Mac Pro Refesh in March - Page 9

I think you will have to wait till June if it happens at all.

This is pretty typical with Boxx. When things don't scale well with more cores, they overclock the machines. They're supposed to be quite good. Given that it's a smaller vendor, they're probably able to get earlier orders on cpus and things. They are obviously smaller than Apple. Their notebooks aren't manufactured here like many of their towers. I don't know if they've changed. The GoBoxx was rumored to previously be a rebranded Sager. If you're looking at Windows, Lenovo's notebooks look quite nice. One thing I should mention is that whether it's Apple or another brand, notebook displays are annoying as hell to calibrate. They've all switched to LED. LED is tough to calibrate with most colorimeters. I've considered switching myself at times as I deal with bootcamp part of the time anyway. Much of it is related to hardware options and driver performance. For me it wouldn't be until the next thing + Mountain Lion come out for a fair comparison. I should mention it has nothing to do with Apple's silence or refresh cycle.

Just to reiterate, my biggest issues are centered around gpu selection and drivers. Apple has always been a bit behind in that regard, but these things have become much more important to me. Personally I don't think a new mac pro will come out until mountain lion and I'm fine with that.


You post way too much rhetoric.

Well. It's a free world.

Depends on what work you're doing. But you have to decide if the Mac platform/hardware is meeting your requirements.

Looking at the Mac Pro. It's the well rehearsed arguments. Apple could have upgraded it in numerous ways other than CPU to make it a compelling purchase along with a price cut. Especially for the entry model. In so many other ways than CPU it has languished.

9 years is a long time in computing. The current Pro looks like time has passed it by. In an era where Apple puts so much energy in an era of portability there's no reason why the Mac Pro's place in the scheme of things can't be re-factored into the paradigm shift.

Lemon Bon Bon.

That's about it. I guess if he's waited this long then why not wait until the Summer is over and Mountain Lion is out. That's a decision for him, I guess. Mind you, I waited ten years for a decent Mac Pro with a decent GPU. I almost plumped for a dual core 3 gig Mac Pro...but waited a bit longer and got the iMac instead.

It's clear that everything will get a revamp this year with Mountain Lion. Laptops, obviously...and the desktops too.

I'd like a cheaper more compact pro with 6-8 core cpus and thunderbolt with a damn fine gpu.

If the Pro doesn't come by the time Mountain Lion launches then it will look ominous for the Pro.

Rhetoric? It's been that long since the Pro was dusted off that every argument about the Pro is rhetoric.

Lemon Bon Bon.

Yes. I am down with that, for sure.

http://www.youtube.com/watch?v=jwBY2...c0RVAAAAAAAABw

Those were the days.

What was that? An entry model for a £1000?!!! The current entry Pro is twice that...wondered why sales sucked.

Lemon Bon Bon.

That's right, Mr. Bon Bon.

...and the top end...

2,999.00 USD = 1,873.18 GBP!

Apple could price the Pro like that today. And let desktop buyers decide on AIO beauty and power with the lovely screen...or a more powerful upgradable (allegedly) tower without the screen.

Lemon Bon Bon.

The chips alone cost more than that. (well, if I'm looking at the right chips and yes, I know Apple gets a screaming deal on them beyond the price that we can see because they're a manufacturer)

Screw those particular chips. Heh... The Ivy i7s offer more bang for buck. They can take the tower market mainstream again for their 'forgotten' creative pros.

They could go 4 core ivy for the lower and middle with 6 core xeon(?) at the high end with the old G3 prices and put in a juicy gpu in a compact case. You get a kickstart to tower sales.

Anything dual processor Xeon related £2000 and over using the existing case.

Lemon Bon Bon.

I miss the theatrics that went into the old keynotes. These days it's just:

- hey check out this new store we have
- look how many people visit the stores we have
- here's a new device that looks the same as the last one
- here's a new advert we made to explain why you should buy the new device and not the last one

When the performance mattered, they really put in the effort to show it. They could arguably do the same with the Mac Pro now.

Everyone else is going to sell high resource buyers on the incremental improvement over the last model and inevitably, Apple won't rival them on performance per dollar. All they have to do is demonstrate the appropriate use for the appropriate buyer.

Offer a stunning compact design at a better price with decent performance (within reason ) and show off the trump card: Thunderbolt. Nobody else is using it because they don't see the benefit. If Apple could allow you to hook up another machine as a co-processor with zero-config that appears transparently to the Mac OS, that's enough.

Just have a demo of the fastest E5 Xeon PC running a Cinebench render, then have a demo of a chain of 4x Mac Pros hooked up doing the same render. No matter how much of an incremental improvement the PC gets, Apple always has the integrated model to fall back on - the combination of software, interconnect options and selected hardware.

They can make the Mac Pro desirable again for one last run. Either that or put a Xeon in the iMac.

Marvin - So you would like a good portion of the next press conference to focus on tech specs? I can't say I blame you there. I would love to see that myself.

The question I would like to know though is do a majority of people sitting there care about the specs? Or do they think people will buy what they buy and make choices there.

These Sandy Bridge E chips came out a month ago or so? Apple probably had its plans set in place since it had to wait so long.

I just don't believe desktops are dead quite yet.

Great post.

Hey, Marv', where ya been?

I think you have a point with the keynotes when Apple was in 'survival mode.'

I understand that we've had a few home runs with iPod, iPhone and iPad. The iMac had a decent redesign and the Air has been tweaked. Pretty amazing stuff.

But the Pro really stands out as the machine that was once the jewel in the crown, the flagship and it would be nice for it to have an emphatic redesign that puts it back on the map.

I love your idea of a compact pro with thunderbolt and daisy chaining another box for 'co-processor' mode!

Apple have the OS. They have the means. If anyone could make this X-Grid work, it would be them. They have Open CL. Do they have the will?

Imagine you have a 4 core ivy bridge with a decent dedicated GPU. £999. You buy a 2nd box and you have a 'dual processor' Mac Pro. You drive a bit more volume and still get the £2000 price. It created a compelling render farm prospect for Mac heads. When it's time to upgrade? Just add another box instead of getting rid of the other one. You end up with 8 cores instead of 4 instead of being ass-reamed over one box.

I think that would put the Pro tower compact back on the map.

Drool. Imagine 4 compact pros daisy chained. Der-rool.

(I wouldn't be adverse to a 6 core xeon being put in the top end iMac either...)

Lemon Bon Bon .

£999 per box. You get 4 cores. That's pretty decent. Lose the screen but you get better GPU as compensation.

If you pay £4k you get 16 cores. But you build as you need. 1 a year? After 4 years, your Mac 'workstation' grows with you over time. Maybe in a nice stackable design. Lego bricks.

Lemon Bon Bon.

I'd second that. It was a great post.

It completely nails the thread hands down.

Lemon Bon Bon.

In fairness, Apple's decision on the future of the Mac Pro has already been decided in reality.

People do care about price and specs. When Apple did it right with the iPad...people ooh and ahhh at the amazing £399 price when most surmised Apple would charge £1000!

And the link to the original G3 tower which had an amazing price! Hard to believe in those days the G3 tower occupied the iMac price range and the imac pretty much the £500-£1200 range.

I'd like to see Apple drive the Pro back into the mainstream again...

I think it's a case of the iPad stealing the limelight for now and then the Mac Tsunami will begin...

Lemon Bon Bon.

I don't think they should necessarily focus on technical specs but usage scenarios. In the past, it just happened to be the case that both coincided because everyone struggled for better performance. Steve Jobs always knew best how to describe these things and it's about areas of emphasis. Products are built and designed for their target audience.

I think the iPhone is the device that should always be with you as your personal assistant and communication device.
The iPad will eventually replace the laptop. I don't think the laptop design is right but for now, it is necessary for performance and flexibility.
The Mini is a desktop that is cheap and very small - ideal entry model desktop, easy to upgrade and fits lots of usage scenarios.
The iMac is simple, one plug and you are away and good value when you take into consideration the displays.

The Mac Pro is for high resource users but the current design is for high resource users of a past era that is long gone. Optical storage is dead, almost no-one is buying PCI cards and the prices are way too high.

These users have demonstrated time and time again that no upgrade is good enough. They will pay whatever their budget allows to achieve the performance they need so the design should accommodate that.

WWDC is an event for developers so software should be the subject of the event but these users need good hardware too and a scalable hardware model would go down well. I think the main focus should be resolution independence and the further convergence of iOS and Mac OS and the hardware can be the surprise 'oh and one other thing'.


When it comes to emphasising the performance, I'd like to see the Mac Pro move exclusively to higher-end chips. Intel keeps building Xeons as they are the best chips they can offer and Apple keeps building Mac Pros around them as that's how they can build the best workstations. The problem they have is that with an old design, they don't offer the best chips for the money you pay.

I think a realistic expectation is to have a 6-core Xeon compact Mac Pro for £2,000. High margins for the low volume but good value for the target audience and scalable. Buy 3 of them and you have an 18-core/36-thread render farm for £6,000 that is easy to upgrade at any point in time, regardless of what Intel releases and easy resale for old machines.

Value has been an issue there for some time. With the initial mac pro, for $2500 you could purchase a machine that was firmly seated above the imac in raw performance. There are a lot of other features that you gain, but at that transition, the value is quite weak. You really need to spend quite a lot more to approach anything that is worth buying. In terms of eliminating the current PCI standard (which I don't expect will happen yet in this segment) you have to ensure that there is some kind of path forward for these companies or the functions they provide. It doesn't matter if it's thunderbolt or something else. It just has to actually work, which has always been a big focus for Apple. The problem has been their trend toward making whatever they feel like, and letting everyone else work around it. They can do better in that regard. I could "personally" make due with thunderbolt assuming adequate ports for a couple displays and extra storage, or by keeping the internal bays and just running a NAS via 10G ethernet or something of that sort. It shouldn't be any stretch of the imagination though for a company that produces computers to look at how they're being used, and ensure that some path forward will exist even if there's a gap. Usually you can just stretch the old hardware an extra half or full generation while you wait. If something is not likely to surface, that's what drives customers away. I can think of several Adobe and Autodesk applications that appeared in 64 bit or with new features on Windows a generation before Macs, yet this did not trigger a mass migration. Users just waited it out. The same will happen here as long as there can be a bridge in the workflow within a reasonable time period. If it's a case of a big job coming up next month, you buy now. You don't wait for updates. It might suck, but in those kinds of situations, you just accept it, as even if something new comes out, it's generally a good idea to wait for bug and limitations to be assessed.


I keep an eye on both Windows and Macs. When I respond to some of those guys, it's clear that they aren't completely examining the other side in depth. Sometimes it's cheaper. Usually the base configurations are much much cheaper. My concerns have kind of gone away from the fastest cpus to a minor degree. They help, but they're not the only thing that helps. My issue is with gpus and the driver features available. It's not one of x machine hasn't been updated in a long time so I'm leaving. I'm sure you get this. It's not that likely that I'll switch, but it's reasonable to see how things look once both are upgraded, and I did check out what it would cost to swap several software licenses. When you're not in a rush, you can just email the company and wait for a response.


The base chips aren't expensive. I suspect sales are weak on the low end of the mac pro line given the poor value there. It's kind of awkward. They went with a semi-exotic internal design, yet put in a cheap cpu, possibly to control costs, but the line is clearly balanced toward its higher end making it somewhat of a niche product. It's quite a bit more in line with the average price points at the upper end. Apple is still typically a bit more expensive, but it varies. Usually when others make these comparisons, they look at something isolated that happens to favor their argument.


The imac wouldn't be a bad solution there if it wasn't designed contrary to what any of these people require. Contrary to Wizard's theory, matte screens are heavily favored. The problem with glossy is it's too difficult to control the results even in moderate lighting. I think he just hated older laptop displays, but at the higher end, some anti glare coatings were far better than what he may have experienced. Even the perception of deeper black levels isn't a huge advantage for those guys, as it's more about control and detail than absolute contrast ratio. The problem with high contrast is that the steps become too great in their spacing. I could go into talking about dithering, but that would make this post even more boring.


They do that with posted benchmarks on their site.



First thunderbolt isn't a trump card. The PC oems can implement it if they wish given that Ivy Bridge supports it. Second problem with the logic. I can't find a proof of concept that thunderbolt can be used in such a manner. It had problems initially just with target disk mode as an interconnect (initially terrible speeds at release). You could do this via ethernet, but at some point it becomes a better idea just to have a render farm. This seems more like an idea for an old machine. Right now the gpu has been increasing in importance, and this has historically been a weak area for Apple. We all know that. We accepted it and bought them anyway. I'd like to see them really address this, and I feel like they should pay more attention to the Mac line top to bottom because customers are being leveraged in by phones and tablets. The experience should be consistent. I'd like to see NVidia cards. Even better would be Quadro options with the same driver features that Windows gets. I'm just missing the point where they have a more sophisticated interconnect than a typical PC. I've searched for some verification of this.

With 3d content creation and editing in general, most of those guys would be better served by going the small node based direction if the goal is solely faster renders. Faster workstations are made to benefit you on the things where an almost real time response is expected. When newer software chokes the older hardware, or you start having to make compromises on settings, the tendency is to upgrade. Trends toward smaller hardware simply comes up when the growth of such needs is flat at the top and everything below it slowly catches up. Right now the biggest issue with a mac for such work would be that their gpu support is terrible. To me that's a bigger irritation than any other individual issue. It's one of those things where you can often work around it, but workarounds aren't always fun. As I mentioned I'm waiting to see how Mountain Lion looks in this regard. It's not like Kepler based cards that I'd purchase are really available yet anyway, so it's pointless for me to look at hardware unless it's the older thing.

As some one else said it is a fee world, but you really look foolish and rash. I wouldn't even bother until all the major manufactures have laid their cards on the table.



And you act surprised? Really. Performance leaks and other info have been on the net for months, this is squarely the function of the new Sandy Bridge E hardware. Thus every manufacture will be able to demostrate solid performance gains when their hardware debuts.

As to Boxx I see no wisdom at all in building a business around their hardware.

So has everybody else building this type of hardware!!! Why? Because Intel screwed up and dragged out the debut of Sandy Bridge E for almost a year. In your imagination you see this as an Apple issue but frankly you are not be totally honest with yourself. I'm expecting Apple to take what is currently the Mac Pro in a different direction, that may impact release of an new hardware, but to imply that the current lateness is Apples fault just isn't demonstrating a solid grasp of where Intel put Apple and all manufactures building such hardware.

Now I'm not saying Apple has been perfect here as Mac OS and yes the Pro need more work for professional use. However it isn't like Apple has given up here, as each release of Mac OS has been catching up with standards. The Pro could have used an upgrade to th GPU but let's be honest the payoff won't be as great as simply waiting for NVidia or AMDs latest generation of GPU wonders.

Beyond all of that frequent upgrading isn't worthwhile like it was a few years ago. Even the most progressive GPU and CPU manufactures gain little yearly compared to a few years past. Often a new years GPU is a rebadging of last years hardware with some minor tweaks. One can fall for the marketing if they want but rapid upgrading is more a mental benefit than a practiced and rational way to spend money.

Don't believe me. Take a look at how long it took Intel to deliver the performance upgrade in Sandy Bridge E. Sure it is a good chip but look at how long it's predecessor has been at the top of the heap.

Of course there are good ones out there. However Apple currently sells class leading laptops so I'm not to certain that shopping else where is going to gain you much.

In any event your post is like many here, you stomp your feet like a child saying Apple is late and I'm going to go play elsewhere. No where in your post are there sound reasons for leaving the platform, just the morbid belief that because something isn't delivered on your schedule it is somehow justification for throwing a tantrum.

Honestly what are you going to do if Apple refactors the entire desktop lineup and replaces the Pro with something much different? That is a refactored machine targeting advanced users and professional desktop users. Like it or not the writing is on the wall for huge tower type computers on the desktop. We are just starting to see the work going into the hardware to produce such machines in the near future. Intel will be shipping SoC far beyond what we have today, 3D memory hardware is being explored and SSD functionality is coming along nicely. So yeah big boxes on the desktop will become a thing of the past. More so engineers have little choice as with increasing clock rates, distance inside that box becomes a significant issue.

This part is especially fascinating to me as I could definitely see it happening.

Do you see it replacing all laptops or just low end ones?

I didn't want to quote the entire post though I actually favored it to kind of look back on.

I don't personally use them, but I don't see anything wrong with their hardware design. They've got a reasonably good list of clients. They tend to do a lot of very specific testing, so in the end you're probably paying more for support than anything else.




Looking at the last AMD generation, it wouldn't have offered a meaningful upgrade from the 5870. A major complaint seems to be paying the same price late in the cycle, and that always sucks. Even the latest gpus that are just coming out haven't pushed things that much further. NVidia's somewhat popular Quadro 4000 came out in 2010. It's still at the top of the pile. I don't expect Quadro drivers on kepler for a bit.

Here are some thought on AMD. I know you don't really care for Anandtech, but this isn't exactly a sensationalized article. The link is to the conclusion page.


I agree with this. Part of my reasoning for holding off on an upgrade is waiting to see what Apple does on the driver and feature side. Part of it is that I would like the latest hardware since new hardware is just starting to trickle out even if it's nothing revolutionary. The last bit would be that I have to wonder given that these are really tweaked 2009 models similar to the mac pro 2,1 relative to the 1,1, how long after discontinuation it will be before they hit Apple's legacy list. Anyway... my reasoning on things is different from the other person you quoted.



I don't think anyone should concern themselves with physical size so much as functionality. Anyway part of the problem may be that when people look at anything other than the mac pro in terms of desktops, they're confronted with the imac and mini. If they're going from a mac pro to one of those, there are a few gotchas even if cpu power is adequate, so there's a tendency to cling to the mac pro as it does work, and it's familiar. I think all of this depends on what Apple has actually been doing. Internally it did see a few significant changes in 2009, and that was the last major refresh cycle with 2010 being more of a minor one. It's also definitely a custom design. It doesn't match any of Intel's reference designs. I am not sure what they'll do this round.

Yeah, I agree they need to get better at this. They have enough resources that they don't have to create scenarios like this any more.


In a way, though they tend to compare to their own previous model instead of using absolute scores. Years ago, they were convincing people to buy a Mac instead of a PC, now it's more about convincing current owners to upgrade.


PC manufacturers can and will implement it but only Apple can do it through their whole eco-system and their OS. Hardware manufacturers could never do a zero-config render farm because they don't build the OS. They also dismiss Thunderbolt in favour of USB 3 but the superiority of Thunderbolt is obvious when you see things like this:

http://www.youtube.com/watch?v=143IpdBS4xA

That simply cannot be done using any other currently available interconnect.

Ethernet connectivity for processing is an option but it is 1/10th the speed and via Thunderbolt, the idea would be that other computers don't show up as network nodes. They would be co-processors, which means not having to install software on each machine and hopefully not having to pay for multiple licenses. You could essentially build a 60-core (120-thread) render station out of 10 office Mac Pros. During the day, they can be used by artists and then overnight, one artist submits a render and without even thinking about it, it starts using all 10 machines and the render is ready the next day. There may only be a single copy of AE or Maya or whatever installed.


I see the iPad replacing all laptops because the only difference between them is that the computing parts and battery go behind the screen instead of the base. When I owned a 12" powerbook, I remember wishing I could just take the 12" screen off and not have the base but at that time, it wasn't clear it could be done as no touch implementations were good enough to replace the keyboard.

The big problem right now is the battery. Once they figure out how to improve the energy density of the battery and improve charging times, they can use bigger displays without impacting the weight. One major reason for iPads replacing laptops is portrait mode - a laptop design cannot allow you to use the device in portrait and it makes such a big difference to certain applications.

I could see the first incarnation of this sort of thing being a dual Macbook Air and iPad. OLED displays allow you to make extremely thin displays so say they flattened the MBA display and it had a screen on both sides, touch screen on the outside with a smart cover over it. When the MBA is open, you can use it as normal and when closed, it goes into a low-power mode possibly using an ARM chip and runs like an iPad. The 11" Air is 2.4lbs vs the iPad's 1.5lbs so it can be held in a similar way to the iPad, though mostly used on a desk I expect.

Decent post. Hard to argue with. Particularly noted your last paragraph re: socs and the morphing paradigm shift that desktops are faced with.

If we look at the new iPad, there's an immense amount of power in such a tiny slither of a device.

It makes you wonder what Apple could really do with a re-factored Mac Pro if they put their back into it.

Lemon Bon Bon.

It is a fascinating prospect. But Apple gave the market the iPad and crushed the 'Apple has got to make a netbook' analysts and the Acer net book market under a crushing 'low end laptop' crushing paradigm shift.

Low end laptop makers and PC Wintel marketshare is wondering what hit them.

To say that Apple's portability marketshare is crushing the opposition is an under statement.

To outsell all PCs with Macs and iPads included?

Simple astonishing.

Put a Rogue class GPU in the iPad 4 with a quad core cpu?

Sorry, but anybody who hasn't been taking the iPad seriously with this gorgeous retina release are going to be slapped around and dragged around a gravely car park until they get the idea.

Much less what the iPad 4 (presumed) above will do to the opposition. M$'s tablets are going to run head long into train as the iPad 4 hits them.

As for middle to upper tier laptops. They'll survive for a while yet...

Lemon Bon Bon.

I take your point about the Cinebench scores. Hard to argue with plain old facts...but...

If it was me, I'd be happy with a fully loaded iMac. If you've been used to a dual processor Mac Pro then you'll be used to that, I guess. So? Wait until the Pro hits for a more level playing field and make your decision then?

Rather than move my entire eco-system over to PC land I'd wait for the Mac Pro release which I'd say could either be imminent after iPad has had it's month in the sun or WWDC or a Mountain Lion release. I think that's worth pausing for.

I have an iMac, Lightwave 3D, Poser, Manga Studio etc. ie software investment, iPhone, iPad, ATV. I guess I'm emotionally invested in the Mac ecosystem more than I am re: specs alone. I have Steve Jobs to thank for that!

My first Mac was a Tower. I remember that feeling of being at the 'top of the heap' for a while. (And then 3+ years later I had a big, slow box... My adobe software investments got trampled over in the OS X race for desktop supremacy. And even when I bought again...ten years of OSX improvements got me 'done' again.

I tried PC for a while with an Athlon and then upgraded the box. But I just hated 'Winblows.' It sucks. I hate it. OSX. It's like a disease as Wizard puts it.

Let's face it. We're far away from the dark years of of Power PC stagnation and the near collapse of the Mac creative market.

The top end iMac is far from a modest machine fully loaded with ram, SDD drive and a thunderbolt external HD.

If anybody puts the boot into the Mac Pro I'm not going to say it's not justified. The cinebench scores don't lie. And while Intel remain culpable for the delay to Sandy Bridge E, I find Apple more culpable for the things in their grasp: more ram, updating to what GPUs are available (regardless of whether they are incremental or not - that's not an argument for not including a bump in spec), bigger hard drives AND cutting the price or including two cpus on the base model to get sales jumping again.

Short memories people have. It wasn't so long ago Apple offered 'two' cpus for the money and at cheaper prices and made much boast and fanfare of it. They can price the pro's ridiculously high and offer stale specs (all bar the cpu which they can't help...but they could offer two on the entry model...like they used to!) and whinge about low sales or make the Pro more value added with a price cut and get units moving again.

How about comparing the entry price of a Macbook Pro laptop to the entry Mac Pro desktop? How come Mac desktops are so expensive compared to their laptops? It didn't quite used to be that way.

Anyway. Back to your decision. Given your eco-system investment. I'd urge you to wait a bit longer. You've waited so long any how. Why cause so much upheaval for a few minutes on Cinebench? (I guess it depends on what you're rendering...)

It's pretty obvious the Mac floodgates are going to open once Intel pulls their fingers out with the chips. Look to the skies from April onwards I'd say. Educated guess and all that.

I'd still like Apple to offer a re-factored desktop/mini-tower in the iMac price range. You know. Like they used to with the brilliant G3 blue and white classic and bundle a decent gpu with it.

Sure their would be a bit of iMac cannibisation but many of the iOS developers may find a home with the 'tower/refactored' desktop if they didn't like the iMac. At least Apple would offer people like Wizard, myself etc. the choice.

Lemon Bon Bon.

F*ck me pink. (Excuse me...) Wow. Impressive link. That was over thunderbolt? Was that mid-level card? Could it power a high end desktop card? That was MSI, eh? (I've always thought of them as a bit ambitious. I looked at them alot during my Athlon PC days for upgrades. I think I went with MSI for an ATI gpu upgrade.)

So...the 'hardware is ready' according to the MSI rep but Apple's OS control is tighter, do we have to wait for Apple's say so? Or can MSI implement it anyhow?

Basically docking your Macbook Pro to an external GPU when you get home. That's kick ass. I like that.

But also for the iMac. If I could augment an iMac through a thunderbolt connection - POW! That sure would power a HiDPI image alot better than an 'M' class card.

Hmm. If the thunderbolt cable can do even a modest mainstream desktop gpu (ie it sees the laptop as a 'desktop'...) then surely seeing a 'pro' box or a mini or a laptop...they could all be co-processors with thunderbolt daisy chaining and...Apple showing the will to take all that power and run it through Open GL processing 'capsules' of data regardless of source?

If you invest alot of money in an Apple eco-system, you could get the value added performance of staying loyal to the Apple system, eg Macbook Pro and iMac/Pro pooling their power as co-processors over thunderbolt rather than sitting separate. Saves on render farm node price gauging with, in my case, a Lightwave license per box.

Intriguing stuff.

Keep it coming.

It makes the wait for the Mac Pro go a bit faster...

Lemon Bon Bon.

Yeah this is the point my friend made last night when I linked him to this topic. I have actually taken the iPad seriously since the beginning however only the iPad 3 just recently has made me consider buying one due to the retina display. I am still happy though with my Mini hooked up to my HDTV for the time being though.

I can't wait to see what the next iPad brings.

Having said that, I still want to see what happens with the Mac Pro. I still feel there's life in it.

It says it was the Radeon 5770/6770 (same GPU, different badge) - same GPU as the entry Mac Pro, although presumably with PC firmware. The highest it will take is a GPU that uses 2x 75W power inputs. This won't take the highest cards but it should support the 6950:

http://www.amd.com/uk/products/deskt...verview.aspx#2

562GFLOPs double precision compute with 2GB VRAM on a laptop isn't bad. This is the 12th fastest GPU on this chart:

http://www.videocardbenchmark.net/high_end_gpus.html

If their dock is priced at $100-150 like the guy said and the GPU is $290 (there is a 6950 for $250 on Newegg), a $400-440 upgrade for that level of performance would be a nice option, especially for computation. Given Apple's use of OpenCL, you could do an FCPX edit on your laptop without the GPU, plug in the GPU before export and get over 500GFLOPs of extra compute power.


Whatever they plug in shows up as PCI so it will be the same problem as plugging a PC GPU into a Mac Pro. If OS X doesn't have the drivers, the GPU won't run. It might work fine on Mountain Lion as there are new drivers:

http://9to5mac.com/2012/02/21/hackin...acks-required/


The 6970M is a pretty decent card so an external GPU wouldn't offer much more and the GPU output couldn't go to the iMac display anyway. To be honest, I would have preferred that the Thunderbolt dock used an MXM slot and only supported the M class cards as they are much smaller and lower powered but it does restrict the use cases. This dock for example could run any PCI card like a capture card or audio cards and the iMac can run two of them.


It should be able to send any kind of data to each machine so it could use the CPU and GPU of each machine. Mostly, this would be for long computation and not interactive rendering but it depends. They'd have to experiment with it and see what the limitations are. I think it would be a pretty neat feature because it removes that technical barrier for artists who need as much compute power as their wallet can handle.

This is a quote from the comments section of the link you gave.

Interesting...

So...Mac users would have a greater choice of GPUs if PC GPU makers supported the UEFI standard?

Still, a moot point, as we only have the 'Maxi' Tower Pro starting at £2000+ to support it. If we had a mini-tower Blue and White G3 price range tower under neath the 'pro' it would substantiate an upgradeable GPU market on the Mac. (ie if Apple sold more 'tower' units that could be upgraded...)

Lemon Bon Bon.

The 6970M looks a pretty decent card to me. I think the first one since the 8800GS(GT?) worthy of the name. It's given the top end iMac some much needed clout. And the 2 gigs of Vram is wow. In short, I'd tear somebody's arm off, sell my gran and exchange this Core 2 Duo iMac for current top end iMac.

Your point about MXM is noted as well as the bang for buck portable gpu cards are now giving. But it's the same with cpus too? The gap between 'laptop' and 'desktop' has closed markedly in recent years. To be honest. I remember when Apple had easy access to the iMac and made much song about it. I thought when the MXM standard was issued I was surprised Apple didn't support it with the iMac and Mac Mini in terms of you being able to open 'em up and just slot eg a 6970M class gpu in there.

Pity.

Could you imagine being able to open up an i7 Mac Mini and stick a decent 'M' class card like a 6970M in there?

That would be cool.

*Notes the original Cube design that could just be pulled out.

Apple seemed to have abandoned those accessible designs that the Cube, G3 B+W tower promised. Well, in terms of price range. The desktop units seem sealed off bar adding a bit of ram and external HD.

Mind you, if the 6950 works natively with Mountain Lion as the link suggests...could...COULD the 6950 work via the Mini via it's Thunderbolt cable? If the power of the GPU draw isn't too great? What's to stop MSI shipping a 6950 for the Mac market? There's millions of laptops and a substantive number of Minis that could create a killer gpu market for a brave entrepreneur. Hmm. Does that give us our 'mini-mini' tower? (...an i7 based Mini plus 6950 plus the external gpu case all in for about a £1000? You'd then have your monitor of choice. I wouldn't be able to resist the overpriced Apple badged 27 incher though. So it would be £1800 all in? About the same as a top end iMac but with a slightly better gpu? 12th on the list you linked. That's a decent card that 6950..!)

Could it be the holy grail Mac 'tower' buyers are looking for?

Lemon Bon Bon.

Like a pre-compute array? Parcelling up the processing into Open CL chunks? Fed back to the host application?

Like virtual co-processors. I thought X-Grid was some cluster Apple software Apple had a while back? What became of it? Between X-Grid and Thunderbolt, could Apple make OS X recognise these daisy chained machines and give them 'silent partner' co-processor status.

Like a local iCloud/Lan hardwire compute? Or all the Power Rangers forming together to make one Mega-Robot Ranger?

If they're trying to sync iPads, iPhones, iPods and ATVs...

....surely syncing and pooling compute resources on a Mac can't be rocket science for Apple?

Lemon Bon Bon.

As a side note, if Apple really wants to push openCL, they better starting to put good GPUs in the Macs ;-P

It seems to be in Apple's DNA at this point. The concern could be retaining high margins. It's just that the way it works tends to break down as soon as you require something even slightly non standard.



I think a couple PC oems have set up external graphics cards in the past. Sony comes to mind. Anyway I was thinking 10G ethernet as it's likely to come down in price. Regarding thunderbolt, behind the scenes it would still see a PCIe bridge. I don't know of any programming that changes this. This is the kind of thing where I'll believe it when I see it tested, as I don't plan on pioneering such a setup on Macs.


It's a little early to call right now. I don't know what it would be like for business travelers and people who need to do quite a lot of computing on the road as opposed to on the go. I could see such an Air design, but I'm not sure wha that would take in terms of engineering. On a side note my desktop displays can do portrait mode. The stands allow rotation and they support it internally (suck it thunderbolt display).



That's the same thing I'd suggest. I find the idea that using it for work means buying the first one off the line to be silly. Usually on software and hardware updates, I wait for bug fixes and an all clear or at least mostly clear. I don't even install a new 10.7.X release without checking that it doesn't break anything to avoid having to restore from the backup. Usually a day or two after it comes out I know if it'll be fine. I've mentioned some of the reasons I don't care for the imac option. If the display worked for me as a primary display, and the ssd remains user replaceable, that might be just enough. Sadly the display thing won't happen. I don't have room for a 27" display that's not suitable for work, and when you take the display out of the equation, it's not such a great value. Previous issues with their displays also make me not want one.


I do some of my stuff under bootcamp, so I'm not 100% invested into OSX based software. I don't run anything that only has an OSX version. I started off with Macs because they were the norm for what I did at the time, and it would have been awkward not being 100% comfortable with the keyboard shortcuts, command differences, etc. I used a PC for a few years before that. It's been roughly 10 years on the Mac. If you note my comments, my concerns are different from the others. On a side note... I've decided to learn nuke. I have zero use for it currently, but there's a good possibility I may want to know my way around it in the near future. It doesn't look that intimidating either. Many of its functions are similar to things I've used in other programs, and I've worked with node based software in the past so the mindset isn't a stretch in any way.


Better value toward the lower end of it would be nice. Regarding gpus, look up the 5870 compared to the 6870. The 6870 looks a bit less power hungry. The 6970 could have been an option. I'm not sure what its power draw is like. It would be frustrating to spend that much late in the life of a computer. The 7800 or 7900s aren't likely to be as big of a performance jump as you might hope.


I remember this stuff quite well. The one thing that offset it somewhat was cheap ram as opposed to the older finned stuff, but I don't think the other oems are using finned ram either. What annoyed me was how much they dropped down on the spec sheets from 2008-2009, while going away from the intel reference board design to something more exotic (and probably more expensive to build). The 6 core is actually an excellent machine. It's just quite expensive. I'm getting off track here. Originally the mac pro was kind of a half workstation. It lacked some of the features of other comparable ones, but the price was extremely competitive relative to specs and performance. I don't feel this is really the case anymore, and that has most likely contributed to slowing sales. I mean any manufacturer can look at something and say we can predict X sales at this amount. The goal seemed to be to push anyone who doesn't absolutely need a mac pro in the direction of the imac. People still do buy it anyway, and while it's more expensive, it's only certain configurations that are really really weak. If you look at Windows versions, quite often they'll start lower and have higher upgrade costs.

Anyway.... I wish to make due until all of the new hardware comes out so that I can make a rational decision.

Some of the first computing hardware I had where 6502 and 8088 based. Every time I pull out my iPhone or iPad I'm a bit shocked. To get faster computers, hardware will have to get smaller; intel is already at work on this with 3D memory technologies and the like.

And we are only on generation three. More so Apple hasn't even pulled out the stops on process technology.

I can imagine vastly different approaches all leading to a very different approach.

Even modest GPUs can have dramatic Impact on certain classes of Applications. However you need to also realize that both AMD and NVidia have only recently been focusing GPU architecture development on compute. This has lead to entirely be architectures making their way to the buying public. Compute on GPUs is still very new technology.

Well, they're going to have to make them compatible later this year so let's hope we get more options:

http://arstechnica.com/microsoft/new...we-know-it.ars


That's the plan.


Apple don't really publicise it much these days but it's still in OS X Server.

https://developer.apple.com/hardware...rid_intro.html

It would be like XGrid but require less input and wouldn't require OS X Server. I could see it being just a sharing option in system preferences like web sharing or printer sharing, it would be compute sharing. Turn it on and connected machines via Thunderbolt could be used as slaves or controllers based on configured options.

I'd like to see how some of this stuff actually works in practice rather than on paper. I'm not sure that some of your suggested uses will work that well in practice.

The problem with Apple is their seeming unwillingness to promote some of the technologies they develop. I just don't see scientist and engineers running to Apple for compute hardware. It is much easier to build a small cluster with Linux machines, stick them in a closet someplace and connect a Mac to them with a fast pipe. To really compete well as a cluster solution Apple would need a very turn key system that dramatically improves their odds relative to a Linux cluster.

Where Apple has the most potential in my mind is the single box compute machine. That is a Mac Pro or XMac with the hardware required to produce a high performance workstation.

As for these external adapter boxes for GPUs I have a very hard time grasping the wisdom in such hardware. Looking at GUS in that video linked above just had me shaking my head in disapproval. I'm just not sure whom they are trying to reach with such a box. The box takes an external 150+ watt brick to power it which is a mistake right there. Then the box is fairly expensive but yet only handles midrange cards. So while I'm sure such a box will work I just can't see people jumping all over themselves to buy one as you might as well buy a whole PC.

In any event this is one of the reasons I'm so hot over the idea of an XMac. Apple really needs a low cost desktop that can handle this sort of GPU card. A very much midrange machine. I wouldn't even be bothered if the GPU was soldered right on the motherboard as long as suitable video RAM was supplied. Better yet is an XMac with a soldered in GPU and a PCI Express slot capable of supporting another 200 watts of capability. Such a box could be very capable but low cost as much of the Mac Pro is deleted from the implementation. Well relatively low cost as this would likely still be a $1500 computer.