This may have been answered previously or is a little off topic, but is it reasonable to expect the bump up to 1333 RAM in the next Mac Pro?
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
How is RAM frequency controlled in 2009 Mac Pros with QPI?
- Thread starter gugucom
- Start date
- Sort by reaction score
No idea, really. Hopefully Yes, but nothing to confirm that, from what I've seen.
I'd base it on the possibility that Apple can get a good enough deal to make it viable for their accountant dept. to approve it, and it would also be a way for the marketing dept. to sell it - better memory throughputs. It should be possible, but will be up to the financial side of the equation when worked backwards from MSRP minus margins if it falls within the manufacturing cost per unit.
I think gugucom's going to try 1333 in his system, so wait a bit, and check back for the results.
Hopefully, it will be sucessful, and you can get the 1333 parts if you wish, and not have to worry about returns due to non operation at the expected speed (bad sticks are always a possibility, but statistically lower than other components).
Current information is that Gulftown Xeons only support 1066MHz memory officially, so no.
I'd expect it to follow the current Nehalem parts though. 1066 on the lower clocked parts, 1333 on the upper most part group. No details yet, but I don't see why they can't, as it's already designed in the IMC, and thermals won't be a real issue either, as it's going on a smaller die.
I'm expecting delivery during the next couple of days. It isn't a big deal to fit the RAM so one hour later I wiil post the result of the test.
Kingston is typically decent memory, and I've never had problems with it. I do recall an occasional RMA being mentioned here and there, but they do seem to support their products well.
The test failed. OS X recognizes the 1333 MHz RAM with 1066 MHz only.
Lavalys Everest Home in Win7 cannot read out SPD. I guess that means that bloody Apple have hardwired the RAM frequency in the EFI. I cannot even get the additional bandwidth in Windows then unless I find a way to tweak it.
On the upside Win7 tells me that the pair of W5590s has two Turbo Boost cores capable of running at 3,43 GHz. I wasn't aware of that.
Lavalys Everest Home in Win7 cannot read out SPD. I guess that means that bloody Apple have hardwired the RAM frequency in the EFI. I cannot even get the additional bandwidth in Windows then unless I find a way to tweak it.
On the upside Win7 tells me that the pair of W5590s has two Turbo Boost cores capable of running at 3,43 GHz. I wasn't aware of that.
That blows.
Thanks again, 
. Thanks for giving it a shot though, as at least we have a definite answer.
So they just set a value in the firmware rather than use SPD.
Bastards. At least you've not totally given up. I wish you luck with it.
A nice little surprise then, and as Eidorian posted, a single core can go a tad higher.
Rather nice, but somehow seems wrong to have an 8 core system only running on one...
I will run 4 of the new and four of the old RAM sticks for a capacity of twelve GB. Two of the new sticks I will return for a refund and two of the old sticks I will sell on ebay. So in the end pushing my system from 6 to 12 GB will cost me 140 which isn't all bad as a result.
Unfortunately there is little hope that Apple will do an EFI update for SPD enabling, a real bummer.
Unfortunately there is little hope that Apple will do an EFI update for SPD enabling, a real bummer.
At least you've gotten the capacity up to where you wanted, without paying through the nose.
Sucks that it's stuck at 1066 (for all 1333 under for testing). Obviously in SPD capable systems, mixing would default to the lowest setting (though some systems <general> only use one of the DIMM's per channel to determine the SPD settings). Not an issue in this case though.
Dumb
I see no reason why Apple set's artificial caps like this.
Thank you for doing the test.
Hmm... I'm curious why Everest can't read the SPD? Could that mean there's a problem with the SPD programming on those DIMMs? Is the firmware having the same difficulty and defaulting to 1066? Or is Everest junk or having difficulty with the Apple firmware? I would get to the bottom of this before throwing in the towel.
Have you tried other Win7 utilities? CPUZ?
OCZ also has an SPD programming utility... SPD-Z that might work as another way to verify the SPD.
Yep, I tried CPU-Z and it can't read the SPD either. It shows the DRAM at 532 MHz.
I havn't tried SPD-Z yet. But there seems to be little point in doing this if the firmware blocks everything. Both OS X and Win7 cannot set anything but 1066 which to me indicates that EFI has blocked the multiplicator at 8 and will not give me 10. EFI steals 20% of my bandwidth! Piece of sh°t firmware once again.
Hmm... that does suck. Perhaps CPU-Z and Everest use BIOS to access the SPD tables which would explain why they don't work on the Mac.
Dumb
I see no reason why Apple set's artificial caps like this.
Thank you for doing the test.
I think I see a reason. Apple has decided for whatever reason to use 1066. They therefore locked at this speed so that it's even less incentive for people to upgrade on their own, rather than buy apples own ram, or upgrade upon purchase.
They might not think the ludicrous ram upgrade overpricing is not enough to dissuade buyers.
BTW for another thread I had to do some reading on RAM technology and the 5500 Intel CPU design.
The design is flexible for max capacity of 192 GB of RAM at 800 MHz or max bandwidth with 48 GB of RAM at 1333 GHz. Other manufacturers (Sun) have managed to run two slots on one channel at 1333 GHZ. At 1066 the Intel design can run 96 GB which isn't possible in the Mac Pro because it lacks the double RAM slots per memory channel to do this. Intel devide their CPU SKUs into three classes, of which the highest W and X denominations are the only ones able to run the 1333 MHz DIMMs. The very low denominations can only run 800 MHz. All of Apples CPU can at least run 1066 MHz.
ECC RDIMMs are available at (16), 8 and 4 GB density. The 16 GB density is probably still under development. The other sizes are commercially available but expensive. The bigger modules are not available in 1366 MHz.
ECC UDIMMs are competitively priced available at 1, 2 and 4 GB density. All speeds of 800, 1066 and 1333 are available. You cannot mix UDIMMs and RDIMMs.
The design is flexible for max capacity of 192 GB of RAM at 800 MHz or max bandwidth with 48 GB of RAM at 1333 GHz. Other manufacturers (Sun) have managed to run two slots on one channel at 1333 GHZ. At 1066 the Intel design can run 96 GB which isn't possible in the Mac Pro because it lacks the double RAM slots per memory channel to do this. Intel devide their CPU SKUs into three classes, of which the highest W and X denominations are the only ones able to run the 1333 MHz DIMMs. The very low denominations can only run 800 MHz. All of Apples CPU can at least run 1066 MHz.
ECC RDIMMs are available at (16), 8 and 4 GB density. The 16 GB density is probably still under development. The other sizes are commercially available but expensive. The bigger modules are not available in 1366 MHz.
ECC UDIMMs are competitively priced available at 1, 2 and 4 GB density. All speeds of 800, 1066 and 1333 are available. You cannot mix UDIMMs and RDIMMs.
The 16GB versions have been announced some time ago, but still haven't shipped. The 8GB densities are out, but expensive. (~$760 - 800USD per).
So 32GB in a base Quad is as much as actual system.
The Octads definitely make more sense for that level of memory capacity, even if the additional cores aren't really needed.
As discussion in another thread evolved graphic designers using Photoshop and the forthcoming Photoshop-64 will need 32-64 GB RAM for optimum speed of operation. In the octad such RAM capacities will be more cost effective than in the quad because you can use twice the number of RAM slots or half the RAM stick density. High density is the premium cost factor. As CPU speed isn't so important for Photoshop the advise for graphic designers is to go for a slower octad rather than a faster quad and put the money into as much RAM as they can afford.
This being the case, when do you think the boys in Cupertino will realize that they need to spike the RAM clock (more than 333 MHz) to make their actual Mac users happy
The last information I spotted didn't mention 1600MHz capability (lots of info was still missing; specifically the actual P/N's, and related clocks and pricing), but I'm not surprised (gives the new parts another edge). It's less of an issue these days for most SP systems, compared to previous architectures, as the boards can already OC quite well (well over 1600MHz for some). DP is another story, and quite important. Unless we finally get an enthusiast DP board, but as it's not out yet (or even an announcement - previous units planned where dropped
), I've a feeling it won't.I'd expect a similar memory speed breakdown over the part line for Gulftowns as it is now (i.e. bottom parts at x, mid can go to y, and the top end z; where x < y <z). Current parts offer 800, 1066, and 1333MHz over the models. Just sub in larger values in the Gulftowns.
Could you elaborate on the source of this info? I thought that 5600 was only a die shrink on Nehalem + adding 2 cores. There were not supposed to be other architectural changes. To run a faster IMC sounds like a sensible idea but it isn't in line with the previously communicated features, unless they simply held back on the 5500 to be able to top it.
Edit: Found it with Google
http://www.atomicmpc.com.au/Feature/159549,cpu-and-gpu-now-the-convergence-goes-on.aspx
So the W5690 is supposed to top out at 3,66 GHz without turbo boost.
I'm looking forward to see what Apple will do to castrate this new chip. Continue four DIMMs/socket and keep locking the multiplier at 8 instead of 12?