Mixed memory types in PBG4?

[londonweb]

I've just purchased a 1GB stick of memory with the intention of upgrading my PBG4 (rev e) and just as I was reviewing the order I noticed that the stick I bought is faster than the ones currently installed, so I did a search here to check if it would work or not, and read that RAM has to be installed in identical pairs...? Now I'm quite confused and wondering if someone could clarify (did a Google search also but just got bombarded with cheap memory offers).

Here's my current set up:

2 x 512MB SODIMM PC2-4200S-444 (not sure what the 444 is - I think they're 533mhz)

And my proposed set-up:

1 x 512MB SODIMM PC2-4200S-444
1 x 1GB SODIMM PC2-5300-667Mhz

So, is this going to work, or do I need to buy another 1GB stick to match?


Update: Have just read a couple of things that suggest that matching sizes is not essential in the PBG4, and the performance loss is very small compared to matching. Still can't find anything about the different speeds though.

 

[CanadaRAM]

RAM does NOT have to be in matched pairs in the PowerBook G4 1.67 GHz DL Superdrive / HD screen model -- in fact there is no difference in speed at all between matching and non-matching, because the PBG4s do not implement Dual Channel memory

As far as the 667 MHz module clocking down -- it probably will and you're probably fine, just can't offer an absolute guarantee of it. I would make sure you get RAM that is guaranteed by the vendor to be compatible with your model machine. Choose a reputable seller with a lifetime warranty and a no-cost return (no shipping, no restocking) if it doesn't work.

 

[londonweb]

OK, thanks.

I read a report (sorry, didn't keep the link) that said some cpu-intensive processes came back slower by around 3% when run on a machine with non-matched memory as compared to one with the equivalent amount of matched memory (not sure how they managed to configure it on the pb with only 2 slots though), but the consensus was that the benefits of the increased amount of RAM always far outweighed the deficit caused by non-matching pairs.

As far as the speed goes, I have a 7 day money back guarantee so will take my chances.

 

[CanadaRAM]

OK, thanks.

I read a report (sorry, didn't keep the link) that said some cpu-intensive processes came back slower by around 3% when run on a machine with non-matched memory as compared to one with the equivalent amount of matched memory

Yes, just that report is not applicable to your machine at all coz' it doesn't implement either Dual Channel or 128 bit access.