New Mac Pro - Page 6

Sorry, these numbers are for which chips- the current Mac Pros, or the upcoming release?


To be honest, I wasn't surprised. In my experience, the Radeon XT1900 was a dog from from the start- it ran hot and sounded like a hair dryer, and while nominally an upgrade (the only one available at the time), it wasn't much of an improvement at all. Frankly, it's put me off ATI- if they're the only option available, I'll take it, but I'd rather try something else.


I miss Freehand! What tablet do you use? I don't do much vector work, but I think I might get a tablet to take advantage of the gesture support- anything to take the strain from the wrist.


Yes, this is what I'm curious about- while I'm all set regarding storage for the next few years (I hedged my bets and made sure that all my external drives are both Firewire and USB 3), I wonder which of these will take off, and which will slowly fade away... Faster storage would be nice- it's not bad now, but 6Gb/s SATA would considerably speed up Photoshop.

I'm not sure why that specific generation would put you off ATI/AMD. NVidia wasn't doing any better at the time, in a nut shell the designers where coming up with GPU designs faster than process shrinks could make them practical.

At this point in time though AMD has a significant lead over NVidia when it comes to low power performance. Honestly nobody rejects Intel chips because of the shoddy crap they built a few years ago.

My artistic capabilities never developed! . I can draw pretty nice straight lines with a CAD system though.

I see a future where USB and Thunderbolt survive for a very long time. They really don't compete at this point and likely won't for some time. When you start to see IP stacks for TB ports in ARM based SoC then TB might start to replace USB in midrange implementations. Even then I don't see it as a low end port.

We really don't have a clear picture of Apples vision for TB. However we currently only get one port per machine, that should tell us something right there.

These are processors that follow up what is currently in the mac pro. I can't recall if the current ones are 130 Watts though. It seems kind of high.


Apple has had a number of graphics card problems. The Quadro ones from nvidia have had some complaints too. Currently Apple doesn't offer any non ATI bto options, but I haven't heard of any of the current ones dying. It sucks that you had so many problems. Those were an expensive upgrade at the time. The biggest issue I can think of in the current lineup is that there are a few programs that don't run as well without CUDA, but they're limited and even that will probably change eventually.



I didn't mean the program blah. I meant I like to sketch and paint without doing it on paper first. Rather I prefer to block in rough shapes with colors and and then bring in shading via different layers. Works fine but sometimes you have to do pieces of it at higher resolution then bring them down just because photoshop doesn't handle sub 10 pixel brushes very well.

I still have an intuos3 9x12. The intuos 4's are wide screen formats instead. If you're picky like me, you want a large one, and you want it tapered down in mapping so that pen movement to screen cursor movement is 1:1 or at least somewhere near that. I like it to feel to natural. If your cursor feels too jumpy you're constantly having to rely on what you see on the screen to know when to stop, rather than doing so by feel. I don't rely on having the ability to go back because it's digital. I prefer things be exact so that I don't have to redo the same action several times, or stack slight inaccuracies into the work as I go. If you do design work I'm amazed you're using a mouse. It would drive me crazy. If I absolutely had to I'd pick an ergonomic mouse and disable any cursor accelerations + set the mapping speed low. Mice are just way too slow though.


6Gb/s SATA would be a factor only with the fastest possible SSDs (currently, they keep getting faster) or if you're using them in a raid. In its current version having a lot of ram helps. If it still scratches a lot a small SSD scratch drive would have you running fast. Thunderbolt may soon be another viable option for drive enclosures, especially multi-drive types. Really there are so many emerging IO standards right now that haven't made it into the mac pro yet.

Personally I've done a 180 on processor upgrades. These days the benefits weighed against the cost seem to indicate that waiting for a next generation machine is the smarter move. RAM and other hardware are a different story. The problem with spending money on a processor upgrade is that you mis out on the rest of the architecture improvements.

Yeah the Mac Pro is a raw deal. In fact it is the only machine where the deal gets worst with every hardware revision. It is kinda pathetic on Apples part. Of course that is part of why this thread is becoming as long as it is. The Mac Pro has become a very expensive Yugo.

At the rate that Intel is going you might be retired by the time XEON class Ivy Bridge based CPUs arrive. Seriously Intel has been very slow to push workstation/server hardware forward. This is one of the reasons why I'd love to see a XMac with 65 to 95 watt (maybe even 45 watter) processors. That is a mid range machine that uses parts that Intel actually pays attention to. Many of us simply have no need for the Mac Pros hardware nor price structure.

I've never been that big on upgrading processors. Usually by the time it became a realistic consideration the rest of the machine becomes a significant bottleneck.


It is.... in 2006 the most popular mac pro seemed to be the 2.66ghz woodcrest. It required a dual socket motherboard and 2 $700 processors. It cost $2500 for the machine. Today the machine at that price point uses a much cheaper logic board and a $300 processor from a previous generation. It's beyond stupid that's it's beaten in raw cpu power by the imac. Apple isn't run by idiots. They know this is dumb. It's just a matter of if what they will do about it and when.



The pricing structure there is terrible. Many of the processor/configuration choices are terrible. As you go up the mac pro ladder, each configuration has significant issues that shouldn't be there. For example the 8 core machine uses slower processors than it should at that price point. No user only runs stuff that takes advantage of 8+ cores yet they chose lower clock speed variants when it should have had the manufacturing budget for something better. At this point the six core should be the baseline model, with an updated graphics card (current one debuted on the PC side in september 2009).

Yes, I know it's irrational, but I still get angry just thinking about the pain in the neck that card turned into. It was a pain from the start- not only was it expensive for such a small boost, it took months to arrive. And things went downhill from there...


Well, as I said, I don't do much vector or freehand work, and I guess I've just gotten used to a mouse after so many years. I know what you're talking about, though- in my experience, the best mice have turned out to be high-end gaming mice: they're very sensitive, and can be very precise with the right driver. But there are times when I miss my old pad. I see that Wacom has come out with some new models that support touch gestures- I might have to try one out.


All of these apply to me: I'm planning on an SSD for my boot/app drive, and use a raid for scratch and data. And yep, Photoshop still loves its scratch.


I feel the same way- the last time I upgraded a processor, it was '92 or '93, when everything cost so much that $750 for a Sonnet upgrade made a lot of sense.


I agree. I'm sympathetic to the people who want something else "in the middle"- in fact, I'm probably one of them. Lately it seems to be a choice between paying a small amount for something that doesn't have what you need and paying too much for things you won't use. I don't know if Apple needs to introduce a mid-range line, though: I would probably be happy if there were simply a few more choices when I build my order.

While I would normally agree with this assessment, I think that the single-CPU '09 model is an exception to that rule, which is part of the reason I bought it.

For $300 I can drop in a W3565, which is the exact same processor as in the current $2899 model. No tweaks required, a straightforward upgrade and for less than 2K I am within spitting distance of the current quad upgrade (except for the gpu but that is of minor importance to me).

For $600 and a firmware tweak I get a W3680, which is the cpu used in the current hexacore. So for less than the price of a new base model, I get a machine that is comparable to the op-of-line model. That is a pretty good upgrade in my book.

At those prices, that does seem to be a good investment, and it pokes some holes in Apple's pricing structure as well. Although you're not getting the absolute bleeding edge for your $300-$600, you're getting a very comparable machine at a third of the cost.

I know that if I had such an option, I'd take it and wait another year or so while all these questions get resolved.

On the other hand, these kinds of upgrades are the bread and butter of third party companies- while I know that Apple has a record of making things difficult for them, I think the major factor here is the size of the market.

I thought it over and concluded that Ivy Bridge is gonna be much bigger leap forward than Sandy Bridge. Early next year I'll upgrade the cpu and I should be good until Ivy Bridge is here. If Apple does not offer a new MacPro I'll keep the '09 and get a PC.

What led to that conclusion?

It's really not that irrational. They were asking a premium price for a midrange card (at best) there. When you're paying a premium for what you get at the very least it should be expected to run well. Did you call about it when it first started experiencing issues?


Yes I'll agree with that, but I still prefer intuos pens to mice. The easiest way to test the feel of a device is to write your name and draw shapes without looking at the screen. If you're good at drawing and the device is working well for you, it's not that hard.




The way the line is set up is dysfunctional in far too many ways. I've listed enough of them in this thread. It hasn't really evolved in a good way.

Yes, there's no question that in some situations, a pen is the best way to go. That's why I'm looking into these new midrange Bamboos, hoping to find one that gives me more sensitivity (but not on the level that you need it) and lets me use the new touch gestures as well.


I agree- I sympathize with Mac Pro users who feel neglected, because we have been for a while now. It's clear that Apple's focus has been in other areas (and I can't blame them for that, it's where the money is), and as a result the line seems fragmented and incoherent.

The real shame of it is that it's the Mac Pro users who are most likely to have remained loyal through thick and thin. We're the people who made Mac the standard in certain industries, and while we may represent only a fraction of users, we seem to be taken for granted.

They know these kinds of users are a semi captive audience, and it feels like they've been trying to guide them toward imacs and laptops for years. There are a lot of larger businesses that do purchase a number of mac pros at a given time. Single users that buy them tend to also own a lot of other Apple devices.

I'm completely confident that Apple could do a better job of implementation on a line aimed at being powerful. Even Aperture and Final Cut Pro X haven't scaled well with the power offered by the 12 core machine. Snow Leopard was still criticized for poor scaling with a large number of cores available. They need to actually think about what would be a good successor to the line rather than just using the PC vendor method of throwing components in a box.

On the other side of it, they need to offer options that truly improve performance over a wider range of applications, and consider what they are offering relative to pricing. Right now it goes from quad core i7 imac at $2200 to slower quad core mac pro with no screen at $2500 just for better IO. Think the gpu difference matters? It's starting to more than it did in the past, but it's still a very modest boost in performance relative to other things, and Apple doesn't offer a single gpu option that is strong enough to be a deciding factor in purchasing decisions. I can cite evidence of this for final cut pro x, aperture, photoshop, etc. Many people buy mac pros and gpu upgrades only to be disappointed.

I can't stress this enough but an expensive machine needs to justify its price in real performance. Xbench, geekbench, etc. scores are meaningless (I'm looking at you and your jpeg Marvin).

Semi captive or not all users are impacted by economics. If the competition can beat you with far lower cost hardware then you won't be captive for long. The point here is not that the Mac Pro is a bad machine, but rather it is more machine than a lot of users need.

For things like Aperture the software will likely never scale well across all the cores possible in a Mac Pro. The lack of core usage should not reflect negatively upon the hardware as that will vary based on the specifics of the software package in use.

Well anybody can criticize but they might not know what they are talking about. Snow Leopard actually enabled significant increases in core utilization for software that was properly written for Grand Central Dispatch. Before and after examples abound but some apps like ray tracers really leveraged the cores well. People can blame Snow Leopard if they want but it is rather foolish and likely highlights a misunderstanding of what is possible with parallel processing.

That is about all they can do. When it comes right down to it all they get to work with is a set of parts from Intel. The trick is in building a box that will sell in large numbers, something the Mac Pro has failed to do. That failure to sell is almost directly related to price / performance ratios. To put it simply the Mac Pro is grossly over priced for what a lot of people need and what their software can get out of it.

Stinks doesn't it? Again I'm not knocking the market where a machine like the Mac Pro makes sense. The real problem is that Apple doesn't have a reasonably priced desktop machine. More importantly a machine built with mainstream chip technology that allows it to stay competitive with other Apple hardware.

I'm certain that the Mac Pro will get revved sometime soon when all the hardware is in place. I'm just as sure that the machine will then sit in the line up for another year or more without an upgrade. It is the nature of the beast, modest sales and obscure chipsets result in long production cycles.

This thing with the GPUs is interesting. For one thing Apple is very conservative with their drivers. That is likely a good thing as for the most part Apple machines are very stable.

The other use for GPUs, that is computation on the GPU, is new technology. As such adoption is spotty. When fully utilized though a good GPU can make a significant difference. People really should understand how their software makes use of the GPU before making assumptions, because disappointment is assured otherwise.

Most assuredly! However it is just as meaningless to knock a high performance machine because a specific software package can't use all the cores in that machine. What you have there is a mismatch in capabilities. The Mac Pro is not to blame nor the operating system. Some apps will simply perform better on a machine with fewer cores clocked at a higher speed.

This highlights another reason to support a more mainstream machine, sometimes cores are not the answer. At least when looking at the performance of a single app.

I should have indicated that was directed more at Final Cut Pro X. Capture One and Lightroom seem to scale better in terms of bulk raw file processing than Aperture. They're all completely different code though. The difference is more noticeable when you're dealing with higher resolution files like 30 megapixel + cameras rather than the cheaper ones.


You made me realize I actually don't know the full details of Grand Central Dispatch myself. Ray tracers do love cores, but with rendering it's pretty highly variable how much you gain going to 8, 12, etc.








Yeah... I wish they would spend some time thinking on what might make for a good mainstream machine. There are plenty of areas where people are not fully satisfied with the options that Apple gives them. Your only real alternative in a desktop with any reasonable level of power is the imac, where the amount of power you can obtain is relative to the display size. For me the imac display is a secondary display at best. The glossiness bugs me to no end and the quality control on the display is aimed at a prosumer level at best. It's just not on the level of some other options. I also worry about heat because I run these things quite hard. No matter what computer I'm on, activity monitor always indicates a high processor/ IO usage. I tend to turn off desktop animations close out web browsers and tweak application settings to pull out every last bit of performance. When I buy a new computer it's the same thing as the old one. It just seems to make use of as much power as it can at any given time.


In spite of this they've had some rough cards that have had long complaint lists. X1900XT, Quadro 4000, etc. Many of the upgrade cards have had driver issues and shorter lifespans.


You know whenever a manufacturer talks at all about starting to leverage the gpu, everyone jumps on gpu upgrades at the time of purchase. Really most applications including raster based image programs don't make enough use of them to warrant seriously changing your purchasing direction just for a better gpu. Once you're out of the realm of intel integrated graphics, the differences across the mac line are less severe. Applications that involve heavy rendering (including games) do still seem to benefit from the upgrades offered.


Quad core is basically the norm these days for consumer machines so going below that hardly seems like an option. It's not like those processors are expensive anyway. There are plenty of people who have purchased the mac pro for stability, decent IO, and/or desktop level graphics. These things shouldn't be hard to include. As mentioned if growing the mini a bit was an option, that could likely provide a more cost effective alternative to the current mac pro. I realize their margins may not be too insane on that machine currently, but the price to performance ratio of desktop components tends to be better than those designed for laptops. Even now I tend to wonder how well the 21.5" imac is doing. It doesn't boast truly impressive screen real estate by today's standards. The performance margin over the laptops isn't anywhere near what you get from the 27" i7, and it lacks the dual thunderbolt ports. If it's all about the mobile devices this one must be starting to slow down too.

Not entirely meaningless. They test loops that use the raw performance of the CPU - the same raw performance used by other apps. I personally consider Cinebench to be one of the best as it's not purely synthetic but an actual real-world rendering benchmark.

I agree that a high-end machine should justify its price, which is exactly why the current entry Mac Pro is very bad - you can get the same performance for $1000 less from the same company.

Everyone is making some very good points here, I hope the discussion continues.

But, since I'm not up-to-date on current trends in chip design or computer architecture, I have a couple of questions:

Is there, speaking in terms of engineering, a reason that we couldn't have more options available to us when we order our machine?

I'm thinking mainly of things like gpus: if we don't count the hyper-expensive Quadros (and I don't count them because they offer me little advantage over much less expensive gpus), why are we limited to two choices? Would it be so hard to design a system that allowed us to access some of the other gpus out there? I realize that part of the answer has to do with the companies that produce the gpus, and how they design their cards, but is there something specifically unique to the Mac Pro that prevents our using these cards?

Similarly with cpus: I understand that different architectures have different requirements, but are they really so dissimilar as to justify the huge differences in price (I realize that partly this is due to Apple's tradition of charging a lot for everything, but still...)? In other words, if I wanted a 6-core westmere instead of a quad-core nehalem, I would have to pay $1200 more. Does the swap really involve the kind of customization that the price suggests?

Snap! I sucked you back into the thread I get you on these things, but they still don't tell the full story. Cinebench doesn't always give an accurate look at performance in real applications as it can't really track the use of core utilization or gpu leveraging (mostly thinking of things like rendering video here). The ram and overall IO are considerable bottlenecks too (one thunderbolt port for data and display bandwidth). Integrated graphics on the only quad core model chop its market potential further. If they at least built it into an actual desktop, it might be an eventual successor. Thunderbolt can't really replace things like SAS and fibre channel standards today, but depending on intel's performance roadmap it could happen eventually. I've noticed some cool stuff has started to trickle out for the standard, but more than one port to be shared with the display would be a good thing.

Another issue is overall machine stability and heat. The mini is packed in pretty tight but it's been suggested that they've improved on this aspect of the machine. I'd have to play with the most recent model, but that was always a huge turnoff for me. As it is, it's still hard for me to see purchasing a mini as opposed to just upgrading the laptop and using that as my sole machine (which I probably wouldn't do).

I upgrade computers when necessary, but no matter how new the computer, I tend to push it pretty hard, so heat is always a consideration. I could see photographers, graphic designers, video editors, prepress workers, professional photo labs, ad agencies, and possibly animators and audio engineers benefitting from a midrange machine like the possibilities that have been discussed. A lot of these guys are going to own laptops regardless, and this simply provides a more appropriate desktop option for when they're at the office with better ergonomics, speed, stability, and IO integration. Gamers are another market for such a machine. The imac is a lot of resolution to cover with a mobile graphics chip for something like gaming as it renders on the fly (I'm not a gamer but I know the logistics of it).

Well anyway I hope they do something with this market segment. They just need something that fits well with the line. I've commented on the 21.5" imac. It's in kind of a weird spot right now. It's a bit better than a laptop with a larger screen. It's a little small for use as a television for content viewing even though the resolution covers it. The display isn't bad but it's not great. I tend to wonder what people think when they see the lower tier of imacs next to the 15" macbook pros. You do spend a bit more to get to a quad core laptop, but that may change in the next couple revisions.

Yes, I'm one of those- I won't have any use for a 12-core machine until Adobe starts writing Photoshop (and InDesign, and Illustrator) to use those cores effectively. In fact, according to some tests I've seen, the six-core Westmere outperforms the eight-core when it comes to processing large images.

Laptop? Forget it, I hardly use mine for anything other than downloading and previewing during a shoot.

In other words, I need the IO and GPU options (as well as the ergonomics and other points you mention) more than I need a simple CPU upgrade: the current iMac would be almost perfect if only it allowed me to add another graphics card (or two) and keep my internal RAID configuration.

I understand that, but you own one. I'm trying to say that most people that own a mac pro or those who would buy an xmac would own one in addition to that macbook air or pro. Also regarding desktops, I've heard people claim they were on their way out for more than a decade, but the software becomes more resource intensive, and the use is still there. I'm somewhat curious what kind of stuff you shoot now. Photoshop doesn't actually make that much use of the gpu outside of a few functions. It has opengl enabled drawing and leans on it when working with 3d elements and during a few other things. Processing images doesn't seem to use it much in any software package.

If you ever do have to use a laptop for programs like the ones described, there are a lot of things you can do to speed up the workflow. In fact if it had as little as a second thunderbolt port to provide a data only channel and the option for an external display, you could get reasonable performance if you set it up just right. Like in photohop turning off all the thumbnails, setting the cache as low as possible (you have to test how long you can go without screen redrawing lag), custom ram setting, external scratch disk over thunderbolt with scratch compress disabled, dashboard and overall animations (like the dock) disabled, etc. There's a fairly long list but I've applied custom settings for people in the past that make it fly on less than ideal machines. If I see the beachball too much, disk warrior cures it. Many people haven't used it in a very long time, but it still helps me out so I continue to use it.

In simple terms there are few limitation in a Mac Pro when it comes to video cards.

Support and engineering expenses.

I don't know how much truth there is to this but it has been said that Apple develops it's own drivers. If this is the case then they can control expenses by limiting the cards they support.

The problem I have with this idea is that GPU cards are very complex, as such you would need a manufactures involvement. On top of that it looks like things have changed with the move to AMD GPUs. That is there appears to be a greater involvement on AMDs part, but understand that is me reading the tea leaves.

In the end I just don't think Apple wants to support an endless number of video cards. Their arraingement with the GPU suppliers seems to enforce this as you don't see third party cards offered for Apple hardware with drivers independent of Apples.

Well there is no BIOS in the Mac so in the past the GPU cards had to have Apple specific firmware. Hardware wise I don't see an issue they have a very hefty power supply in the Mac Pro.

This is a most difficult question to ask and answer. For example if a platform supports buffered memory you can have much more installed in a machine. If that machine supports Error Correction and Detection it might make the machine suitable for engineering where mistakes are not acceptable. As to Intel processors well they can be had with all sorts of features which some may value while other don't.

The hard part here is your question about price, for some like me the price isn't worth it. This in part is why I'm gung hoe on XMac, all I really need is an expandable Mac with a decent desktop class processor. I'm not talking expandable like the Mac Pro either, all I want is a couple of slots and storage bays. Note that I dont really want to give up to many cores but right now I know I don't need twelve.

Processor architecture itself is very interesting. Our good friends at AMD are going through a very interesting change to their lineup of hardware. They are taking dramatically different paths than Intel and in some ways outperforming Intel. Bulldozer (a code name for a new AMD core) is dramatically different than Intels latest. It will be interesting to see Intel feel a little heat again. The thing is it is not impossible to find situations where different architectures have different weaknesses or strengths. So what does this side trip to AMD land mean? In the context of Intel hardware it means that the various features are ways to create pricing teers, with the user deciding if he wants a specific feature.

In the context of many users on this forum what you get from the Xeon line is virtually worthless. On the other hand as hmm has pointed out, Apple has really cheapened the Mac Pro with some of the lowest end Xeons Intel makes. People might argue with this point but right now you are paying first class rates in exchange for steerage class accommodations. I still believe that Mac Pros incredible bad value is the result of poor sales. Apples laptops, Minis and IMacs aren't that bad of a value, the mac Pro on the other hand is a joke value wise.

On the topic of GPUs I've heard all sorts of things regarding who makes the drivers and everything. Apple used to be a smaller company than it is today, so it's always kept things a bit trimmed down. The last mac pro actually got a couple decent card options. Considering the starting price of the mac pro I liked that they went a little bit lenient on the upgrade price for the 5870 card. That card is about a generation behind now. I don't know if we'll see a card currently out or an upcoming one in the next mac pro revision. 2GB of vram has become pretty much the norm in the newer ones. Going over as many tests as I could find it still seems it's further ahead of the 27" imac than I might have expected, but this performance gap isn't going to show up in all applications.

I've accepted that Apple will probably sit on this for now. There's a pretty good chance that they want to see how thunderbolt will play out and if SSD prices drop. If we have the option for two to three faster (newer generatithunderbolt ports (fast enough to displace things like SAS) or more ports and the ability to multilink devices, it could go somewhere. Wizard69 also mentioned PCIe 3.0 based SSD technology which is a really interesting point, and a really solid use of the PCI bus. If you had the pci devices limited to SSD cards and however many lanes dedicated to discreet graphics. If we see more quality designs for thunderbolt products from what used to be pci type devices (here's an example http://www.blackmagic-design.com/pro...ultrastudio3d/ ) it could be a viable design. Right now a big issue that I can see for power users is that thunderbolt can still throttle disk IO for video editors. With another generation or two this could become less problematic. The big advantage that you'd gain from these kinds of products is that your laptop would be able to hook up to similar devices.


The TDP on the base mac pro isn't that much higher than the imac. The cpu and gpu combined are about 40W higher in total. As for the rest of the machine the ports are fairly comparable. A couple people have mentioned the PCI lanes in the mac pro which require power so overall it probably does have a beefier power supply. You could still collapse it down quite a bit and put those PCI lanes to good use even without leaving things as they are now. I don't really mind thunderbolt. I think we just need the ability to spread out the IO a little rather than have everything shared. With higher bandwidth devices like multi drive enclosures or displays, I'd like to be able to dedicate a single port to the device, without it being the only one on the machine.

Bleh I forgot to respond to this earlier. The G5 was for the most part a dual socket machine. In order to maintain that it meant using Xeons which for the first generation ended up also meaning a very expensive custom ram spec. Over the first few generations they did however inch up the price and lower the budget of parts to be used. 2008 was the first time a single socket configuration was available. The baseline configuration involved 2x quad core processors for $2800, but you could order a single socket configuration with just one of these for $2300. Apple was really sneaky here. With the next hardware generation they changed to a single socket logic board and claimed it started at a lower price when you were really getting the cheapest possible xeon in a version made for single socket boards only (lowering its cost considerably) for $2500 (another price increase). It limited you on ram a little at the time but with the current price of 4GB dimms it's not as big of a deal anymore.

So anyway some xeons enable dual socket configurations which you can't really do with the desktop variants. ECC ram essentially tracks bit flipping. It's not something that happens often and it will not fix multi bit errors. Without ECC ram you don't really have a way of knowing when this happens. It's popular for engineering applications that are absolutely error critical and mission critical servers. Beyond that it's not necessary. ECC ram carries a slight performance penalty but you'd never notice it.

Wow, thanks for the responses, guys, this is great information! I owe you one.


Right. Sorry if I seemed dismissive, I wasn't trying to be. I own a laptop because it is an intermediate step between camera and desktop that lets me review shots in the field. On a different note, I also teach, so the laptop is useful for recording notes when I'm at the library. However, I think that I may not have a laptop in a few years, because the iPad appears to be capable of performing both functions more than adequately.

I think you're right: the desktop is not dead... yet. The way I see it, desktops will increasingly become a niche market for people who need the extra power/space/whatever: engineers, graphic artists, gamers, etc., and everyone else will switch to portables, tablets, or something similar.


It's not so much about software using the gpu as it is about real estate. I have a 27" and a 30" display, and I often work on oversize layouts and images, so it's really a question of being able to see what I'm doing.


This makes a lot of sense to me- in the past, Apple was very selective with the manufacturers it supported, but also worked very closely with them. As a result, Mac peripherals "just worked"- they were true plug and play- but they also cost more than their PC equivalents.


Oh, I didn't know this. So it goes back to the point you made above, that Apple is selective with its vendors.


Yes, this is something that would interest me as well; I'm just trying to understand why it's so difficult.

It's not that I have a problem with paying more, it's that I need to have a good reason to do so. As I mentioned earlier, buying any of the current Pro line would be a huge upgrade for me, but once I accept that, it becomes a question of price vs. performance, and as you say, it doesn't seem worth it.


This is interesting, and I love the idea: why not give customers a few more options when it comes to the card you want? Do you work in video? Then card X is for you. Do you work in 3D? Serious gamer? Get card Y. As Apple already knows, customers like to have the ability to customize their order; it only makes sense that the Pro line, because it caters to users with very specific needs, should have a few more options than the rest.

But, as you suggest above, it all depends on whether Apple decides to support it. Since I don't know what kind of investment Apple would have to make, it's impossible for me to say whether it's something they should do.


On the other hand, we should consider the possibility that the current line is too specific.

Consider: not counting the server, the Mac Pro line offers eight options for cpus. Of those eight, four have 12 cores, one has 8 cores, one has 6 cores, and two have 4 cores. Now, only a few programs will take full advantage of 12 cores, so for the rest of us, anything more than 8 (or even 6) is a waste: we'll see no difference (and in a few cases,even lower performance).

In other words, half of the entire Mac Pro line caters only to a select portion of the users who need the Pro's power. As for the other half of the line, only two of those models (the 6-core and dual quad-core Westmeres) outperform the i7 iMac- which also comes with a high-quality display and yet still costs at least $200 less.

So, when it comes down to the wire, the Mac Pro line really only offers two choices to the majority of people who use them, and when those choices are compared with other options, such as the i7 iMac or a tricked-out Mini, the Pros simply don't stand out as a good value.

We know that the 6-core Westmere is by far the best-selling option of the entire line. That alone should tell Apple something: they are neglecting the lower end of the high-end user spectrum. The demand is out there, but the majority of the line offers too much of the wrong thing.

It may be that we'll see an increasing reliance on macbook air or ipad like machines when on the go. Smart phones like the iphone have been pushing things in this direction too. Essentially devices like this go beyond portable to the point where they require very little effort to transport and of course battery life has been improving. The air is becoming like a mobile phone where you just put it in the charger at the end of the day.

I'm not totally sure on this but if people do start to replace their laptops with tablets, we may see some users end up owning an ipad + a desktop and foregoing the laptop completely. It could go any number of ways but I'm thinking of people who need functions beyond web browsing and basic office applications.




I was going from the fact that you ordered an upgraded graphics card last time. Unless I'm mistaken the basic stock graphics cards are capable of refreshing both of those displays without any kind of banding. I have no idea if that is beyond the capability of the imac (I know one of them would work fine).


Ehhhh...hmmm how to explain this... Well on the PC end you have your typical graphics cards which tend to be optimized toward gaming. Then you have workstation card options. These often employ the same or similar hardware unless you're talking about the most expensive ones which are often several thousand. The drivers (and possibly firmware) are different in that they tend to be optimized for for modeling/animation applications. With some applications these workstation cards don't necessarily do a better job. You're best off searching for information before buying. On the mac end the only workstation cards we've seen have been quadros, and for the most part they haven't been well supported. They provided some decent options the last couple rounds. I'm not sure what we'll see with the next revision.


I actually didn't know that. Where did you find this info? I know the hexacore model was well reviewed on a number of sites, but I would've thought the pricing would turn a lot of people off. I think I'd research its use in specific applications before dropping that much. The thing that sucks with the Apple store is even though I found out you can apparently install applications to test them there, those machines remain stuck with stock ram. It seems like it would make it more difficult to gauge the performance vs. the quad machines. The problem solving you present to each machine would need to be heavy in computations yet small enough that it wouldn't generate a lot of pageouts.

Yes, I think this will happen, too- at least for me. I also think there are people who will use laptop + tablet combos, with the laptop taking the place of the desktop at home.


This is true, but there are other considerations as well: first, I'm not interested in the bare minimum, and two large monitors at highest resolution is pushing the limit- the ability to a second card therefore represents an option to do more, and do it more quickly. Second, I have other interests as well- I mess around with 3D, watch the occasional video, even play games from time to time. Again, having a second card means I'm able to do these things without the performance suffering.

I was looking at the Apple site and saw under the "best selling Mac" tab...

1. MacBook Pro
2. iMac
3. Magic Trackpad
4. Mac Mini
5. Mac Pro

Interestingly missing are the MacBook Airs. The Mac Pro hasn't been on the list recently... could there have been a huge buy from someone? I noticed the refurbed 2010 Mac Pros were out of stock.

It's possible. I wouldn't trust that list without knowing how it was compiled. I would find it difficult to believe that the macbook air would be out of the top five and beaten out by the mac pro and mini in terms of volume. It wouldn't make sense, but I guess it also depends on how this data was compiled. It might sound cut and dry but do they reset it with product transitions? Is the data replaced daily? weekly? monthly? How is it averaged out overall? Obviously these things tend to rise and fall based on seasonal factors and refresh cycles, but it's just simply weird not to see the Air on there. It can be a really nice machine for lighter computing. It's not so great for games, but it's adequate for a pretty wide range of stuff. Even graphics applications run well enough on it up to a certain file size. The smaller hard drive isn't ideal for anything that generates a massive pagefile. Oh and the refurb site is all over the place. Mac pros are a lower volume machine with a fairly low failure rate, so not that many show up there. Most of the time it seems to be current generation returns.

I think it reflects the online stores sales. However I'm not surprised at all that the AIRs aren't on the list. AIRs certainly attract early adopters, but even the most recent models have limitations that some userscant accept.

AIRs being very portable can move through various other channels. As you note it isn't clear how the list is compiled, but in the past I've found that over time it can give you a sense of what is selling well. You certainly don't want to take data from any one week and make decisions on it.

Even more interesting is that the old Mac Book showed up often on the list. The thing here is a simple corporate buy could significantly shift these numbers.

One thing people seem to mis out on is the idea that those extra CPUs can be used effectively for multiprocessing/multitasking. With a Mac Pro it is easy to fire off a long running job and still have a usable machine. Laptops certainly don't handle such usage well. The point is even if your favorite software can only max out four cores an six or eight core machine can still be an advantage.

The whole problem with Apple and their Macs is that that means an iMac. That is to get a machine with what I call midrange performance in an affordable desktop. All in ones aren't my cup of tea though so I'm screwed. As are many of us that want something that is a step up from the Minis.


I've been using my iPad more and more but it comes up short currently. To really get the versatility that I'd like the iPad needs an easy to use USB port. I know this causes some to get a bit excited but I don't think many grasp how important ports are to technical users.

That is in comparison to previous vendors.


Still a decent performer though. At the start of this reply I had this thought that Apples next Mac Pro revision is waiting on AMDs newest GPUs to come out.

You would expect that performance gap to show up in all apps. Not all of your apps are GPU bound.

The problem with the "you don't need slots" people is that they said that about USB, FireWire and a host of other ports. I dont dismis the need for USB and even TB but I don't understand why people are so quick to dismiss PCI-Express slots. Slots solve an entirely different set of problems.

Neither do I, I think it is great technology. However a TB port simply isn't capable of reaching PCI express speeds.

Same here. Too many limitations with an iMac. So it is either settle for the mini which is less than I want or go overkill in size and price and get a Mac Pro.

Even things like photoshop (I use it because it's an easy generic example) didn't use the gpu to redraw images until CS4 came out. That was maybe 2008? Even then it was barely noticeable. We're starting to see more and more stuff lean on the gpu, but it really varies from application to application.



You already know I'm not really one of those people. Intel seems to have plans to increase its bandwidth, but I don't know how fast that'll happen anyway. The current design for as large as it is, isn't really that ideal. Hard drives and pci cards get some significantly uneven temperatures which has always annoyed me. If they built the mini outward a bit though and put in all desktop parts (except for maybe an SSD drive), I would probably buy one. If Apple surprise with a nicer update than expected on the mac pro, I will definitely buy one . I'm really not sure why Intel has been so slow with rolling out Xeons. They should at least have healthy margins on some of those, so it seems a bit odd.

They have a single-core benchmark and a GPU benchmark as well as multi-core.


The highest-end mobile GPUs outperform some of the high-end desktop GPUs. The 6990M should outperform the desktop 5770 in the Mac Pro. Mobile just means it fits in a certain thermal limit.

Consider an Ivy Bridge Mac Mini. No thermal issues as Intel has cut the power draw in half with better performance. This means Mac Pro performance in the quad i7 with enough room to allow a 28nm AMD 7000-series GPU too. Throw in a 256GB SSD and plug in a Thunderbolt or USB 3 drive/RAID and what more do you need from a mid-range machine? This is under $1000 and fits in your hand as well as being one of the nicest looking Macs and most reliable.

No, it won't be on par with a mid-range tower but it's not going to be significantly slower either so it's all Apple needs to do.

Www.hardmac.com had an interesting bit on the Mac Pro but it sounded like they where blowing wind. However we should start to hear rumors or hints of a refresh soon.


I probably didn't get my point across well in that statement. What I was trying to get at is the expectations that some have that GPU usage should lead to massive performance increases. It can for a minute set of problems but the user shouldn't expect that for every package they run. That variance from app to app is not always something that can be over comed.


I really haven't tracked Intels chips much over the last couple of years, but Hardmac seems to believe the coming machines are going to be very nice. My goal isn't another Mac Pro class machine, at least not at the expense/performance ratios of the current machines.

Yeah I've had people try to tell me what a great thermal design the Mac Pro is. The Mini is actually a great thermal design, I especially find the current design interesting. The Mini however is not a XMac nor Mac Pro class machine.

In a XMac class machine I suspect that minimizing the number of slots would vastly increase the ability to cool hardware in those slots. The big problem with PCI-Express in a PC card format is that the system can't really cool the cards like a rack based system.

Actually I'm very tempted to buy one. Adding the GPU to the one version has made it a far more desirable machine. I still think Apple screwed that machine by adding to little GPU RAM though. I have nothing against a low end Mini but if I'm going to go up scale at least give me something for it!! As it is I'm out of the market for a computer this year anyways, this actually might be a good thing as an Ivy Bridge Mini might be very compelling.

Your idea with building the Mini outward is interesting. Frankly they don't need to even use PC card format PCI-Express cards, there are a number of very useful card format standards that could work well in a Mini. Obviously this wouldn't go over well with the mainstream PC / desktop world but let's face it the PC card format has been around a long long long time. These days ports, I/O cards, even low end video cards can be had extremely compact. Somedays I wish Apple would adopt a mid way standard like Compact PCI.

In one word AMD. I suspect they have serious concerns about AMDs Bulldozer based chips and want to be able to develop a clear communications regime to compete with Bulldozer. Bulldozer has some very interesting approaches that should allow it to perform really well in certain workloads. I suspect that Intel wants to be able to address that in their marketing. They might also be trying to hit much higher clock rates as a result of AMDs chips.

Margins mean little if you can't successfully market your chips. Right now I suspect that everyone is having problems moving hardware for servers and such into corporate America. It isn't just that the economy sucks, which is a big issue, but there is a real feeling that the industry is in transition. Who in their right mind would invest in IT infrastructure with product from HP or Dell? I can actually see a transition back to Sun/Oracle or IBM hardware in data centers, probably running UNIX too. Building IT infrastructure on commodity hardware is proving to be a shaky way run the back office unless you have the will to do much of the work yourself. In the end I think Intel is being cautious, maybe excessively so, but not without multiple reasons. AMD literally has nothing to loose. I think the question to ask is if you need data center hardware, and you where looking at i86 who would you select as a vendor. A vendor you would hope would be around in 5 years.

The problem is with it's size and the capability restrictions that brings.


At a given process node though more power generally means better performance.

Yes it almost makes me glad I'm not in the Mac Market this year. The problem isn't the chips though it is what Apple does with them. If they went out and designed a Mini with true midrange performance, for the technology of the time it would be great. However Apples history here sucks. I don't mind the low end Mini but if I'm to pay extra for a high end machine let's really make it a mid range machine. Apple doesn't do this apparently castrating the Mini on purpose. That would be fine if they had a desktop machine (not an iMac) that sat between the Mini and the Mac Pro.

Depending upon who you are and your needs, any of the following:

1. Slots for expansion cards.
2. Slots for RAM. Note we are talking midrange here, so three slots for a triple channel machine would do fine. The Mini is problematic for RAM expansion beyond a minimal limit.
3. Drive bays / Storage expansion slots. I'm actually looking forward to the day solid state storage comes on plug in cards.
4. Redundant power supplies. If Apple keeps trying to pass off it's machines as servers this should make sense.

Further note that none of these wants require a Mac Pro sized machine.

It isn't always about looks though. For me the Mac Pro is ghastly but that doesn't stop sales. Reliable I don't know about, but that is an issue of design.

Actually though the Mini is always significantly slower. This isn't a huge problem as low end machines are always needed. The problem is the massive capability gap between the Mini and the Pro. The gap can't be dismissed reasonably. Capability isn't always about speed though, which is what a lot of people mis in this discussion.

The quad core machine actually starts at $1000 sadly, and really plenty of single applications make solid use of four cores. On GPUs I'm having a slightly difficult time finding as many comprehensive tests as I'd like. Barefeats has a gaming comparison which isn't completely relevant, but games do rely pretty heavily on the GPU. Their results suggested that the 5870 was still ahead of the 2GB VRAM version of the Radeon 6970M. All of their tests performed better under Windows 7 than OSX but that's unsurprising because the games they test are probably built to run under Windows and then ported to OSX.

If they brought out a mini with two thunderbolt ports, usb3, and decent discreet graphics (with at least 1-2 GB of dedicated VRAM) I might take the machine seriously. It's still a bit throttled on ram but I don't know. You think they'll put a quad core and discreet graphics in at the current pricing structure (even considering that they don't even bundle a keyboard and mouse with those models)? I think it'll probably get shot down because it would lower their margins on the machine or bump the price into imac territory.

The comment about all Apple really needs to do just relates to the fact that to run OSX we don't really have other options. Anyway since when are the Apple keynotes about doing just enough? I'm not trying to twist your words around here, but I think it's a poor mentality to get stuck on, and it's the same reason you often hear people spite companies like Adobe.

They fit everything inside a 45W limit (used to be 35W with the optical in there). Over time, this limit will contain more and more to the point that everyone is satisfied with it and Ivy Bridge will be the first big development that highlights this.


I'm still behind Thunderbolt on this one.


I wouldn't call 8GB a minimum but a 3rd slot wouldn't go amiss.


It depends what you call significant. The quad i7 Mini is at most 30% slower than the high-end iMac. We know the GPU is significantly lower but as I say, the Ivy Bridge development should help here.


I don't know if they'd put a dedicated GPU in the server model but even if they didn't, the middle model with Ivy Bridge will match the current i7.

I think it's good that they keep HDMI on there too. I'd personally be happy with a dual-core i7 that performed the same as this year's quad i7 combined with a 512MB Radeon 7000-series GPU and USB 3. With a 3rd party SSD, it will be a very fast machine.

If you have a set of tasks that constantly demand the highest performance then the mid-range is not suitable. If you have a set of performance goals that have a minimum requirement, the Mini should have already surpassed them and year after year, it will keep getting better until you don't even care about it.

I'm well aware of Ivy Bridge. However you mis my point, unless Apple has a change of heart, the Mini will always be a low performance machine relative to any other hardware available at the time. Beyond that there is more to user satisfaction than just the raw CPU performance. The Mini can't and never will satisfy everyone.

Thunderbolt has it's place but it is by no means a replacement for slots in a desktop machine.

Well I never really thought that 2GB would be too little on my old MBP. However it has become very painful just trying to run Apples software.

note that you are comparing old technology to new here. Further I suspect you are being very selective about how you arrive at that 30% differential. All you really need is a pattern of usage that loads all of those cores in a Mini which really isn't that hard to do these days. My old MBP displays significant performance problems just trying to run XCode and Safari at the same time. Part of that is the need for more RAM, but today's software may incorporate many threads and even secondary processes. I've seen significant performance issues doing what should be simple tasks.

So when looking at a desktop machine what is important to me. Well number one is the ability to easily expand RAM, as this is critical to the long term viability of the platform. Ideally RAM expansion can be done without wasting the current installed RAM. The second thing is a surplus of cores to avoid running out of performance one OS update down the road. Closely ranked to item two is the ability to add significant amounts of internal secondary storage. The common theme here is buying hardware that isn't immediately obsolete, and provides better than low end performance out of the box.

In that regard the Mini simply comes up short. In many cases it comes up short by design. For example they have a nice quad core unit that could have been very desirable if they had continued with the discreet GPU offering from the middle of the road machine. Nope, instead they decided to call it a server and delete the discreet GPU. Maybe that was due to a issue with the power budget, I really don't know, but it takes the machine out of the running for many of us. And NO Ivy Bridge won't fix that.

Is that a surprise? I mean by that tomorrow's low end hardware matching the performance of yesterdays higher end hardware. I do think you are overly confident in Ivy Bridge and Apples willingness to build a Mini that is something more than bottom end.

HDMI is good. However I suspect you are giving the coming Ivy Bridge CPU more credit than it possibly deserves. For one no matter how fast the dual core chip is it is a step backwards from a quad core. Second by the time they get to a discreet GPU with 512MB of RAM that will be a very outdated configuration. Really think about what you are paying right now to get a 256MB supported GPU in a Mini. It is a joke.

This all come back around to my point made earlier that the Mini is intentionally castrated. It isn't the low end embedded video model that is the problem. Rather it is what you get if you are willing to pay more. Sadly it isn't much and what it is keeps the Mini at the lower end of the performance spectrum. Ivy Bridge simply won't matter if Apple intentionally cripples the machine.

That is a very self serving statement. Basically you are trying to say that if one can't use the Mini buy the Mac Pro. A good portion of this thread though highlights why that is a terrible idea. The gap between the Mini and the Pro is a lot wider than you seem willing to acknowledge.

As to performance getting better and better, again no big surprise here. Mind you some of my first computers where eight bit devices. In every case the old computers went away for one reason - performance. The Comodore went away as did my Heatkit, my Mac Plus, my series of homebuilt Linux machines and now my MBP has serious issues running today's software. The only real difference these days is that you do have the option of buying hardware that lasts a bit longer than they did in the past. That requires buying smart though. These days that means a minimal system is a quad core with a discreet GPU and the ability to easily upgrade the installed memory. Note carefully I say "minimal system", I can make a very good argument that even quad cores are not enough if you want the system to last through a couple of software iterations.

To put it plainly I'm not willing any more to update hardware constantly. There was a time when doing so every six months to a year was not unheard of. I'm not willing to play that game anymore. So what we have in the Mini is a machine where it becomes obsolete to fast and an unwillingness on Apples part to offer a version that addresses that. Sadly this looks to be intentional on Apples part.

The XMac desire is just a way to express dissatisfaction with Apples rigging the market.

I've always thought so. It's kind of the Apple way. They build whatever they want and if you require a slightly better feature set, you're locked into a very high minimum sale even if it's something that's very common/mainstream in hardware terms. It's not that everything has to remain like the mac pro forever, but it annoys me that they'd make a new desktop line then intentionally throttle it. This has been the case ever since it was the G4 mini.


It looks like a nice evolutionary step, but what everyone wants tends to be a moving target. Today a computer from the 1990s could barely load most websites. My point is that software isn't completely static. The migration toward 64 bit applications also ups the requirements for ram since it's forced to address information in larger chunks.


It'll be interesting to see how that plays out. Right now I don't agree with you but for my own purposes it could work if it grew a bit in bandwidth. Having 1 port on a machine for display + data though is just a joke. More panels are coming out with 120Hz refresh rates. Combine that with an external SSD or two and you'll be able to throttle that pretty easily.


The thing is, even cheap ass Windows machines often have four slots. It's not a high end feature and should not be marketed as such. If the thermal envelope is that tight that could be a reason. They're using laptop parts, so they probably don't want to put in the design time here or figure out how to fit it in. That's the thing I dislike about the mini. It's not actually engineered as a desktop. It's engineered to be as small as possible at the cost of other things.


I'd have to test this myself. Barefeats put it all over the board. Some of their tests show some homogenized performance among the line, but mostly those dealing with smaller problem sizes. It fell way behind on gaming no matter what model you viewed.

http://barefeats.com/fcpx01.html
http://barefeats.com/aper313.html

Their testing methods aren't always 100% perfect but it's hard to find as much data as I'd like. The point was to get reasonable performance even against the laptops you need to go with at least the mini server, but then you're back to integrated graphics and still only have a single thunderbolt port.


See I've read words like that before. I'd be happy with this or that. It changes based on what is released. We get faster hardware, more ram, and the programs we run become more resource intensive. It's worked like that for quite a long time. No matter how much you spend on hardware, there will be room for gains. Even what you consider a basic machine appropriate for basic office applications and internet use is a moving target. We have a lot of that on our phones today. Pretty soon any phone that can't run web content as easily as a laptop today will no longer feel acceptable.

I wouldn't expect the next dual core to equate to the current quad in performance. I think that's a bit optimistic. There isn't really hardware they're leaving out due to thermal concerns. If anything it's a cost issue and that's something I hate about a desktop with laptop hardware. You get the performance/cost factors of a laptop in a stationary form factor (like a desktop). I can honestly say I hate their product offerings at the moment. It's like each of them is missing something. The mac pro is probably the best overall choice for me, but the price to performance ratio is quite off there. If they replaced that machine in a month or two with a truly updated model, people would call it an old machine even if it was the most recent just a day or a week previous to the statement.

One more thing (I used that phrase long before I heard Steve say it) even the ipad and iphone have dual core processors these days. Quad is basically the norm for desktop computing. It's not exactly a high end feature. I'm trying to think of something more to add here to make the discussion interesting, but it seems we're just at odds on a lot of these points.