A7M266-D and upgrade from Athlon MP 1500+ to 2800+

[Guest]

Archived from groups: alt.comp.periphs.mainboard.asus (More info?)

I have been running happily with an A7M266-D for two years. I originally
had MP 1500+ (Palomino). Today, I upgraded to 2800 + (Barton), retail
AMD chips, clearly marked MP 2800 models on the chip and box.

The motherboard has the 1010 BIOS .

When it boots up, it complains that the CPUs are not MP capable . It
also doesn't display the frequency. Is there a BIOS for this motherboard
that supports this model ?

Also, I did some benchmarking before and after the upgrade, and the
results are mixed :

Here are the CPU results with 1500+ :

CPU integer
Dhrystone : 2248.300 VAX 11/780 MIPS equivalent
Hanoi : 2350.000 moves/25 microseconds
Heapsort : 1469.800 Million Instructions Per Second
Sieve : 1842.700 Million Instructions Per Second
-----------------------------------------------------------------------
Total : 2020.848 CPU integer-marks

CPU float
Linpack : 330.250 MFLOPS
Flops : 991.570 MFLOPS
Fast Fourier Transfrm : 119.700 VAX FFT's
-----------------------------------------------------------------------
Total : 620.800 CPU floating point-marks

And with the 2800+ :

CPU integer
Dhrystone : 5855.200 VAX 11/780 MIPS equivalent
Hanoi : 3635.000 moves/25 microseconds
Heapsort : 2411.200 Million Instructions Per Second
Sieve : 2922.100 Million Instructions Per Second
-----------------------------------------------------------------------
Total : 4153.007 CPU integer-marks

CPU float
Linpack : 534.180 MFLOPS
Flops : 275.680 MFLOPS
Fast Fourier Transfrm : 140.090 VAX FFT's
-----------------------------------------------------------------------
Total : 310.672 CPU floating point-marks


These benchmarks were done under OS/2 Warp Server SMP using Sysbench, a
program that runs CPU tests in parallel, once for each CPU in the machine.

I'm pretty happy with the integer performance increase of the 2800+
chip. However, as you can see, the floating point marks went down in
half ! The "Flops" test dropped from 991.57 to 275.68 ! And the FFT only
got a 17% improvement.

Has anyone ever seen a drop in FPU performance with the recent AMD chips ?

I'm investigating with the benchmark's author to find out if this is
something wrong with the program or with the new AMD chips ...

 

[Guest]

Archived from groups: alt.comp.periphs.mainboard.asus (More info?)

"Julien Pierre" <madbrain@rawbw.nospam.com> wrote in message
news:xx_cc.1305$HU5.301@fe03.usenetserver.com...
> I have been running happily with an A7M266-D for two years. I
originally
> had MP 1500+ (Palomino). Today, I upgraded to 2800 + (Barton), retail
> AMD chips, clearly marked MP 2800 models on the chip and box.
>
> The motherboard has the 1010 BIOS .

I have1011.003 bios and am running two 2100+ that were mod'ed from xp to
mp. It processes seti units and does 13 - 15 per day with 3h29m per
unit. I would be interested in any info you got on seti with those 2800
chips.

>
> When it boots up, it complains that the CPUs are not MP capable . It
> also doesn't display the frequency. Is there a BIOS for this
motherboard
> that supports this model ?
>
> Also, I did some benchmarking before and after the upgrade, and the
> results are mixed :
>
> Here are the CPU results with 1500+ :
>
> CPU integer
> Dhrystone : 2248.300 VAX 11/780 MIPS equivalent
> Hanoi : 2350.000 moves/25 microseconds
> Heapsort : 1469.800 Million Instructions Per
Second
> Sieve : 1842.700 Million Instructions Per
Second
> ------------------------------------------------------------------
-----
> Total : 2020.848 CPU integer-marks
>
> CPU float
> Linpack : 330.250 MFLOPS
> Flops : 991.570 MFLOPS
> Fast Fourier Transfrm : 119.700 VAX FFT's
> ------------------------------------------------------------------
-----
> Total : 620.800 CPU floating point-marks
>
> And with the 2800+ :
>
> CPU integer
> Dhrystone : 5855.200 VAX 11/780 MIPS equivalent
> Hanoi : 3635.000 moves/25 microseconds
> Heapsort : 2411.200 Million Instructions Per
Second
> Sieve : 2922.100 Million Instructions Per
Second
> ------------------------------------------------------------------
-----
> Total : 4153.007 CPU integer-marks
>
> CPU float
> Linpack : 534.180 MFLOPS
> Flops : 275.680 MFLOPS
> Fast Fourier Transfrm : 140.090 VAX FFT's
> ------------------------------------------------------------------
-----
> Total : 310.672 CPU floating point-marks
>
>
> These benchmarks were done under OS/2 Warp Server SMP using Sysbench,
a
> program that runs CPU tests in parallel, once for each CPU in the
machine.
>
> I'm pretty happy with the integer performance increase of the 2800+
> chip. However, as you can see, the floating point marks went down in
> half ! The "Flops" test dropped from 991.57 to 275.68 ! And the FFT
only
> got a 17% improvement.
>
> Has anyone ever seen a drop in FPU performance with the recent AMD
chips ?
>
> I'm investigating with the benchmark's author to find out if this is
> something wrong with the program or with the new AMD chips ...
>

 

[Guest]

Archived from groups: alt.comp.periphs.mainboard.asus (More info?)

Hi,


Beemer Biker wrote:
> "Julien Pierre" <madbrain@rawbw.nospam.com> wrote in message
> news:xx_cc.1305$HU5.301@fe03.usenetserver.com...
>
>>I have been running happily with an A7M266-D for two years. I
>
> originally
>
>>had MP 1500+ (Palomino). Today, I upgraded to 2800 + (Barton), retail
>>AMD chips, clearly marked MP 2800 models on the chip and box.
>>
>>The motherboard has the 1010 BIOS .
>
>
> I have1011.003 bios and am running two 2100+ that were mod'ed from xp to
> mp. It processes seti units and does 13 - 15 per day with 3h29m per
> unit. I would be interested in any info you got on seti with those 2800
> chips.

I haven't run SETI. It appears the benchmark issues I got were related
to an overflow in the program - it didn't handle CPUs as fast as the
2800+

So I think my CPUs are running at the proper speed. The BIOS setup shows
the MHz at 2133 which is correct, but the chips are reported as 2600+ (I
hadn't paid attention during the first boot - it says 2600+ ; subsequent
boots with CTRL-ALT-DEL show no speed).

So I think I just need this BIOS 1011.0003 update to fix this cosmetic
issue. Where can I download it ? I can't find it anywhere ?

As for SETI, I'm not running it. I'll see if I can find a client for
OS/2 and let you know.

 

[Guest]

Archived from groups: alt.comp.periphs.mainboard.asus (More info?)

"Julien Pierre" <madbrain@rawbw.nospam.com> wrote in message news:4074ADCD.1090808@rawbw.nospam.com...
<snip>
> So I think I just need this BIOS 1011.0003 update to fix this cosmetic
> issue. Where can I download it ? I can't find it anywhere ?

sorry, i should have said it is a beta bios, you have to select beta when at their download page (select beta in the following url)
http://tinyurl.com/2y6dr
I have had no trouble with the new bios system. Be aware that updating a bios can result in a dead motherboard. I suggest you boot a dos floppy with the bios update dos (award flash i think) http://www.asus.it/support/download/item.aspx?ModelName=Tools
I had a problem with their live update program when it first came out and on two occassions ended up with a corrupted bios on a CUV4X-D. Fortunately, I had a 2nd CUV4X-D and was able to do a live bios swap to recover. There seems to be a very new live update at the top of that page, the one I had was 2 years old and maybe they got all the bugs out. If you do update your bios, be sure to verify that you can read and store the old bios. That (saving the old bios) is a good check that you got the correct bios burner.

>
> As for SETI, I'm not running it. I'll see if I can find a client for
> OS/2 and let you know.
>

I didnt see your list OS/2 as I didnt scroll down far enough. i do not think there is a version but i could be mistaken.

FWIW, I do not understand why ASUS quit producing that motherboard. I tried to get another one once about a year ago but the only source seemed to be ebay.

 

[Guest]

Archived from groups: alt.comp.periphs.mainboard.asus (More info?)

Hi,

Beemer Biker wrote:
> "Julien Pierre" <madbrain@rawbw.nospam.com> wrote in message news:4074ADCD.1090808@rawbw.nospam.com...
> <snip>
>
>>So I think I just need this BIOS 1011.0003 update to fix this cosmetic
>>issue. Where can I download it ? I can't find it anywhere ?
>
>
> sorry, i should have said it is a beta bios, you have to select beta when at their download page (select beta in the following url)
> http://tinyurl.com/2y6dr
> I have had no trouble with the new bios system. Be aware that updating a bios can result in a dead motherboard. I suggest you boot a dos floppy with the bios update dos (award flash i think) http://www.asus.it/support/download/item.aspx?ModelName=Tools
> I had a problem with their live update program when it first came out and on two occassions ended up with a corrupted bios on a CUV4X-D. Fortunately, I had a 2nd CUV4X-D and was able to do a live bios swap to recover. There seems to be a very new live update at the top of that page, the one I had was 2 years old and maybe they got all the bugs out. If you do update your bios, be sure to verify that you can read and store the old bios. That (saving the old bios) is a good check that you got the correct bios burner.

Thanks. I have flashed BIOSes many times before - and occasionally the
wrong one too - I know very well what happens when you do that. I just
needed the download link . The beta BIOS worked and the board shows
2800+ CPUs now, and no longer complains about MP.


>>As for SETI, I'm not running it. I'll see if I can find a client for
>>OS/2 and let you know.
>>
>
>
> I didnt see your list OS/2 as I didnt scroll down far enough. i do not think there is a version but i could be mistaken.
>
> FWIW, I do not understand why ASUS quit producing that motherboard. I tried to get another one once about a year ago but the only source seemed to be ebay.
>
>

I found a command-line OS/2 client. I installed it. Looks like it is
single-threaded, so I have to run two copies at a time to make use of
both CPUs. How do I tell how fast it's going ?

 

[Guest]

Archived from groups: alt.comp.periphs.mainboard.asus (More info?)

"Julien Pierre" <madbrain@rawbw.nospam.com> wrote in message
news:5fbdc.1753$HU5.1448@fe03.usenetserver.com...
<snip>
> Thanks. I have flashed BIOSes many times before - and occasionally the
> wrong one too - I know very well what happens when you do that. I just
> needed the download link . The beta BIOS worked and the board shows
> 2800+ CPUs now, and no longer complains about MP.

Asus seems to keep their motherboards up to date even after out of
production. An old P5 socket 7 I had got upgraded to ACPI and large
disks and supports windows xp even though they quit producting it before
win2k came out. I had a compaq desktop that would turn on anytime the
phone rang. They never provided a bios where you could disable the
modem interrupt and their support said it could not be done. I found
that gigabyte actually made the motherboard and downloaded a bios that
did all the stuff that compaq had left off. I notice that compaq is not
in business anymore but gigabyte is.

>
> I found a command-line OS/2 client. I installed it. Looks like it is
> single-threaded, so I have to run two copies at a time to make use of
> both CPUs. How do I tell how fast it's going ?
>

Just run one thread, i dont think that os2 seti client is properly coded
for hyperthreading and probably does not support smp either. The linux
version I have ( RH9 SMP) clearly does not use the 2nd cpu efficiently
and I am getting 4 per day where I used to get 9 on windows 2000. I
have a removeable disk and can boot either win2k or RH9 for various
developments we are working on and have discovered that the seti client
for linux is not working properly. Here are some photos of some of the
systems we got http://tinyurl.com/2eut6 . We are using RH9 linux on
some dual AMD64 tyan mombos as the enterprise edition cost $799 USD.
The RH9 is free. I am not into OS2, what is an OS2 SMP version going
for now-a-days?

To check completion, look at the file "state.sah". There is a
completion % (prog=0.31741765 for example) and you can get a general
idea of how fast it is progressing from reading that file.