Re: Athlon patches for 2.2.x kernels

Dieter =?iso-8859-1?Q?N=FCtzel?= (dieter.nuetzel@arcormail.de)
Tue, 21 Sep 1999 01:18:44 +0200

> > Is there any patches for the 2.2.x kernels for the athlon? I've noticed
>
> Nope
>
> > that it can be selected in 2.3.18ac6, but since some features I need to
> > use are not working at the moment, I need to stick with the 2.2.x
> > kernels for now.
>
> The changes are optimisations, they also rely on the 2.3.x FPU stuff. I have
> no plan to back port them to 2.2.x. The only thing that does matter - MTRR
> support - is sorted.
>
> Alan
>
>
I have my Athlon running with 2.2.13pre9 and 2.3.18ac6.

Alan, I found some problems with my Voodoo3 3000 and my MSI MS-6167 main
board (BIOS).
When ever I set the MTRRs for the Voodoo3 I get in trouble (only with
3D/Glide). Yes, I get some big trouble under Win98/98 SE, too. It never
runs under Win at this time!
I think it is under Win MTRR related, too. But I can't get access to the
MTRR registers under Winbloze.

Which Athlon main board have you got, Alan? Reversion? BIOS?
Have you got a Winbloze tool to fiddle around with the MTRR registers?
I am still in contact with MSI/3dfx/AMD.

Regards,
Dieter

BTW Here are some benchmarks:

BYTE UNIX Benchmarks (Version 3.11)
System -- Linux SunWave1 2.3.18ac2 #1 Sun Sep 12 20:59:51 MET DST 1999
i686 unknown
Start Benchmark Run: Mon Sep 13 04:20:25 MET DST 1999
1 interactive users.
Dhrystone 2 without register variables 1164344.7 lps (10 secs, 6
samples)
Dhrystone 2 using register variables 1156970.0 lps (10 secs, 6
samples)
Arithmetic Test (type = arithoh) 34861680.9 lps (10 secs, 6
samples)
Arithmetic Test (type = register) 91612.2 lps (10 secs, 6
samples)
Arithmetic Test (type = short) 97030.2 lps (10 secs, 6
samples)
Arithmetic Test (type = int) 91587.7 lps (10 secs, 6
samples)
Arithmetic Test (type = long) 91551.2 lps (10 secs, 6
samples)
Arithmetic Test (type = float) 214766.2 lps (10 secs, 6
samples)
Arithmetic Test (type = double) 214943.0 lps (10 secs, 6
samples)
System Call Overhead Test 302421.0 lps (10 secs, 6
samples)
Pipe Throughput Test 371559.1 lps (10 secs, 6
samples)
Pipe-based Context Switching Test 115875.5 lps (10 secs, 6
samples)
Process Creation Test 3125.7 lps (10 secs, 6
samples)
Execl Throughput Test 621.1 lps (9 secs, 6
samples)
File Read (10 seconds) 1151294.0 KBps (10 secs, 6
samples)
File Write (10 seconds) 166564.0 KBps (10 secs, 6
samples)
File Copy (10 seconds) 15059.0 KBps (10 secs, 6
samples)
File Read (30 seconds) 1152364.0 KBps (30 secs, 6
samples)
File Write (30 seconds) 167910.0 KBps (30 secs, 6
samples)
File Copy (30 seconds) 10397.0 KBps (30 secs, 6
samples)
C Compiler Test 473.6 lpm (60 secs, 3
samples)
Shell scripts (1 concurrent) 1176.5 lpm (60 secs, 3
samples)
Shell scripts (2 concurrent) 605.1 lpm (60 secs, 3
samples)
Shell scripts (4 concurrent) 301.0 lpm (60 secs, 3
samples)
Shell scripts (8 concurrent) 150.0 lpm (60 secs, 3
samples)
Dc: sqrt(2) to 99 decimal places 63349.4 lpm (60 secs, 6
samples)
Recursion Test--Tower of Hanoi 15369.9 lps (10 secs, 6
samples)

INDEX VALUES
TEST BASELINE RESULT
INDEX

Arithmetic Test (type = double) 2541.7 214943.0
84.6
Dhrystone 2 without register variables 22366.3 1164344.7
52.1
Execl Throughput Test 16.5 621.1
37.6
File Copy (30 seconds) 179.0 10397.0
58.1
Pipe-based Context Switching Test 1318.5 115875.5
87.9
Shell scripts (8 concurrent) 4.0 150.0
37.5

=========
SUM of 6 items
357.7
AVERAGE
59.6

---
Dieter Nützel
Graduate Student, Computer Science


University of Hamburg
Department of Computer Science
Cognitive Systems Group
Vogt-Kölln-Straße 30
D-22527 Hamburg, Germany


email: Dieter Nützel <nuetzel@kogs.informatik.uni-hamburg.de>
@home: Dieter Nützel <dieter.nuetzel@arcormail.de>


_________________________________________________________


Vote for DRI (Direct Rendering Infrastructure) for Linux.
Because I do 3D medical visualization as my thesis.
_________________________________________________________




-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/