xgman
macrumors 603
I just don't see it. Recent history has not shown Apple to be on the cutting edge of new motherboard based platforms. I'd love to be surprised though.
Mac Pro update is taking so long becuase...
- Thread starter UltimaLink
- Start date
- Sort by reaction score
I just don't see it. Recent history has not shown Apple to be on the cutting edge of new motherboard based platforms. I'd love to be surprised though.
I was only commenting on the parts availability dictating when LP will show in general, not specifically on Mac systems.
Apple is interested however, as the first demo Intel performed was run on OS X. That means they're already developing OS X for LP simultaneously with the hardware development. The only reason for doing this, is to be able to implement LP quickly (i.e. OS X will be ready at the time an LP equipped system is).
This doesn't mean Apple would be the first system out with LP, but ready when/if they decide to do so (they could still decide to scrap it, as some unknown deficiency/issue may surface before the public is aware of it).
Hellhammer
Moderator emeritus
The quad core versions of Westmere EP aren't exactly cheap. The E56xx line is reasonably priced, being 100-200$ cheaper than Gainestown but there is HUGE gap between X56xx and E56xx series. For example, there is no 2.8GHz or 2.93GHz versions, they are jumping like bunnies. From 2.66GHz to 3.06GHz and from that to 3.46GHz.
Sure that would mean 130MHz bump in clock speed in low-end octo (assuming they keep the same pricing as E5620 costs about the same as E5520) and lower price in 2.66GHz octo but SP hexa core would be cheaper (at least 800$ for two quad core CPUs) and possibly faster due clock speed than any octo system so I don't really buy that octo point.
Why would anyone want two extra core at lower frequency for extra $? Octo would be fairly useless. It would top out at 2.66GHz, having slower clock speed than low-end hexa core would.
My guess is that Apple concentrates on hexa core by giving plenty of CPU options, but gives the option for 12-core for the people who need it. Low-end hexa core will likely be faster than base octo currently is, let alone what the W3680 would be... Then 12-core to start from 2.66GHz and likely to have only two options, that and 2.93GHz.
Sorry for confusing post, I somehow lost my point in the middle of writing so that's the result
What I mean is that octo core would be fairly useless in terms of price and performance (being outperformed by hexa and 12-core being only some hundreds more. The two extra cores aren't that much increase in performance, especially when operating at low clock speed)BTW, are the clocks of W3620/40 known? Just thinking so I didn't talk übercrap here
I assume they are fairly high anyway, start from 2.66GHz?
xgman
macrumors 603
Re: the usb 3.0 thing, this sounds like it might be doable coming up?
"While Intel is still waiting for installing USB 3.0 by default on its motherboard, many manufacturers decided to add an additional chips on their motherboards to offer USB 3.0. In this field, the best component is currently the µPD720200 chips from Renesas, however, its power consumption was too high when computer were idle.
A new version, known as µPD720200A fixes this problem. It drains only 50 mW when idle, a perfect specification for being used on notebooks. So, it is possible to have 2 USB 3.0 ports to share a bandwidth of 5 GBits/s on the same PCI Express 2.0 line
So, even if Intel does not plan to integrate USB 3.0 in its forthcoming mobile architecture, with this chips Apple will have no more excuse not to add USB 3.0 support to our Mac."
"While Intel is still waiting for installing USB 3.0 by default on its motherboard, many manufacturers decided to add an additional chips on their motherboards to offer USB 3.0. In this field, the best component is currently the µPD720200 chips from Renesas, however, its power consumption was too high when computer were idle.
A new version, known as µPD720200A fixes this problem. It drains only 50 mW when idle, a perfect specification for being used on notebooks. So, it is possible to have 2 USB 3.0 ports to share a bandwidth of 5 GBits/s on the same PCI Express 2.0 line
So, even if Intel does not plan to integrate USB 3.0 in its forthcoming mobile architecture, with this chips Apple will have no more excuse not to add USB 3.0 support to our Mac."
Hex Core 56xx:
Xeon X5680 (3.33GHz, 130W, 6.40GT/s QPI): $1,663
Xeon X5670 (2.93GHz, 95W, 6.40GT/s QPI): $1,440
Xeon X5660 (2.80GHz, 95W, 6.40GT/s QPI): $1,219
Xeon X5650 (2.66GHz, 95W, 6.40GT/s QPI): $996
Quad Core 56xx:
Xeon X5677 (four cores, 3.46GHz, 130W, 6.40GT/s QPI): $1,663
Xeon X5667 (four cores, 3.06GHz, 95W, 6.40GT/s QPI): $1,440
Xeon E5640 (four cores, 2.66GHz, 80W, 5.86GT/s QPI): $774
Xeon E5630 (four cores, 2.53GHz, 80W, 5.86GT/s QPI): $551
Xeon E5620 (four cores, 2.40GHz, 80W, 5.86GT/s QPI): $387
Please note, that I didn't list the L series parts, as they're clocked lower, and in the case of the Hex core, it's the same price as the X5650.
As you mentioned, there's gaps in the clock speeds in the Quad core models, but it's to help push the sales of the Hex core parts (look at the pricing from the top down rather than core count). It should make some sense as to why (prevent parts from not being sold, and more importantly, also works out in terms of production capabilities - fewer P/N's means they can better meet production quotas).
As far as "price collision", there's only one instance of it, which is the $1440 mark (2.93GHz Hex and 3.06GHz Quad).
The problem with Dodeca ONLY DP systems, is cost.
Doubling up the processor cost, you'd get:
X5670 = $2880 CPU cost alone
X5660 = $2438
X5650 = $1992
Now lets look at the current system costs:
DP systems
2.26GHz = $746 ($373 * 2), MSRP = $3299. Other costs, including profit = $2553
2.66GHz = $1916 ($958 * 2), MSRP = $4699. Other costs, including profit = $2783
2.93GHz = $2772 ($1386 * 2), MSRP = $5899. Other costs, including profit = $3127.
Just adding in the difference between the existing systems and the cost of the newer CPU's, it's not pretty for the base model @ $4545USD, and in reality, it would be a bit higher, given the drastic cost difference between processors (remember, the margin is applied to the total cost associated with it, and $1992 is noticeably higher than $746. To get a more accurate cost, take the CPU difference, and multiply it by say 0.40, then add that to the $4545.
You get $5043.40. Rounding to something a bit more fitting, you'd get $5049 USD. For a BASE model.
It gets less drastic as you move up in clock speeds (i.e. $6050.20 for the top end X54670 based model, so figure a $6049 price point here; only a $1000 difference at this point), but it should be reason enough to realize that they can't just offer Dodeca's for DP models.It would make more sense to offer a single Hex (SP), then at least one Octad as the base model to meet the price point ($3300 - 3500 target, based on the existing MP's pricing). It would have to be a pair of E5620's, as that's the closest CPU price point that would be in the right range.
The W3680 based SP system should stay at the $3699 USD price point as the current model (W3580), given the CPU's are the same cost. As it's higher, you avoid collision. User's would need to decide which is the better system for their needs between these two, just as is currently the case. A tad murkier IMO though, given the additional pair of cores in the W3680.
I can't locate a source ATM, but that's about right.BTW, are the clocks of W3620/40 known? Just thinking so I didn't talk übercrap here
W3620 = 2.66GHz
W3640 = 2.93GHz
This is what I'd expect, given Intel's recent clocks and number schemes any way (i.e. W3530 = 2.8GHz mark, so xx20 and xx40 would be the respective clock below and above this).
Cost. There's a PCB redesign involved for the main logic board, not just the additional part.
Hellhammer
Moderator emeritus
A wall of text, just like I expected
What would octo offer over hexa core? A simple, though not so accurate calculation:
2.66GHz * 6 = 15.96GHz (~600$)
2.93GHz * 6 = 17,56GHz (~800$)
3.33GHz * 6 = 19.98GHz (999$)
2.4GHz * 8 = 19.2GHz (387$*2 = 774$)
That's about 20% performance increase when all cores can fully be used. Octo would be also about 20% more expensive (~2800$ for low-end hexa - ~3400$ for octo). Well, that somehow fights again my statement that octo is useless
BUT, that's only when all cores are used and as you know, hexa core will blow the head out of the octo in limited thread performance (less than 8 threads). I just don't find the market for it. For couple hundreds more, you can get 3.33GHz hexa which is faster than low-end octo would be. Not to forget higher power draw in dual CPU config (130W vs 80W*2)
I think the price of low-end goes up a bit (~2800$) and includes W3620. For 3200$, you get W3640 and for 3600$ you get W3680. So simply, W3680 would just replace the low-end octo, in price and in performance (if there is no 3.06GHz or 3.2GHz W36xx). Then 12-core for about the same prices as current BTO octos are. It would be too much CPUs for Apple to handle and too many cores. 6, 8 and 12 core, too confusing!
Again, sorry for confusing post, it's getting late here and all these numbers are confusing me
I just can't find market for it and it would hassle things even more. All those prices are just my guesses so feel free to correct, this is just speculation I hope they could keep the low end at 2499$, that would make even more sense to drop the octo IMO. It's not impossible, not at all but is there need for it?
Hellhammer
Moderator emeritus
Westmere EP uses B1 steppings while Gainestown uses D0 steppings so it would require a BIOS update which does not exist. Anyway, you would be paying almost 700$ for 130MHz
Because you're asking questions that need detailed facts just to explain it, let alone just to support anything.A wall of text, just like I expected
That's the real question, isn't it.
For all but SMP that can utilize all the cores, the SP Hex core systems will make more sense, as little software can actually utilize the existing machines. But those that do say rendering the vast majority of the time, would benefit from the Octads, or Dodeca's, and is their biggest target IMO, given the available OS X software. At this point, budgets would be the dictating factor, as to whether or not they choose an Octad or Dodeca machine.
For those earning a living while using their systems for such work, the Dodeca actually makes sense, as they could get additional projects accomplished during a fiscal year = more profit in that year.
But not everyone would be in the position to get one (either the budget is too limited, or the workload is insufficient to justify such a system, so the Octad would be a better choice over a SP Hexa core). One instance where the extra cores do matter. Which is exactly what occured with the Nehalem systems.
Utlimately, it all comes down to cost. Specifically what CPU's they can use to achieve their price points. Apple may actually have 4, 6, 8, and 12 core systems as a result. Yes, it's a bit confusing, but we'll have to wait and see if the MSRP targets have changed, as the information posted previously is based on keeping the MSRP in the same range as they are now, as I expect system buyers won't bite if they're increased by much more, save perhaps corporate buyers that can justify it based on a cost/performance analysis (i.e. it can produce more profit per employee than would be spent on the hardware).
JesterJJZ
macrumors 68030
Hellhammer
Moderator emeritus
But W3680 would be faster than dual E5620 machine, even if all cores of both could be utilized (if my equation above is used though it's not that accurate and I'm too lazy to look at some actual benchmarks
(no HT or Turbo)). Dual E5630 would already be more expensive than W3680 machine and would only provide 1GHz more than W3680 (again, using my equation above). Knowing Apple, they would just say "Well, there is 12-core for you, it's suits you the best!" if you needed speed
Mac Pro hasn't got much love lately so I somehow doubt Apple would start offering huge range of CPUs for Mac Pro, Apple likes to keep things simple, thus I'm sticking with 6-core + 12-core guess 1999$ quad core Mac Pro would be very cool and I of course hope Apple would offer as many options as possible, I just don't believe on it
once again.. the blind leading the blind..
Rumors are rumors.. doesn't mean its gonna happen.. personally, i could give a flying... since I am happy with what I have.. I have no need for westmere at this point.
What i am after however, is copying the microcode over and flashing it to the 2009 firmware.. I am hoping to begin this project shortly, but need assistance with the coding.
Rumors are rumors.. doesn't mean its gonna happen.. personally, i could give a flying... since I am happy with what I have.. I have no need for westmere at this point.
What i am after however, is copying the microcode over and flashing it to the 2009 firmware.. I am hoping to begin this project shortly, but need assistance with the coding.
I am working on this..
I plan to write a flash bios program or better worded to be an efi flash utility which will ultimately flash the efi from the 2010 to the 2009, thus enabling the microcode for B1 stepping processors to run..
This is what makes me want to do this..
I plan to write a flash bios program or better worded to be an efi flash utility which will ultimately flash the efi from the 2010 to the 2009, thus enabling the microcode for B1 stepping processors to run..
This is what makes me want to do this..
Just using your own calculations,But W3680 would be faster than dual E5620 machine, even if all cores of both could be utilized (if my equation above is used though it's not that accurate and I'm too lazy to look at some actual benchmarks
Knowing Apple, they would just say "Well, there is 12-core for you, it's suits you the best!" if you needed speed
1999$ quad core Mac Pro would be very cool and I of course hope Apple would offer as many options as possible, I just don't believe on it
The Octad is close to the W3680 in terms of performance, even though it's 2 fewer cores. But the system should actually be cheaper (look at the actual MSRP, as you have to account for profit margin). Granted, this is only applicable with SMP software that can actually utilize all the cores, not capped (true n cores).
For usage other than this, the Octad won't make as much sense, as fewer cores would benefit from the faster clock speeds (i.e. Photoshop, email, browsing, word processing, spreadsheets,...). Things like Handbrake (capable of n cores IIRC) for example, would need to be analyzed in terms of % of the time spent.
In some cases (i.e. Handbrake or After Effects = more than 50% of the time spent), then more cores could be of benefit, as those systems would shave enough time that additional projects could be taken on during the year, meaning more profit for that year.
Hellhammer
Moderator emeritus
But then you should jump for the 12-core anyway as it's a lot faster than octo is
I guess the time will show us... If Apple could just decrease their ridiculous profits and have reasonably priced BTO options!
BTO pricing
Apple is all about the big bucks.. and I don't see them lowering their stupid, ridiculous bto pricing at all.. remember: the mac pro's are not consumer based machines.. hence why they will not lower the pricing.. unless enough people bitch about it, then maybe.. but i don't see that happening.
Apple is all about the big bucks.. and I don't see them lowering their stupid, ridiculous bto pricing at all.. remember: the mac pro's are not consumer based machines.. hence why they will not lower the pricing.. unless enough people bitch about it, then maybe.. but i don't see that happening.
But then you should jump for the 12-core anyway as it's a lot faster than octo is
I guess the time will show us... If Apple could just decrease their ridiculous profits and have reasonably priced BTO options!
Mac'nCheese
Suspended
You can have as many lunches with her as you want. I think it would be cooler to bang her.
Ideally, I agree (no budget limitations, just based on performance). Unfortunately, budgets are a major issue for most, especially if they're independents or SMB sized entities. And keep in mind, this is a specific type of usage (primarily SMP that is 50% or better of the time spent on the system, and that can utilize n cores, not fixed, even if it's fixed to 8 cores).But then you should jump for the 12-core anyway as it's a lot faster than octo is
The n core aspect may be more of an issue, as some software that can currently utilize 8 physical cores may actually be shown to be fixed (presumed to be n as it currently runs on the current core limit), not truly n cores afterall. Hopefully this won't be the case, but it's possible.
Hellhammer
Moderator emeritus
Neither do I. That was just pathetic hoping. Apple loves the profit. Actually, that's the only thing they care about
True. The octo would be good for ~3300$ segment as W3680 will likely settle for ~3700$. However, IF there will be a Xeon version (W3660?) of the rumored i7-970 (six core @3.2GHz) for $800ish, that would solve the octo issue as it provides the exact same 19.2GHz (again, using my inaccurate calcs
) and would likely go for ~3400$ in Mac Pro. I should have included this in my ealier posts already but I forgot, as usual 
Chilla Frilla
macrumors member
Hey Chilla..
What do you think about all this hype about the 2010 mac pro?? Do you think one will come out? I am putting 100 dollars up on a bet that we won't see an update for a long time, possibly in the NEAR future.
What do you think about all this hype about the 2010 mac pro?? Do you think one will come out? I am putting 100 dollars up on a bet that we won't see an update for a long time, possibly in the NEAR future.
I lol'd.
dgookin
macrumors newbie
how did u do that??? upload a pic?
I;'d really like to know
I;'d really like to know
What a "Mac" might look like:
