Page 1 of 1

2 RAM options : 1 Gb PC2-5300 or 1.2 Gb PC2-4200 ?

Posted: Wed Nov 18, 2009 4:48 pm
by pierre04
Hi,

I just bought a X60 (with one stick of 1 GB PC2-5300 5-5-5-12)

A friend gave me a spare 256 Mb stick of PC2-4200 4-4-4-11

I have to decide what is better for me : Keep my actual config, or plug the 256 Mb and have my whole system run at lower ram speed, but with more RAM. (will it work as dual channel ?)

I mainly use this laptop for websurfing (firefox), word processing, spreadsheets (OOo) et sometimes do some photo editing ~10 Mpx each file (gimp). I use Debian Linux (stable)

I will soon do a few memtest session to have some numbers, but I already know they do not reflect every-day use.
I might try to measure some kernel compilation time also

Do you think of any other test to have real world numbers ?

Re: 2 RAM options : 1 Gb PC2-5300 or 1.2 Gb PC2-4200 ?

Posted: Wed Nov 18, 2009 6:06 pm
by RealBlackStuff
AFAIK you machine could run with 2x 2GB PC5300, although under Windows 32-bit it only recognizes 3-3.2 GB.
The extra 256MB doesn't bring much, certainly not if it brings the bus-speed down.
Dual Channel only works when 2 equal RAM modules are in it.

Re: 2 RAM options : 1 Gb PC2-5300 or 1.2 Gb PC2-4200 ?

Posted: Thu Nov 19, 2009 7:35 am
by pierre04
I know I could buy some RAM for 30 € or so, but 1Gb is enough for me.

Here are some memtest results.

1 GB
Bandwidth = 1985 MB/s
Clock speed : 609 Mhz (DDR 1219) @ 5-5-5-15
FSB : 304 Mhz - Single chanel 64bits
1DB + 256 MB
Bandwidth = 1881 MB/s
Clock speed : 508 Mhz (DDR 1016) @ 4-4-4-12
FSB : 304 Mhz - Dual chanel asymetric
I plan to compile a kernel before this week-end.

Re: 2 RAM options : 1 Gb PC2-5300 or 1.2 Gb PC2-4200 ?

Posted: Fri Nov 20, 2009 6:05 pm
by pierre04
After some linux kernel compilations I can say the time is exactly the same !

I also ran some benchmark, such as Fibonacci, MD5, SHA1, Blowfish or raytracing and the difference is less than 1%

Looks like there is no big difference, so I will probably keep 1.2 gb to avoid some swapping to occur.

Any other benchmark ideas ?