2016 nMP

[Mago]

Thunderbolt 3 is not fast enough to feed a CPU with data. Any rumor from the "dark net" that Thunderbolt 3 is going to be used to link CPUs from two different machines together is a bunch of nonsense. You can't even add a standard CPU on PCI Express so it's definitely crazy that you could add one on Thunderbolt.

I come with two different rumors from "dark net", one is about new macbook pro target mode (sort of KVM/USB/DAS), the other rumour account on some "appliance offering co-processing emulating a networked compute node", which seems to me more like an Xeon Phi Knights Landing on an external cage connected to a mac thru TB3. The the thread speculated about interconnecting mac's cpu on hsa schemes etc, nothing to do with the actual rumor which is consistent with Intel MIC platform development targeting OSX.

The rumors speak about new target mode feature on macbook pros, and something like Razor's External GPU but on MIC nodes moreless like putting a Xeon Phi KL card on an external PCIe cage.
[doublepost=1464308045][/doublepost]

For those that like read rumours, on the same source, speak about the new mac pro to come on Q3, on E5v4 as expected with revised TDP (500W+) and as GPUs the base model will be available on AMD FirePro 310 (Polaris Elsemere XT GDDR5X), 510 (same Polaris GPU on HBM2) and D710 AMD Vega 11 on HBM2, the latest will be the only with FP64 Compute 2:1, total TFLOPS from 9/13/18 FP32 0.5/0.75/9 FP64, ram 16GB ddr5/ 16 GB HBM2/ 32 GB HBM2, 6 TB3 Ports, 4 USB-C 3.1, 1 HDMI2, dual 10.000/1000/100 MBPs lan on RJ45, 2TB 2.5 GBps m.2 nvme with passive cooler, upto 16 CORE Cpu, Ram Upto 128GB.

Everything consistent with expectations, No Zen Macs this year (sniff), no RoseGold Mac Pro either.

Thunderbolt 3 Display will arrive with new iMacs and an radical all-new mac mini on Q4/Q1'17, new macbooks will arrive sooner than everything, despite latest rumors new MBP will be available earlier Q3 at same time as the new macOS rolls.

No Xeon MBP either, neither MBA(as expected), new MBP on 14/16" with same footprint, slimmers but not that slimmer as the rMB12, touch ID std, 4 core only on 16" model, only with dGPU option, MBP 14 availbe on 8/16 GB and rMBP16 available on 16/32 GB Ram, same storage options on nvme, bit cheaper too.

 

[MacVidCards]

I really need (want) to find out where the Unicorn Tea is coming from.

 

[Mago]

I really need (want) to find out where the Unicorn Tea is coming from.

The Unicorn tea, said abouth the new rMBP the oled display on top the keyboard, isn't a graphic display, just an touch sensitive stripe with OLED backligt, only speicial is that it will enable gestures, as to control screen luminosity you touch the screen icon and scroll left or right, same for volume, etc.

This feature will arrive on new wireless keyboards too.

 

[hollyhillbilly]

For those that like read rumours, on the same source, speak about the new mac pro to come on Q3, on E5v4 as expected with revised TDP (500W+) and as GPUs the base model will be available on AMD FirePro 310 (Polaris Elsemere XT GDDR5X), 510 (same Polaris GPU on HBM2) and D710 AMD Vega 11 on HBM2, the latest will be the only with FP64 Compute 2:1, total TFLOPS from 9/13/18 FP32 0.5/0.75/9 FP64, ram 16GB ddr5/ 16 GB HBM2/ 32 GB HBM2, 6 TB3 Ports, 4 USB-C 3.1, 1 HDMI2, dual 10.000/1000/100 MBPs lan on RJ45, 2TB 2.5 GBps m.2 nvme with passive cooler, upto 16 CORE Cpu, Ram Upto 128GB.

Good gosh, I would be just happy with the dual 10GBE and TB3 ports.

 

[ManuelGomes]

If this is true, it will be awesome.
Still, I have some doubts.
The PS seems reasonable, and not that difficult to accomplish.
The difference between D310 and D510 seems short to me, just the mem type? Nah. I'd go for 8GB GDDR5 on D310 and possibly based on Pro instead of XT. D510 and D710 seem possible although Vega 11 is not even taped out as far as I know, so unlikely too. Vega 10 is more likely.
Also, in the current setup D500 is "similar" to D700 and not to D300, and this would mean a different lineup, with the mid card not having decent DP performance. For this machine, having n entry card with weak DP is ok, in case you don't need it. But the more powerful cards seems lame.
But if it comes out to be true, this is very good news, and worth the wait.
Also, the port count doesn't quite add up. How on earth will you get 6 TB3 and 4 USB-C 3.1, and dual 10Gbps eth? That would require loads of PCIe 3 lanes, which aren't there. Or switches, but real estate is short.
TB3 display is a must...
Exciting news, even if just rumors.

By the way, anyone buy the 1080 yet?
[doublepost=1464340457][/doublepost]Thinking about this port count issue, I could see an all USB-C nMP, 10 ports in total.
4 PCH USB 3.0 with USB-C connectors.
6 USB-C ports for either 6 TB3, or 4 USB 3.1 plus 2 10G Ethernet. And you could use these 6 ports with whatever TB3/USB/Eth config you want. Awesome hem?
But that would still require some switching.
And there's the SSD too. It would hang off the PCH. Since the Gbe is not needed anymore, and moving the BT/WiFi to unused USB 2 ports, the whole 8 PCIe 2 lanes could "become" 4 PCIe 3 lanes for the SSD.
Wishful thinking.

 

[MacVidCards]

The Magical Mystery Mac, it's waiting until you turn gray...

(Quit Bogarting, guys)

 

[koyoot]

Mago, If Apple will use Polaris and Vega GPUs there is absolutely no need for using 500+ PSU.
Secondly: Polaris will not have HBM2, and I do not think AMD would want to build semi-custom chip on this, because of the costs of interposer, and HBM chips, and manufacturing at Amkor them. That single chip alone would cost over 1000$!

If anything: cut down Vega.

Thirdly I have heard few weeks ago a rumor, that I did not want to post here, because I did not know how credible was and it said that Vega will have around 90 GFLOPs/watt, and 1/4 FP64 ratio.

4096 GCN4 core chip with 1.65 GHz core clock would have 13.5 TFLOPs of compute power and... 150W of TDP to achieve that. And this is more possible than anyone of you think. Fury Nano with HBM already is 850 MHz, GPU and has 180W of TDP. And we are looking on a node that brings GIGANTIC power savings along higher core clocks.

2560 GCN cores, 1.25 GHz - 6.4 TFLOPs, 4 GB of GDDR5X - D310?
3584 GCN cores, - 1.45 GHz - 10.3 TFLOPs, 8 GB of HBM2 - D510?
4096 GCN cores, - 1.45 GHz - 11.9 TFLOPs, 16 GB of HBM2 - D710?

And here are single GPUs...

 

[Mago]

Mago, If Apple will use Polaris and Vega GPUs there is absolutely no need for using 500+ PSU.
Secondly: Polaris will not have HBM2, and I do not think AMD would want to build semi-custom chip on this, because of the costs of interposer, and HBM chips, and manufacturing at Amkor them. That single chip alone would cost over 1000$!

If anything: cut down Vega.

Thirdly I have heard few weeks ago a rumor, that I did not want to post here, because I did not know how credible was and it said that Vega will have around 90 GFLOPs/watt, and 1/4 FP64 ratio.

4096 GCN4 core chip with 1.65 GHz core clock would have 13.5 TFLOPs of compute power and... 150W of TDP to achieve that. And this is more possible than anyone of you think. Fury Nano with HBM already is 850 MHz, GPU and has 180W of TDP. And we are looking on a node that brings GIGANTIC power savings along higher core clocks.

2560 GCN cores, 1.25 GHz - 6.4 TFLOPs, 4 GB of GDDR5X - D310?
3584 GCN cores, - 1.45 GHz - 10.3 TFLOPs, 8 GB of HBM2 - D510?
4096 GCN cores, - 1.45 GHz - 11.9 TFLOPs, 16 GB of HBM2 - D710?

And here are single GPUs...

rumors from no-where require some salt ...

What you said has some logic, but as long I know, Original AMD planned to build Polaris around HBM2, then switched to GDDR5, so it's very likely this Polaris on HBM2 Silicon actually exists but AMD keep it for later release once HBM2 ships on quantity and the market needs it.

Given Vega it's aimed at Server/Compute WS I doubt AMD will offer less than 1:2 Fp64, 1:4 pf64 dont care both to gaming enthusiast and falls short for HPC.

One things its clear no new Mac Pro on WWDC, maybe only an annoucement.
[doublepost=1464345997][/doublepost]

If this is true, it will be awesome.
Still, I have some doubts.
The PS seems reasonable, and not that difficult to accomplish.
The difference between D310 and D510 seems short to me, just the mem type? Nah. I'd go for 8GB GDDR5 on D310 and possibly based on Pro instead of XT. D510 and D710 seem possible although Vega 11 is not even taped out as far as I know, so unlikely too. Vega 10 is more likely.
Also, in the current setup D500 is "similar" to D700 and not to D300, and this would mean a different lineup, with the mid card not having decent DP performance. For this machine, having n entry card with weak DP is ok, in case you don't need it. But the more powerful cards seems lame.
But if it comes out to be true, this is very good news, and worth the wait.
Also, the port count doesn't quite add up. How on earth will you get 6 TB3 and 4 USB-C 3.1, and dual 10Gbps eth? That would require loads of PCIe 3 lanes, which aren't there. Or switches, but real estate is short.
TB3 display is a must...
Exciting news, even if just rumors.

By the way, anyone buy the 1080 yet?
[doublepost=1464340457][/doublepost]Thinking about this port count issue, I could see an all USB-C nMP, 10 ports in total.
4 PCH USB 3.0 with USB-C connectors.
6 USB-C ports for either 6 TB3, or 4 USB 3.1 plus 2 10G Ethernet. And you could use these 6 ports with whatever TB3/USB/Eth config you want. Awesome hem?
But that would still require some switching.
And there's the SSD too. It would hang off the PCH. Since the Gbe is not needed anymore, and moving the BT/WiFi to unused USB 2 ports, the whole 8 PCIe 2 lanes could "become" 4 PCIe 3 lanes for the SSD.
Wishful thinking.

You're rigth the rumours its about Vega 10, vega 11 was a typo.

About ports and cpu lines, maybe one GPU is on 8lines or Apple is using again a PCIe Switch (more likely).

Another thing I miss to copy from the darknet rumor is that ALL ports on the New Mac Pro Will be Video Enabled, 4K max on USB-C only, and 5K on thunderbolts.
[doublepost=1464346253][/doublepost]WWDC will give us only new retina Macbook Pros, maybe a Thunderbolt Display and a all New Mac Mini.

 

[koyoot]

rumors from no-where require some salt ...

What you said has some logic, but as long I know, Original AMD planned to build Polaris around HBM2, then switched to GDDR5, so it's very likely this Polaris on HBM2 Silicon actually exists but AMD keep it for later release once HBM2 ships on quantity and the market needs it.

Given Vega it's aimed at Server/Compute WS I doubt AMD will offer less than 1:2 Fp64, 1:4 pf64 dont care both to gaming enthusiast and falls short for HPC.

One things its clear no new Mac Pro on WWDC, maybe only an annoucement.

That is actually partially correct .

HBM2 plan was for Greenland GPU. AMD was planning to release 3 GPUs from Arctic Islands: Baffin, Ellesmere and Greenland. Baffin - Polaris 11, Ellesmere - Polaris 10, and Greenland - the missing piece .

As we know, Greenland died, and morphed into something else, maybe Vega, maybe not. Baffin and Ellesmere are alive as Polaris 10 and 11.

So everything is actually pretty logical how things turned out.

Why AMD would offer 1/4 DP ratio? Save power. Dual Vega 10 if that calculation and TDP estimate, based on unconfirmed rumor( ) is correct, dual Vega 10 might have same TDP as GP100, but hell of a lot more compute power, both FP32 and FP64. 300W for dual GPU: 27 TFLOPs of compute power FP32 with 6.75 TFLOPs FP64.

Apart from all this you know what blows my mind? Compute power numbers... completely different level to what we know was possible on 28 nm... And this is first version of 14/16nm silicon. Nvidia was able to get such gigantic boost with their GPUs while using not top level silicon. GP104 uses 16FF, not 16FF+. 16FF+ has another 20% better efficiency AND 10% higher core clocks compared to 16FF. So the game is still on.

Another news about Nvidia: Next-Gen SLI technology is coming with Volta. It will be hugely similar to... XDMA. Without SLI-bridge needed.

P.S. If you have the GPUs connected while using internal coherent fabric like NVlink, or Intel Omnipath, you can run the GPUs on slower PCIe speeds, because internal connection between the GPUs allows them to share the data directly. For example: Xeon has 40 PCIe lanes. Use double 8 PCIe for GPUs, and you end up with 24 lanes completely free. And the GPUs are connected together by internal fabric.

Edit: http://www.24liveblog.com/share/194237198?url=http://www.24liveblog.com/live/1331452
Looks like Macau AMD event is reality.

 

[Mago]

That is actually partially correct .

HBM2 plan was for Greenland GPU. AMD was planning to release 3 GPUs from Arctic Islands: Baffin, Ellesmere and Greenland. Baffin - Polaris 11, Ellesmere - Polaris 10, and Greenland - the missing piece .

As we know, Greenland died, and morphed into something else, maybe Vega, maybe not. Baffin and Ellesmere are alive as Polaris 10 and 11.

So everything is actually pretty logical how things turned out.

Why AMD would offer 1/4 DP ratio? Save power. Dual Vega 10 if that calculation and TDP estimate, based on unconfirmed rumor( ) is correct, dual Vega 10 might have same TDP as GP100, but hell of a lot more compute power, both FP32 and FP64. 300W for dual GPU: 27 TFLOPs of compute power FP32 with 6.75 TFLOPs FP64.

Apart from all this you know what blows my mind? Compute power numbers... completely different level to what we know was possible on 28 nm... And this is first version of 14/16nm silicon. Nvidia was able to get such gigantic boost with their GPUs while using not top level silicon. GP104 uses 16FF, not 16FF+. 16FF+ has another 20% better efficiency AND 10% higher core clocks compared to 16FF. So the game is still on.

Another news about Nvidia: Next-Gen SLI technology is coming with Volta. It will be hugely similar to... XDMA. Without SLI-bridge needed.

P.S. If you have the GPUs connected while using internal coherent fabric like NVlink, or Intel Omnipath, you can run the GPUs on slower PCIe speeds, because internal connection between the GPUs allows them to share the data directly. For example: Xeon has 40 PCIe lanes. Use double 8 PCIe for GPUs, and you end up with 24 lanes completely free. And the GPUs are connected together by internal fabric.

Edit: http://www.24liveblog.com/share/194237198?url=http://www.24liveblog.com/live/1331452
Looks like Macau AMD event is reality.

I Doubt the nMP's GPUs will have enabled some fabric unless AMD managed to use the "sli" bridge existent on the nMP as a full compilant fabric, or Apple added lines to this connecto to make it useful.

Whatever, I found Intel X540T2 dual 10GbT controller plugs into 8 lines PCIE2, the 3 Thunderbolt 3 controller maybe hooked to a PXE to enable them share the remaining 8 PCIe3 lines from the CPU, and the 4 USB-C ports remains as 3.0 using the included usb3 on the PCIh, but this wont allow enable all them as video output (which is something I dont believer, its too overkill having 6tb3 +1hdmi2 ), I would be happy if those ports remains as classic usb3, wi-fi could be connected to an internal usb3.

 

[koyoot]

WWDC will give us only new retina Macbook Pros, maybe a Thunderbolt Display and a all New Mac Mini.

NDA for AMD event expires at... 29th June 2016. It is for reviewers of new hardware. Very interesting in this context.

 

[flat five]

If you spent a significant portion of the day waiting for the computer to produce results - there will be a strong correlation between watts and work. More watts, less waiting, more work.

this 'if' isn't reality though.. who sits around at work waiting on a computer all day every day?
computers are fast now.. this isn't the 90s anymore.

also, if you happen to be computing in a way that requires a calculator to just sit around churning numbers for hours on end then you would generally have an alternate terminal whose resources aren't bogged down so you can continue working on something else instead of watching paint dry on the bogged computer.

i think some (or a lot) of you all value/judge a computer in a similar manner as you would an air conditioner.. you can either turn it on or off and maybe push a button but after that, you don't actually use the thing.. it just sits around making the air cold.. the specs etc of the a/c will determine how much cold air it can generate.. and you all sit around arguing about these types of specs.

but apple doesn't make the kind of computers that just sit there unattended day after day churning numbers.. they sell products that are meant to be used, by a person, day after day..
apple wouldn't sell an a/c unit because it's inanimate.. they simply couldn't compete, price wise, with G.E.s offering (well, the couldn't compete against GE with their pricing style.. apple's a/c would be too expensive for an object that just sits around in the background and isn't used by a person)

apple's whole schtick is selling things (they feel) can be an integral part of someone's life.. something this person will actually use. they sell personal computers.. not servers and renderfarm components.. i don't know why many here seem so proud to point this out.. apple tells you this stuff themselves.

 

[AidenShaw]

this 'if' isn't reality though.. who sits around at work waiting on a computer all day every day?
computers are fast now.. this isn't the 90s anymore.

also, if you happen to be computing in a way that requires a calculator to just sit around churning numbers for hours on end then you would generally have an alternate terminal whose resources aren't bogged down so you can continue working on something else instead of watching paint dry on the bogged computer....

Of course, efficient people will not watch the "calculator churn" (this isn't the 70's anymore) - they'll do other tasks not dependent on the system run.

At the end of the day, though - whether you're doing a large render, or a machine learning training run, or data analytics on a 20 TB dataset - you can't complete your "work" without the results of the run. Faster systems will let you do more work.

Open your eyes to all the ways in which professionals use computers, and stop trying to dismiss any of the many use cases that don't match your own lightweight workflow.

 

[Zarniwoop]

WWDC will give us only new retina Macbook Pros, maybe a Thunderbolt Display and a all New Mac Mini.

Is it sure that it is not 13-14" Macbook that will be released?

Macs' with Iris Pro 580 will be most likely the machines we'll see first. That is, some iMac models, Mac mini and maybe part of the MBP models.

And why wouldn't they release Mac Pro with two GPU option for launch and third option when it arrives at the end of the year? And there could be special models of Polaris series so Apple could diffrentiate from the competition. And to earn the Firepro name. HBM for 4Gb model, GDDR5/X and HBM2 for better.

Biggest surprise would be, that Apple will sell these new GPU's also for the current model at Apple Authorized Service Provider... but would they?

 

[Stacc]

Lets pick apart some of these rumors.

and as GPUs the base model will be available on AMD FirePro 310 (Polaris Elsemere XT GDDR5X), 510 (same Polaris GPU on HBM2) and D710 AMD Vega 11 on HBM2, the latest will be the only with FP64 Compute 2:1, total TFLOPS from 9/13/18 FP32 0.5/0.75/9 FP64, ram 16GB ddr5/ 16 GB HBM2/ 32 GB HBM2

As koyoot has mentioned, its unlikely Polaris can use both gddr5(x) and HBM. Its likely one or the other. I don't think they would waste silicon space on something they won't use and its likely gddr5x is more than enough bandwidth for Polaris' performance target. Also, the FP64 performance numbers are low, as its rumored to be a 1:3 ratio although its also possible amd reserves the unlocked chips for its workstation cards. Another inconsistency is if both the 310 and 510 are Polaris 10 XT then it would have the same GFLOPS as memory bandwidth does not factor in to this.

Vega also means that we are waiting until late 2016 or early 2017 for this machine. Would Apple wait this long to introduce a new one?

6 TB3 Ports, 4 USB-C 3.1, 1 HDMI2, dual 10.000/1000/100 MBPs lan on RJ45

As brought up elsewhere, this is a huge bandwidth requirement. 40 Gbps for each TB3 controller plus 10 Gbps for 2 USB-C controller. Possible only if Apple drops the GPUs to 8x PCIe.

Another thing I miss to copy from the darknet rumor is that ALL ports on the New Mac Pro Will be Video Enabled, 4K max on USB-C only, and 5K on thunderbolts.

This would be a lot of displays unless they cap it at a max of 6 total displays. USB-C should be able to do 5k over displayport 1.3.

upto 16 CORE Cpu

Why would they limit it to 16? Broadwell-EP goes up to 22.

Mago, If Apple will use Polaris and Vega GPUs there is absolutely no need for using 500+ PSU.

Sure there is room for a little bit more power. Right now the GPUs are limited to roughly 125 W. Bump the PSU up to 500 W and that gives you 150 W for both GPUs simultaneously. Guess what Polaris 10 is rumored at, 150 W. Sure you can downclock, but that also brings a performance penalty. As Vega will be a bigger die, it will also consume more power.

NDA for AMD event expires at... 29th June 2016. It is for reviewers of new hardware. Very interesting in this context.

Yeah, interesting. I am not sure how this factors in to a potential mac pro announcement at WWDC. I don't see Apple being the one to announce the technical specs of Polaris. So if AMD hasn't talked about Polaris performance in detail (i.e. GFLOPS, GCN cores, clock frequencies and price) by WWDC I don't think we see a new mac pro. AMD holds a press conference next week so we will see what they talk about.

Will we see a Mac Pro at WWDC? The tech is there now for Apple to release it. Broadwell-EP, Polaris and Thunderbolt 3 would be the headlining features. Could they wait for Vega, sure, but there will always be something to wait for and if they want to push thunderbolt 3 it helps to have the mac pro now vs in 6 to 12 months. Remember, they could always release a skylake-EP/vega machine in 18 months if they want to.

 

[koyoot]

http://www.macg.co/mac/2015/08/des-mac-4g-et-4k-dans-la-derniere-beta-dos-x-el-capitan-90354

Copied from Macbook Pro thread. Interesting news in the context of coherent fabric's .

 

[flat five]

Of course, efficient people will not watch the "calculator churn" (this isn't the 70's anymore) - they'll do other tasks not dependent on the system run.

At the end of the day, though - whether you're doing a large render, or a machine learning training run, or data analytics on a 20 TB dataset - you can't complete your "work" without the results of the run. Faster systems will let you do more work.

Open your eyes to all the ways in which professionals use computers, and stop trying to dismiss any of the many use cases that don't match your own lightweight workflow.

my eyes are open.

a decade or so ago, my workflow required the biggest baddest computer I could afford. today, at these forums and elsewhere, when people speak generically of what type of user requires high end computers, they use my work as an example.

that wiki link I gave you the other day:
https://en.wikipedia.org/wiki/Personal_computer

under 'types' there is 'workstation'..5 of those 7 examples of uses requiring workstations are things I use my computers for every single day of my life.

my workflow isn't 'lightweight' by any means.. at least not historically.

---
what's weird is that you all love arguing about latest and greatest and fastest tech. that's fine, fun even.. but you've been arguing that for 13 years on this forum. surely you can realize that tech has actually advanced, drastically in many cases, within that time and the things you were arguing about last decade are now mainstream tech.

computers (and just as importantly - software) have gotten so much better that yesteryear's demanding workflow can today be handled ,fairly easily, by a much wider range of computers.

I'm applauding this advancement. I wish for improvements in speed and then when the improvements come, I acknowledge and celebrate them.

you (not exactly 'you'), instead, will just try to undermine and ridicule me or my work as 'lightweight'.. 'you're not a real pro' etc.

guess what.. nothing is ever going to be good enough for you. while you may speak like you want advancements, that's just talk.. once the advancements come, you'll be on to the next negative thought while conveniently forgetting all about the arguments of speed etc you've spent the last decade yelling about. they're no longer relevant to the real reasonings of why you like spending time at a tech forum.

 

[tuxon86]

my eyes are open.

a decade or so ago, my workflow required the biggest baddest computer I could afford. today, at these forums and elsewhere, when people speak generically of what type of user requires high end computers, they use my work as an example.

that wiki link I gave you the other day:
https://en.m.wikipedia.org/wiki/Personal_computer

under 'types' there is 'workstation'..5 of those 7 examples of uses requiring workstations are things I use my computers for every single day of my life.

my workflow isn't 'lightweight' by any means.. at least not historically.

---
what's weird is that you all love arguing about latest and greatest and fastest tech. that's fine, fun even.. but you've been arguing that for 13 years on this forum. surely you can realize that tech has actually advanced, drastically in many cases, within that time and the things you were arguing about last decade are now mainstream tech.

computers (and just as importantly - software) have gotten so much better that yesteryear's demanding workflow can today be handled ,fairly easily, by a much wider range of computers.

I'm applauding this advancement. I wish for improvements in speed and then when the improvements come, I acknowledge and celebrate them.

you (not exactly 'you'), instead, will just try to undermine and ridicule me or my work as 'lightweight'.. 'you're not a real pro' etc.

guess what.. nothing is never going to be good enough for you. while you may speak like you want advancements, that's just talk.. once the advancements come, you'll be on to the next negative thought while conveniently forgetting all about the arguments of speed etc you've spent the last decade yelling about. they're no longer relevant to the real reasonings of why you like spending time at a tech forum.

Again, that is only your own personnal workflow and it isn't representative of the vast majority of workstation users out there. Here at work our engineer don't work on only one thing at a time. If their computer are over solicited by any of those tasks it will have a negative impact on their overall work output for the day. In other word, the more powerfull their workstation are, the more work gets done each day.

 

[rockyromero]

Will we see a Mac Pro at WWDC? The tech is there now for Apple to release it. Broadwell-EP, Polaris and Thunderbolt 3 would be the headlining features. Could they wait for Vega, sure, but there will always be something to wait for and if they want to push thunderbolt 3 it helps to have the mac pro now vs in 6 to 12 months. Remember, they could always release a skylake-EP/vega machine in 18 months if they want to.

Great news!

Fully agree with this.

I will get in line with this mentality and it provides me with workable options.

The MP will expand with various points of computing entries.

 

[flat five]

Again, that is only your own personnal workflow and it isn't representative of the vast majority of workstation users out there.

maybe my exact workflow isn't indicative of the vast majority of workstation users out there but my hardware requirements are.

go to any pro software forum.. people there are generally concerned with working faster and seeking improvements in the usability of the software and/or their own methods of using the software.. very rare are hardware topics seen but every once in a while, a thread will pop up asking 'what computers do you use?' (or similar).

the way you guys like to talk, everybody would answer with "i have the brand new x9000 with 3255423blahblahs and 3543gigalurks.. only the best for me because i'm a real pro!"
reality is, very very few even use a new computer (like- people don't buy a computer every year or every time an update is available)... few even use the best that were available at their time of purchase.

the vast majority of workstation users are on computers that a maxed out imac will out spec..

i think you should change your statement to 'Again, that is only your own personnal workflow and it isn't representative of the vast minority of workstation users out there.'

that would be more in tune with real world.

[edit]
for example.. at Mcneel's forum..
a benchmark thing was made for Rhino.. look at the computers people are using.
http://discourse.mcneel.com/t/holomark-2-released/8040

maybe not everybody in that thread is a 'real pro' but i can assure you, a considerable amount of those people are most certainly professional CAD users.. industrial designers, architects, engineers, (lots of developers in that thread too but i suppose the content is attractive to that type of mind)

 

[tuxon86]

maybe my exact workflow isn't indicative of the vast majority of workstation users out there but my hardware requirements are.

go to any pro software forum.. people there are generally concerned with working faster and seeking improvements in the usability of the software and/or their own methods of using the software.. very rare are hardware topics seen but every once in a while, a thread will pop up asking 'what computers do you use?' (or similar).

the way you guys like to talk, everybody would answer with "i have the brand new x9000 with 3255423blahblahs and 3543gigalurks.. only the best for me because i'm a real pro!"
reality is, very very few even use a new computer (like- people don't buy a computer every year or every time an update is available)... few even use the best that were available at their time of purchase.

the vast majority of workstation users are on computers that a maxed out imac will out spec..

i think you should change your statement to 'Again, that is only your own personnal workflow and it isn't representative of the vast minority of workstation users out there.'

that would be more in tune with real world.

[edit]
for example.. at Mcneel's forum..
a benchmark thing was made for Rhino.. look at the computers people are using.
http://discourse.mcneel.com/t/holomark-2-released/8040

maybe not everybody in that thread is a 'real pro' but i can assure you, a considerable amount of those people are most certainly professional CAD users.. industrial designers, architects, engineers, (lots of developers in that thread too but i suppose the content is attractive to that type of mind)

Again that's only your limited perception of how things work in the field and not the reality for the majority of workstation users in the industry. You keep pushing your use case as if it mattered. You are a single guy working on a single project at a time. I have thousands of engineers working on many major energy project at a time. They aren't working on a single thing at a time and they can't wait for one task to finish before being able to work with their machine. In short, their needs in performance vastly out pace yours.

 

[Zarniwoop]

maybe my exact workflow isn't indicative of the vast majority of workstation users out there but my hardware requirements are.

go to any pro software forum.. people there are generally concerned with working faster and seeking improvements in the usability of the software and/or their own methods of using the software.. very rare are hardware topics seen but every once in a while, a thread will pop up asking 'what computers do you use?' (or similar).

the way you guys like to talk, everybody would answer with "i have the brand new x9000 with 3255423blahblahs and 3543gigalurks.. only the best for me because i'm a real pro!"
reality is, very very few even use a new computer (like- people don't buy a computer every year or every time an update is available)... few even use the best that were available at their time of purchase.

I suppose Linus Torvalds can be regarded as a Pro user. Check out the first two minutes of this interview:

The toys really don't tell how Pro one is. But I like it too that what ever it is, it's quiet.

 

[tuxon86]

I suppose Linus Torvalds can be regarded as a Pro user. Check out the first two minutes of this interview:

The toys really don't tell how Pro one is.

He's a programmer... He can do is job with an atom based PC if GCC and VI is install on it...
Not the same as running some heavy engineering and design software package...

 

[Zarniwoop]

That's not a Pro task? And a compiler doesn't benefit of a speed at all?

Yes, there are a lot of different Pro needs. HP meets them with a huge army of possible combinations. Apple chose a different path. That is of course sad from our point of view.. for some Pro needs Apple is not the right choice. At least not today. But that's what we're investigating here, that maybe in near future...

 

[tuxon86]

And that's not a Pro task? And a compiler doesn't benefit of a speed at all?

Sure it does, but you don't really need a workstation to code. I compile most of my application including the kernel on my Arch Linux box at home with an i5 and 8gigs of ram and it finish compiling in minutes... Also, the main job of Linus these day is to approve/QA the different patches that other produce, not really coding them by himself. The linux foundation as a video of his home setup and you'll see that he's using a simple PC for this.