?  S    09:50   0:00
gpm -R -m /dev/mouse
root       135  0.0  0.4   836   312  ?  S    09:50   0:00
ncpserv LARGE c0.a8.2
root       153  0.0  0.9  1156   576  ?  S    09:51   0:22
sh /usr/news/bin/innw
root       397  0.0  0.7  1096   500  ?  S N  09:55   0:00
sh
root       427  0.0  0.4   824   296   2 S    09:57   0:00
shuffle ABBA - Knowin
root       428  0.0  0.4   856   288   2 S <  09:57   0:00
xargs -x -l1 -n 1 amp
root      1139  0.0  1.0  1180   692  ?  S    10:24   0:00
nmbd
root     14993  0.0  0.7  1108   504   1 S    20:56   0:00
sh /usr/X11R6/bin/sta
root     14994  0.0  0.9  1872   624   1 S    20:56   0:00
xinit /usr/X11R6/lib/
root     14995  2.0  6.4  8080  4048  ?  S    20:56   1:04
/usr/X11R6/bin/Xwrapp
root     14997  0.0  1.5  1764  1004   1 S    20:56   0:00
fvwm95
root     15000  0.0  1.0  1520   668   1 S    20:56   0:00
/usr/X11R6/lib/X11/fv
root     15002  0.0  1.0  1508   692   1 S    20:56   0:00
/usr/X11R6/lib/X11/fv
root     15004  0.0  1.7  2168  1128   1 R N  20:56   0:00
xload -nolabel -bg gr
root     15005  0.0  1.7  2128  1112   1 S    20:56   0:00
xclock -bg #c3c3c3 -g
root     15006  0.0  0.9  1472   600   1 S    20:56   0:00
/usr/X11R6/lib/X11/fv
root     15008  1.8 22.0 19464 13904   1 S    20:56   0:58
netscape
root     15010  0.1  2.5  2524  1592   1 S    20:56   0:06
xterm -ls -sb -fn 7x1
root     15018  0.0  4.0 12568  2528   1 S    20:56   0:00
(dns helper)
root     17955  0.0  2.3  2412  1460   1 S    21:19   0:00
xterm -ls -sb -fn 7x1
root     19414  0.0  0.7  1104   492   2 S <  21:48   0:00
shellplay ./amplay Bu
root     19415 42.8  3.9  3076  2512   2 S <  21:48   0:08
mpg123 -b 2048 Buddy
root     19416  0.6  3.9  3076  2480   2 S <  21:48   0:00
mpg123 -b 2048 Buddy

--
-----------------------------------------
| And when our worlds                   |
| They fall apart                       |
| When the walls come tumbling in       |
| Though we may deserve it              |
| It will be worth it  - Depeche Mode   |
-----------------------------------------

 
 
 

Linux Vs NT performances

Post by Thad Phetteplac » Sun, 31 Dec 1899 09:00:00




>>I'm writing this for pure understanding and is not my intention to
>>have any flame here. My point is about Linux performances. I recently
>>built a new system totally dedicated to Linux where I'm learning, i
>>want to say first that I've been in PC as a professional for 20 years.

>>Well while I'm totally in love with Linux and I like both the product
>>and the ideology behind it, I don't see all those promised
>>performances out of it. My other machine is a Win NT machine, a little
>>better but not too much, I actually feel that is much  faster.

I've worked with Linux and NT on identical hardware.  Linux definetly
performs better when you load on the tasks.  Under a light load, NT did
have the edge sometimes, but that reversed in a big way as soon as
we cranked up the load.  NT seems to be optimized toward running a few
big processes at one time while Linux (and UNIX in general) is optimized
toward running a great many smaller processes.  This translates to
NT doing fine as a workstation OS but sucking rocks big-time when it
tries to act as a server.  Linux really shines as a server, and does well
as a workstation too.

Note that if you really push your workstation, that server performance edge
begins to shine through.  I have a Linux workstation that runs some server
tasks in the background, always has a copy of Netscape up, and is often
downloading files from the Internet while I compile a software project.
It stands up to this punishment quite well... and this is only a P100 / 32MB
system.  I've tried to do similar with an NT box with 64MB of RAM, and it
was noticeably slower.  Stick with it.  The more you use Linux the more you
will notice and appreciate its performance edge.

Thad

 
 
 

Linux Vs NT performances

Post by Thad Phetteplac » Sun, 31 Dec 1899 09:00:00



> You are looking at over 20Mb for X & NN alone. Being
> statically linked is why Netscape is so big compared to the
> Windows version, but knowing the reason, doesn't change the
> fact.

This is why I'm looking forward to the new NGLayout based browsers.  From
what I've seen, they should have a much smaller memory footprint and much
better performance.

BTW, has anyone seen a decent web browser for Linux that talks directly to
svgalib or GGI?  I'm attempting to set up Linux netsurfing stations on
extemely old hardware and want to avoid the memory penalty of loading X.

Thanks,

Thad

 
 
 

Linux Vs NT performances

Post by Tim Kelle » Sun, 31 Dec 1899 09:00:00





> >I'm writing this for pure understanding and is not my intention to
> >have any flame here. My point is about Linux performances. I recently
> >built a new system totally dedicated to Linux where I'm learning, i
> >want to say first that I've been in PC as a professional for 20 years.

> >Well while I'm totally in love with Linux and I like both the product
> >and the ideology behind it, I don't see all those promised
> >performances out of it. My other machine is a Win NT machine, a little
> >better but not too much, I actually feel that is much  faster.

> >I don't have too many hard data, is more personal feeling but the way
> >the Linux machine is "answering" to my commands seem slow. (I work
> >only in X11)

Well, you have to realize that usually when people are talking about
linux performing better than NT, they are referring to linux acting in a
server role, not a workstation.  As a workstation, linux in console mode
is of course much faster that NT in any case, and it is definitely
better at handling multiple tasks than NT.

Memory is the biggest performance factor in Linux.  Using X on less that
64MB you will start swapping if you use any large programs (like
Netscape and gimp at the same time, or Star Office by itself.) and this
will be slow.  Notice that Linux never swaps to disk unless it has used
up every last byte of RAM. Not so with NT, where the swap file is being
used constantly no matter what you have open.  This alone makes linux
faster.

 
 
 

Linux Vs NT performances

Post by Anthony O » Sun, 31 Dec 1899 09:00:00


On 18 Mar 1999 16:50:29 -0600, Thad Phetteplace



>> You are looking at over 20Mb for X & NN alone. Being
>> statically linked is why Netscape is so big compared to the
>> Windows version, but knowing the reason, doesn't change the
>> fact.

>This is why I'm looking forward to the new NGLayout based browsers.  From
>what I've seen, they should have a much smaller memory footprint and much
>better performance.

>BTW, has anyone seen a decent web browser for Linux that talks directly to
>svgalib or GGI?  

There is a DOS (really!) web browser called Arachne which is
supposedly going to be ported to SVGALib. I have not heard
anything further on it though.

Quote:>I'm attempting to set up Linux netsurfing stations on
>extemely old hardware and want to avoid the memory penalty of loading X.

Wise move.

Quote:>Thanks,

>Thad

Regards

Anthony
--
-----------------------------------------
| And when our worlds                   |
| They fall apart                       |
| When the walls come tumbling in       |
| Though we may deserve it              |
| It will be worth it  - Depeche Mode   |
-----------------------------------------

 
 
 

Linux Vs NT performances

Post by Todd Bandrowsk » Sun, 31 Dec 1899 09:00:00


Quote:> As a workstation, linux in console mode
>is of course much faster that NT in any case, and it is definitely
>better at handling multiple tasks than NT.

It really depends on what you are doing with it.  If your intent is to serve
web pages, then yes.

Quote:

>Memory is the biggest performance factor in Linux.  Using X on less that
>64MB you will start swapping if you use any large programs (like
>Netscape and gimp at the same time, or Star Office by itself.) and this
>will be slow.

If you are going to run X on Linux, do make sure that you have a graphics
card that is supported with hardware acceleration.
Quote:>Notice that Linux never swaps to disk unless it has used
>up every last byte of RAM. Not so with NT, where the swap file is being
>used constantly no matter what you have open.  This alone makes linux
>faster.

 
 
 

1. Linux vs. NT performance / Mindcraft results

Bruce Weiner wrote in his rebuttal:
 "Setting the Record Straight: Where ABCnews.com Got It Right and Wrong":

            "2.Mindcraft used a server with a MegaRAID controller
               with a beta driver (which was the latest version
               available at the time of the test) while the PC Week
               server used an eXtremeRAID controller with a fully
               released driver. The MegaRAID driver was single
               threaded while the eXtremeRAID driver was
               multi-threaded."

That means that these Mindcraft guys were aware that they have tested
the LINUX performance with a BETA DRIVER for the MegaRAID controller and
they were aware that this beta driver was just a SINGLE-THREADED driver!!

It's evident that the multi-threaded NT driver for the MegaRAID controller
has a much better performance under havy load as a single-threaded
driver ... so their so called test results are very dubiuos and rigged.

As Lincoln said:"One can cheat the poeple .. but only for a limited time"

Armin

2. Router Setup

3. Linux vs OS2 vs NT vs Win95 vs Multics vs PDP11 vs BSD geeks

4. Specifics on kernel updates? (where)

5. performance: linux/samba vs nt vs novell

6. 4.4 Network Setup?

7. performance of linux vs nt

8. processes

9. JDK1.2.2 performance, Linux -vs- NT

10. Performance comparison of nt vs linux

11. FPU performance of Pentium vs. K6 vs. Ppro vs. PII

12. nfs performance, linux vs freebsd vs solaris

13. NT NT NT NT NT NT NT MT