Mac Pro 6 core RAM upgrade

[krankyone]

I currently have 6 G of RAM (2x2x2) in my hexacore 3.33G machine.

I'm planning to upgrade my RAM to 12G by getting two 4G sticks from OWC. I'm planning to put the 4G chips in slots 1 and 2, and moving the 2G sticks into slot 3 and 4.

Will this cause any problems?

Thanks in advance.

 

[beto2k7]

someone more educated on this can confirm or deny but I think you will lose triple channel when going from 3 sticks to 4 sticks.

 

[philipma1957]

it may like it or not depends on if the sticks have quad tiering I have links give me a minute.



http://www.superbiiz.com/detail.php...-1333-16GB-ECC-REG-Samsung-Chip-Server-Memory


this stick is quad rank type.

four of these will change bandwidth so if you buy sticks that use quad rank your bandwidth will drop.


http://www.superbiiz.com/detail.php...8GB-512Mx4-ECC-REG-Samsung-Chip-Server-Memory


another quad rank



another solution is use this link



http://www.superbiiz.com/detail.php...-DDR3-1333-4GB-256x8-ECC-Micron-Server-Memory


buy 3 of these for 144



http://eshop.macsales.com/item/Other World Computing/1333D3W4M08K/


owc wants 148 for 2 4gb sticks


I have used superbiiz ram for years it has been good for me. I would prefer 3 matched sticks of 4gb = 12gb myself . In fact that is what is in my mp

 

[krankyone]

I had my two 4G sticks overnighted to me from OWC. Installed with no problems in a 4-4-2-2 configuration using the 2G sticks I already had. The memory configuring utility came up happy. Techtool Pro came up happy with memory testing. Geekbench shows no real performance difference, but Xbench showed some real decrease in performance (14% decrease) over the 2-2-2 configuration.

Not sure what to make of this. I could leave it the way it is, or I could get one more 4G stick.

 

[DeeEss]

I had my two 4G sticks overnighted to me from OWC. Installed with no problems in a 4-4-2-2 configuration using the 2G sticks I already had. The memory configuring utility came up happy. Techtool Pro came up happy with memory testing. Geekbench shows no real performance difference, but Xbench showed some real decrease in performance (14% decrease) over the 2-2-2 configuration.

Not sure what to make of this. I could leave it the way it is, or I could get one more 4G stick.

Hmmm...my understanding is it needs to be triple channel. I have 3x 8 and one empty slot. I think if I put a spare 3GB that I have in it will slow it down so I left it out. I thikn the way it works is if the system needs the RAM then it will be better off for it, but otherwise in cases where it uses less ram then it will be slower. Which is what it seems your test have showed.

I would phone or instant chat with OWC (on their site), they are pretty good with that kind of thing and know what they are talking about and might offer you some kind of solution or advise if it's best to buy another 4GB stick etc.

It's probably because it's not triple channel, but just wondering is it all 1333MHz ram?

 

[krankyone]

Yeah, everything is 1333. I did a little search on Xbench and decided to not pay too much attention to these results.

Although Apple recommends that you use sets of three matched chips for optimal performance (the triple channel thing), they state you can mix chip sizes. In this case I'm running in dual channel mode using slots 1/2 and 3/4.

So going 4-4-2-2 instead of 4-4-4 will decrease my RAM performance, it's not clear by how much. Some people here have stated it's negligible, maybe around 2%. But so far no kernal panics or smoke coming from my MP, and my Geekbench score nudged up a bit: a rocking 15578.

 

[krankyone]

Also talked to tech support at OWC. He reiterated that the optimal solution is to go triple channel in groups of three, but there would be no problem running in my current configuration besides losing the triple channel advantage over dual channel. The support guy was knowledgeable but he couldn't say what the difference would be between triple and dual channel operation performance wise.

 

[Riot Nrrrd]

Also talked to tech support at OWC. He reiterated that the optimal solution is to go triple channel in groups of three, but there would be no problem running in my current configuration besides losing the triple channel advantage over dual channel. The support guy was knowledgeable but he couldn't say what the difference would be between triple and dual channel operation performance wise.

I just bought a Mac Pro like yours (single 6-core 3.33 GHz). It came with 8 GB (2-2-2-2).

So due to this "triple channel" effect I should either remove the 4th stick and go down to 6 GB/2-2-2 or replace them all with 12 GB/4-4-4 or 24 GB/8-8-8?

 

[DeeEss]

I just bought a Mac Pro like yours (single 6-core 3.33 GHz). It came with 8 GB (2-2-2-2).

So due to this "triple channel" effect I should either remove the 4th stick and go down to 6 GB/2-2-2 or replace them all with 12 GB/4-4-4 or 24 GB/8-8-8?

Triple Channel works best. Having 4 sticks makes it, from what I hear, 2% slower. You probably won't even notice that. And it won't make it slower if you actually need the 8GB of RAM.

 

[toxic]

the performance benefit of triple channel is hardly worth thinking about. just buy whatever RAM fits your needs and budget.

 

[DeeEss]

the performance benefit of triple channel is hardly worth thinking about. just buy whatever RAM fits your needs and budget.

The thing is though with a 14% reduction in performance (like the OP has stated) then I would say it matters. 5% is apparently where you first start to notice difference. 14% is therefore rather significant.

And while not going with the triple channel is suppose to only be a reduction by 2% or so, there is defo something here not to be ignored.

 

[johnnymg]

The thing is though with a 14% reduction in performance (like the OP has stated) then I would say it matters. 5% is apparently where you first start to notice difference. 14% is therefore rather significant.

And while not going with the triple channel is suppose to only be a reduction by 2% or so, there is defo something here not to be ignored.

The is NOTHING "here" because the there is no significant real world speed difference. You'll be much better off when you stop believing bogus benchmarks.

JohnG

 

[DocNo]

The thing is though with a 14% reduction in performance (like the OP has stated) then I would say it matters.

14% in applications you use or 14% in a synthetic benchmark that has no bearing on reality?

In my experience with the applications I use there was no difference in speed when I had three sticks of memory to get the triple channel mode vs. four sticks to get more overall RAM.

However, their was a SIGNIFICANT difference in having more RAM overall.

Performance is a complex combination of all factors on your machine. Focusing on just one factor, such as artificially limiting yourself at all costs to maintain triple channel mode, may slow you more than if you added more RAM and weren't necessarily optimized for triple channel mode.