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.
Anandtech Mac Pro Review - Final Section
- Thread starter jroad
- Start date
- Sort by reaction score
Interesting comments they have about the FB-DIMM's, but I really wonder how much real world non-gaming performance is effected. Surely Apple must've considered this type of thing as well when choosing the Woodcrest/FB-DIMM setup?
It's not like they had a choice. The 5000x chipset is based around FB-DIMM. My Dell Precision 690's have it too and the same "limitation" applies to all current 5000x implementations. What could end up an Apple-specific issue is the cooling of the RAM.
Can anyone recommend a good modestly technical review that discusses the system ability to utilize multiple cores compared between OS X, some flavor of Windows (preferably Vista), and optionally anything else?
Specifically, how able the operating system is able to distribute relatively intensive API calls across cores, and how well it is able to manage distributing processes to the cores? That is, everything outside of the individual application that contributes to a machine pratically benefitting from multiple processors?
Specifically, how able the operating system is able to distribute relatively intensive API calls across cores, and how well it is able to manage distributing processes to the cores? That is, everything outside of the individual application that contributes to a machine pratically benefitting from multiple processors?
I liked the way they easily summarised the pro's and con's of FB RAM.
Pro's
Fantastic at doing 50 things at the same time.
Con's
No better (and sometimes worse) at doing only a few things at a time.
Pro's
Fantastic at doing 50 things at the same time.
Con's
No better (and sometimes worse) at doing only a few things at a time.
Of course they did - it's called releasing a dual-core Conroe desktop. A system that would (given the cost and latency of DDR2 vs. FB-DIMMs) serve some of their core constituencies (photography, graphic design, prosumers of varying kinds) better than the quad setup.
This is not a difficult concept.
milozauckerman said:
A Conroe setup would not serve those functions as well as the Xeon setup now.
1: Conroe cannot be placed in a DP setup. So Apple would be taking a step back from its Quad G5 and only have Dual Processors
2: It would not offer the stability that the Xeon setup provides.
3: A speed hit in the RAM area has also been the case with workstations. The workstation more than makes up for that in its stability.
4: A few gaming benchmarks most assuredly does not mean that the Mac Pro is a slouch. Fact is, there arent many apps that can take advantage of more than 2 processors. I mean, look at the Windows benchmarks. There was even a speed hit when running benchmark + quad core in Windows. Once apps are designed to take advantage of multiple cores, the Mac Pro will slaughter any single Conroe setup.
Everyone is whining about 64bit when we havent even managed to properly utilize multiple cores in everyday apps.
So yes, right now the Mac Pro in certain applications offers nothing more than a Conroe, however in the future it will. Those with just two cores will be scrambling to catch up.
Nah, really! I didn't know that!
Uh, you realize that the Quad G5 was the highest of the high-end, right?
Throughout the G5 era there were dual-processor and then dual-core models. I know this will be shocking but there was even, for a time, a SINGLE-CORE SINGLE-PROCESSOR SYSTEM. Don't faint.
But as Anandtech's numbers illustrate, the Quad Xeon doesn't offer any measurable improvement to a dual Conroe setup in many tasks - like the memory-heavy arenas I referred to earlier.
So we just pay more - initially, for the second Xeon and later for FB-DIMM RAM - for no benefit.
Of course it would. Conroe systems are no less stable than any other platform. This is just a stupid argument.
If we want to talk hypotheticals, people are putting FB-DIMMs in their Mac Pros that don't meet Apple's heat requirements because 'official' RAM is too bloody expensive. This may lead to a shorter life for the RAM, perhaps for the computer - or simply a number of ECC errors in the meantime - making the Xeon Mac Pro less stable than a Conroe Mac Pro running cheaper, cooler and more usable DDR2 RAM.
This is also an absurd argument - desktops are stable. Kinda like the G3, G4 and G5 towers used for the last near-decade.
But in any case, the Mac Pro isn't a 'workstation' - it's Apple's only genuine desktop. It can work as a 'workstation' but in the vast majority of consumer situations it's going to be called upon in the exact same way G5s were before, or Dell desktops or any other machine.
No one suggested that it was - however, the FB-DIMMs are crippling in desktop tasks that the Mac Pro will often be used for - games, Aperture, Photoshop - compared to what the system could be otherwise.
You're paying for quad-core Xeon goodness, and paying out the nose for FB-DIMM 'stability' - but getting the performance of a dual-core system.
If I placed an order for a Mac Pro 2.66/4GB tonight for $3000 - it would perform not noticeably better than a Conroe 2.4/2.66/4GB coming in at $1500-2250. That's a major problem.
All of those points, however, are irrelevant, because you don't seem to grasp a basic concept:
Apple can produce a Quad Xeon Mac Pro for two groups (people who need it, Mac fanboys who can't help but slobber on Steve's latest offering) and also produce a dual-core Conroe for everyone else at a reasonable price-point.
While I agree with a majority of your argument, I do need to point out that most graphic designers use more than one app at a time and having two dual-core Xeons provides a designer the ability to run his/her most critical apps all the same time at full speed. I think the question comes down to, do you need this hardware? No, then stop complaining about it. Either you need something or you don't. Complaining that your perfect computer is not being made does not change anything. Apple is one company. They cannot produce a new line of computer for every processor that Intel releases. Also, they have to bid against companies that have the same need and produce a lot more income for Intel. It would be nice to have an offering that would fulfill everyones individual needs but maybe it just isn't economical for Apple or feasible. No one here knows their margins or bottom lines so we have no clue how bringing in a completely new model could effect their sales. There research obviously has not validated the introduction of this "mid-range" model that I keep hearing some wish for. So lets take the models we have, which is a pretty solid line-up, and do what we do best, use them.
Your assumption is that these multiple apps peg the CPU needle at or near 100% each - highly unlikely - and that they're all used at the same time - almost impossible, in order to overload two cores.
No, I don't need FB-DIMMs (75% of Mac Pro buyers don't), and I don't need four cores (ditto). But I have no other options, genius. That's the point of raising the issue - Apple's product matrix skips a great number of people, who are either forced to go to Windows, use a handicapped iMac - or blow $3500 on a reasonable Mac Pro.
Well, actually, they can - but in this case they don't need to. They have the internals and the case already done. No 'new line' needed, any more than the Dual and Quad G5s were 'separate lines' or single and dual-processor G4s before them.
Apple research isn't the most reliable (iBoob iMacs...) and always subject to the Jobs RDF.
milozauckerman said:
So please tell me why someone would buy a Mac Pro over a Conroe setup if it offers nothing more than the Conroe?
People buy workstations because they are indeed MORE STABLE. You call it a stupid argument but you didnt prove why it is. Things like ECC is a major one. THe chipsets are more stable. I am not saying a Conroe isnt stable, but a Xeon setup would prove to be more stable.
Thanks for the history lesson, as if I didnt know there was a Single CPU desktop. Step off the horse man...
I never argued against the basic concept of a Conroe Mac for everyone else. So please step down from your high horse buddy.
Also, I find it quite odd that you can make grand statements as to how apps would perform on a Conroe based on benchmarks. Benchmarks mean nothing. Considering the knowledge you like to flaunt, maybe you would know this.
Let me see a FCP benchmark where a Conroe beats a Mac Pro. How about Aperture. Thats right, it wont. Just because Windows is incapable of handling Quad cores does not mean it useless to those doing actual work on the Mac Pro, like myself.
If you are buying a Mac Pro to play games you would be a fool. Considering Most games cant even utilize 2 cores properly.
Hell the AMD 3700+ was killing everything in gaming performance over the X2 line.
So again, Apple may very well release a Conroe Mac. I am sure it will be well recieved. However it will not offer the performance that the Mac Pro offers in optimized applications(which I discussed in the original post).
It is also foolish to base your judgements on some Windows benchmarks. Heck, on benchmarks in general.
Because Apple doesn't offer a Conroe? Gee. Maybe.
Or if, on the off chance they do need a server or workstation, one is available. Or if they're using programs that benefit from four cores more than they are hampered by latency.
No, certain benchmarking programs mean nothing. Real-world benchmarks - app performance, you know? - are highly meaningful. And real-world benchmarks indicate that there is no Xeon advantage in speed.
Well, gee, since there are no Conroe Apples... do you realize how dumb that argument is?
We do, however, have real-world benchmarks on a 2.16GHz Merom - not even as fast as a Conroe system - that easily rival the Mac Pro - with a slower clock speed and FSB. Between a 2.66 Xeon and 2.16 iMac, you lose 30 seconds of Photoshop speed (which is similar to the gap between the 2.0 Xeon and 2.66) - now, move that to a 1066 FSB and 2.66 clock (and, of course, have a Conroe system cheaper than either the iMac or the Mac Pro), and you've narrowed the gap to within a few seconds, if not less.
A lot of people are - do you read these forums at all? Games are part of the computing experience, and once again this is the only Apple desktop in existence. Between Boot Camp and the X1900XT, Apple knows they're selling to people who will play games.
If you are talking about the Macworld benchmarks, they even note that the benchmarks they ran arent optimized to take advantage of the 4 cores.
Which is all I am arguing. Sure the FB-Dimms cause a drop in performance. The Mac Pro will more than make up for that again in its stability and when more apps are optimized to utilize 4 cores properly.
Sure the iMac may be close, it still gets spanked by the Mac Pro. I also didnt see FCP or Aperture being tested there. I know for a fact that FCP utilizes the Quad Xeon power, I am sure the iMac wouldnt be able to touch it.
If you get your Conroe Mac I am sure you will be happy. However, the Mac Pro will blow it away when the apps are optimized.
Which is all I am arguing. Sure the FB-Dimms cause a drop in performance. The Mac Pro will more than make up for that again in its stability and when more apps are optimized to utilize 4 cores properly.
Sure the iMac may be close, it still gets spanked by the Mac Pro. I also didnt see FCP or Aperture being tested there. I know for a fact that FCP utilizes the Quad Xeon power, I am sure the iMac wouldnt be able to touch it.
If you get your Conroe Mac I am sure you will be happy. However, the Mac Pro will blow it away when the apps are optimized.
Partially true - but a Mac Pro problem nonetheless. If application benchmarks don't 'take advantage of the 4 cores' - it's because the applications (which will be used, presumably) don't take advantage of the 4 cores.
The Mac Pro doesn't 'more than make up for that' - it doesn't quite make up for it at all. But it does so for an extra $1000-1500 (between the basic costs and RAM costs).
You keep banging on about stability - but that's an issue for a very small section of the Mac Pro buying populace (and it's not as if DDR2 RAM is prone to failure at any minute - or else all those G5s were worthless, no?), and you don't even seem to understand what stability means in such a way that you can articulate exactly what it means to the consumer, say, a wedding photographer running Photoshop and Aperture. Or a small magazine startup that needs a few good desktops for production.
Not really - it gets beaten. Not spanked - and doing so with half the cores, half the FSB and 4/5 the clock, lesser graphics, a laptop optical drive, etc..
[EDIT: I should note that on the flip side, this speaks very well of the 24" iMac)
Since we'll be three years into the product cycle and both of our computers will be almost ready for upgrade or replacement by then, c'est la vis!
Ok there Milo, I came into the discussion just to make a couple of basic points. I am a graphic designer and my workflow can require more than what one dual-core can provide, ok chief. If you could relax for a minute you would realize that I said I mostly agreed with you, genius. Instead of spending your time acting like a toughguy on these threads why don't you become more productive so you could maybe then afford a Mac Pro.
suneohair said:It is really great that we may have the potential to upgrade to Clovertown.
Yummy 8-core goodness.
You mean when they dropped in the engineering samples of Clovertown's that they had lying around?
AnandTech article said:
What's with the attitude?
Gee, Milo, it ain't like anyone here is making decisions at Apple.
Would a mid-tower dual-core Conroe setup benefit some Apple users?
Sure. However, there are some things we don't know.
In order to meet pricepoints for the MacPro, they may have had to order in a quantity that meant the entire Pro and Prosumer lineup had to be Dual proc/dual core setups.
As well, you get $300 back on stepping down to 2.0 GHz and another $75 for moving down to the 160GB drive. That makes the Mac Pro only $2124 not including monitor. And that's with 1GB Ram.
Now, for Apple to offer the Prosumer you suggest, they'd have to design a new enclosure, new internals, get Intel to design a different EFI mobo for the enclosure and all the other associated costs.
If I'm Jobs, I stick with this lineup. I mean, remember the days of the Performa days with literally dozens of configurations? Jobs has been clear that there will be ONLY 3. Three levels of Pro. 3 levels of iMac. The mini makes the 3 levels of machine. This way, they get the best deal on the Woodcrests, are upgradeable to the (insert Englishtown here) with umpteen gazillion cores and can apply those economies of scale to lowering production costs, marketing costs, development costs and can then price them accordingly.
And as Schoolhouse Rock sings, "three....is a magic number...yes it is...it's a magic number..."
I understand your frustration. The iMac isn't what you want (although that 24" model is pretty sweet for the price) and you don't want to pay workstation prices for a prosumer machine. I get it. And I think with some creativity, you can get close to your pricepoints with the existing configurations.
The attitude, tho, really isn't necessary. Few here aren't at least somewhat knowledgeable and no one has been rude to you in this thread, nes pas? There's no reason to be rude because people happen to disagree with you (and from what I read, it is only a partial disagreement at that). I think it is possible to disagree without being disagreeable.
Gee, Milo, it ain't like anyone here is making decisions at Apple.
Would a mid-tower dual-core Conroe setup benefit some Apple users?
Sure. However, there are some things we don't know.
In order to meet pricepoints for the MacPro, they may have had to order in a quantity that meant the entire Pro and Prosumer lineup had to be Dual proc/dual core setups.
As well, you get $300 back on stepping down to 2.0 GHz and another $75 for moving down to the 160GB drive. That makes the Mac Pro only $2124 not including monitor. And that's with 1GB Ram.
Now, for Apple to offer the Prosumer you suggest, they'd have to design a new enclosure, new internals, get Intel to design a different EFI mobo for the enclosure and all the other associated costs.
If I'm Jobs, I stick with this lineup. I mean, remember the days of the Performa days with literally dozens of configurations? Jobs has been clear that there will be ONLY 3. Three levels of Pro. 3 levels of iMac. The mini makes the 3 levels of machine. This way, they get the best deal on the Woodcrests, are upgradeable to the (insert Englishtown here) with umpteen gazillion cores and can apply those economies of scale to lowering production costs, marketing costs, development costs and can then price them accordingly.
And as Schoolhouse Rock sings, "three....is a magic number...yes it is...it's a magic number..."
I understand your frustration. The iMac isn't what you want (although that 24" model is pretty sweet for the price) and you don't want to pay workstation prices for a prosumer machine. I get it. And I think with some creativity, you can get close to your pricepoints with the existing configurations.
The attitude, tho, really isn't necessary. Few here aren't at least somewhat knowledgeable and no one has been rude to you in this thread, nes pas? There's no reason to be rude because people happen to disagree with you (and from what I read, it is only a partial disagreement at that). I think it is possible to disagree without being disagreeable.
milozauckerman said:
I didn't say it had anything to do with RAM. Althought the ECC does help achieve stability. Also, generally desktop chipsets are configured more liberally , meaning a little more loose for maximum performance whereas server/workstation chipsets are configured tighter for maximum stability.
This is coming over from my PC experiences. For instance in building an Opteron box vs. a X2 box. The X2 mobo chipset allows for one to tweak however much they want. Whereas on the Opteron chipset (the 100 series opteron can be used with regular chipsets, I am speaking in a chipset like the nForce Professional) you are limited and in some cases you cant do any tweaking. This translates to less crashing, so when you working on that photoshop file the chances of losing it because of a poorly configured (read: less stable) machine are slim on a workstation chipset.
This may not translate to Apple hardware, in which case you would be correct. But with the Xeon and Xeon chipsets being workstation/server grade hardware, I am sure it offers more stability than a Conroe chipset would. I will need to do research on this. I am speaking from my experiences however.
Workstations are more stable than a general desktop box. In addition to what I mention, a workstation used more high quality parts in general, like the power supply to ensure more stable operation. This could verified by testing the rails in the Mac Pro to ensure they are offer stable power to the components.
As with the G5, you can offer dual and quad models in the same enclosure. Is the cheese-grater tower overkill? Absolutely - but it works.
The Mac Pro's motherboard appears to be a stock, or slightly modified, Foxconn part - hardly a great hurdle to overcome, given the wider variety of single-socket/Conroe boards available.
A single-processor, dual-core system is not a big leap for Apple to make even in the same form - I believe it's entirely a marketing decision, with a little Steve thrown in. All Quads! Bestest Macs ever. Look how much cheaper we are compared to Dell X (even though we're also competing against Dell Y - that comes in at $1699).
I'm not bothered by Apple's pricing for myself - the developer discount starts out at $1999. That's perfectly reasonable, just a couple hundred more than I would expect to pay for a comparable mid-tower Mac. As I've said - I fully expect to pay the Apple Tax forever, and I don't mind so much. I save plenty by not bashing my head against the wall using Windows. Sheetrock and ER visits add up. Apple was even smart enough to offer an excellent video card for $200 my price (though the 7300GT stock is still bollocks), a real improvement over the old days.
The memory costs (in terms of dollars and performance), though, are a huge issue, brought about entirely by Apple's decision to skip over the desktop parts bin. Needing $900-1000 to get an adequate amount of RAM for running Aperture and Photoshop is absurd, when those are two of your core audiences.
Well, I agree about the memory prices. Especially since the added heatsink parts only cost at most $10. Charging $100-200 more for a $10 part is absurd.
That said, I refuse to speculate whether the Intel certified memory is good enough for the box. There may be other considerations which justify the need for the added heatsinks (the price gouging notwithstanding). And since I'm not inclined any further to do fine motor work (shaky hands due to seizure disorder), I'll not be adding heatsink parts to the bladed, Intel-certified memory.
For those wishing to take the chance, they can get up to 4GB for under $750. I'm going the 8GB route and with the additional fins, it's $1899 in the 4x2GB configuration. That's a ton of money and it will depreciate madly after 6 months, but I'm hoping to start a business and have to pay for certain guarantees.
I appreciate your take on this. It is similar to the idea of having a Cloverton OctoMac running at 2.4GHz. For most apps, it'll ALWAYS go slower since they won't stress all 8 cores. Not even 4 cores. And by the time those 8 cores would matter, it'd be time to buy a new box...
Still, I have a feeling that much of this was as much economies of scale based to ensure the lowest price (and give Intel a boost in their dual-Xeon numbers) on not Dell volume versus a Jobsian "fastest PC on the planet is a Mac" thing. Essentially, I think it came down to raw numbers. Performance, dev costs, manufacturing costs and keeping the product lines simple. Apple is always gonna have holes in its product line compared to Dell.
Anyway, I hear your points. And appreciate the civil tone.
That said, I refuse to speculate whether the Intel certified memory is good enough for the box. There may be other considerations which justify the need for the added heatsinks (the price gouging notwithstanding). And since I'm not inclined any further to do fine motor work (shaky hands due to seizure disorder), I'll not be adding heatsink parts to the bladed, Intel-certified memory.
For those wishing to take the chance, they can get up to 4GB for under $750. I'm going the 8GB route and with the additional fins, it's $1899 in the 4x2GB configuration. That's a ton of money and it will depreciate madly after 6 months, but I'm hoping to start a business and have to pay for certain guarantees.
I appreciate your take on this. It is similar to the idea of having a Cloverton OctoMac running at 2.4GHz. For most apps, it'll ALWAYS go slower since they won't stress all 8 cores. Not even 4 cores. And by the time those 8 cores would matter, it'd be time to buy a new box...
Still, I have a feeling that much of this was as much economies of scale based to ensure the lowest price (and give Intel a boost in their dual-Xeon numbers) on not Dell volume versus a Jobsian "fastest PC on the planet is a Mac" thing. Essentially, I think it came down to raw numbers. Performance, dev costs, manufacturing costs and keeping the product lines simple. Apple is always gonna have holes in its product line compared to Dell.
Anyway, I hear your points. And appreciate the civil tone.
A MacPro makes as much sense as an 1080p HD TV. By the time you can fully utilize it, it's time to upgrade. That's the problem with being at the absolute bleeding edge.
Yes an xMac makes 110% perfect sense, it's absolutely needed, but with the introduction of the 24" iMac I highly doubt Apple will ever come out with one. They've pretty much closed the retail price gap with it, all we're left with is a huge performance vs cost gap.
I wonder how hard it would be to build a Conroe hackintosh with as many similar parts to what a mac uses as possible...
Yes an xMac makes 110% perfect sense, it's absolutely needed, but with the introduction of the 24" iMac I highly doubt Apple will ever come out with one. They've pretty much closed the retail price gap with it, all we're left with is a huge performance vs cost gap.
I wonder how hard it would be to build a Conroe hackintosh with as many similar parts to what a mac uses as possible...
epitaphic said:
But, if you use FCP, Photoshop (come CS3) and, to a lesser extent maybe, Aperture, then you will utilise it. Video encoding also benefits from every extra core and Mhz you can throw at it.