It is definitely the workstation to dream about, we've been waiting for a tech leader in the high-end professional space even if the market is not that significant anymore (relative to the $$$ Apple makes on consumer, anyways).
Why do you want OS X? I'm really genuinely curious and flabbergasted by people wanting apple stuff . What is it that you cannot get cheaper with more flexibility and choice on linux or windows?
Adobe Creative Suite + Unix without dual-booting or virtualization. Not to mention a multitude of rather nice indie apps for creative and technical professionals.
* Runs Office, which means it runs Excel (and Word). Gotta love Excel. I'm sure there are unholy combinations of calendering software that mean I can interact sort-of sometimes with the Exchange server in the office, but Outlook Just Works too, on a Mac.
* I never ever ever ever have to interact with XF86Config. Literally days of my life wasted on that.
* Runs Omni(Outliner|Graffle|Plan|Focus). I guess I could cobble together something from org-mode, some terrible Visio clone, etc.
* It does sleeping properly when I close my laptop. Always. Reliably.
Your hardware support experience is exactly opposite to mine. My laptop runs Linux just fine and all hardware works correctly, but OSX doesn't even want to boot on it. Weird.
I love developing in ubuntu, but ubuntu and windows seem to have gone crazy with their user interfaces lately. Ubuntu switched to unity, an interface designed for netbooks. Who has a netbook these days? And wants to run linux on it? Canonical should be targeting their main userbase, developers or advanced users. They should concentrate on making gnome as stable and simple as possible, and include vmware/virtualbox drivers out of the box. If developers have a simple, stable platform then the platform will grow organically.
Windows 8 can't make its mind up about if it should be used with a mouse or a touchscreen. Discoverability is hard, it's really hard to work out that you drag an app to the bottom of the screen to close it, because their are no visual clues. I like the look of metro, but it's really a polished turd.
That limitation is due to Apple's licensing of OS X, and their unwillingness to support (or to even let others support) non-Apple hardware. It's not due to problems with the HP workstations.
We have both z620s and Dell T7600s (http://www.dell.com/us/business/p/precision-t7600/fs) at work. Apart form the badge you'd be hard pressed to tell them apart. Certainly neither one is better than the other.
No, there is nothing really special here with respect to performance. It just makes me think back to the SGI O2, Next Cube, the iCube, the Amiga 2000...when computers were more works of art than they are now and not just built from a boring specifications. I'm being a bit sentimental here.
If you want the best bang for the buck, or even the best bang, it still makes sense to go with a boring HP box and not a Mac Pro. Honestly, the only ones who will probably go with the Mac Pro is the high-end photoshop/toaster crowd, or those who want big bangs in small spaces (e.g. Japan), as well as those with too much money and some amount of sentiment, but this is not even mainstream for the workstation market.
I agree with the idea, this is Apple having a radical rethink of what it means for a pro level workstation going forward. The majority of criticisms have been levelled at the lack of expansion, not at what the potential for the machine is.
I'm sorry, but there are only two things that are new with this mac:
1) its a funny shape
2) it comes with two GPU boards by default.
I'm glad that you are enthused by a new platform, but there is really nothing new. Really. If you want to start writing GPGPU programmes, I congratulate you, its going to keep you busy for a good few months. But you can do that now, without a $10k mac "pro"
Over the next few months, we'll be hearing about how the CPU is dead and that all you need is GPU. However much like the initial flash/iphone argument it over looks the main point. The Mac pro only has one CPU because that's all they could fit in. It wasn't a bold visionary move, it was a compromise because of the form factor. (much like the original iphone, to get the battery life up, they had to use a tiny(by the standard o the time) CPU, which meant no flash)
There is one large elephant in the room when it comes to GPGPU and that is bandwidth. Getting stuff from main memory to GPU memory take a lot of time. Yes PCI3 is wonderfully fast, but compared to RAM<=>CPU its like reading a floppy disk.
That asside, there is really nothing new or special about the macpro. In fact for VFX land, its rather a disappointment. Yes Mari runs really really fast on it. But then its got two large GPUs and a stonking SSD. Everything other programme is going to be CPU constrained(unless they've been pushing stuff to the GPU).
Please excuse my rant, I've been inundated with people wanting new macpros, despite them already having a far more powerful z620 with fusion IO cards and quadro k5000s. cult of the shiny and new I suppose.
It actually makes me mad!
If you really need mac software then why not built a far more powerful and cheaper hackintosh. But why would you even want that? what software is there for mac OS that there is a not an equivalent to for linux or windows?
Yes.
So it is a combination of laziness/lack of time to retool and learn a new program, and, a case of, well I've already thrown X amount of money down this hole so I'll just keep going, to stop now I'd have to admit I made a bad choice.
Most pros I know will happily swap software if it will give them access to new features they need or will make them more efficient and productive. They won't switch just to save a few bucks. Even saving $10000 on hardware isn't enough to compensate for even a small drop in productivity.
You probably do very little dev. Of all the development environments I have seen so far, nothing beats the Mac. Even Emacs for Mac is light years better than plain old Emacs.
Developer is a very broad term. What kind of developer are you? I'm a games developer.
Emacs is great but it is essentially a text editor. there are literally 1000's of similar things of varying complexity depending on your preferences available for linux and windows.
I develop a lot on Common Lisp. SLIME is Aquamacs beats any other so called IDEs and REPLs. If you think Emacs is just a text editor, you should come out of your shell (snicker, snicker).
It could be also a question of how much your time costs, if cost of the new MacPro is relatively small why should you waste time learning new tools? You'll be literally losing money.
It could be also a question of how much your time costs, if cost of the new MacPro is relatively small why should you waste time learning new tools? You'll be literally losing money.
It could be also a question of how much your time costs, if cost of the new MacPro is relatively small why should you waste time learning new tools? You'll be literally losing money.
It could be also a question of how much your time costs, if cost of the new MacPro is relatively small why should you waste time learning new tools? You'll be literally losing money.
It could be also a question of how much your time costs, if cost of the new MacPro is relatively small why should you waste time learning new tools? You'll be literally losing money.
I don't think you can have a legit Apple developer's account using hackintosh though if you wish to put apps into the OSX or iOS app store. Don't quote me on that though, but at the very least, it's going to complicate things. I'd like to be wrong about this though as it would simplify things (please provide links if you correct me :) ).
However, a mac mini is good enough for such things normally. I'd stick with my custom built PC for anything else.
The writer of this article is obviously very ignorant and does not know what he is talking about in terms of pc architecture. But he thinks he knows as, here he is writing a public article on the subject.
My current theory for mac popularity is that is is down to the consumers ignorance and good marketing (read: shiny box).
>If you really need mac software then why not built a far more powerful and cheaper hackintosh
I don't think it should make you mad. Different strokes for different folks.
If I go to a car forum I bet I can find a lot of people who love BMW, and more still who love Mercedes Benz. They'll probably happily argue with each other about who makes the better cars, and why one brand is far superior to the other. And then there's probably also some people who are happy with either car.
For you, there's no software on OS X that doesn't have an equivalent on Linux or Windows. But that might not be true for others (for example, I don't consider GIMP to be 'equivalent' to Photoshop...but you might). Another example - I used to work in live events, and used Mac Pros to drive video rigs. Could I have built a more powerful and cheaper hackintosh? Probably. But I doubt the build quality of what I'd put together would be superior. The cheese-grater Mac Pro could take a beating.
Again, my needs are very different to yours. Please don't get angry on my account!
Well it just seems like such a waste of money. Thats why.
But on a more rational level, apple having more users and hence more clout is a bad thing as they have always used their influence to try and drive a wedge between different pc users by intentionally not supporting things.
If a $100/hour developer averages 30 minutes/week more productivity because they're using a Mac they're comfortable with, that's a savings of $7,800 over the course of a 3-year upgrade cycle.
Try actually doing the math before you declare other peoples' considered choices a "waste of money".
No. Just saying "well I earn so much, so money is no object" does not make things less wasteful. Paying 2 to 4 times the price for the exact same components is wasteful.
This is like the fella above basically saying "high-end professionals" dont need to be logical anymore.
> why not built a far more powerful and cheaper hackintosh
Because I have better things to do with my time than assemble a poor substitute for a cohesive just-works experience.
> what software is there for mac OS that there is a not an equivalent to for linux or windows
As a developer, most of the exact same software I use every day on OS X runs on Linux just fine. And yet, Linux doesn't provide me with a good user experience, and it doesn't just work[1].
But fundamentally, you are just mad because other people like something you don't. Dwell on that for a while.
[1] If you're going to try and argue with me about how great Ubuntu or Mint or whatever else are these days, let me save you the time: You're wrong, and there is absolutely nothing at all that you can say that will convince me otherwise, because I do use them, but they continue to fail me as primary desktops. Your words conflict with reality, and more words won't change that.
That might've been true in the mid-1990s, but the various modern Linux distributions are quite straightforward to install, especially if it'll be the only OS running on the system.
Not if you have a large infrastructure to integrate it into. Sadly we have to triple boot (win 7, RHEL5 & RHEL6)
I'm a sysadmin and I make things just about bearable for my users. I install VLC, gitk, flash and chrome. I make sure that the email client is setup and talking to ldap/AD. Granted we use RHEL5/6, so there is a lot of work for me to do to make things nice for them. However RHEL/centos/SL whilst old and not trendy is a loverly slow moving target for our customers to jump onto.
Granted mint and the like make installation easy, but integration is still hard.
If you have a "sysadmin to set it up for you", that probably means you have a standardized environment, and the developer won't have root. This cripples developer productivity.
> The Mac pro only has one CPU because that's all they could fit in. It wasn't a bold visionary move, it was a compromise because of the form factor.
Where did you get that? It sounds entirely implausible that a company renowned for its design would decide on the form factor first, and then try to fit stuff in there as well as possible...
Because that's how design works. The end goal of the iPhone, (to take a very good example) was designed to fit in your hand. Everything else had to be crammed in to accomplish this.
to give a non technical example, the london underground map sacrifices geographic accuracy for simplicity. A very large, and successful compromise.
Also, where could they fit another CPU? Each one kicks out almost 180 watts of heat, and must have RAM and power to make it work.
It also complicates GPU<=>CPU and GPU<=>GPU communications. You either have to shove all the GPUs on one CPU and waste lots of QPI cycles on moving RAM back and forth, or have one GPU on each CPU and build some very clever glue to allow the two GPUs to share memory.
You talk specs I know little about. But I got to say I'm just surprised they actually even released the damn thing. Many seem convinced (me included) that Apple would go the way of Sony, producing pithy consumer electronics and diluting their brand values of being 'forward-thinking'. I thought VFX world was mostly PC/Hackintosh based?
Naa, hackintoshes are far to hard to maintain. Its either real Macs, Dell optiplexes (although I've seen a few of those hackitoshed) and HP z series(win/linux) Most of london is HP.
Macs are still needed because of quicktime, and final cut (plus a few other niche products)
People use the Laptops a lot, because they are actually quite good. Certainly longer lasting than anything from Sony or Toshiba.
Many production houses still use Final Cut Pro 7 and will continue to do so for quite some time to come. Most smaller teams and freelancers I know are switching to Final Cut X, now that it has support for FCP7 compatibility and multi-cam editing. Features like audio syncing are a lot better in Final Cut X, and once you get used to the new interface, the workflow is a lot more convenient.
One of the biggest challenges our engineers faced with this new design was how to join the front to the back. The enclosure is so thin, it’s not possible to weld the pieces using traditional methods.
Formfactor first, everything else second. Hence why apple makes the products they do.
I am of course willing to be convinced otherwise, perhaps you have some sources to backup your claim?
I think this belief that form factor comes first or somehow dictates component use is a common misconception (doesn't make it a dumb one).
Apple believes that design is essentially three things: form factor + hardware inside form factor + software. The three have to play together seamlessly for the overall design to be good. We can debate on how effective they are at this, but that is what Apple believes. Steve Jobs talked about the interplay of hardware and software many times.
Steve Jobs also frequently discussed compromises. He believed that consumers paid Apple to make decisions about what was useful and what wasn't. So including Flash in iOS would have meant that users could watch all web video (only 25% at launch!), but they looked at the stability and the battery life and made the decision that, no, Flash was not worth it. So Apple believes in tradeoffs.
Sometimes these tradeoffs are at the cost of the aesthetic. For example the black plastic on the back of the original iPhone, or the black plastic used on the entire iPhone 3G when they wanted to increase margins and go for market share. This is design. This is a company making tradeoffs.
Sometimes the tradeoffs do come to preserve the form factor. I struggle to type on my iPhone, but I was so quick on my BlackBerry that the software lagged behind me. But Apple made the tradeoff.
The antennae on the iPhone 4 is a great example of Apple design. It's accomplishing so much. It means there's no need for black plastic on the device, so the aesthetic is improved. It also provides the structural rigidity for the device, which means they can make it feel super sturdy (love Samsung's aesthetic, but if you hold an iPhone 5 and a Galaxy, there's a real difference to the feel). That sort of outcome isn't the product of a guy in a turtle neck with Auto CAD putting lens flares on everything. That's serious design.
Jony Ive talked a lot about the process for making the MacBook Air. They wanted to build a fast, ultraportable computer without making it one of those shitty 9" Taiwanese pieces of crap. So they had to work with Intel to build a 60% smaller Merom. Because they didn't want to compromise the internal hardware by shipping a slow processor. They had to develop Unibody so they could strip out internal fixtures that would add weight and reduce the structural integrity of the product. Commitment to the aesthetic.
You've cited (it's cited, btw, not sighted) the iMac. Apple wanted to make it thinner and lighter. At the same time they wanted to ship better components inside it. So they had to keep the space inside the same size or make it bigger, or they weren't going to be able to ship the product. So they spent time developing ways to make it thinner and more aesthetically appealing without compromising the hardware components. In this instance I'd argue that deference to the hardware components came first. Much like the MacBook Air, where so much of the work went into developing a tiny logic board so the device could be full powered.
tl;dr Apple makes tradeoffs on behalf of consumers.
> Hence why apple makes the products they do
A quickie on this: hence why is never correct. Just use hence.
My post is overtly black and white, as you've quite correctly pointed out. Design is a series of compromises; be they aesthetic, political, financial or technical.
This mac pro is all about GPU. The second CPU would require a larger volume and would add too much cost(in both style, support components and thermal envelope). This was deemed unacceptable and was therefore cut to keep the final design within spec.
It deemed an acceptable risk to release a product that is notably less powerful than the equivalent workstation.
> That's pretty much how every Apple product since 1998 was developed.
This feels really lazy and borderline trollish.
Over the last decade or so Apple (and others!) have fairly consistently demonstrated a belief that design is multifaceted: the form factor, the software, the components inside the form factor.
Increasingly we see these intersecting. The form factor for the iPhone is also the hardware: the screen and the antennae. The unibody manufacturing process for MacBook Pro means that the trendy aluminium form of the product also provides its rigidity and reduces the number of components required to hold it together. More room for batteries, which increases the battery life. Apple and Intel worked together to develop a Merom chip 60% smaller than the standard in order to ship the MacBook Air, which was at the time the thinnest notebook around.
Apple doesn't design the form factor and then work out what components need to go inside it. Very often the overall design (form factor + software) is where Apple can innovate, because Apple doesn't typically manufacture components, but they frequently make compromises of design in order to accommodate hardware (and vice versa). IR receivers on the original MacBook Pro (horrible black spots on the bezel), the home button on the iPhone (an absence of hardware buttons on the front of the device would be more pleasing). The black plastic at the base of the original iPhone for Wifi (since engineered out by a combination of component and product design).
People will look back on the last two or three decades as a renaissance for design. Apple has contributed, but across the board the importance of design -- and by design I mean the aforementioned intersection of form, software, and hardware -- is visible in products from Samsung, Sony, HTC. Heck even Microsoft has realised that design is more than commissioning Brian Eno to write the Windows boot music.
Back to the Mac Pro: I'm sure Apple threw out a dozen designs which had space for more RAM, or which had more CPU sockets, or which had internal expansion. But Apple makes tradeoffs for the overall goal. Sometimes the tradeoffs are such that hardware suffers (I can still type way faster on a BlackBerry than on an iPhone), and some times the aesthetic is diminished (I bet the visible sensors on the iPhone 5 really rankle with Jony Ive). But to suggest that since 1998 Apple has consistently developed the form factor for products first is ludicrous.
> design is more than commissioning Brian Eno to write the Windows boot music.
It's always funny when people use obscure(?) but interesting anecdotes to make a joke. It sort of shows the breadth of one's knowledge. Well, at least I had never heard of this.
> It sounds entirely implausible that a company renowned for its design would decide on the form factor first, and then try to fit stuff in there as well as possible...
Design has nothing to do with specs, and something to do with form factor. So yes, it sounds plausible that a company renowned for its design would focus on the design over the specs.
> much like the original iphone, to get the battery life up, they had to use a tiny(by the standard o the time) CPU, which meant no flash
I think you might be misinformed. The first-generation iPhone shipped with a 620 MHz ARM CPU that was under-clocked to 412 MHz. Even at the under-clocked speed, it was considerably faster than most other smartphones released in 2007.
For example, the Nokia N95 (322 MHz ARM, which admittedly was dual but took very little advantage of that), Palm Centro (312 MHz XScale), HTC Touch (200 MHz), and the BlackBerry Curve 8300 (312 MHz XScale) all had slower clock speeds, and were all released around the same time as the iPhone.
A 620 MHz ARM chip, even one that was underclocked, was in 2007 speedy. It was certainly not 'tiny by the standard of the time'. I remember a lot of people complaining about various deficiencies in the iPhone back then (such as the lack of 3G, which was a battery saving measure), but I really can't recall anyone complaining about the chipset.
For a pro computer, where performance is absolutely the most important consideration, do you really care about the shape of the case?
Another thing most of people are not bringing up and driving me nuts, is for pro computer, being able to upgrade components at will is one of the most important considerations. From the looks of it, this looks like its not going to be possible with the macpro. With the old Macpro, it was possible to upgrade components. As a matter of fact, being able to upgrade was a big part of the design.
> For a pro computer, where performance is absolutely the most important consideration, do you really care about the shape of the case?
It’s a tiny 10"×6"×6" workstation. That means you can take it anywhere in a backpack. Lots of pros are going to love that, from photographers to DJs to forensic detectives to doctors without borders.
The new cooling system (and thus the case’s round shape) is what allows the new Mac Pro to be so ridiculously small.
I love carrying around my computer with no monitor, no mouse, no keyboard, no power cables and no accessories. I like even more if it is in a delicate housing with an exposed and breakable fan. I think I will throw that in my bag!
Screw The Macbook pro, I am on board with this new and massively inconvenient alternative.
Plenty of pros have to bring Mac Pros to worksites in big flightcases (think live events, shoots on location, outdoors experiments, etc.). The current Mac Pros are ginormous and they weigh over 50 pounds. If they can be replaced by something 1/8th of the size, those people will gladly do so.
When I need to edit video on the road, I bring a MacBook Air. Final Cut works alright on it, but it’s nowhere near as fast as on a Mac Pro. I could bring a MacBook Pro on these trips, but I decided that the little gain in speed compared to the MacBook Air is not worth the extra weight and mass (for me). Taking a current Mac Pro with me to a shoot would mean I need help from others (I don’t drive). I can take the new Mac Pro with me on location, and that will make my work a lot more pleasant.
Really? Last time I saw a "pro" carry a mac pro to an event was 2006, right after they were released, and it was mainly to show it off. Shortly there after even he realized how stupid it was an got a Macbook Pro (released the same year).
If someone in 2013 is lugging around the basically abandoned Mac Pro line ... I question their sanity.
EDIT: The rotting Mac Pro has long since been eclipsed in performance by the Macbook Pro line, so unless you did custom upgrades to it, you likely are carrying 50+ pounds + monitor + (other) for worse performance than a 4.5 pound laptop.
I point is I can't think of any single task where a "photographers to DJs to forensic detectives to doctors without borders" (as you mentioned), needs to tug along a workstation to accomplish said task that can't be done by a macbookp pro, or any medium to high end laptop.
As a matter of fact, I have never seen or heard of instance of any tugging around their powerful station to do any special purpose work. Esp, nowadays. That just sounds like a ridiculous thing to do.
>As a matter of fact, I have never seen or heard of instance of any tugging around their powerful station to do any special purpose work
Live events work, video work, anything where you need >2 external display outs and top performance...you're thinking in terms of the single person, where lugging a Mac Pro around is an inconvenience. I'm thinking of the Mac Pro getting thrown on the back of the truck with the rest of the lighting rig and truss, in which case it's one very small component in a larger travelling system.
I used to work in live events, it wasn't unusual to load Mac Pros in with the rig.
> I used to work in live events, it wasn't unusual to load Mac Pros in with the rig.
Yeah, maybe 8-10 years ago, but now it would seem ridiculous. Or the person who is moving around with a macpro, is not tech savvy enough to know he has better options.
That doesn't mean its a reasonable thing todo.
I have seen my share of live events, managed by couple of laptops/MBpro or at worst case scenarios AIOs.
I think the new macpro is a case of designer's dictating over practical applications. Most of the time it works well for Apple, but I am afraid, this might not be one of those case.
For a seriously pro user, this is nothing more than a toy.
> Yeah, maybe 8-10 years ago, but now it would seem ridiculous.
No, it still happens and it isn’t ridiculous. Try driving a high resolution video wall with a MBP, especially if realtime rendering is involved.
> For a seriously pro user, this is nothing more than a toy.
I already gave you an example of what I’m going to be using it for: video. A MBP can edit regular HD footage (like from a 5D) but it simply doesn’t have enough power to edit footage from a professional camera like a RED Scarlet, Arri Alexa, or Canon C500.
Fashion photographers have the same problem. For instance, a Leaf Aptus camera shoots at 80MP. That means you get RAW files of 80MB in size. Those are not fun to import or edit on a MacBook Pro. Fashion photographers and their crews don’t wait around for that, they bring Mac Pros.
"Beige box" ? No innovation for the past 30 years ?
It saddens me to see people compare a super expensive mac book pro case with the low end $30 cases. I'm not sure if its ignorance or blindness, but you should inform yourself on the high end market for desktop cases (and not just the external look of the case, the internals are what matters). It will cost between $150 and $300 for a really good model, but then again you're comparing it to a mac pro.
Companies like Antec, Lian-Li, Noctua, CoolerMaster, Thermaltake, ... Have spent the last decade or so making fans and cases that can cool down a maxed 2/3 GPU SLI running the latest game while still being silent for the high end gaming and power users market.
High end PC cases are mostly rubbish. I've had many of them, I've built many workstations over the years. None of them compare favourably to the (original) Mac Pro case. There has been very little "innovation" in that market segment. None of those companies have anough money or the market big enough to warrant R&D expenses.
Considering its shape, location of its fan, I see no reason why Dyson cannot come out with a air multiplier extension that mounts to the top.
Silly aside.
I don't care what the hardware inside the Mac Pro is because I cannot upgrade it from what we have seen already. If I am bound by what I buy then its not a traditional "Pro" machine. Its a design statement and nothing more.
Let alone I don't want a family of these enclosures on my desk, I do fully expect the after market to come up with matching towers you can tbolt connect all across your desk. Perhaps one or two will double as monitor stands.
guys, Apple has always operated with the ethos that design matters. Clearly there are a lot of people who don't care about design, who don't care about the way things look. Apple does and it has made the bet that there are a lot of people out there who do too, and who are willing to pay for good design. I would say Apple's bet has paid off, wouldn't you?
If you don't care about design, about the way things look, if you don't believe that form can facilitate function, then that's OK dude. You don't have to buy their stuff. Just stop criticizing others who do. Lots of people care about form and function, lots of people are willing to pay for good design. It doesn't make them naive or stupid or sheep or shallow... it just makes them happy.
If you don't believe in the premise that good design, well-thought out form+function can have an impact on mood, emotion, workplace (or home) atmosphere, etc, then dude, I have to say, you are mistaken.
Also remember that Apple is a for-profit business, and they are not selling "computers" any more than BMW is selling "personal transportation devices".
The Mac pro only has one CPU because that's all they could fit in.
I suspect the real reason is that Apple still {can't be bothered|can't figure out how} to add NUMA support to OS X. The existing 2-socket Mac Pros have been crippled by the interleaved-mode NUMA. (which, interestingly enough, is even hardcoded in the firmware, so when running OSes that support NUMA on a Mac Pro, you can't benefit from that support, presumably so the benchmarks don't shine a bad light on OS X - of course it's blindingly obvious when compared to a same-spec workstation from HP, Dell, etc.)
In fairness, adding NUMA to xnu is probably quite an undertaking. I wouldn't be surprised if they've previously tried and got in a terrible mess. The task/thread management is scattered across the Mach and BSD subsystems, and the virtual memory system handles some conditions (such as low memory) terribly, and seems to scale badly to large amounts of RAM and 64 bits.
Except there are Thunderbolt-to-PCIe enclosures, which come with their own set of problems (latency, hotplugging, I/O memory range allocation, bandwidth, not to mention cost). But I guess Apple can just say "we don't support those, take it up with the enclosure manufacturer".
Drivers for Thunderbolt devices still are necessarily kexts and can thus take down the system without an obvious culprit. For this reason alone I'm surprised they went with it rather than supporting USB3 early on or that they didn't come up with something similar but altogether new, like they did with Lightning. They seem to be trying to cut down third party kext development, and aren't opening the sources for most of their own new kexts. Of course, with IOMMUs available in all Thunderbolt systems, they could have brought back the microkernel aspect of Mach for device drivers, but that would have been even more work.
What does NUMA have to do with SMP setups? NUMA would be about CPUs with separate physical memories acting as one weird processing resource, not a Mac Pro with multiple processes connected to the same memory. What am I missing here?
All modern CPUs have memory controllers built in as opposed to using a distinct "northbridge" chip. The RAM attaches directly to the CPU socket (or soldered BGA). In multi-socket systems, this means there are memory buses coming off each CPU, and for one CPU to reach memory attached to the other CPU, it needs to go through that CPU via the interconnect. Here's some ASCII art of a system with 3 memory channels from each CPU (old Mac Pro):
Going via the interconnect introduces extra latency and consumes its bandwidth. Therefore, it's more efficient to pin threads and/or processes to specific CPU sockets, and to allocate memory for it from that socket's associated physical memory. This requires kernel support, and for absolute top performance you also want a user space API, so you don't need to rely on the in-kernel heuristics.
Lacking software support means you have to fall back to whatever the hardware does automatically. Current multi-socket x86 CPUs all offer an "interlaced" mode for this (sort of like RAID0), where they dispatch to physical socket based on some physical address bits, e.g.:
and so on. This spreads the extra latency evenly. Makes for consistent but poor memory performance. The mode is selected during system initialisation, i.e. is the responsibility of the Firmware (BIOS/EFI). Non-Apple workstations and servers typically let you select this as part of the pre-boot system setup UI.
It depends on the motherboard layout, but you'll find that there are groups of DIMM slots located near each CPU socket in multi-socket systems. Physical proximity helps keep speeds high.
The other thing to mention here is with PCIe 3, the controller for the bus is now in the CPU, which means that if you have two GPUs, you'll start to see GPU affinity problems as well. In previous systems you'd have one bus attached to a southbridge, which is then shared by both CPUs*
I've seen this with linux and nvidia cards, with two CPUs we would get a maximum CPU->GPU transfer of around 7-8gigs, take out a CPU and it'd bump up to 10-13. The z620 has the second CPU on a daughter card, so its simple to do.
I get it, I haven't studied NUMA much since 10 years ago, when we were looking in it in terms of Sparc interconnects and expensive workstation stuff.
From what I've heard from my colleagues who do HPC, NUMA just isn't that effective. The sweet spot right now is either with GPUs or MPI; the shared memory abstraction that NUMA provides is just not very tenable. Where is NUMA being used as a best (rather than legacy) solution right now?
The question is not so much "where is NUMA effective," but, "given that for highest speeds memory needs to be attached directly to the CPU, how do we deal effectively with the case of multiple CPUs?" If you want multi-socket SMP, these days you need to deal with its inherent NUMA architecture. Or sweep it under the rug as Apple did with the old Mac Pro and sacrifice performance.
The fanboyism in the article is just ridiculous, but I think new chasis deserves more credit than just being a "funny shape", I think it is genuinely innovative in a space that hasn't seen much innovation for a long time. Just like the MacBook finally shook up the "ugly grex box" status quo of laptops the new MacBook Pro finally shakes up the "big ugly grey box" status quo of workstation computers. I am curious how many watts/m^3/dB will Apple be able to pack with this approach, it seems to a be a very interesting direction.
Yes. It's strong in GPU terms, but no stronger than (some) currently-available PC workstations. At the same time it's irremediably weaker than (some) currently available PC workstations in terms of CPU, RAM and bandwidth to specialised hardware. Presenting this as a new paradigm of GPU computing seems a bit spinny.
EDIT: However, if the Mac Pro (despite the name and the heritage) is really priced and sold as something more like a high-end desktop than a full-blown workstation, then the new-GPU-centric-paradigm thing might be borne out to an extent - it would be an unusually GPU-heavy machine compared to PC desktops, if not to the workstations of VFX professionals.
I am not interested in a Mac Pro, but I hope it marks the beginning of widely available computers with Cell-like architectures where a moderately powerful, general purpose CPU acts as a frontend scheduler for other parallel or specialized chips with negligible context switching costs.
No price has been announced. Apple may surprise us. Before they came out with the iPad, journalists and analysts thought Apple’s tablet would cost $1000 or more[1].
The current entry level Mac Pro costs $2500, while the most expensive iMac costs $2000. I expect the new Mac Pro to cost between $2000 and $2500.
Apple's big advantage here is that it certainly hasn't paid rack rate for those "pro" graphics cards. How much of that saving it will pass on to buyers is the question.
OSX has been moving progressively further from desktop beasts and towards tablet/phone fusion. The hardware for the mac pro does have a striking design, but I suspect it will wind up being the same red-haired step-child Mac Pro's have been for years unless Apple does a huge about-face on the development of OSX. This certainly wouldn't be the first time Apple produced a cool-looking desktop machine that turned out to be a sales-dud (see G4 cube).
Eh, not really. That's just a popular delusion. Mavericks just further cements that Apple has no plans whatsoever of unifying OS X and iOS on a UI level. To me it even looks like OS X and iOS are moving further apart.
I think sometimes people confuse interoperability with some sort of codebase merge. It makes sense for each release of osx and ios to be more interoperable, and even take lessons learned from one back to the other. It doesn't make sense to merge them, because they are different platforms with different needs.
I find it easier to think of it as an upper mid class workstation. It's not a high end workstation. Exciting and weird thing about is that all extensions to it are planned to be external, we'll see how that works out. Also, my concern is about the cooling design, looks like it will suck dust from below, like a really inefficient vacuum cleaner. Doesn't matter it's supposed to go on desk and not on the floor. Dust is always an issue, no matter where you are. Also, ATI graphics is not something I am excited about.
Is this a Mac fanboi's idea of writeen(1) porn? It reads like a diary of a boy about to go out on a first date, while already dating... Actually, it's kinda cute. Or is it creepy? Lets hope the writer doesn't end up, er, "injuring" his or her self on this new computing phallus.
Anyway, I think the writer needs a cuddle.
(1) Well, yes, initially it was a typo, but on second thoughts I like how it worked out.
If it really was, why do you think you need to tell people? Do you think people are a bit stupid know and need telling? Do you need to tell every one for your own sense of well being?
When it comes to GPGPU there is one big problem with the Mac Pro: Lack of CUDA support. CUDA is the de facto standard for GPGPU programming in HPC. Software support for it is years ahead of OpenCL. I doubt the new Mac is going to change that, if anything it might give some impulse into new workstation design for the likes of Dell and HP, sporting NVIDIA cards. Of course there is still hope that Apple recognizes this and that it offers an NVIDIA configuration. A dual Titan Mac Pro could be a very compelling HPC workstation for those who prefer this to working on a cluster. Even more so since Unix/Linux is the standard OS for HPC and most software tools therefore have OSX/darwin support.
Apart from the initialization code which I'd agree is significantly less painful for CUDA, I haven't seen that much difference in terms of exploiting a GPU's compute units. I never wrote complex applications though, so am I missing something?
The bulky initialization is no longer a problem with a library like PyOpenCl - everything becomes much more elegant...
First of all, nobody is going to port few KLOC of CUDA to OpenCL or vice versa because of one developer's whim. Hence no choice of GPU is bad for developers.
And there are some differences which may be relevant for large projects: CUDA supports ahead-of-time compilation of GPU code down to the lowest level assembly (compilation of several KLOC project takes minutes), direct data transfers between GPUs or even GPUs and network adapters, some subset of C++ (classes without virtual methods, function/class templates, operator overloading) and probably few more useful features I haven't listed.
128 comments
[ 0.29 ms ] story [ 192 ms ] threadWhen the next gen Xeons come along with the release of the new macpro, it'll be literally twice as fast. in side by side comparisons*
*there maybe some apple GPU special sauce for inter GPU communications
The other awesome thing about the z series is that they have a three year next day engineer on site warranty.
although there are work arounds for quicktime.
* Runs Office, which means it runs Excel (and Word). Gotta love Excel. I'm sure there are unholy combinations of calendering software that mean I can interact sort-of sometimes with the Exchange server in the office, but Outlook Just Works too, on a Mac.
* I never ever ever ever have to interact with XF86Config. Literally days of my life wasted on that.
* Runs Omni(Outliner|Graffle|Plan|Focus). I guess I could cobble together something from org-mode, some terrible Visio clone, etc.
* It does sleeping properly when I close my laptop. Always. Reliably.
I just don't have time to run Linux on my laptop.
OS X works better on Macs than Linux distros and Windows do. OS X rarely works well on computers other than Macs.
I love developing in ubuntu, but ubuntu and windows seem to have gone crazy with their user interfaces lately. Ubuntu switched to unity, an interface designed for netbooks. Who has a netbook these days? And wants to run linux on it? Canonical should be targeting their main userbase, developers or advanced users. They should concentrate on making gnome as stable and simple as possible, and include vmware/virtualbox drivers out of the box. If developers have a simple, stable platform then the platform will grow organically.
Windows 8 can't make its mind up about if it should be used with a mouse or a touchscreen. Discoverability is hard, it's really hard to work out that you drag an app to the bottom of the screen to close it, because their are no visual clues. I like the look of metro, but it's really a polished turd.
But Im using Windows 8 and Im in the desktop mode 99÷ of the time (I have 0 metro apps installed).
If you want the best bang for the buck, or even the best bang, it still makes sense to go with a boring HP box and not a Mac Pro. Honestly, the only ones who will probably go with the Mac Pro is the high-end photoshop/toaster crowd, or those who want big bangs in small spaces (e.g. Japan), as well as those with too much money and some amount of sentiment, but this is not even mainstream for the workstation market.
Very small niche but I used to work in media and event installations, and this will be great for that. Especially the tri 4k display output.
I'm sorry, but there are only two things that are new with this mac:
1) its a funny shape 2) it comes with two GPU boards by default.
I'm glad that you are enthused by a new platform, but there is really nothing new. Really. If you want to start writing GPGPU programmes, I congratulate you, its going to keep you busy for a good few months. But you can do that now, without a $10k mac "pro"
Over the next few months, we'll be hearing about how the CPU is dead and that all you need is GPU. However much like the initial flash/iphone argument it over looks the main point. The Mac pro only has one CPU because that's all they could fit in. It wasn't a bold visionary move, it was a compromise because of the form factor. (much like the original iphone, to get the battery life up, they had to use a tiny(by the standard o the time) CPU, which meant no flash)
There is one large elephant in the room when it comes to GPGPU and that is bandwidth. Getting stuff from main memory to GPU memory take a lot of time. Yes PCI3 is wonderfully fast, but compared to RAM<=>CPU its like reading a floppy disk.
That asside, there is really nothing new or special about the macpro. In fact for VFX land, its rather a disappointment. Yes Mari runs really really fast on it. But then its got two large GPUs and a stonking SSD. Everything other programme is going to be CPU constrained(unless they've been pushing stuff to the GPU).
Please excuse my rant, I've been inundated with people wanting new macpros, despite them already having a far more powerful z620 with fusion IO cards and quadro k5000s. cult of the shiny and new I suppose.
Most professionals don't want equivalent software, they want the software that they know and that is supported by all their plugins and hardware.
You probably do very little dev. Of all the development environments I have seen so far, nothing beats the Mac. Even Emacs for Mac is light years better than plain old Emacs.
http://aquamacs.org/
However, a mac mini is good enough for such things normally. I'd stick with my custom built PC for anything else.
I don't think it should make you mad. Different strokes for different folks.
If I go to a car forum I bet I can find a lot of people who love BMW, and more still who love Mercedes Benz. They'll probably happily argue with each other about who makes the better cars, and why one brand is far superior to the other. And then there's probably also some people who are happy with either car.
For you, there's no software on OS X that doesn't have an equivalent on Linux or Windows. But that might not be true for others (for example, I don't consider GIMP to be 'equivalent' to Photoshop...but you might). Another example - I used to work in live events, and used Mac Pros to drive video rigs. Could I have built a more powerful and cheaper hackintosh? Probably. But I doubt the build quality of what I'd put together would be superior. The cheese-grater Mac Pro could take a beating.
Again, my needs are very different to yours. Please don't get angry on my account!
Try actually doing the math before you declare other peoples' considered choices a "waste of money".
This is like the fella above basically saying "high-end professionals" dont need to be logical anymore.
I feel like you don't understand actual business or finances at all. Are you perchance quite young?
Because I have better things to do with my time than assemble a poor substitute for a cohesive just-works experience.
> what software is there for mac OS that there is a not an equivalent to for linux or windows
As a developer, most of the exact same software I use every day on OS X runs on Linux just fine. And yet, Linux doesn't provide me with a good user experience, and it doesn't just work[1].
But fundamentally, you are just mad because other people like something you don't. Dwell on that for a while.
[1] If you're going to try and argue with me about how great Ubuntu or Mint or whatever else are these days, let me save you the time: You're wrong, and there is absolutely nothing at all that you can say that will convince me otherwise, because I do use them, but they continue to fail me as primary desktops. Your words conflict with reality, and more words won't change that.
otherwise, it's going to take a while.
I'm a sysadmin and I make things just about bearable for my users. I install VLC, gitk, flash and chrome. I make sure that the email client is setup and talking to ldap/AD. Granted we use RHEL5/6, so there is a lot of work for me to do to make things nice for them. However RHEL/centos/SL whilst old and not trendy is a loverly slow moving target for our customers to jump onto.
Granted mint and the like make installation easy, but integration is still hard.
http://paulgraham.com/mac.html
Where did you get that? It sounds entirely implausible that a company renowned for its design would decide on the form factor first, and then try to fit stuff in there as well as possible...
to give a non technical example, the london underground map sacrifices geographic accuracy for simplicity. A very large, and successful compromise.
Also, where could they fit another CPU? Each one kicks out almost 180 watts of heat, and must have RAM and power to make it work.
It also complicates GPU<=>CPU and GPU<=>GPU communications. You either have to shove all the GPUs on one CPU and waste lots of QPI cycles on moving RAM back and forth, or have one GPU on each CPU and build some very clever glue to allow the two GPUs to share memory.
Ergo, compromise.
Macs are still needed because of quicktime, and final cut (plus a few other niche products)
People use the Laptops a lot, because they are actually quite good. Certainly longer lasting than anything from Sony or Toshiba.
I'll cite apple's own website in this:
One of the biggest challenges our engineers faced with this new design was how to join the front to the back. The enclosure is so thin, it’s not possible to weld the pieces using traditional methods.
Formfactor first, everything else second. Hence why apple makes the products they do.
I am of course willing to be convinced otherwise, perhaps you have some sources to backup your claim?
edit, syntax error
I am sorry, but I just don't know where to start. That's not how industrial designers work, it's a caricature.
Apple believes that design is essentially three things: form factor + hardware inside form factor + software. The three have to play together seamlessly for the overall design to be good. We can debate on how effective they are at this, but that is what Apple believes. Steve Jobs talked about the interplay of hardware and software many times.
Steve Jobs also frequently discussed compromises. He believed that consumers paid Apple to make decisions about what was useful and what wasn't. So including Flash in iOS would have meant that users could watch all web video (only 25% at launch!), but they looked at the stability and the battery life and made the decision that, no, Flash was not worth it. So Apple believes in tradeoffs.
Sometimes these tradeoffs are at the cost of the aesthetic. For example the black plastic on the back of the original iPhone, or the black plastic used on the entire iPhone 3G when they wanted to increase margins and go for market share. This is design. This is a company making tradeoffs.
Sometimes the tradeoffs do come to preserve the form factor. I struggle to type on my iPhone, but I was so quick on my BlackBerry that the software lagged behind me. But Apple made the tradeoff.
The antennae on the iPhone 4 is a great example of Apple design. It's accomplishing so much. It means there's no need for black plastic on the device, so the aesthetic is improved. It also provides the structural rigidity for the device, which means they can make it feel super sturdy (love Samsung's aesthetic, but if you hold an iPhone 5 and a Galaxy, there's a real difference to the feel). That sort of outcome isn't the product of a guy in a turtle neck with Auto CAD putting lens flares on everything. That's serious design.
Jony Ive talked a lot about the process for making the MacBook Air. They wanted to build a fast, ultraportable computer without making it one of those shitty 9" Taiwanese pieces of crap. So they had to work with Intel to build a 60% smaller Merom. Because they didn't want to compromise the internal hardware by shipping a slow processor. They had to develop Unibody so they could strip out internal fixtures that would add weight and reduce the structural integrity of the product. Commitment to the aesthetic.
You've cited (it's cited, btw, not sighted) the iMac. Apple wanted to make it thinner and lighter. At the same time they wanted to ship better components inside it. So they had to keep the space inside the same size or make it bigger, or they weren't going to be able to ship the product. So they spent time developing ways to make it thinner and more aesthetically appealing without compromising the hardware components. In this instance I'd argue that deference to the hardware components came first. Much like the MacBook Air, where so much of the work went into developing a tiny logic board so the device could be full powered.
tl;dr Apple makes tradeoffs on behalf of consumers.
> Hence why apple makes the products they do
A quickie on this: hence why is never correct. Just use hence.
This mac pro is all about GPU. The second CPU would require a larger volume and would add too much cost(in both style, support components and thermal envelope). This was deemed unacceptable and was therefore cut to keep the final design within spec.
It deemed an acceptable risk to release a product that is notably less powerful than the equivalent workstation.
This feels really lazy and borderline trollish.
Over the last decade or so Apple (and others!) have fairly consistently demonstrated a belief that design is multifaceted: the form factor, the software, the components inside the form factor.
Increasingly we see these intersecting. The form factor for the iPhone is also the hardware: the screen and the antennae. The unibody manufacturing process for MacBook Pro means that the trendy aluminium form of the product also provides its rigidity and reduces the number of components required to hold it together. More room for batteries, which increases the battery life. Apple and Intel worked together to develop a Merom chip 60% smaller than the standard in order to ship the MacBook Air, which was at the time the thinnest notebook around.
Apple doesn't design the form factor and then work out what components need to go inside it. Very often the overall design (form factor + software) is where Apple can innovate, because Apple doesn't typically manufacture components, but they frequently make compromises of design in order to accommodate hardware (and vice versa). IR receivers on the original MacBook Pro (horrible black spots on the bezel), the home button on the iPhone (an absence of hardware buttons on the front of the device would be more pleasing). The black plastic at the base of the original iPhone for Wifi (since engineered out by a combination of component and product design).
People will look back on the last two or three decades as a renaissance for design. Apple has contributed, but across the board the importance of design -- and by design I mean the aforementioned intersection of form, software, and hardware -- is visible in products from Samsung, Sony, HTC. Heck even Microsoft has realised that design is more than commissioning Brian Eno to write the Windows boot music.
Back to the Mac Pro: I'm sure Apple threw out a dozen designs which had space for more RAM, or which had more CPU sockets, or which had internal expansion. But Apple makes tradeoffs for the overall goal. Sometimes the tradeoffs are such that hardware suffers (I can still type way faster on a BlackBerry than on an iPhone), and some times the aesthetic is diminished (I bet the visible sensors on the iPhone 5 really rankle with Jony Ive). But to suggest that since 1998 Apple has consistently developed the form factor for products first is ludicrous.
It's always funny when people use obscure(?) but interesting anecdotes to make a joke. It sort of shows the breadth of one's knowledge. Well, at least I had never heard of this.
Design has nothing to do with specs, and something to do with form factor. So yes, it sounds plausible that a company renowned for its design would focus on the design over the specs.
I think you might be misinformed. The first-generation iPhone shipped with a 620 MHz ARM CPU that was under-clocked to 412 MHz. Even at the under-clocked speed, it was considerably faster than most other smartphones released in 2007.
For example, the Nokia N95 (322 MHz ARM, which admittedly was dual but took very little advantage of that), Palm Centro (312 MHz XScale), HTC Touch (200 MHz), and the BlackBerry Curve 8300 (312 MHz XScale) all had slower clock speeds, and were all released around the same time as the iPhone.
A 620 MHz ARM chip, even one that was underclocked, was in 2007 speedy. It was certainly not 'tiny by the standard of the time'. I remember a lot of people complaining about various deficiencies in the iPhone back then (such as the lack of 3G, which was a battery saving measure), but I really can't recall anyone complaining about the chipset.
I've been saying that for years. http://chris.pirillo.com/are-gpus-the-future-of-cpus/
Somehow.
In some way.
The shape and style is nice, and I think innovative in how it allows for better cooling of the system. But what do I know, I still run xp :P
Another thing most of people are not bringing up and driving me nuts, is for pro computer, being able to upgrade components at will is one of the most important considerations. From the looks of it, this looks like its not going to be possible with the macpro. With the old Macpro, it was possible to upgrade components. As a matter of fact, being able to upgrade was a big part of the design.
It’s a tiny 10"×6"×6" workstation. That means you can take it anywhere in a backpack. Lots of pros are going to love that, from photographers to DJs to forensic detectives to doctors without borders.
The new cooling system (and thus the case’s round shape) is what allows the new Mac Pro to be so ridiculously small.
Screw The Macbook pro, I am on board with this new and massively inconvenient alternative.
No one is going to carry this around with them.
When I need to edit video on the road, I bring a MacBook Air. Final Cut works alright on it, but it’s nowhere near as fast as on a Mac Pro. I could bring a MacBook Pro on these trips, but I decided that the little gain in speed compared to the MacBook Air is not worth the extra weight and mass (for me). Taking a current Mac Pro with me to a shoot would mean I need help from others (I don’t drive). I can take the new Mac Pro with me on location, and that will make my work a lot more pleasant.
If someone in 2013 is lugging around the basically abandoned Mac Pro line ... I question their sanity.
EDIT: The rotting Mac Pro has long since been eclipsed in performance by the Macbook Pro line, so unless you did custom upgrades to it, you likely are carrying 50+ pounds + monitor + (other) for worse performance than a 4.5 pound laptop.
Eh, no. A topped out Mac Pro is a LOT more powerful than a topped out MacBook Pro. Let’s compare specs, I think you will see what I mean:
MacBook Pro: 2.8GHz Quad-core Intel Core i7, 16GB 1600MHz memory, 768GB Flash Storage, NVIDIA GeForce GT 650M
Mac Pro: Two 3.06GHz 6-Core Intel Xeon (12 cores), 64GB DDR3 ECC SDRAM, 2048GB Flash storage, ATI Radeon HD 5870
As a matter of fact, I have never seen or heard of instance of any tugging around their powerful station to do any special purpose work. Esp, nowadays. That just sounds like a ridiculous thing to do.
Live events work, video work, anything where you need >2 external display outs and top performance...you're thinking in terms of the single person, where lugging a Mac Pro around is an inconvenience. I'm thinking of the Mac Pro getting thrown on the back of the truck with the rest of the lighting rig and truss, in which case it's one very small component in a larger travelling system.
I used to work in live events, it wasn't unusual to load Mac Pros in with the rig.
Yeah, maybe 8-10 years ago, but now it would seem ridiculous. Or the person who is moving around with a macpro, is not tech savvy enough to know he has better options.
That doesn't mean its a reasonable thing todo.
I have seen my share of live events, managed by couple of laptops/MBpro or at worst case scenarios AIOs.
I think the new macpro is a case of designer's dictating over practical applications. Most of the time it works well for Apple, but I am afraid, this might not be one of those case.
For a seriously pro user, this is nothing more than a toy.
No, it still happens and it isn’t ridiculous. Try driving a high resolution video wall with a MBP, especially if realtime rendering is involved.
> For a seriously pro user, this is nothing more than a toy.
I already gave you an example of what I’m going to be using it for: video. A MBP can edit regular HD footage (like from a 5D) but it simply doesn’t have enough power to edit footage from a professional camera like a RED Scarlet, Arri Alexa, or Canon C500.
Fashion photographers have the same problem. For instance, a Leaf Aptus camera shoots at 80MP. That means you get RAW files of 80MB in size. Those are not fun to import or edit on a MacBook Pro. Fashion photographers and their crews don’t wait around for that, they bring Mac Pros.
It saddens me to see people compare a super expensive mac book pro case with the low end $30 cases. I'm not sure if its ignorance or blindness, but you should inform yourself on the high end market for desktop cases (and not just the external look of the case, the internals are what matters). It will cost between $150 and $300 for a really good model, but then again you're comparing it to a mac pro.
Companies like Antec, Lian-Li, Noctua, CoolerMaster, Thermaltake, ... Have spent the last decade or so making fans and cases that can cool down a maxed 2/3 GPU SLI running the latest game while still being silent for the high end gaming and power users market.
http://www.techeta.com/wp-content/uploads/2011/01/Lian-Li-PC...
http://www.overclockers.com/lian-li-pc-v1020-case-review
Silly aside.
I don't care what the hardware inside the Mac Pro is because I cannot upgrade it from what we have seen already. If I am bound by what I buy then its not a traditional "Pro" machine. Its a design statement and nothing more.
Let alone I don't want a family of these enclosures on my desk, I do fully expect the after market to come up with matching towers you can tbolt connect all across your desk. Perhaps one or two will double as monitor stands.
If you don't care about design, about the way things look, if you don't believe that form can facilitate function, then that's OK dude. You don't have to buy their stuff. Just stop criticizing others who do. Lots of people care about form and function, lots of people are willing to pay for good design. It doesn't make them naive or stupid or sheep or shallow... it just makes them happy.
If you don't believe in the premise that good design, well-thought out form+function can have an impact on mood, emotion, workplace (or home) atmosphere, etc, then dude, I have to say, you are mistaken.
Also remember that Apple is a for-profit business, and they are not selling "computers" any more than BMW is selling "personal transportation devices".
and that's OK.
I suspect the real reason is that Apple still {can't be bothered|can't figure out how} to add NUMA support to OS X. The existing 2-socket Mac Pros have been crippled by the interleaved-mode NUMA. (which, interestingly enough, is even hardcoded in the firmware, so when running OSes that support NUMA on a Mac Pro, you can't benefit from that support, presumably so the benchmarks don't shine a bad light on OS X - of course it's blindingly obvious when compared to a same-spec workstation from HP, Dell, etc.)
In fairness, adding NUMA to xnu is probably quite an undertaking. I wouldn't be surprised if they've previously tried and got in a terrible mess. The task/thread management is scattered across the Mach and BSD subsystems, and the virtual memory system handles some conditions (such as low memory) terribly, and seems to scale badly to large amounts of RAM and 64 bits.
Drivers for Thunderbolt devices still are necessarily kexts and can thus take down the system without an obvious culprit. For this reason alone I'm surprised they went with it rather than supporting USB3 early on or that they didn't come up with something similar but altogether new, like they did with Lightning. They seem to be trying to cut down third party kext development, and aren't opening the sources for most of their own new kexts. Of course, with IOMMUs available in all Thunderbolt systems, they could have brought back the microkernel aspect of Mach for device drivers, but that would have been even more work.
Lacking software support means you have to fall back to whatever the hardware does automatically. Current multi-socket x86 CPUs all offer an "interlaced" mode for this (sort of like RAID0), where they dispatch to physical socket based on some physical address bits, e.g.:
and so on. This spreads the extra latency evenly. Makes for consistent but poor memory performance. The mode is selected during system initialisation, i.e. is the responsibility of the Firmware (BIOS/EFI). Non-Apple workstations and servers typically let you select this as part of the pre-boot system setup UI.It depends on the motherboard layout, but you'll find that there are groups of DIMM slots located near each CPU socket in multi-socket systems. Physical proximity helps keep speeds high.
I've seen this with linux and nvidia cards, with two CPUs we would get a maximum CPU->GPU transfer of around 7-8gigs, take out a CPU and it'd bump up to 10-13. The z620 has the second CPU on a daughter card, so its simple to do.
*my memory is a bit hazy, I'm open to corrections
From what I've heard from my colleagues who do HPC, NUMA just isn't that effective. The sweet spot right now is either with GPUs or MPI; the shared memory abstraction that NUMA provides is just not very tenable. Where is NUMA being used as a best (rather than legacy) solution right now?
I think you mean the Vaio?
EDIT: However, if the Mac Pro (despite the name and the heritage) is really priced and sold as something more like a high-end desktop than a full-blown workstation, then the new-GPU-centric-paradigm thing might be borne out to an extent - it would be an unusually GPU-heavy machine compared to PC desktops, if not to the workstations of VFX professionals.
Who says that? For all I know the cards are using AMD CrossFire.
Not true, the GPUs are connected with a bridge and so share the workload efficiently.
> "New"
New to mac maybe but my desktop has 3 GTX 670s which provides some pretty serious computational power.
No price has been announced. Apple may surprise us. Before they came out with the iPad, journalists and analysts thought Apple’s tablet would cost $1000 or more[1].
The current entry level Mac Pro costs $2500, while the most expensive iMac costs $2000. I expect the new Mac Pro to cost between $2000 and $2500.
[1] http://appleinsider.com/articles/10/01/19/minor_issues_could...
I think sometimes people confuse interoperability with some sort of codebase merge. It makes sense for each release of osx and ios to be more interoperable, and even take lessons learned from one back to the other. It doesn't make sense to merge them, because they are different platforms with different needs.
Anyway, I think the writer needs a cuddle.
(1) Well, yes, initially it was a typo, but on second thoughts I like how it worked out.
Or are you just a karma whore?
The bulky initialization is no longer a problem with a library like PyOpenCl - everything becomes much more elegant...
And there are some differences which may be relevant for large projects: CUDA supports ahead-of-time compilation of GPU code down to the lowest level assembly (compilation of several KLOC project takes minutes), direct data transfers between GPUs or even GPUs and network adapters, some subset of C++ (classes without virtual methods, function/class templates, operator overloading) and probably few more useful features I haven't listed.