Re: performance of cygwin32

Re: performance of cygwin32

[Earnie Boyd]

The slowness comes from the cygwin.dll emulation of unix.  If you 
install Sergey's coolview update you will notice an increase in speed;  
but, it still won't match solaris.  Win32 appears to be slow anyway, 
especially Windows 95.

-        \\||//
---o0O0--Earnie--0O0o----
-earnie_boyd@hotmail.com-
------ooo0O--O0ooo-------

Check out these great sites:
ftp://ftp.cygnus.com/pub/gnu-win32/latest/           (ftp site)
http://www.cygnus.com/pubs/gnupro/                   (Comercial Page)
http://www.cygnus.com/misc/gnu-win32/                (Project Page)
http://www.cygnus.com/ml/gnu-win32                   (Mail Archives)
http://www.itribe.net/virtunix/winhelp-man-pages/    (HTML Man Pages)
http://www.lexa.ru/sos                               (Sergey Okhapkin)
http://www.fu.is.saga-u.ac.jp/~colin/gcc.html        (Colin Peters)
http://www.xraylith.wisc.edu/~khan/software/gnu-win32/ (Mumit Khan)

>From: epeschko@den-mdev1.tci.com (Ed Peschko)
>Subject: performance of cygwin32
>To: gnu-win32@cygnus.com
>Date: Thu, 1 Jan 1998 17:06:06 -0700 (MST)
>
>hey - 
>
>I've found after installing cdk.exe taht the performance of cygwin32 is 
>about 5-10 times slower than my solaris box. 
>
>Is this due to the underlying win32 api, or is it due to cygwin32, or 
is it 
>just me not doing something correctly?
>
>Thanks much,
>
>Ed
>-
>For help on using this list (especially unsubscribing), send a message 
to
>"gnu-win32-request@cygnus.com" with one line of text: "help".
>


______________________________________________________
Get Your Private, Free Email at http://www.hotmail.com
-
For help on using this list (especially unsubscribing), send a message to
"gnu-win32-request@cygnus.com" with one line of text: "help".

Re: performance of cygwin32

[Ed Peschko]

> 
> The slowness comes from the cygwin.dll emulation of unix.  If you 
> install Sergey's coolview update you will notice an increase in speed;  
> but, it still won't match solaris.  Win32 appears to be slow anyway, 
> especially Windows 95.

Win32 is slow, but network drives are *glacial*, at least with cygwin32 tools.
I've installed the coolview update (when will it be incorporated into the 
'vanilla' cygwin32 tools anyways?) and it is going a bit faster.

Just curious on why it would be so slow over a network drive... I don't seem 
to remember having the same problem over NFS mounts..

Ed
-
For help on using this list (especially unsubscribing), send a message to
"gnu-win32-request@cygnus.com" with one line of text: "help".

Re: performance of cygwin32

[Geoffrey Noer]

Ed Peschko wrote:
> 
> Win32 is slow, but network drives are *glacial*, at least with cygwin32
> tools. I've installed the coolview update (when will it be incorporated
> into the 'vanilla' cygwin32 tools anyways?) and it is going a bit faster.

The "coolview" sources represent a state of the actual Cygnus development
tree as of whenever that was made.  It's made more progress since then;
right now I'm battling race conditions in the signal code...

Right now I can do a complete configure of the compiler tools (including
tcl/tk) in about 20 minutes and build them in an hour 15 min. which is a 
huge improvement performance-wise.  But signal handling isn't quite as
robust as it needs to be...

-- 
Geoffrey Noer
noer@cygnus.com
-
For help on using this list (especially unsubscribing), send a message to
"gnu-win32-request@cygnus.com" with one line of text: "help".

Re: performance of cygwin32

[Ed Peschko]

> 
> Ed Peschko wrote:
> > 
> > Win32 is slow, but network drives are *glacial*, at least with cygwin32
> > tools. I've installed the coolview update (when will it be incorporated
> > into the 'vanilla' cygwin32 tools anyways?) and it is going a bit faster.
> 
> The "coolview" sources represent a state of the actual Cygnus development
> tree as of whenever that was made.  It's made more progress since then;
> right now I'm battling race conditions in the signal code...
> 
> Right now I can do a complete configure of the compiler tools (including
> tcl/tk) in about 20 minutes and build them in an hour 15 min. which is a 
> huge improvement performance-wise.  But signal handling isn't quite as
> robust as it needs to be...

Well, something must be seriously wrong with my configuration, because I was
seeing performance times of about 6-7 hours to do the same. 

1 hour 15 mins is what I call slightly slow. 6-7 hours is what I call glacial.
I'd hit 'ls' and it would take about a minute to respond....

I could live with an hour to build all the tools, believe me. Its just the 
going to dinner, then taking a nap, then playing two games of pool, and finally
coming back to the keyboard to see that the tools are *still* compiling 
which is getting annoying.

Ed
-
For help on using this list (especially unsubscribing), send a message to
"gnu-win32-request@cygnus.com" with one line of text: "help".

Re: performance of cygwin32

[Geoffrey Noer]

Ed Peschko wrote:
> 
> Well, something must be seriously wrong with my configuration, because I was
> seeing performance times of about 6-7 hours to do the same. 
> 
> 1 hour 15 mins is what I call slightly slow. 6-7 hours is what I call
> glacial.

Beta 18 was much slower.  The latest stuff is much better, performance-wise
although there are still improvements to be made.  The other thing is that
my compile includes packages that were not present in beta 18 (such as the
full distributions of tcl, tk, itcl, and tix).

I should be in a better position to give a timeframe for beta 19 as soon as
the configure process is more stable again.  (As I mentioned earlier, we're
in the process of getting signal handling working better and there are
still bugs to overcome).

-- 
Geoffrey Noer
noer@cygnus.com
-
For help on using this list (especially unsubscribing), send a message to
"gnu-win32-request@cygnus.com" with one line of text: "help".

Re: performance of cygwin32

[Earnie Boyd]

Is it possible that we are comparing w95 to wNT??  I would expect there 
to be a "glacial" difference between these two OS.

-        \\||//
---o0O0--Earnie--0O0o----
-earnie_boyd@hotmail.com-
------ooo0O--O0ooo-------

>From: epeschko@den-mdev1.tci.com (Ed Peschko)
>Subject: Re: performance of cygwin32
>To: noer@cygnus.com (Geoffrey Noer)
>Date: Mon, 12 Jan 1998 00:59:17 -0700 (MST)
>Cc: gnu-win32@cygnus.com
>Reply-To: ed_peschko@csgsystems.com (Ed Peschko)
>
>> 
>> Ed Peschko wrote:
>> > 
>> > Win32 is slow, but network drives are *glacial*, at least with 
cygwin32
>> > tools. I've installed the coolview update (when will it be 
incorporated
>> > into the 'vanilla' cygwin32 tools anyways?) and it is going a bit 
faster.
>> 
>> The "coolview" sources represent a state of the actual Cygnus 
development
>> tree as of whenever that was made.  It's made more progress since 
then;
>> right now I'm battling race conditions in the signal code...
>> 
>> Right now I can do a complete configure of the compiler tools 
(including
>> tcl/tk) in about 20 minutes and build them in an hour 15 min. which 
is a 
>> huge improvement performance-wise.  But signal handling isn't quite 
as
>> robust as it needs to be...
>
>Well, something must be seriously wrong with my configuration, because 
I was
>seeing performance times of about 6-7 hours to do the same. 
>
>1 hour 15 mins is what I call slightly slow. 6-7 hours is what I call 
glacial.
>I'd hit 'ls' and it would take about a minute to respond....
>
>I could live with an hour to build all the tools, believe me. Its just 
the 
>going to dinner, then taking a nap, then playing two games of pool, and 
finally
>coming back to the keyboard to see that the tools are *still* compiling 
>which is getting annoying.
>
>Ed
>-
>For help on using this list (especially unsubscribing), send a message 
to
>"gnu-win32-request@cygnus.com" with one line of text: "help".
>


Check out these great gnu-win32 related sites:
ftp://ftp.cygnus.com/pub/gnu-win32/latest/           (ftp site)
http://www.cygnus.com/pubs/gnupro/                   (Comercial Page)
http://www.cygnus.com/misc/gnu-win32/                (Project Page)
http://www.cygnus.com/ml/gnu-win32                   (Mail Archives)
http://www.itribe.net/virtunix/winhelp-man-pages/    (HTML Man Pages)
http://www.lexa.ru/sos                               (Sergey Okhapkin)
ftp://www.lexa.ru/pub/domestic/sos/                (Sergey's ftp site)
http://www.fu.is.saga-u.ac.jp/~colin/gcc.html        (Colin Peters)
http://www.xraylith.wisc.edu/~khan/software/gnu-win32/ (Mumit Khan)
http://gnu-win32.paranoia.ru                         (Chuck Bogorad)
ftp://ftp.deninc.com/pub                       (Den Internet Services)


______________________________________________________
Get Your Private, Free Email at http://www.hotmail.com
-
For help on using this list (especially unsubscribing), send a message to
"gnu-win32-request@cygnus.com" with one line of text: "help".

Re: performance of cygwin32

[Wei Ku]

I do notice a boost in performance when I upgrade the main memory from 32 MB
to 32 + 64 = 96 MB. Maybe this can explain the different result between your
machines.

Sincerely,
Wei Ku

-----Original Message-----
From: Ed Peschko <epeschko@den-mdev1.tci.com>
To: Geoffrey Noer <noer@cygnus.com>
Cc: gnu-win32@cygnus.com <gnu-win32@cygnus.com>
Date: Monday, January 12, 1998 7:35 AM
Subject: Re: performance of cygwin32


>>
>> Ed Peschko wrote:
>> >
>> > Win32 is slow, but network drives are *glacial*, at least with cygwin32
>> > tools. I've installed the coolview update (when will it be incorporated
>> > into the 'vanilla' cygwin32 tools anyways?) and it is going a bit
faster.
>>
>> The "coolview" sources represent a state of the actual Cygnus development
>> tree as of whenever that was made.  It's made more progress since then;
>> right now I'm battling race conditions in the signal code...
>>
>> Right now I can do a complete configure of the compiler tools (including
>> tcl/tk) in about 20 minutes and build them in an hour 15 min. which is a
>> huge improvement performance-wise.  But signal handling isn't quite as
>> robust as it needs to be...
>
>Well, something must be seriously wrong with my configuration, because I
was
>seeing performance times of about 6-7 hours to do the same.
>
>1 hour 15 mins is what I call slightly slow. 6-7 hours is what I call
glacial.
>I'd hit 'ls' and it would take about a minute to respond....
>
>I could live with an hour to build all the tools, believe me. Its just the
>going to dinner, then taking a nap, then playing two games of pool, and
finally
>coming back to the keyboard to see that the tools are *still* compiling
>which is getting annoying.
>
>Ed
>-
>For help on using this list (especially unsubscribing), send a message to
>"gnu-win32-request@cygnus.com" with one line of text: "help".
>

-
For help on using this list (especially unsubscribing), send a message to
"gnu-win32-request@cygnus.com" with one line of text: "help".

RE: performance of cygwin32

[Kevin Hughes]

Dear all,
	I have been chasing performance problems with our build system which ran on unix and is
Snail like on NT (i.e. it is unusable). I have just discovered more or less the same BUT  think
the problem is the NT networking not gnuwin32. 

To support this I timed ls to local and networked drives - wow the network took 10 times as 
the local drive and dir took around the same ratio (if not more). 

Does anyone know how to asses NT network peformance?

Kevin 

-----Original Message-----
From:	Geoffrey Noer [SMTP:noer@cygnus.com]
Sent:	08 January 1998 05:42
To:	Ed Peschko
Cc:	earnie_boyd@hotmail.com; gnu-win32@cygnus.com
Subject:	Re: performance of cygwin32

Ed Peschko wrote:
> 
> Win32 is slow, but network drives are *glacial*, at least with cygwin32
> tools. I've installed the coolview update (when will it be incorporated
> into the 'vanilla' cygwin32 tools anyways?) and it is going a bit faster.

The "coolview" sources represent a state of the actual Cygnus development
tree as of whenever that was made.  It's made more progress since then;
right now I'm battling race conditions in the signal code...

Right now I can do a complete configure of the compiler tools (including
tcl/tk) in about 20 minutes and build them in an hour 15 min. which is a 
huge improvement performance-wise.  But signal handling isn't quite as
robust as it needs to be...

-- 
Geoffrey Noer
noer@cygnus.com
-
For help on using this list (especially unsubscribing), send a message to
"gnu-win32-request@cygnus.com" with one line of text: "help".


-
For help on using this list (especially unsubscribing), send a message to
"gnu-win32-request@cygnus.com" with one line of text: "help".

Re: performance of cygwin32

[Jeffrey C. Fried]

What did you do to determine performance?  What CPU on your PC are you
comparing to which CPU on your Solaris box?

... jeff
 
At 05:06 PM 1/1/98 -0700, Ed Peschko wrote:
>hey - 
>
>I've found after installing cdk.exe taht the performance of cygwin32 is 
>about 5-10 times slower than my solaris box. 
>
>Is this due to the underlying win32 api, or is it due to cygwin32, or is it 
>just me not doing something correctly?
>
>Thanks much,
>
>Ed
>-
>For help on using this list (especially unsubscribing), send a message to
>"gnu-win32-request@cygnus.com" with one line of text: "help".
>
>
--
Jeffrey C. Fried      jcfried@ix.netcom.com

   Because a liar tells the truth does not mean the truth is a lie.

NOTICE: I charge $500.00 for each unsolicited advertisement i receive as email
to cover the cost of my time to review and possibly respond to your
advertisement.
-
For help on using this list (especially unsubscribing), send a message to
"gnu-win32-request@cygnus.com" with one line of text: "help".

performance of cygwin32

[Ed Peschko]

hey - 

I've found after installing cdk.exe taht the performance of cygwin32 is 
about 5-10 times slower than my solaris box. 

Is this due to the underlying win32 api, or is it due to cygwin32, or is it 
just me not doing something correctly?

Thanks much,

Ed
-
For help on using this list (especially unsubscribing), send a message to
"gnu-win32-request@cygnus.com" with one line of text: "help".