[jchen]

I just ordered my MacBook two days ago - in preparation, I also need to purchase additional RAM. I've seen so far that the new Macbooks (MBP included) seem to be particularly picky when it comes to accepting RAM - I was wondering what people thought of transintl.com? I remember seeing a promo they had on xlr8yourmac for 1GB PC5300 SO-DIMMs - $99. Also, I just want to confirm this but the new Macbooks do require same size pairing, correct?

[icruise]

I ordered a hard drive upgrade from transintl.com about 5 years ago ($498 for 48GB!). I don't think I had any problems. Haven't ordered RAM from them. I've had good luck with OWC RAM, although it's a bit more expensive.

The MacBooks don't *require* the same size pairing, but there is apparently a noticeable drop in performance if you don't use it.

[uicandrew]

Originally Posted by Icruise

I ordered a hard drive upgrade from transintl.com about 5 years ago ($498 for 48GB!). I don't think I had any problems. Haven't ordered RAM from them. I've had good luck with OWC RAM, although it's a bit more expensive.

The MacBooks don't *require* the same size pairing, but there is apparently a noticeable drop in performance if you don't use it.

i have a friend who is very snobby about ram and he's going to order his mbp ram from transintl for the following reasons:

1. they are a mac "specialist" of sorts
2. price (the 1gb stick was somewhere around $105-110

in terms of pairing them, i don't think you'll notice it at all. i'm currently running my imac core duo with 2gb of ram. 1 stick was factory installed by apple. the 2nd stick is a 1gb pc 4200 533mhz made by corsair. the 2nd stick meets the specs for the last revision of the powerbooks (the hi-res ones) and i haven't had any problems. and performance hasn't been an issue.

barefeats report that there is roughly a 10% gain in performance for getting completely matched ram. but since the intel core duo chips are so powerful, i've never used more than 1.3gb of ram (according to istatpro - i heavily recommend that widget for everyone) but then again, i'm using the ilife suite, and other universal binary programs.

[icruise]

First of all, we were talking about pairing different sizes of RAM, not different brands. And the Mac mini and MacBook are supposed to suffer a much greater speed hit from having unpaired RAM because of their integrated graphics. Your experiences with your iMac don't really apply.

[uicandrew]

Originally Posted by Icruise

First of all, we were talking about pairing different sizes of RAM, not different brands. And the Mac mini and MacBook are supposed to suffer a much greater speed hit from having unpaired RAM because of their integrated graphics. Your experiences with your iMac don't really apply.

i'm guessing your last sentence sounds harsher than you probably intended, but oh well.

in order to take advantage of the dual channel and the extra 10%, the ram specs must be identical as well as the same size (pc4200 vs pc5300 and/or 533mhz vs 667mhz)

the fact that they are different sizes is analogous to having different speeds in the sense that if either of them don't match, you don't get the 10% speed boost.

want "proof"?

http://guides.macrumors.com/Matched_RAM_on_Intel_Macs
www.barefeats.com (under quick takes)

my experience is relevant.

edit: i realize that the imac has dedicated vram, and that the macbook has shared ram. intel's website shows that the macbook video card can take up to 250mb of the system's memory. so i understand that all things being equal, the macbook needs more ram. i'm just saying that it is better to get more ram (1.5gb of unmatched ram) than getting identical ram.

[icruise]

Sorry, I really didn't intend to sound curt.

But I still think that you're not taking into account the differences between the iMac and the other models with integrated graphics.

[Eug Wanker]

The iMac is not a good example for the MacBook, for the reasons Icruise stated. Here is real-world proof:



First line (red): Core Duo 1.66 GHz Mac mini with 2 x 1024 MB
2nd line (grey): Core Duo 1.66 GHz Mac mini with 1 x 256 MB and 1 x 1024 MB

ie. At least for this particular game, matching RAM gives a 50% performance boost.

[azt33]

Eug, I don't think that only matching RAM gives you the perfomance boost, but also the amount of RAM you have. I think it is hard to do a comparison like that, unless you have the same amount of RAM asymmetrically aligned compared to the same amount of RAM, matched. But that is very difficult, if not impossible to do.

[icruise]

Yes, the more relevant comparison would have been to use 2x 512MB sticks versus a 1GB stick + 256MB stick. That way the paired case would have less memory. If it still won out, you could tell for sure that it was a result of the pairing.

Still, I kind of doubt that Quake 3 really uses 2GB of RAM.

[Eug Wanker]

azt33, the amount of RAM is irrelevant here. 1280 MB vs 2048 MB makes absolutely no difference because Quake 3 doesn't use much RAM, and 1280 is already way more than enough to make RAM a non-issue (unless there were 10 other apps open or something). This isn't Photoshop we're talking here. (You might have been right if it had been 256 MB vs 512 MB, but that's not the case.)

The speed difference is simply due to GPU speed differences, which is in turn dependent upon memory speed differences. That's no surprise, since with single-channel system RAM the graphics memory bandwidth is only 5.3 GB/s (hence the name PC2-5300). With dual-channel system RAM, the graphics memory bandwidth is a theoretical 10.7 GB/s.

This video memory bandwidth difference is not seen on systems which have dedicated video RAM, like the iMac. On the iMac, dual-channel system RAM has no bearing on GPU memory bandwidth. It will still make a bit of difference on other things, but usually not enough to worry about in most usage.

That's why the iMac is basically useless as a gauge of how well paired vs. non-paired RAM will perform on the Mac mini or the MacBook.