Solved POST RAM size changing after CD/RW drive access

September 5, 2015 at 17:46:31
Specs: Mandrake Linux 8.1, Pentium(R) 133/48M to 128M
I have a P5-133 Gateway 2000 machine I am trying to bring back to life. I couldn't find any documentation online for it. The most I could figure out from surfing was that it may have a motherboard with "Thor" Bios in it. I went to the 4allmemory website. They had a listing for Gateway P Series P133-XL that I selected to upgrade the memory. I believe I chose the XL because of the number of expansion slots it had (3 ISA, 4 PCI, and 1 AGP). The machine only had 48MB installed and I was trying to get more. They showed the maximum memory for this system to be 128MB, and that it was to be installed in pairs. I got 4 units from them "32MB 72p EDO SIMM 5v 60ns 2K Refresh Rate" for 128MB total. The RAM arrived separated into pairs, but all 4 appeared to be identical. I installed it, assuming the pair slots are right next to each other, and POST reported 128MB. It shows a string of numbers on the screen that overwrites itself up to the amount of memory it sees installed. It stopped at 128MB, but there was an extra leading zero making me think this mother board may take more. I'm not even sure if this is the original motherboard. I couldn't find any definitive markings on it. Anyway, it recognized the total memory size installed correctly so everything appeared fine. I went to load Mandrake Linux 8.1 on it using the CD/RW drive installed. It gets as far as reformatting the hard drive, then chokes each time. I powered it off and back on and the POST stops only reports up to ~29MB of RAM. If I power of and back on again, it still shows 29MB. If I hit the reset button, the 128MB comes back during POST. But if I let the system run with the Linux install disk in the tray, the POST memory reported drops from 128MB to 29MB right before the Linux install screen is displayed. It doesn't seem to do this if the Linux disk is not in the drive tray. I put the original 48MB of memory back in and don't see this happening. It looks like Linux is installing now. But it reports low memory and doesn't seem to be relying on "RAM disk" to do it. Although it is my first thought, I am not convinced that I received the wrong or bad memory just yet. I'm thinking it might also be the way the CD/RW drive is loading or the way the Linux install is running. Anyone have any ideas?

message edited by ISAmad


See More: POST RAM size changing after CD/RW drive access

Report •

✔ Best Answer
September 19, 2015 at 16:33:10
The memory check on posting is not a sophisticated check. As long as the bios can tell it's there it'll show it. It doesn't test to see if it's any good. For that you'd need to run something like memtest. I don't know why the cdrom spinning up should be a factor but in any event swapping in a good memory stick should fix it.


#1
September 6, 2015 at 00:35:34
No, I've no idea what is causing the problem.

I would abandon this project and just buy a new PC or build your own. I think the Gateway is not worth struggling with.


Report •

#2
September 6, 2015 at 00:55:34
That's odd. I've never heard of that happening either. That gateway wouldn't have come with a cd-rw drive in it. The fact it's sucking up some of the ram when the drive spins up after you power it may be some memory caching feature peculiar to that drive. If possible try another. Most of those old gateways came with really dandy Toshiba cd drives. They seemed to work forever.

You actually 'bought' simm memory? I've got boxes full of that stuff around here somewhere. Lots on ebay too.

Oh another thing, there was only one intel P-I chipset that could cache more than 64 meg. I don't remember which one that was but what it means is that although the motherboard can take 128 meg, the second 64 doesn't make too much difference.


Report •

#3
September 6, 2015 at 05:56:04
I also have a shoe box full of old SIMM memory. I find it hard to believe anyone would invest any money in an old socket 7 system. The Intel chipset DAVE referred to is the Intel 430HX. It would address up to 512MB RAM, but only if the board was fitted with the correct Tag RAM. I doubt your board is based on the HX though, my guess would be the 430FX in which case it might support 128MB RAM, but only 64MB is cache-able.

You mentioned that the board has an AGP slot...are you sure about that? It's awfully old to support AGP. It's possible you're mistaking a pipeline burst cache slot (aka COASt slot) as AGP; they look very similar.

Here's the slot: http://www.devicelog.com/wp-content...

Here's one populated with a COASt module: http://img.tomshardware.com/us/2004...


Report •

Related Solutions

#4
September 6, 2015 at 08:40:59
I think you are right that it is a cache slot. There is a 256K Pipeline Burst Cache KMM764V40GB-15 installed. I thought it was video ram, but there is no onboard video. Seems strange that its placed right next to the last PCI slot, aligned to an opening for an expansion card, and so far away from the RAM and the CPU. I found a picture of it on ebay.

http://www.ebay.com/itm/Gateway-Sys...


Report •

#5
September 19, 2015 at 12:37:02
I removed all the RAM and put just one pair of the upgrade back in the machine. The paired SIMMs were all labeled identically, so I treated them as such and swapped out one SIMM at a time. This narrowed it down to one offending SIMM, which would consistently fail POST. I figured this out without ever seeing it falsely passing. And as such, I never really figured out why POST would report it correctly and then suddenly not just before the screen changed. I still think that it had something to do with the Operating System not being installed at the time. I was trying to load an OS from CD when it happened. Given that I do have a bad SIMM, I suspect it was placing data into RAM where it actually could not. And then POST was updating itself. As I only had one bad SIMM, the offending SIMM would have likely been installed in the 2nd slot of the first pair for it to drop from 128MB to 29MB. 29MB would have been the equivalent to one 32MB SIMM. And since the SIMM's are installed in pairs, really the machine shouldn't have even boot. I tried booting with just one SIMM installed to see what would happen and the machine just beeps loudly and displays nothing. If what I think is true, I wonder how reliable other initial POST memory tests are and if all would have reacted in the same way.

Report •

#6
September 19, 2015 at 16:33:10
✔ Best Answer
The memory check on posting is not a sophisticated check. As long as the bios can tell it's there it'll show it. It doesn't test to see if it's any good. For that you'd need to run something like memtest. I don't know why the cdrom spinning up should be a factor but in any event swapping in a good memory stick should fix it.

Report •

Ask Question