The MBR can easily be replaced, especially if you have another computer to access the internet.  I would guess that the memory was fried by the same problem.  As some others have said - check them in a memory checker at the local computer store if you can, other wise you may break another computer if you install broken RAM to check it, if it's a junker then don't worry.  But I usually start with the power and move from there.  You would be really surprised at the number of things that originate from either outright power filtering issues, or low voltage issues.  And as they have said - once the memory has issues you start writing bad data sectors back to the disk and it cascades from there.

Alright, so. Sure. If the RAM is showing a compatibility issue you could have both showing as bad due to a non-RAM related issue.  Both could have failed previously but not shown a problem until just now. Or they could've baked electrically at the same time.

If you've proved that both sticks of RAM are bad it's not just reasonable for that to have happened, it's 100% probable.

Before it comes up. Other problems? They're just that, other problems.  But if you're running data through bad sectors of RAM, corrupting it, then overwriting your disk with it?  Sure, could do anything.

if you vad a short circuit, then suspect that your memory controller on the motherboard is broken.

Possibly a address bus driver line.

Test your 8gm memory modules in another computer.

 I got 3 almost identical laptops so I tested the memories: the 8GB modules don't work anywhere now, after more than a year of flawless operation. The original 2GB work nicely everywhere.

I need to find a way to dump my modules' EEPROM so I can compare if it ever happens again.

Sounds like some kind of power issue, something is out of spec killing the parts. Are you sure the test computer gain use 8GB sticks? Most aftermarket memory companies have a lifetime warranty on DRAM.

The computer has been working for more than 1 year and half with 16GB onboard (upgraded from 4GB).
Last week the computer behaved strangely with firefox using 300% of the CPU (3 of the 4 CPU) [hence my inability to load the hacker channel], that I tried to isolate down to a .XPI but after a few days it worked again.
Then it refused to come out of sleep mode.
Then it refused to boot.

Hi Yann, unfortunately, it seems your motherboard is dying (something fishy with your master bus , I believe).

I compared with the spare computer, the 8GB sticks don't work anymore anywhere, but the 2GB work everywhere. I ran many memtest on the old sticks without problem.

MemTest needs to run overnight to give very reliable results. What I find very strange is your HD partition issues... I've seen some cascading failures, but I've never seen a computer still working afterwards. Cascading failures are often due to a motherboard or power supply failure.
Edit : only time I've seen something like this was with a old IBM thinkpad with bad RAM sockets soldering, the RAM was failing when the computer was hot, and after many crashes, it finally triggered HD errors.
Before formatting the HD, try to repair the partition table using fdisk or fsck. I'm almost sure your data is still there.

The RAM is dead, the disk data is dead.

Now tell me I'm paranoid !

My hard disk seems affected :

[root@acer /]# fdisk /dev/sdb
AVERTISSEMENT: données surperflues ignorées dans la table de partition 5
AVERTISSEMENT: données surperflues ignorées dans la table de partition 5
AVERTISSEMENT: données surperflues ignorées dans la table de partition 5
AVERTISSEMENT: fanion 0x6169 non valide dans la table de partitions 5, sera corrigé par w(écriture)

Disque /dev/sdb: 1000.2 Go, 1000204886016 octets
255 têtes, 63 secteurs/piste, 15200 cylindres
Unités = cylindres de 16065 * 4096 = 65802240 octets
Identifiant de disque : 0x000c39d0

Périphérique Amorce  Début        Fin      Blocs     Id  Système
/dev/sdb1   *           1          13      819200   83  Linux
/dev/sdb2              13         561    35176448   82  Linux swap / Solaris
/dev/sdb3             561       14249   879607808   83  Linux
/dev/sdb4           14249      121602  2603521720    5  Etendue
/dev/sdb5   ?      133704      242317  2684476820   39  Plan 9  <= ?????

Malicious activity becomes more probable now.

Are they from the same batch?

Yes.
Now, explain why the partition table is altered (my data partition is type "plan9" instead of ext)

I'm accessing the 'net with another cranky backup computer. I'm pretty miffed.

I've seen whole banks of RAM go in servers before. It's rare, but it does happen. The partition table could have become corrupt from bad writes with bad RAM. Alternatively it could be a faulty chipset on your motherboard causing issues. Are there any weird kernel errors in the log?

Why would the MBR be overwritten ?
I can't mount my partitions from this computer either, I'm **cked.