Welcome to our community

Be a part of something great, join today!

Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

New MacPro predictions? Hopes? Fears?

Well let's see how soon we get actual 10bit support and hopefully HDR as well. All fine and great that a monitor might have it, but if the computer OS's are not brought up to spec, then it becomes a niche monitor to use only in software that specifically supports it and usually in conjunction with specific hardware. ...As 10bit often is now.

Personally, I'm not crazy about 4K at 32" for a desktop monitor. And I don't see this offering anything for grading or playback that we can't already get from other 4K HDR displays already on the market. So I'm going to reserve judgement until I see it, but this just seems to me like they took their existing 32" 4K monitor, which is "meh.." and upgraded the internal DSP and backlight system to support HDR. I'd really like to see HDR ability added to their 5K panel. That's a much nicer usable desktop monitor resolution and size. But we're only now just seeing that being refreshed with TB3 interface, but it's still the same panel they have been selling in their stand-alone unit with dual TB2/DP inputs and also the 5K iMac display.
 
32 would give a nice field of view. If they could standardise on am industry standard, that would help software use it. Frankily, I prefer an 55-65 inch desktop monitor, you can always use the extra realestate instead of seoetate monitors under many circumstances. I would like an 8k 84 inch for that reason to, you get to have 4 42 inc
 
Which decade?

There has been greater desktops before that never saw the light of day yet. Maybe some of them? But they are dated now.
 
At least they still recognise it as strategic, which is what I had been saying un tje past. However not enough has been done with it for a while. All this thing of wgats coming.next. Well Appke has vern whats comong after other co psbues do it. He is sieaking of thungs knocking on from things. But yhe rwality is, when O design something I look to see what strategic features and markets I can hit, not to wait for them to materialuse. If you want a good product, design it so in the first place and put it down for people to expand upon. You have the product featured up as really desirable and targeted for a broader base leading to broader sales. Unfortunately the industry has become to obsessed about fashion sales rather then real work and then make it fashionable at the same time.

I'm sure a cave man that saw a modern IPhone for the furst time might he impressed, but fur is it is old and the feature set so so, not just from age of the style, and a fullness of features, but because the competition offers much of it for a long time.
 
Jarek Zabczynski said:
Timmy says "We Have Great Desktops in Our Roadmap"

http://www.macrumors.com/2016/12/19/apple-tim-cook-desktops-roadmap/
Click to expand...

Within a few hours of that statement, reports from the usual sources claim that Apple is pretty much letting the Mac go into a decaying orbit, due to the focus on iPhones and iOS and an alleged exodus of talent and manpower from the Mac and OS X teams, who have been moved to other Apple projects like Apple TV and their automotive efforts or left the company all-together. At this point, it's clear that the Mac Pro is dead and the tech needed to bump it up just isn't available to Apple and the iMac outsells it by a huge margin, so I'm thinking that Timmy has a new iMac planned for early next year. Perhaps Apple will get smart and drop Intel in favor of AMD, which could give them a huge edge and early access to AMD APUs and GPUs with HBM, which would bring a big performance boost to the iMac and Mac-mini.
 
Lol! People have been complaining they have been using AMD instead of nvidia cuda up till recently. O saw the announcement on the new AMD GPU with major performance jump. His quickly things change.

Apple would have knkwn the projected GPU and CPU design improvements years ago (but just because it is projected xiesnt mean it will come through). So it makes sense for them to do desks with AMD in preparation. Remrmber when they showed the cell processor to Steve, he wasn't really interested, he would have sltwady seen Intel's future tech presentation. However, we know what happened to proposed Intel tech. Things did not turn out. AMD would make sense on low end products as an APU. It's cards might make sense on the high end.
But as for CPU and future Xhi Phi, that is another issue. If AMD did make it work, we might expect to see AMD on low end CPU and Intel onntje high end. Of course there is also ARM and RiscV. But all Apple has to do is start adopting the competition and threatening yo take their business elsewhere, and the supplier is motivated to give a deal. This is a game businesses at. It doesn't mean they want to change (however, as Antony points oit , they haveba good Arm design, so probably have it as a preference to go to, meaning Intelin Apple might suddenly disaoear one day). However, Intel has to go ARM or RISCV in preparation to swap if they become yje industry standard otherwise be stuck with the new power PC processor (however, I imagine that they have a non x86 non ARM etc next generation computational technology, not like current transistor based processors, but the replacement sometime in.the next 20 years, if not within a decade.

If I could make a MQDCA computational block now running between 500mhz to 2Ghz, it could have so much computational power it could take decades for people to manage to effectively use it, while Sony would struggle to compete with new PlayStation's using conventional technologies. It maybe slow against some future 200ghz Sony, but there are so much of the computational performance it wouldn't matter much. All you do is have a high-speed co processing section to handle the exceptional cases (though don't expect QDCA to be stuck at 500-2000mhz with me around, 20 GHz might be possible, even 200ghz, plus each part of a trace line acts as a pipeline, so millions of oeicrs of information can.be loaded up. But you might say what dies that matter. Well, here is something nobody has tried yet on QDCA. A calculation circuit forms a bottle neck, but while that bottle neck is being performed it duesnt mean that the other data inntne pipeline can't be sent to other circuits. This is not ideal due to data and processing intre-dependencies, so speed up a lit of the time maybe limited to less than 1 or 4x. But it occurs to me writing that, normal processors do something like this in super scalar, out of order execution. This is however a different ball game as QDCA works in a completely different way to normal wires and transistors. However, a hybrid direct link could transport data across chip much faster, as I said before
 
Now, what might Apple be up to. Years ago their was supposed to be a 30 inch + lay down touch iMac TV. Such a design is getting their in size to compete with what you can put into a Mac Pro. So Apple could do a z1 G2 like IMac with top end processors, eclipsing the z1 (note Apple was showing off a z1 g2 touch like patent mechanism I think before the Apple TV IMac talk).

So basically, if you make the screen large enough, you can put a workstation under it.

The sort of maximal desk size is around 80 inches down to 50 inches for the whole desk. That is workstation farm like territory.
 
I have been thinking of a script idea in recent weeks. A guy atomically prints a 3D MQDCA block. He trains the block with an AI frame work. He takes the thing around with him in life secretly, talking to it. Eventually the walls close in on him, so he hides the block in the wilderness, where it uses energy scavenging circuit to operate (and idea I had since maybe 10-12 years of age). He dies, but the block goes on deliberately affecting the course of history remotely. It is interesting, as the blocknis so large, it can operate multiple sub personalities in corresponding inline with people, and can be killed off or reset, containing and dumping any emotions thatntje sun personality might develop, whichbarevakso controllable command dthven, and naive as to their true states, even unreality of their condition (believing they are an actual housewife from passadina etc) puppets.

I have been considering where ever to make such a thing. Since primary school, I have had useaa for AI design that still are ahead. Actually, in my novel I'm writing, I wanted to make a difference between the different synthetics, and put in two AI related explanations as to their different styles of thinking, which in reflection, turns out to be too close to the human mechanism I was trying to differentiate them from. Since maybe Ben I heard something that indicates the new research shows it to be true. Its getting a bit Battlestar Galatica 21st century like. I'm also doing some philosophical science on reality.

The reality is that while you can make a sterile intelligence to serve only and not feel or resent, or do for itself outside its service, if things are down to the wire, and you have to make really big extinction level decisions, you don't want to leave it to such an intelligence, you want a human to consider weight it up and make a decision. So, a human like intelliegence is preferable to a sterile intelligence to get the nuances and morality of the decision. Frightening stuff. I've been completely 100% against human autonomous intelligence, as you great non living beings that simulate the living and have choices and express rights and simulations of feelings. While yiy can avoid much of that in a major decision making intelligence, it needs some of it to make it works. Basically, as in the Restaurant at the end of the universe in the Hitch Hikers Guide to the Universe, you want the being that comes to your table on a dinning cart and asks you which part of him would you like cooked up to eat. But with less self awareness etc.

The interesting thing is, that using QDCA what is portrayed in the TV series Human, is possible to do.

However, I have realised writing this, that such a block technology is an excellent way to do my block technology to simulate the entire world in 3D. Something I was calculating around 2001-2004, naively applying a Moore's law timeline to it, to figure out at what stages different sized simulations would be possible. The issue is not only has transistor shrinkage run out before the target is reached, the heat dissipation is the issue, along with power requirements. However, with a QDCA block of low power technology that melts away. You could aromjc level 3D print today with my prininting machine design, but which decade you would finish printing the block would be another thing. So, a higher speed atomic level printer design is needed, but I forget if I did one. See, even if you could print a whole layer in one stroke, an atomic level layer (a bit complicated in that) is so thin you are talking so, say a tenth of a nano meter or less (but maybe much more, doebds on how I can position them into position) is ten million passes per milimeters (hmm I am starting to think I need a new delivery mechanism to supply different types of materials). A few more tricks, I mean 'metthods' can greatly decrease that, especially for areas of homogeneous materials. However, my design is meant for small items rather than blocks. Even a 20cmx10cm+ block is going to take ages. However, doing a QDCA sheet on the back of a table computer, or AIO, would be much faster, as millions could be done at once, but at least the while surface. Are there you ho, that should speed it up millions of times, don't know what is wring with my Braun tuday, I think that might be incorporated in my design already. Anyway, so even a handful of atoms thick gives you the vertical complexity of modern memory, 12-20 layers for CPU etc. Not that QDCA would require more layers for each layer, but I'm secretly looking at magnetic processing and hybrids, which I have a proposal for already, however, that is too speculative to be at all certain. So, doing a massive more conventional like processing array, can be fine relatively quickly. However, 1mm-1cm thick sheets are preferable to get beyond present designs.
 
IF the cylinder is dead, and iMac is the only desktop going forward, I have an idea. Piggybacking.

For most buyers, it's just the next iteration of recent thin iMacs with more up to date internals, USB-C ports, touch bar keyboard, etc

For power users - a "snap on" Power Pack, maybe 3" thick with the same height & width as the main unit, that offers PCIe slots, liquid cooling, secondary PSU, etc

This concept would keep the "regular" unit as the volume selling piece that fits Apple's lineup, but with the option of adding a lot more power - at a fat markup. Yes, a bunch of USB-C/TB3 tethered boxes could accomplish this, but the piggyback form factor could support lower latency interconnects and be more user friendly. Just a thought, not holding my breath ;-)

Cheers - #19
 
I am beginning to think that the 2016 MBP was intended to max out with both 32GB and a slightly snappier GPU - but this all turned South when the (Safety) tests on the larger multi layered (terraced) Lithium battery that perfectly fit (ie contoured) into the MBP space did not pass the (possibly more stringent) test.

With the internal strife, and pulling the iMac technical bods to redo the work (and use the previous gen battery technology to meet the pre-Xmas market), I think Schiller's public defence of 16GB might have been to smooth over the internal upset.

Apples PR team have purportedly leaked that next years Laptop will be what we all wanted. With Apples resources probably already pouring into getting the next gen battery working perfectly, Intel finally 'alllowing' 10nm CPUs, AMD Vega GPU on TSMC 10nm, and probably 4TB SSD, a multitude of Apps that 'excel with Touch Bar, and Apples P3 displays - they may very well be right.

Now that Mr Cook has had his private meeting with Mr Trump, it is also likely that Apple will have an understanding of the (Twitter) fallout if they move (expensive) production of Mac Pro back to the Far East, or if they need a (political) shell company to show that they are building high end gear on the USA. Perhaps nMP terminating production (rather than moving it) would be a happy compromise that suits both of their agendas?

I am still trying to guess when Apple will decide to introduce an MacOS that uses an actual ARMv8 compatible CPU. Maybe they just slipped the references into their Sierra release knowing that people would find and report on it, in order to put pressure on Intel to give them more reasonable volume deals?

Qualcomm's server ARM chip will not be ready in time for Sierra rollout, and is unlikely to be as performant as Apples own (mobile) cores, and also not have the Metal GPU 'goodness' in them - and probably (speculation) cost Apple half way between Intels XEON and Apples own efforts, probably making it unsuitable for Laptops (in perf/watt for low core count).

If Apple do decide to introduce their own desktop CPU, its is going to require 'courage' that surpasses the killing of the headphone Jack, and more like that of killing the giant worms from 'Dune'. The only way I see people willing to 'happily' move off x86 is if they are salivating, and/or the entry price for what they think is a no-brainer.

Could Apple's next step be to push AMD (GPU) and Intel (CPU) out of the way and bring 'fusion to the Desktops?
I am not so sure they'll be able to pull this off in 2017, but 2018 the CES buzzword might very will be '8k'.
8k screens, and 8k GPUs.

Maybe they'll take a leaf out Red's hat and sell them for $8K to? :-)

AJ
 
Last edited:
Good idea. I've been talking about other things like that, assuming Apple would cover the basics of maximum power. I forget what power options I've cobmvered. Maybe Apple should just offer a broad pcie expansion box, and some anchoring bolt holesband let others do the heavy lifting (4xgpu cards).

But honestly, people should support open gl, migrate all the performance code to it and plug a processing unit with drive support to a TB port (or 4). Basically, except for finished display frames, not much needs to touch the TB, as pickup and deposit of processed data is handled by the expansion box. This is similar to the ...... scheme I was aiming to do many years back. As time goes on, the ability to penertrate the market changes, at the time this offered a simple way to get a product into the market. A useful product with much less expense.

Now, let's step back and give your ideas some real muscle: A lot of what makes a computer can be dine on the Mac Pro. The TB connected unit can be a x86 based unit with X pcie cards and drives (even a display capability). If Apple and software companies agree to support it by a standard . Now, you can use a high speed bus architecture to connect multiple units together, or chain them for a processing farm. But taking a step back again, you realise that AMD, INTEL etc, can make fat chips with the extra GPU capacity and prixessir on one chip, with layers of stacked memory. If speed impairments can be gotten around (maybe even stacking, or multiple moduling the CPU and GPU separately) then the on chip interconnect canbe higher speed again, and the combined unit possibly work faster than the separate piece version. Now, AMD, Intel, Nvidia or ARM, or RISC V or MIPS partners, can make their own module units and offer it for sale. Now, people can buy an module and enjoy the superior Mac OS system. Attach it to an imac or an air. Attach it to a Linux or Windows even. The PC becomes less the bottle neck than the module. You can connect it to an atom tablet even and do wirjstatuin stuff. The computer now just offers front end functionality to the user for them to buy. This is the sort of stuff I had envisionaged for the PC industry.

Now, replace this with a hybrid QDCA system. Looks like my ultra low power transistor based model is out dated already.
 
Antony, we are wasting time looking at the wring answers. Atomic level production of QDCA (or other) is a durable long term format, other next generation computing can.be attached to (quantum optical etc). Even at the moment, I'm looking at maybe a 100x advantage using techniques using existing production. 10x is a likely minimum increase. Just looking at other competitors l, like Arm and Rusc V I feel a 100%+ (speed up is possible). I feel, because I don't know enough of their designs, but know they tend to be leaner (but ARM's instruction set expansion makes it more difficult).
 
Antony, a gift:

http://www.tomshardware.com/news/qualcomm-x86-emulation-windows-10,33152.html
http://www.tomshardware.com/news/qualcomm-debuts-10nm-processor-48-cores,33151.html

Windows 10 win32 programs on Qualcomm next year. New 48 core server chip (I don't think it will have win32, but people are doing transliteration programs on existing Arms). 10/14nm. I'm surprises at the battery increase again, but then realised the industry has been changing over to the same sort of low powered silicon scheme as Intel had. So, I wonder where the curve will be inntje.future.
 
I'm still kind of disgusted by Apple's refusal to put some high end hardware into a Mac laptop, especially after this supposed big update to FCPX. I'm all for flexibility and most of my FCP7 plugins will transfer to FCPX and, if nothing else, it gives me a way to encode Prores on the go but the quality investment isn't there even after all this time. I would have jumped in the MacBook Pro bandwagon sooner but if it wasn't hardware failure issues with screens and GPUs then it was the lack of CUDA, the release of Thunderbolt, the promise of Thunderbolt 2, then it was the lack of power compared to PC laptops and now it's just stupidity. I'm actually more interested in a Hackintosh laptop at this point after my luck with desktop Hackintoshes! But even then, I'm kind of out of love with Mac OS and needing a Mac for anything. Adobe, Cineform, and the Avid DNX codec have really taken the place of Final Cut and Prores for the most part but I still miss the Final Cut workflow, even though the interconnected nature of the Adobe products is amazing. The desktop scene really needs to go back to the drawing board and if Apple really wants to put some power in a small machine then make it a smaller desktop unit and include at least one PCI Express slot for high end videocards and have the option for dual Xeons again and a large array of RAM slots, then it's just ports for hard drives, kind of like a mid to mini sized Mac Pro tower is what I imagine. They do that then they've got something, otherwise it's just a waste.
 
Apple's philosophy has been said to be "Skate where the puck is going, not where it is." And while I strongly agree with that for the most part, where they fail is that while they skate, they forget about the team and become arrogant in thinking that they know what we want better than we do. Is USBC the future? Most certainly, but we're not living in the future, give us a couple regular USB ports already. The list goes on...no headphone jack, no pci slots, non-upgradabiliy...ugh...
 
Stupidity iftenbstsrts a long time before, but as Jarek SATs, maybe they ate heading towards something else.

Their form factors support current high end performance with new silicon strategy, such as ARM and low energy processing arrays (even my work I'm projecting much more than 100x performance increase per unit of energy, but they would struggle to get more than 10x (stupidity in the computer market started many decades ago with standardisation on C and Unix instead of higher performance per unit of energy and memory, models like Forth, which I move out from). With next generation silicon replacements we are talking more again, sometimes thousands or millions of times more, so the Mac size is very good.

I've been looking into Quantum Cellar Automata again and contemplating how to reduce energy usage per unit of performance while raising performance, and hybridising it with some of my advanced design proposals from silicon. I got my QCA design and test book out last night (have been too busy to get back to it). The book even has a mechanical test model that reminds me of a mechanical computer proposal I wanted to make using the balls, based on a logical 4k transistor microprocessor circuite design (I suppose ifbI got around to designing one). They even have a reversible computing section. In reversible computing certain circuits can be run backwards to reduce energy loss (sounds mad, but I'm yet to go into it significantly).

According to what I've read QCA has a theoretical maximum energy reduction of a thousand times over silicon, and over a terahertz of switching, but likely a fraction of that due to certain stabilisation procedures. However, reversible computing further reduces the energy. Also they talk about huge losses of converting with electrical signaling in magnetic qca which I immediately thought of a design solution I use, but they have now taken another solution around this. But this sort of doing it badly and not the right thing is too common in computing. However, they quote upto 1000:1 energy reduction (+ whatever reversible computing can bring) but I have not found out what the figures for the different versions of the technologies, or which technologies the 1000:1 represents. But it is a 100w per cm2 for one trillion devices. Definitely not a very big block, and hear can effect the values of the technology I'm lookingnat, but haven't found how much heat.

About my block idea. So, at one million times less heat, thermal dissipation management becomes more practical, but I think it is something like 100mw per layer (had eye infection and could barely read the diagrams last night, so maybe misreading the labeling etc), but say 20 layers represents the middle being in a colum of 2w of heat, basically, if errors are not an issue that is passive cooling territory. If, course, that is only a simplistic idea of the nature of the issue (and frankly put a heat sink on either side and use side electrical, magnetic or optical outputs (if they used that with processors today, it would greatly speed things up, as you can deal with more heat). In blocks you get heat from all around, and you get limitations on how fast that heat can tranafer out, producing hot spots that can destabilise or melt, but I'll just stick to the simplified free space model here. Thinking things over in bed yesterday morning, going through all my design techniques, and technologies, I could use to increase performance, I think the energy difference between silicon and my design could well exceed 1 million to one, if not one billion to one. Going back to blocks that means a twenty centimeter column is 4mw or 40mw, I have to redo the maths, at as close to 1-2 Terra hertz as practically possibly instead of 500mhz, to 1-2 GHz wuth the right technology I was hoping before. So, at 1 million to one, very small blocks become possible, at 1 billion, sizable blocks become doable.

Now, you may say, that is a long way from what I was talking about before, but remember, such an AI block might be a lot of memory doing nothing unpowered and AI circuites. So, less than 1% of the block might be doing anything most of the time. So, very diable onnavailable energy left out in a rocky ledge.

Applying a few things I might be able to protect from magnetic feilds (thinking of long term flight space use) and higher temperatures, and heat removal mechanisms. Just writing that, it occurred to me another way to do QCA that possibly might be applicable, yep I should he able to get fast reaction times.

So, Apple could make a Macpro the thickness of a handkerchief (have a very old 3D display technology proposal that can blitz all others out there for portable use that would be suitable too), that could totally blitz the present ones, but however, it isn't here, and the thinness is not what we need today. Bring back the thick Pro. Look at this wat, even I can see how you could have a second thicker pro model that used most of the parts and even casing of the thin model. Why not have two models of lsotop and desktop.
 
Now, back to this, if they really are serious, they could make a rack with a backplane if CPU cards mounted out of it (like the laptop one I mentioned inna previous thread). You then could use cooling methods like with the Mac Pro can, to cool it. You could mount it above your monitor like a shelf (as it would likely melt the monitir if underneath) like a shelf with 16+ top end cpus, pass a few kilowatts of power through it, as a mini processing farm. But you could also balance it off against GPU's. A bit like a monster version of the intel Phi. So, a CPU with sufficient GPU integration would be better.
 
So guys, as far as doing post in native resolution in redcode only his much of a GPU usung the latest GPU cidec drivers do you need to compete with a red rocket x?
 
You must log in or register to reply here.
Back
Top