Some experiences with the Athlon optimisation problem

Some experiences with the Athlon optimisation problem

[Heinz Deinhart]

Hi,

i might have found some interesting informations regarding the Athlon
optimisation problems. I have two different Athlon versions, the
older one seems to work very table, the newer one crashes always
during system startup.  Both chips where running in exactly the same
machine. (the older one is a 1.133Ghz the newer one a 1.2Ghz.)

I tried 2.4.9, 2.4.9-ac6 and 2.4.3, not difference. The system is a
MSI-K7A Turbo with KT133A from VIA, 256MB PC133 RAM. I have 8 of the
new ones, and all of them do not work.

Because the same system is stable with one kind of Athlons and does
not work with the other, couldn't this mean those Athlons are buggy ?

Of course all troubles vanish when i turn off Athlon optimisation. But
this probably doesnt kill the problem.

The old Athlon reads:
	A1133AMS3C
	AVIA 0115TPAW
	95262550081

The new (non working) one:
	A1200AMS3C
	AXIA 0121RPDW
	95987660990

I hope this is somehow useful.

ciao,
heinz

-- 
Heinz Deinhart <heinz@auto.tuwien.ac.at>
+43 1 58801-18321
Technische Universitaet Wien, Dept. E183/1

Re: Some experiences with the Athlon optimisation problem

[Stefan Smietanowski]

Hi.

 >>The old Athlon reads:
 >>       A1133AMS3C
 >>       AVIA 0115TPAW
 >>       95262550081
 >>
 >>The new (non working) one:
 >>       A1200AMS3C
 >>       AXIA 0121RPDW
 >>       95987660990
 >>
 >
 > The first line can be decoded using AMDs documentation at
 >
 >     http://www.amd.com/products/cpg/athlon/techdocs/index.html
 >
 > In particular

<snip>

 > Anyone know what the second line means?  It is quite mysterious
 > that the word VIA appears on the chip that works with VIA KT133A
 > and not on the other :-)

The first letters, be they 4, 5 or 6 or whatever is stepping.

So the old is AVIA stepping, the new is the AXIA stepping.

The nextcoming four numbers are production week, ie 2001 week 15 and 
2001 week 21. I can't remember the letters.

// Stefan

Some experiences with the Athlon optimisation problem

[Eric Olson]

Oops, it should be "Core" Temperature not "Case" Temperature below.
The AMD Athlon Processor Model 4 Data Sheet summary on page 71 also
made this typo and I blindly copied it.  --Eric

>Dear Heinz Deinhart,
>
>>The old Athlon reads: 
>>        A1133AMS3C 
>>        AVIA 0115TPAW 
>>        95262550081 
>>
>>The new (non working) one: 
>>        A1200AMS3C 
>>        AXIA 0121RPDW 
>>        95987660990 
>
>The first line can be decoded using AMDs documentation at
>
>    http://www.amd.com/products/cpg/athlon/techdocs/index.html
>
>In particular
>
>       A 1133 A M S 3 C
>       |   |  | | | | |
>       |   |  | | | |  \___ FSB (B=200, C=266)
>       |   |  | | |  \_____ Size of L2 Cache (3=256K)
>       |   |  | |  \_______ Case Temperature (S=95C, T=90C)
>       |   |  |  \_________ Operating Voltage (M=1.75V, P=1.7V)
>       |   |   \___________ Package Type (A=PGA)
>       |    \______________ Speed in MHz
>        \__________________ A for Athlon
>
>Anyone know what the second line means?  It is quite mysterious 
>that the word VIA appears on the chip that works with VIA KT133A 
>and not on the other :-)
>
>Have you tried Robert Redelmeier's new program burnMMX2 which has 
>the 3DNow streaming cache bypass load/store instructions that seem 
>to be causing all the trouble?
>
>Does it run indefinitely or terminate with the 1133MHz processor 
>configuration?  What about for the 1200MHz processor with kernel 
>optimization turned off?
>
>--Eric
>

Some experiences with the Athlon optimisation problem

[Eric Olson]

Dear Heinz Deinhart,

>The old Athlon reads: 
>        A1133AMS3C 
>        AVIA 0115TPAW 
>        95262550081 
>
>The new (non working) one: 
>        A1200AMS3C 
>        AXIA 0121RPDW 
>        95987660990 

The first line can be decoded using AMDs documentation at

    http://www.amd.com/products/cpg/athlon/techdocs/index.html

In particular

       A 1133 A M S 3 C
       |   |  | | | | |
       |   |  | | | |  \___ FSB (B=200, C=266)
       |   |  | | |  \_____ Size of L2 Cache (3=256K)
       |   |  | |  \_______ Case Temperature (S=95C, T=90C)
       |   |  |  \_________ Operating Voltage (M=1.75V, P=1.7V)
       |   |   \___________ Package Type (A=PGA)
       |    \______________ Speed in MHz
        \__________________ A for Athlon

Anyone know what the second line means?  It is quite mysterious 
that the word VIA appears on the chip that works with VIA KT133A 
and not on the other :-)

Have you tried Robert Redelmeier's new program burnMMX2 which has 
the 3DNow streaming cache bypass load/store instructions that seem 
to be causing all the trouble?

Does it run indefinitely or terminate with the 1133MHz processor 
configuration?  What about for the 1200MHz processor with kernel 
optimization turned off?

--Eric