On Fri, Apr 3, 2009 at 8:21 PM, Carl-Daniel Hailfinger < c-d.hailfinger.devel.2006@gmx.net> wrote:
On 04.04.2009 01:06, Joseph Smith wrote:
On Sat, 04 Apr 2009 00:26:19 +0200, Carl-Daniel Hailfinger c-d.hailfinger.devel.2006@gmx.net wrote:
[ROMCC->CAR] Great. Will a bad reflash hurt you?
No, as long as it doesn't blow the thing up :-0
Heh. That won't happen unless an endless loop is too much for your cooling setup.
What you need to try this one out:
- a POST card or another way to fetch POST codes.
- tell me whether POST works by default or you need special setup.
- a will to try out v3 with some patches.
Ok, I have a PCI/Parallel Post Card. The RM4100 doesn't have eithor, the IP1000 does have a PCI slot so we can use that for testing.
Good. Can you verify that the PCI POST card works?
The idea is to test the v3 intel CAR code which has not been on real hardware yet, but it is a much cleaner and more readable implementation compared to v2. I'm hoping we don't need chipset specific stuff for POST to work.
Not sure, serial output works fine.
We'll debug very early code, before serial can be set up.
I need info on the cache size of your processor, though. Unreliable CAR is not nearly as much fun as reliable CAR.
512k
Thanks. Can you find out L1 cache sizes as well? The spec page was not clear about that. And is the processor hyperthreading capable?
I'll follow up with a patch on Wednesday or Thursday.
All P2 & P3s should be 32k L1 cache, and between the p2s, p3s, and celerons based on them, they have 128, 256, or 512k of L2 cache (Xeons also had 1 or 2MB). I think if it runs on Joe's system, and can handle the different cache sizes, the different cores are similar enough that it might just work on the whole range. I have a few boards kicking around, 440BX and i810 alike, I can test stuff on if you need me to.
-Corey