RE: Memory >128MB not detected

[Darryl Bowler <darryl e-securenetworks net>]

Thanks for all your help, I ve worked this out.
The motherboard I use is a PC Chips SIS 730S M810LR-H, and uses *shared* memory for built in video card, i.e. uses the physical ram for video memory. This is where 64MB of the RAM went to. Minor changes to the bios fixed the problem

Thanks


You wrote:
> I have a Compaq P3-600 Proliant with 1GB of Ram and Redhat 7.1 does not have 
> any problem
> finding this amount of ram without any tweaks. I dont believe 7.2 changed 
> anything
> in regards to this piece of the kernel. I assume you have more than one Dimm?
> Sdram or SIMM
> memory module in the box. USually you reverse sort the memory sizes and put the 
> largest
> modules in the lowest slots, sometimes you have to put them in matching pairs 
> or banks
> depends on motherboard. I have also seen some motherboards that have 
> jumpers/dip switches
> on them that have to be set for ram > 128 MB. We are all assuming your Bios ram
> test is showing all 192MB of ram during POST. Suggest double checking 
> Motherboard config
> instructions in their manual against how you have memory installed just to be 
> safe.
> You could also have a defective module that might pass Quick Boot memory 
> diagnostics but fail
> Linux memory probes. IF you are Quick boot enabled in the ROM Bios on MB, 
> disable it
> so you will get more extensive memory diagnostics.  I have also seen DOS 6.2 
> boot disks
> reject bad memory that passes BIOS checks when they run the extended memory 
> manager during booting.
> You might want to created a DOS boot system diskette with the memory manager on 
> it or boot a DOS 6.2X installation disk to see if it will test the ram and pass 
> the entire 192MB. I have had bad ram
> past full extensive POST diagnostics crash my system until I booted DOS up and 
> it failed the
> module as bad. (emm386 or equivalent)
> Whats the memory module layout 3 X 64MB, 128 + 64, 6 X 32MB??
> make sure all modules are compatible and of same type - ECC verses Parity 
> verses non-parity.
> try swapping out the last 64MB with another unit. If Linux probes the first 
> module and it is
> ECC say, it will probably assume all ECC memory and continue memory test based 
> on that
> assumption. IF you switch to parity in the second module, a memory probe test 
> will fail.
> Linux may use the BIOS saved memory size from boot in the flash memory as a 
> base (compaq), but
> it may also do a page level probe and write clear of ram as it boots to clear 
> the memory to
> a known state and determine the real amount of ram present. If it gets a write 
> failure at
> the 128MB boundary for whatever reason, it will probably stop figuring the 
> error means no ram
> at that location, ie end of memory.
> Kenneth Goodwin
> Director of IT Operations
> Datamark Technology, Inc.
> >>> -----Original Message-----
> >>> From: enigma-list-admin redhat com
> >>> [mailto:enigma-list-admin redhat com]On Behalf Of Darryl Bowler
> >>> Sent: Thursday, September 12, 2002 4:57 PM
> >>> To: redhat-list redhat com; enigma-list redhat com
> >>> Subject: Memory >128MB not detected
> >>> 
> >>> 
> >>> I still haven't fixed my problem whereby RH7.2 is only 
> >>> detected 128MB of RAM instead of 192MB I have installed.
> >>>
> >>> >From what I have read this is not uncommon for Linux not
> >>> to detect all physical memory, and sometimes this is 
> >>> resolved by passing the 'mem' parm to the kernel on bootup. 
> >>> But all the documentation I've seen is for lilo, but I'm 
> >>> using grub. So how do I do this with grub?
> >>> 
> >>> 
> >>> 
> >>> --
> >>> darryl e-securenetworks net
> >>> 
> >>> JWebMail WebMail/Java v0.7.8, built with JDK 1.4.0-beta3 
> >>> WWW to Mail Gateway
> >>>
> _______________________________________________
> enigma-list mailing list
> enigma-list redhat com
> https://listman.redhat.com/mailman/listinfo/enigma-list

--
darryl e-securenetworks net

JWebMail WebMail/Java v0.7.8, built with JDK 1.4.0-beta3 WWW to Mail Gateway