[OLPC-devel] Upgrading to LinuxBIOS. (problem on a subset of ATest boards).

[supat at supat.eu.org]

Please let me share experiance on animal scice and agricultural limiting 
factor theory.

In animal, the first limiting factor is lysine and the second limiting 
factor is methionine.

In agriculture, the first limiting factor is water and the second limiting 
factor is soil.

After the first limiting factor reach it limit then the second limiting 
factor become important to determine animal/plant performance.

The same theory applied to human and computer performance.

At this moment OLPC and pentium 4 pc in my faculty run much slower than my 
old pentium 1 90 mhz not because it has slow processor but because of 
other limiting factors to inhibit it to run at full speed of processor.

This is serious problem I mention earlier. I am upset while old pc can run 
.swf w/o problem from decentralized technique but OLPC do has problem and 
I think it was VGA/usb8388 problem

Because those two component perform very slow and cause problem in my 
experiment.

I suspect VGA because in my experiance pc that pull VGA memory from SRAM 
on-board cause a lot of problem because RAM that use with VGA alway have 
to run at faster speed than SRAM.

I suspect usb8388 because it cause problem under OLPC with slackware 
lib/bin.

Good hardware should cause no problem to any s/w used.

regards,
supat

On Fri, 8 Sep 2006, Jim Gettys wrote:

> If you are using ATest boards, please read this message carefully.
>
> In the early conversion to LinuxBIOS, we found a hardware problem.  This
> is good: it is the purpose of testing.
>
> To meet demand of high volume components that may be in short supply
> sometimes, such as DRAM, it is normal to "qualify" a number of sources
> for those components.
>
> The ATest boards were therefore manufactured with three different DRAM
> vendor's chips.  Unfortunately, a mistake was made in the selection of
> the proper component from Infineon, and a DRAM chip version was used
> that is too slow for the Geode GX.  This is not a problem with
> Infineon's memory chips; the wrong speed chip was used by mistake,
> rather than the parts themselves not working to specification.
>
> Therefore, as the warning in the "Upgrading to LinuxBIOS" page says in
> more detail: http://wiki.laptop.org/go/Upgrading_to_LinuxBIOS#Warnings
> please do *NOT* install LinuxBIOS right now onto boards with Infineon
> DRAM's, but wait until we say to.
>
> We will probably make a version of LinuxBIOS for use on such boards that
> sets the memory controller slow enough for those boards to work as well
> as they have under Insyde's BIOS, which we believe was running at the
> slower speed.  But that slower speed, which is as slow as the Geode GX
> can run, is still nominally too fast for the memory chips used.  We are
> working on a strategy of how to handle this issue, though clearly most
> boards have been stable at this slower speed (or we would have had many
> complaints by now).
>
> If you have already converted to LinuxBIOS on boards with the Infineon
> memory, and your board appears to be working OK, leave it alone for now
> and continue to run LinuxBIOS.
>
> Please *DO* continue to install LinuxBIOS onto boards you have with
> Hynix or PSC memory chips.  We do need to test both LinuxBIOS and the
> use of those manufacturer's memory chips at the higher intended speed.
>
>                  Thank you all for your efforts!
>                           - Jim Gettys
>
> My great thanks to Tom Sylla, Ron Minnich, Richard Smith, Jordan Crouse,
> Chris Ball, Marcelo Tosatti, James Cameron and Mitch Bradley for helping
> figure out this problem.
>
> This problem is: http://dev.laptop.org/ticket/53
>
> -- 
> Jim Gettys
> One Laptop Per Child
>
>
> _______________________________________________
> Devel mailing list
> Devel at laptop.org
> http://mailman.laptop.org/mailman/listinfo/devel
>