From: c@royle.org (Chris Royle) Subject: Re: How does Linux compare to SUN IPC? Date: Fri, 24 Sep 1993 08:17:37 GMT
In comp.os.linux, Steve Norton (steve@interaccess.com) wrote:
:> a228dhal@cdf.toronto.edu (Dhaliwal Bikram Singh) writes:
:> >It has seemed to me that my Linux system at home (X and GCC running in
:> >a 15mb partition, on a 386-40, with room to spare) is faster than the
:> >SUN IPC workstations I use at school. I can only offer subjective
:> >speculation though, ie. time for a xterm to open, etc...
:> Well, here at the office, we've had 4 machines: One 386-40, 8 MB of
:> RAM, one 486-66, 16 MB of RAM, one Sparc-10 with 64 MB of RAM and
:> one Sparc-2 with 32 MB of RAM.
:> The 386-40 ($1000) will process Usenet news about 2x faster than the
:> Sparc-10 ($20,000). Of course, this is entirely due to the super fast
:> Linux I/O. Processing news is all disk access, and the IDE drives blow
:> away the sun SCSI-2 drives. Unfortunately, the Sparc-10 is faster for just
:> about everything else.
:> The 486-66 ($3000) runs 10-25% faster than the Sparc-2 ($7000) for CPU
:> intensive activities (compiling, crunching numbers, Xlife, etc.) For I/O
:> things (xli on Xwindows) it completely blows the Sparc-2 away.
:> Now, I'm kind of biased. I personally believe Sparcs are junk, and that
:> a 40Mhz motherboard with a 386 is just as good as a 40Mhz Sparc motherboard.
Well.... let me see if I can find some stats :
Times in seconds, description of tests to follow.
(I did have some timings for a SPARC 59MIPS, but have lost them. The DX2/50
Beat it running DOS, anyway).
Linux DOS SUN IRIS Linux DOS
0.99.9 5.0 IPC INDIGO 0.99.10 5.0
DX2/50 DX2/50 IRIX 4.0.5F DX2/66 DX2/66
1. 19 35 72 17 15 26
2. 13 19 49 14 9 49
3. 21 27 63 25 17 22
Test 1 is 50,000,000 Floating point multiplications
Test 2 is 50,000,000 Integer multiplications
Test 3 is 50,000 array copies, each array is 1,000 characters.
We have always felt that the SUN IPC was a slow thing, and this appared to
prove it to us. The SPARC CLassic 59MIPS came in a couple of seconds slower
on each test that the DX2/50 running DOS. The programme is written in C
and compiled under Linux with gcc, DOS with Visual C++ 1.0, on the SUN with
the standard ANSI compiler and ditto on the Iris machine.
Hope that is informative to someone somewhere...
Chris.17 15 26
2. 13 19 49 14 9 49
3. 21 27 63 25 17 22
Test 1 is 50,000,000 Floating point multiplications
Test 2 is 50,000,000 Integer multiplications
Test 3 is 50,000 array copies, each array is 1,000 characters.
We have always felt that the SUN IPC was a slow thing, and this appared to
prove it to us. The SPARC CLassic 59MIPS came in a couple of seconds slower
on each test that the DX2/50 running DOS. The programme is written in C
and compiled under Linux with gcc, DOS with Visual C++ 1.0, on the SUN with
the standard ANSI compiler and ditto on the Iris machine.
Hope that is informative to someone somewhere...
Chris.17 15 26
2. 13 19 49 14 9 49
3. 21 27 63 25 17 22
Test 1 is 50,000,000 Floating point multiplications
Test 2 is 50,000,000 Integer multiplications
Test 3 is 50,000 array copies, each array is 1,000 characters.
We have always felt that the SUN IPC was a slow thing, and this appared to
prove it to us. The SPARC CLassic 59MIPS came in a couple of seconds slower
on each test that the DX2/50 running DOS. The programme is written in C
and compiled under Linux with gcc, DOS with Visual C++ 1.0, on the SUN with
the standard ANSI compiler and ditto on the Iris machine.
Hope that is informative to someone somewhere...
Chris.17 15 26
2. 13 19 49 14 9 49
3. 21 27 63 25 17 22
Test 1 is 50,000,000 Floating point multiplications
Test 2 is 50,000,000 Integer multiplications
Test 3 is 50,000 array copies, each array is 1,000 characters.
We have always felt that the SUN IPC was a slow thing, and this appared to
prove it to us. The SPARC CLassic 59MIPS came in a couple of seconds slower
on each test that the DX2/50 running DOS. The programme is written in C
and compiled under Linux with gcc, DOS with Visual C++ 1.0, on the SUN with
the standard ANSI compiler and ditto on the Iris machine.
Hope that is informative to someone somewhere...
Chris.17 15 26
2. 13 19 49 14 9 49
3. 21 27 63 25 17 22
Test 1 is 50,000,000 Floating point multiplications
Test 2 is 50,000,000 Integer multiplications
Test 3 is 50,000 array copies, each array is 1,000 characters.
We have always felt that the SUN IPC was a slow thing, and this appared to
prove it to us. The SPARC CLassic 59MIPS came in a couple of seconds slower
on each test that the DX2/50 running DOS. The programme is written in C
and compiled under Linux with gcc, DOS with Visual C++ 1.0, on the SUN with
the standard ANSI compiler and ditto on the Iris machine.
Hope that is informative to someone somewhere...
Chris.17 15 26
2. 13 19 49 14 9 49
3. 21 27 63 25 17 22
Test 1 is 50,000,000 Floating point multiplications
Test 2 is 50,000,000 Integer multiplications
Test 3 is 50,000 array copies, each array is 1,000 characters.
We have always felt that the SUN IPC was a slow thing, and this appared to
prove it to us. The SPARC CLassic 59MIPS came in a couple of seconds slower
on each test that the DX2/50 running DOS. The programme is written in C
and compiled under Linux with gcc, DOS with Visual C++ 1.0, on the SUN with
the standard ANSI compiler and ditto on the Iris machine.
Hope that is informative to someone somewhere...
Chris.17 15 26
2. 13 19 49 14 9 49
3. 21 27 63 25 17 22
Test 1 is 50,000,000 Floating point multiplications
Test 2 is 50,000,000 Integer multiplications
Test 3 is 50,000 array copies, each array is 1,000 characters.
We have always felt that the SUN IPC was a slow thing, and this appared to
prove it to us. The SPARC CLassic 59MIPS came in a couple of seconds slower
on each test that the DX2/50 running DOS. The programme is written in C
and compiled under Linux with gcc, DOS with Visual C++ 1.0, on the SUN with
the standard ANSI compiler and ditto on the Iris machine.
Hope that is informative to someone somewhere...
Chris.17 15 26
2. 13 19 49 14 9 49
3. 21 27 63 25 17 22
Test 1 is 50,000,000 Floating point multiplications
Test 2 is 50,000,000 Integer multiplications
Test 3 is 50,000 array copies, each array is 1,000 characters.
We have always felt that the SUN IPC was a slow thing, and this appared to
prove it to us. The SPARC CLassic 59MIPS came in a couple of seconds slower
on each test that the DX2/50 running DOS. The programme is written in C
and compiled under Linux with gcc, DOS with Visual C++ 1.0, on the SUN with
the standard ANSI compiler and ditto on the Iris machine.
Hope that is informative to someone somewhere...
Chris.17 15 26
2. 13 19 49 14 9 49
3. 21 27 63 25 17 22
Test 1 is 50,000,000 Floating point multiplications
Test 2 is 50,000,000 Integer multiplications
Test 3 is 50,000 array copies, each array is 1,000 characters.
We have always felt that the SUN IPC was a slow thing, and this appared to
prove it to us. The SPARC CLassic 59MIPS came in a couple of seconds slower
on each test that the DX2/50 running DOS. The programme is written in C
and compiled under Linux with gcc, DOS with Visual C++ 1.0, on the SUN with
the standard ANSI compiler and ditto on the Iris machine.
Hope that is informative to someone somewhere...
Chris.17 15 26
2. 13 19 49 14 9 49
3. 21 27 63 25 17 22
Test 1 is 50,000,000 Floating point multiplications
Test 2 is 50,000,000 Integer multiplications
Test 3 is 50,000 array copies, each array is 1,000 characters.
We have always felt that the SUN IPC was a slow thing, and this appared to
prove it to us. The SPARC CLassic 59MIPS came in a couple of seconds slower
on each test that the DX2/50 running DOS. The programme is written in C
and compiled under Linux with gcc, DOS with Visual C++ 1.0, on the SUN with
the standard ANSI compiler and ditto on the Iris machine.
Hope that is informative to someone somewhere...
Chris.17 15 26
2. 13 19 49 14 9 49
3. 21 27 63 25 17 22
Test 1 is 50,000,000 Floating point multiplications
Test 2 is 50,000,000 Integer multiplications
Test 3 is 50,000 array copies, each array is 1,000 characters.
We have always felt that the SUN IPC was a slow thing, and this appared to
prove it to us. The SPARC CLassic 59MIPS came in a couple of seconds slower
on each test that the DX2/50 running DOS. The programme is written in C
and compiled under Linux with gcc, DOS with Visual C++ 1.0, on the SUN with
the standard ANSI compiler and ditto on the Iris machine.
Hope that is informative to someone somewhere...
Chris.17 15 26
2. 13 19 49 14 9 49
3. 21 27 63 25 17 22
Test 1 is 50,000,000 Floating point multiplications
Test 2 is 50,000,000 Integer multiplications
Test 3 is 50,000 array copies, each array is 1,000 characters.
We have always felt that the SUN IPC was a slow thing, and this appared to
prove it to us. The SPARC CLassic 59MIPS came in a couple of seconds slower
on each test that the DX2/50 running DOS. The programme is written in C
and compiled under Linux with gcc, DOS with Visual C++ 1.0, on the SUN with
the standard ANSI compiler and ditto on the Iris machine.
Hope that is informative to someone somewhere...
Chris.17 15 26
2. 13 19 49 14 9 49
3. 21 27 63 25 17 22
Test 1 is 50,000,000 Floating point multiplications
Test 2 is 50,000,000 Integer multiplications
Test 3 is 50,000 array copies, each array is 1,000 characters.
We have always felt that the SUN IPC was a slow thing, and this appared to
prove it to us. The SPARC CLassic 59MIPS came in a couple of seconds slower
on each test that the DX2/50 running DOS. The programme is written in C
and compiled under Linux with gcc, DOS with Visual C++ 1.0, on the SUN with
the standard ANSI compiler and ditto on the Iris machine.
Hope that is informative to someone somewhere...
Chris.17 15 26
2. 13 19 49 14 9 49
3. 21 27 63 25 17 22
Test 1 is 50,000,000 Floating point multiplications
Test 2 is 50,000,000 Integer multiplications
Test 3 is 50,000 array copies, each array is 1,000 characters.
We have always felt that the SUN IPC was a slow thing, and this appared to
prove it to us. The SPARC CLassic 59MIPS came in a couple of seconds slower
on each test that the DX2/50 running DOS. The programme is written in C
and compiled under Linux with gcc, DOS with Visual C++ 1.0, on the SUN with
the standard ANSI compiler and ditto on the Iris machine.
Hope that is informative to someone somewhere...
Chris.17 15 26
2. 13 19 49 14 9 49
3. 21 27 63 25 17 22
Test 1 is 50,000,000 Floating point multiplications
Test 2 is 50,000,000 Integer multiplications
Test 3 is 50,000 array copies, each array is 1,000 characters.
We have always felt that the SUN IPC was a slow thing, and this appared to
prove it to us. The SPARC CLassic 59MIPS came in a couple of seconds slower
on each test that the DX2/50 running DOS. The programme is written in C
and compiled under Linux with gcc, DOS with Visual C++ 1.0, on the SUN with
the standard ANSI compiler and ditto on the Iris machine.
Hope that is informative to someone somewhere...
Chris.17 15 26
2. 13 19 49 14 9 49
3. 21 27 63 25 17 22
Test 1 is 50,000,000 Floating point multiplications
Test 2 is 50,000,000 Integer multiplications
Test 3 is 50,000 array copies, each array is 1,000 characters.
We have always felt that the SUN IPC was a slow thing, and this appared to
prove it to us. The SPARC CLassic 59MIPS came in a couple of seconds slower
on each test that the DX2/50 running DOS. The programme is written in C
and compiled under Linux with gcc, DOS with Visual C++ 1.0, on the SUN with
the standard ANSI compiler and ditto on the Iris machine.
Hope that is informative to someone somewhere...
Chris.17 15 26
2. 13 19 49 14 9 49
3. 21 27 63 25 17 22
Test 1 is 50,000,000 Floating point multiplications
Test 2 is 50,000,000 Integer multiplications
Test 3 is 50,000 array copies, each array is 1,000 characters.
We have always felt that the SUN IPC was a slow thing, and this appared to
prove it to us. The SPARC CLassic 59MIPS came in a couple of seconds slower
on each test that the DX2/50 running DOS. The programme is written in C
and compiled under Linux with gcc, DOS with Visual C++ 1.0, on the SUN with
the standard ANSI compiler and ditto on the Iris machine.
Hope that is informative to someone somewhere...
Chris.
-- Chris Royle Cheap mail & news feeds over UUCP from UKP5/mo Managing Director Windows / X-Windows code, 386s from UKP540 Objectronix Limited Desktop publishing Leeds, UK Tel. +44 532 661536