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?

Karim D. Ghantous said:
There is no way that Apple is going to make the mistake of licensing its OS. It tried to essentially appease critics before and it didn't work. Apple's past mistakes had nothing to do with licensing.
Click to expand...

The "it didnt work before" position does not take into account that desktops now make up 10-15% of Apple's profit.

Not trying to be a smart arse. The equation if VERY different with iPhones/iPads/iTunes(services etc) in the mix, which ARE the lion's share of income now.

In other words, it wouldn't really hurt.

Even if Apple made a dumb deal - which they wouldn't - on macOS/Open Source (for a fee).
 
Nicholas Croft said:
If Adobe release CC on Linux Apple's desktop's are doomed.
Click to expand...

I think the point is, that Mac OS offers a superior end user experience to Linux. But the hardware is no longer cutting edge performance is an eta where we need extra performance. We are moving to 192k+ technologies in film making, where is the tech to start on that road today. Sony has 16k for 720p holographic resolution. So obviously, Sony is not too interested in that for general purpose. 24-48k is the sweeter spot. As it is we need more to make high grade video advertising walls of 16k+, let alone holographic versions. Now, Apple is not making cutting effect machines with the tech available off the shelf, so it s further down the list, unless you want a cute little machine for people to see.
 
Nicholas Croft said:
The "it didnt work before" position does not take into account that desktops now make up 10-15% of Apple's profit.

Not trying to be a smart arse. The equation if VERY different with iPhones/iPads/iTunes(services etc) in the mix, which ARE the lion's share of income now.

In other words, it wouldn't really hurt.

Even if Apple made a dumb deal - which they wouldn't - on macOS/Open Source (for a fee).
Click to expand...


Not do much smart ass, but smart (we still cool Karim). I carefully craft my comments to allow Apple to make a profit, and for us to get better workstations. I'm not so interested in scenarios where Apple looses from the deal either. In the end though, it might be that a open source Mac os would be ok if the conosby is going to fold anyway. However, that might he unlikely soon if they diversify.

The issue, the dreadful issue,with Linux, which I have said for decades, is it duesnt develop too well too quickly without money. IBM put a billion dollers into Linux development, when a billion was still worth a lot, not a handful of bridges. That shored up Linux a lit in.the earlier time, as develpwrs were paid to devote their time to it. However,look now, we have Ubuntu looking more like android in privacy options, we have Linux split all over the place, and RedHat branch the orefered professional version, with unbuntu a cooler user feesikn, but we need both together.

When I tell you guys radical ideas, I often leave out key points. One of the key points was that an industry workstation standard might be formed by paying RedHat to develop the Linux component (to a standard beyond Mac OS). Now the industry can fund that through an standards alliance, but that could also be kick started by users.........
 
I didn't get into high performance computer architecture not to have the highest performance machine on my desk. So, I have had a big interest in architecture to achieve it, as well as ultra long battery life products, one proposal is dssigned to be up for decades on its own power (how else do you think I'm going to get market pemertratiin and beat everyone).
 
Nicholas Croft said:
http://www.fanlesstech.com/

http://www.tomshardware.de/cpu-kuhler-prozessorkuhler-cooler-itx-cooling,testberichte-241546-15.html



But, really, you need a f#c# ton of surface area for a 'performance' system.

https://www.youtube.com/watch?v=9PJOrfpiVwE&t=491s
Click to expand...

Way cool Nicholas. This mini case even has a low profile slot:
http://www.cirrus7.com/produkte/cirrus7-one/overview

You should have seen the old artic cooler PC, even dual TV tuners in it and optical drive. Pretty much a standup unit like the Nvidia Android TV. I could have got one secondhand, but dudbt because it was out dated and the HDMI was old.

Here found it:
https://www.arctic.ac/worldwide_en/mc001.html

I asked if they were going to do a new one, but they were not planning to at that stage.

However, I was thinking something that could take highest end processor and cards. I could find some bulkier than these, but don't think one workstation grade.
 
On, got to look at that YouTube link. Bigger than I think you need, I couldn't hear here, but 35db isn't that quiet. However, good effort on their part. Wow. Good find. I am thinking of a fin cooler that fits over the motherboard. Asus might have had something like that. Plus there was that liquid that they used for thermal cooling inntjw 80's, would fe worth looking into (prettyuch whatever works including high conduction carbon based cooling fins. Port and cover with a fine mesh to keep geckos and bugs out. The question is, can you beat the 7.2l if that dsn case, even 4L (if the fins are not enclosed I'll count the siace between them as not in the case :)

Anyway, thus is yje dort of thing Intel should he doing, R&D into this cooling tech to allow higher end CPU's on a hybrid system. For us, we can use a smaller silent pure heatsink solution on the normal high end desktop/workstation cpu's if today.
 
Now, another thing. I've known people to get 10x on a process on inter circuit element trickery inntje old days. So, we ar likely talking about 10-20ghz systems on today's tech using this sort of tech.

So dobtbtrust what you are told or read from companies, if you are willing to break compatibility there is still room for some speedup. Now, let's relate that to Intel and Nvidia. Intel could build a wafer portion chip. This is an idea of mine where you you use a fader or an pietikn of an wafer (as they are so big) as a chip/main board attached here to a giant test sink and maybe a 30cm+ fan, on both sides even. Now you might say they will still melt the circuit in such close proximity. But with modern chip making ability you can now put high speed compact memory on the same wafer as the processor logic. So to thin out the thermal density you distribute your memory in between the cores, and still use some cache probably. You further break up that memory so if a segment of it isn't being accessed it goes into low energy mode. The memory uses substrate capacitance that retains the memory image for a long time without refreshing (basically very static) and use in circuit auto performance throttling to keep heat under control, protecting from hackers that might want to by pass siftwated throttling and burn something. You stack as much memory plane as you can too the chip so as to have a fast storage like ability, with a jumper pin to both internal and stacked storage, and blinking lights and storage IP tracking interface with GUI, separate from the CPU, so that storage usage can be tracked and the pins physically (or by a separated monitur system) pulled if a hack is active. You have configured as many storages as you need, so the is has its own, each program has its own, and as nanny connected lights. Now, when nothing us running but the system but you see other storages are being accessed it can be concluded something might be accessing something it shouldn't. Having all tasks joined to a storage means only the system sturafe and the task storage should he active. Being professionals, of course in this senario other thyrd party back ground programs/spyware won't he running ( I'm being funny).

Anyway, looks like I've overcome my difficulties, and am back.
 
Apple most likely has insurance against Adobe, just as they had insurance against Motorola (It's tru[SUB]e[/SUB]!).

In any case, releasing CC on Linux won't mean much to Apple, even in the professional space. Apple may have felt the pinch back when Windows was dominant, when Adobe released Photoshop for Windows. Only a bonehead would really use PS on Windows back then, but there are a lot of boneheads in the world and they are a lucrative market.

Actually Photoshop is one of the least useful applications that everyone loves to brag about using. One of the most useful is Automator, which is like AppleScript without the script, albeit with limitations. Another useful but overlooked application is ImageMagick.

To say that Apple didn't do licensing properly is just a fudge.

BTW Wayne, yes, we're still cool. It's tru[SUB]e[/SUB]. ;-)
 
Wayne Morellini said:
Way cool Nicholas. This mini case even has a low profile slot:
http://www.cirrus7.com/produkte/cirrus7-one/overview

Here found it:
https://www.arctic.ac/worldwide_en/mc001.html

I asked if they were going to do a new one, but they were not planning to at that stage.

However, I was thinking something that could take highest end processor and cards. .
Click to expand...

My favourite, no (least?) compromise sff is this:

http://nfc-systems.com/s4-mini/

Joshua, the creator of the SF Mini, is active here:

https://smallformfactor.net/forum/

You're always walking a tightrope of thermals (TDP) vs performance though with SFF builds.

Basically, once you go above stock cpu speed, you're dealing with a quadratic increase with heat per 100mhz/millivolts.

The way around this is, in the bios, to dedicate one or two cores to maximum speed - say 4.9ghz (x 2 cores) on a 'golden' cpu, and the other cores/threads way below their stock (ie undervolting), to - say - 3 to 3.5 ghz on a 6700k, by way of example.

Even a GTX 1060 is more than 'adequate' for a lot of workflows:

https://www.pugetsystems.com/labs/articles/Adobe-Premiere-Pro-CC-2015-3-Pascal-GPU-Performance-840/#Exportingto4K-SingleGPU

Apologies in advance if its sacrilege to post a non Red specific article.
 
Last edited:
In reality, most all non ultra low powered processes are overclocked to get the GHz. I don't know where the natural speed lies at 7nm but I imagine it will be up to 2ghz, bit maybe 1ghz. I have not mechanism data to judge it as to how exponentially decreasing the natural.speed increase is per generation. The companies have been holding dien the GHz to get power down. The reality is GHz 5 GHz, is suitable for many purposes with go GPU processing assistance in parallel. This means that QDCA is potentially in the range to serve the same task at a smidgen of the power, heat etc. Personally I'd prefer it also do Thz, but at least it has it's purpose. But you can combine it with some future 100's GHz, the, tech to do its stuff.
 
Nice links Nicholas, but I'd still prefer to work on low compressed raw 8kp60 16 bit live even 240fps which.they are talking.about for high speed and 3D. So, what do we need to do just the p60 part? Would the dual gpu MSI mini trash can unit come even come close?
 
Wayne Morellini said:
Nice links Nicholas, but I'd still prefer to work on low compressed raw 8kp60 16 bit live even 240fps which.they are talking.about for high speed and 3D. So, what do we need to do just the p60 part? Would the dual gpu MSI mini trash can unit come even come close?
Click to expand...

Depends on the apps/codecs you had in mind?*

Specifically, app 'optimisation'. Will your app effectively use more than 4 cores? Does it 'see' more CUDA cores/ROPS?


- as you saw in the Puget article there is a cut off - for 4 k - with the "lowly" GTX 1060, otherwise expressed as Law Of Diminishing Returns.

ie the $300 1060 will export in the same time as a $1500 Titan X Pascal.

But that's the Adobe Suite (as a rough proxy).


What apps did you have in mind?



* ps. Octane Render is one of the few apps (that I know of) that will use ALL of the CUDA cores - perfectly scaled - you throw at it.

https://render.otoy.com/octanebench/results.php
 
Everything. Let's be real. There are a big varirnce in app performance, but the thing to judge by is how well it can handle the whole tool chain compared to 60p8k 16bit (in the coming couple of years you might not be able to get a Red less than 8k, unless its a phone). The other thing is how well it can handle on minimum that you might use for local TV but at 4k.

Now, resolve is free so is included, but what software is going to handle a raw workflow at maximum performance using cheap acceleration (ussuslly GPU and not rocket),l and maybe all cores. Is there a high performance tool chain thread here (needs a sticking like that).

The answers are there and easy to tell, but usually there is nobody that lnkws that will lay it out comprehensively , instead of vague bits and prices.


Thanks.
 
I wonder why AMD have brought the release of Vega out early?

I wonder why the Vega is now referenced in MacOs Sierra.

If new VEGA GPUs are 3 x faster than the D700 .. That a lots of power that could be packed into a iMacPro.

In the current US Climate where Apple have (possibly under pressure form GOP-elect) reached out to FoxConn to consider manufacturing in the US, is it possible that Apple are now More likely to keep their Mac Pro factory alive and are now keep to prove this in 2017 Q3?

RRX on TB3 + 2 x Vega + 10nm Xeon + FCPx = A machine that eats 8K raw & will be competitive for at least 2 years?

AJ
 
Whats upbeith the 1000 usd 5k lg 10bit screens whith usb-c connectors thst coems out in dec 2016?
Seams like a bit much of a screen to use with the laptop only. Is it a sign that somthing is comming?

Secondly whag do you think of those screens?
 
The screen should be excellent as it's the same 5K IPS panel that's out in the iMac right now, but this one doesn't have the glossy glass over it. The 4K version is nice too -- I had one briefly as demo unit. Not sure what the hold-up is for the 5K units, Apple took a small handful of orders a couple weeks ago, but not sure when they'll open up again. The screen aside, the design of the rest of the unit leaves a lot to be desired. It seems the new MBP drives the 5K screen rather well. The AMD 460 GPU, even the 455, push a lot of pixels very well. Actually push more pixels than the Curren nVidia mobile offerings. They don't have the 3D/ computational performance of the nVidia GPUs however.

Thunderbolt 3 is starting to look like it could replace DisplayPort as the preferred connection type and Thunderbolt 4 will arrive in about 2.5 years and could very well bring us 8K support.

I'm hoping that Apple opens up orders on the 5K screen before their special pricing expires. I'll pick one up for the current < $1K price and give it a try. That said, I don't really use my MBP with a monitor very often, or really at all. So maybe I should reconsider that thought of picking one up. I do think Apple will have an updated Mac Pro offering of some sort. I'm thinking it will arrive Q2 next year when we see the next round of Xeon chips. I'm betting we get updated CPU with DDR4 memory, option to have more internal flash storage and new AMD GPUs. I'm expecting to see 8 of those TB3 ports on there and probably nothing else. Apple will probably give us some goofy unveil at WWDC and show how the new trash can will drive 4 of those 5K screens and how their new keyboard will have the Touch Bar.
 
Stuff it, how much longer before I can get a 200 48 core arm chips in liquid submersion coolant:

https://en.m.wikipedia.org/wiki/Server_immersion_cooling

You guys are all CONNED, I was working on technology that replaced the CPU GPU DSP in less than 10 k transitorrs per core (estimate only). We are still waiting for a bunch of no hopers to do next to nothing improvements so far off the bar of what is possible it was not funny.
 
You must log in or register to reply here.
Back
Top