[ecking01]

I'm thinking of getting the lowest model mbp if they they go memron in sept and I don't expect apple will bump up the ram on it. So I want to buy a 1gb stick and drop it in giving me 1.5gb ram.

Now the site still says get matched ram even though it has a dedicated gpu. Why? Is there a gain? Because I personally don't think I'll feel a difference between 1.5 and 2 gigs of ram and I don't want to spend money uselessly.

[mduell]

Yes, there is still a bandwidth difference with matched vs unmatched modules. With integrated graphics the performance differential is more obvious because it can be easily measured with framerate or similar metrics. With dedicated graphics, you'll notice the bandwidth difference in memory intensive applications.

[fdmendez]

whoa, whoa, whoa. let's slow down here. I have 1.5GB of ram (the original 512 and a 1gb stick). The computer did end up performing better than it did with just half a gig of memory.

How exacly does this work whole unmatched and matched thing work and why was it a bad idea for me to do this?

[ecking01]

Yeah what fdmendez did is what I want but his bandwith is less? As in memory bandwith? What's the performance hit he's taking if he was to use matched ram?

And does the matched ram need to be the same manufacturer or just the same size (ie: 512+512)

[ghporter]

The issue is not "amount of RAM" and how it affects performance, but rather HOW it affects performance. You'll get better performance from matched RAM because the matched parts will perform together precisely, while unmatched RAM may cause tiny pauses in performance-not anything you'll notice as a pause in program execution, but they'll add up to a slightly slower computer overall. Slower than one with the same amount of matched RAM, that is.

mduell's point is that when you run something that really stresses the memory (like manipulating a very large image in a graphics program, for instance) you will see somewhat less speed with unmatched RAM than you would with matched RAM. How much less depends on how different the parts are. In most user situations, the difference is technical, and not something you'd notice at all.

[fdmendez]

ah, cheerio

[mduell]

There is definately a quantity vs bandwidth tradeoff, and with the Mac Pros, also quantity vs latency.

1GB+512MB will outperform 2x256MB in the vast majority of apps.

[ecking01]

But do they have to be the same manufacterer or just size?

[RevEvs]

So, say we can use 3Gig in the new MBP (as with the new iMac), you will then have unmatched ram.

so what to go for, 2gig matched? or 3 gig unmatched?

[Tuoder]

Originally Posted by ecking01

But do they have to be the same manufacterer or just size?

Strictly speaking, they don't <I>have</I> to be anything in particular. Size and speed are important. They should be the same size and frequency for maximum performance.

<QUOTE=RevEvs>So, say we can use 3Gig in the new MBP (as with the new iMac), you will then have unmatched ram.

so what to go for, 2gig matched? or 3 gig unmatched?</QUOTE>

3GB unmatched will outperform 2GB in the vast majority of applications, especially considering that those computers have dedicated VRAM.

[ghporter]

Tuoder, not HTML tags, BOARD CODE tags...

[CanadaRAM]

To get Dual Channel Access, according to Apple, the modules need to be the same size, speed and composition. The first 2 are easy, the last one, not so much. Composition refers to the number of chips, the organization of the rows and columns of memory within the chips, and the Serial Presence Detect settings that describe the RAM to the machine.

So although it is possible to have matching RAM with two different brands, the only way to know in advance is to buy 2 of the same brand at the same time.

However: More RAM trumps Matched RAM. a 1.5 Gb setup (512 + 1 Gb) unmatched gives 512 more RAM to applications than a matched 1 Gb pair of 512's. In most cases, the extra RAM will have more benefit than the speed improvement (6 - 8 % I have heard) of matching RAM. Especially with programs running under Rosetta, which struggle with only 1 Gb of total RAM (and less on the MacBook with the shared graphic RAM)

Rev: I have never seen verification that 2 Gb RAM modules will work in the first generation intel machines. We have tried, and it was not recognized.

Thanks
Trevor
CanadaRAM.com

[phazedowt]

If I remember from a barefeats.com benchmark about this (I can't remember where the actual article was, but I'm sure some googling will bring it up), the improvement for matched SO-DIMMS on the MBP was around 3% for memory bandwidth intensive apps. That being said, I think the advantage of dual-channel DDR for the intel core machines is only apparent for ones using integrated graphics... which is why Apple insists on only selling memory in matched pairs for the mac mini and macbook, but not for the imac or MBP.

[Eug]

I have 2 GB only in my 24" iMac. I think I'll stick with that for now. If I find I'm paging out to disk I'll consider getting 3 GB in a few months. 2 GB DIMMs are now coming down in price it seems. Last week they were insanely expensive. This week they may actually be affordable if Buy.com is to be believed.

I suspect I'll be happy with 2 GB though, even using the iMac with Aperture, since I'm not usually a heavy user of that app.

BTW, this is the MacBook test, which is highly dependent on my system memory speeds:

DO MATCHED PAIRS OF MEMORY HELP?
We've done some matched versus unmatched memory tests on the Intel Mac mini (which has virtually indentical architecture to the MacBook 13"). In the case of the Intel Mac mini, Quake 3 saw a 51% gain from matched memory pairs while Doom 3 saw an 18% gain. As for Productivity Apps, matched pairs gave iMovie HD a 3% boost and Photoshop CS a 4% boost.

It's a completely different kettle of fish for systems with discrete GPUs and dedicated graphics memory.

EDIT:

Hmmm... The description for the MacBook Pro MUST be an error:

MacBook 13" versus MacBook Pro 15/17

DO MATCHED PAIRS OF MEMORY HELP?
We've done some matched versus unmatched memory tests on the Intel Mac mini (which has virtually indentical architecture to the MacBook 13"). In the case of the Intel Mac mini, Quake 3 saw a 51% gain from matched memory pairs while Doom 3 saw an 18% gain. As for Productivity Apps, matched pairs gave iMovie HD a 3% boost and Photoshop CS a 4% boost.

MacBook Pro 2.0 versus MacBook Pro 2.16

DO MATCHING MEMORY PAIRS MAKE IT GO FASTER?
We tested the MacBook Pro with both matching pairs of 1GB SDRAM and non-matching pairs (one 1GB and one 256MB module). There were gains: 51% in Quake 3, 18% in Doom 3, 2.5% in iMovie, and 3.5% in Photoshop CS.

[Steve Bosell]

Is there any way to see if you are in dual channel mode?