Sidebar Sidebar
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

Will there be a new Mac Pro in 2015?

Status
Not open for further replies.
I might sound skewed for what are my needs. Indeed some more people will also only need a 1S machine without a gigantic number of cores, TBs of memory, RAID or loads of disk space, melting powerful GPUs and whatnot.
The nMP was not designed for that crowd I guess. It's good enough for me and others like me. Perfect in fact.
I can settle with one CPU, 2 GPUs, 1 SSD with enough room and a good single display or 2. That's why to me they could stick only one TB3 controller in the CPU, the SSD also there taking up the rest of the lanes, and even include a TB2 controller off the PCH for other eventual stuff. The 6 USB3 ports on the PCH would also be enough.
The more power hungry guys will unfortunately have to go somewhere else I believe, or so it seems that is what Apple is hinting. Or maybe a surprise will be coming in the future, and they'll release a Mac Server Pro or something, but I actually don't believe it anymore.
 
Give me:
- a slightly higher baseline capacity SSD
- room to add a second SSD
- ONE graphics cards option at a fair price
- and marginal updates on everything else
- dual CPU slots would be a huge plus since it extends the lifespan slightly BUT not a requirement for me.

And I will buy in a second. Me and many many other people.

The lack of secondary SSD and the crazy price tag to have two obsolete graphics cards is what caused me not to buy.

For my needs I need one graphics card and tons of CPU power with more hard drive space then they provide. I can deal with not having my 4 internal TB drives. It's not ideal. but with 2 500 gb SSDs I could stomach it.

My early 2008 mac pro is getting really old. For what I do its a choking hazard. It's still manageable but I cannot wait for this new update! But I guess I have to. I'll give it until the end of the year. Then I'm going PC if nothing shows.

In the end, I just kind a wish they had never thought of the garbage can. Would have loved a giant silver box that was up to date.
 
Stacc said:
When it comes to the retina display, it may be possible for some kind of adapter to exist that would convert 2 thunderbolt 2 channels to thunderbolt 3.
Click to expand...

BTW, such an adapter is entirely possible. Not necessarily from two TB2 to TB3, but from two DP 1.2 to DP 1.3. There are already displayport concentrator chips that do this for DP 1.2 and one thing that Apple loves is being able to sell overpriced adapters. So, it's entirely possible we see a single cable 5K monitor that has an adapter so that recent Macs can use it too.
 
thedommer, 1TB SSD is as much as we get right now. Not sure Apple will provide a 2TB SSD with the next update, unlikely though. OWC provides one but with reduced performance if I recall correctly.
For a stupid amount of money you can get a 12 core :)
But like you I'm playing the waiting game, but it's getting boring... without solid info on when it can be expected.
 
TB3 display might have an embedded GPU, based on Antiqua? Or even Fiji?
Does it make sense for Apple going the eGPU right now?
I'd figure an eGPU would drive the cost even higher and render integrated GPUs almost useless, or they could be both used for compute. OR maybe this is the time for a dual CPU and one GPU nMP?
Nah...
 
I really only have until mid-september before I have to buy. I'm hoping something comes out around then...but I'm probably going to get stuck with the old tech. *sigh
 
  • Like
Reactions: JamesPDX
ManuelGomes said:
TB3 display might have an embedded GPU, based on Antiqua? Or even Fiji?
Does it make sense for Apple going the eGPU right now?
I'd figure an eGPU would drive the cost even higher and render integrated GPUs almost useless, or they could be both used for compute. OR maybe this is the time for a dual CPU and one GPU nMP?
Nah...
Click to expand...

External GPUs are certainly a cool idea and could be a nice solution for laptops. It seems like one of the major hold ups was what happens when the GPU is unplugged while in use, and Intel has been talking about having a solution for this in Thunderbolt 3. However, I don't really see Apple going this route. The biggest reason is probably how small the market would be, especially given the likely high cost of adding the GPU+thunderbolt solution. I am sure there would be technical challenges too. Additionally, I believe the external GPU is only able to drive an external screen, further limiting its applications and market.
 
ManuelGomes said:
TB3 display might have an embedded GPU, based on Antiqua? Or even Fiji?
Click to expand...

An iMac GPU perhaps ( i.e., mobile focused one). The Display docking station is to some extend an gutted iMac. That will likely continue to be the case in the future (both to meet the component pricing and margin targets Apple wants ).


Does it make sense for Apple going the eGPU right now?
Click to expand...

Depends upon if the bulk of Display Docking stations are being sold to MB ( CoreM GPU clocks ) or MBA ( ULV GPU clocks ) or not. For the former set of users, the iMac GPU would be a set up. For the desktop Macs ( including the Mac Pro) it is a dubious value add. I don't think the bottom half of the laptop space can drive 5K/Retina 27" displays very well so eGPU might be pragmatically necessary if a "Retina only" solution is targeted.

A Retina 21.5-23" Display docking station would be both less costly and likely not need a eGPU to comfortably drive it. The question remains though whether Apple will provide more than just one single "display" product.

If eGPU is going to be made part of the TB v3 certification process then yes it would make sense for Apple to have an internal lab product to test out the driver software on. That doesn't necessarily mean it will ever see light of as a product for sale.


I'd figure an eGPU would drive the cost even higher and render integrated GPUs almost useless, or they could be both used for compute. OR maybe this is the time for a dual CPU and one GPU nMP?
Nah...
Click to expand...

For compute the eGPU utility is limited to how much data and how much "loop" on a fixed set of data. The bandwidth to the eGPU is relatively low. With a mobile GPU at best you are likely looking at 2-4GB tops in terms of VRAM space.

eGPU do little to offset what the dual GPUs can do in a nMP. Dual CPU doesn't work both physically or as noted above synch up with the direction Intel is going with 2S CPU designs. Workstations are generally heading toward 1S targeting. ( there will be still be some "outer limits" fringe 2S Workstations but they are likely to increasingly shrink as a percentage of the market. )
 
Last edited:
I was referring to a certain podcast supposedly from someone at Apple, which I haven't seen, that confirms this "new" display. Find it odd though.
As a docking station for laptops, ok, that could be a solution.
But for the Mac Pro it's kind of a silly way to go.
I was, of course, joking with the dual CPU comment. It will never happen, not with this form factor. For this to be possible the mobo would have to be a lot bigger, housing both CPUs, the PCH and all the mem. Won't happen, unless there's a new redesign.
 
ManuelGomes said:
I was referring to a certain podcast supposedly from someone at Apple, which I haven't seen, that confirms this "new" display. Find it odd though.
As a docking station for laptops, ok, that could be a solution.
But for the Mac Pro it's kind of a silly way to go.
I was, of course, joking with the dual CPU comment. It will never happen, not with this form factor. For this to be possible the mobo would have to be a lot bigger, housing both CPUs, the PCH and all the mem. Won't happen, unless there's a new redesign.
Click to expand...

The mobo (the round thing at the bottom) could be the same (different PCH, but same form factor). You'd need either a new CPU daughtercard, or two daughtercards with many connections. It's also complicated by the fact that you have twice as many PCIe v3 lanes but half come from each CPU.
 
Yes, the motherboard is the round thing at the bottom, but as you can imagine a 2S system requires an additional QPI/UPI link between processors, and that might not be the easiest thing to do between the CPU boards. I don't think the approach "just put a flat cable between the boards" would work very well, although it seems to work fine with the current setup.
However, the extra PCIe lanes would have to go there anyway, and I very much doubt the size would remain the same.
Still, the best solution would be having everything on the same board I suppose.
But that would ruin the overall design and size.
 
Leaked benchmarks put the Fury Nano at the same level as a 290X. It's 10w higher than GTX980.

Not a bad card, good for compute, but not as good as Nvidia for general use.
 
SoyCapitanSoyCapitan said:
Leaked benchmarks put the Fury Nano at the same level as a 290X. It's 10w higher than GTX980.

Not a bad card, good for compute, but not as good as Nvidia for general use.
Click to expand...
http://www.overclock.net/t/1565417/dglee-new-r9-nano-pictures-and-unigine-heaven-results-from-amd

Where did you get that its the same level of performance as R9 290X? WCCFTech I guess. Full Fiji chip clocked at 900 MHz should be at performance levels of Fury. 157W. I think if Full Fiji was able to be squeezed to this level, Fury can be even more ;).
 
  • Like
Reactions: SoyCapitanSoyCapitan
And yeah, I've checked a few results from 4K Unigine. Yep, it is as fast as Fury. Also its worth noting that Asus Strix Fury TDP rated is at 216W. That means, all of the cards can be squeezed to 125W while maintaining 85-90% of nominal performance. AMD did the same thing again. Pushed the chips to the absolute limits.

There is one way to move around it. Put the GPU at 850 MHz on core. And overclock the HBM to 625 MHZ and get 640 GB/s of bandwidth. Overclocking results on Fury X are so far very promising.

P.S. What is funny is that you could get insane performance levels with quadfire setup of Fury Nano, while having only 1000W power supply :D

Around 30 TFlops of raw compute power. That is insane.
 
Last edited:
SoyCapitanSoyCapitan said:
Leaked benchmarks put the Fury Nano at the same level as a 290X. It's 10w higher than GTX980.

Not a bad card, good for compute, but not as good as Nvidia for general use.
Click to expand...

Apple does not care about general use. They care about use which they're selling (i.e. FCP X, Logic Pro X etc).
Mac Pro isn't their flagship product anymore (since a few good years). It has to be "amazing enough" to keep them in their "most inventive company" image and has to be the top performing machine for their software. Nothing more IMHO.
Real income comes from elsewhere (cpt obvious ;)).
And they know, that OS X simplicity is addictive and some amount of professionals will have a hard times to leave it for good. Not to mention Mac-only software like Smoke.

Besides, this whole thread is starting to run around in circles ;)
 
Nano would certainly fit perfectly on the nMP, except for DP.
And it sure would make it "amazing enough" in my opinion.
Innovative (HBM) tech, new silicon to last for the next couple of years (although HBM2 will be out next year), good performance with low power draw, what else do you need.
It's not perfect for everything ("low" DP performance), no "single cable" 5K output solution, but those improvements are still about a year away anyway.
At the time, it's as good as it gets.
OK, NVidia GPUs would be great too but it seems it costs more to have them, and if they're not willing to bend over it's their right not to.
 
koyoot said:
Overclocking results on Fury X are so far very promising.
Click to expand...

I want what you are smoking.

This is the EXACT OPPOSITE of reality.

The reason they went with water cooling on first release is that they had to max it out to even get a glimpse at Nvidia's taillights. The card is pretty much "overclocked" as far as it is going to go.

Expect a short life for it as well.

"We’ve seen reports of up to 10 percent GPU core overclocks on Fury X around the web, but our particular sample proved unstable even with a mere eight percent overclock, despite maintaining low core temperatures. We ended up with an even less impressive seven percent stable overclock, which, as you might imagine, means real-world performance is only 3–5 percent faster than stock."

http://www.maximumpc.com/gtx-980-ti-vs-fury-x-overclocking-showdown/
 
koyoot said:
http://forums.anandtech.com/showthread.php?t=2440030

Sorry MVC.
Click to expand...

For what?

I'm going to guess you didn't actually read that thread.

One guy puts a peltier on a Fury and makes some claims in a blurry Youtube video. Meanwhile a dozen reputable Tech reviews came to same conclusion I did. The Fury-X came with a water cooler for the same reason Apple put water cooling on some G5s, they were losing the Mhz war and needed to pull out all the stops to stay even remotely competitive. (Steve Jobs and the 3Ghz G5)

Nobody, and I mean NOBODY has gotten the GPU up more then 10% on Fury, meanwhile Nvidia 980s and Titans easily pull 20% or more, on air.

Read the thread, by the last page everyone is coming to same conclusion. Here is one poster from today:

"The difference is Titan X could still OC well at stock voltages. Not hard to hit 20% OC on the reference cooler either, so long as you are willing to deal with noise.

Modded bios were also figured out pretty quickly for those that wanted to push the voltage on water.

It's been a month since Fury X release now with seemingly no progress on the OC front. Apart from the updates from Unwinder, AMD is completely silent."

And another:

"I'll say this again (and again, until someone gives an explanation telling me my line of thinking is wrong):

The OEMs would have had custom BIOSes if AMD had allowed them to. It should not be hard for OEMs to modify the voltage in BIOS. That the OEMs did not do this tells me that they're blocked either from editing the BIOS at all, or specifically blocked from changing the voltages (either by lack of tools, which the OEMs should be rushing to fix, or by policy/agreements with AMD preventing them from doing so.)

It is VERY telling to me that no OEM is upping the voltages. Maybe that's because AMD doesn't want to deal with more failures. Maybe it's because they know that overvolting the card will result in problems. I don't know."

AMD already has that card maxed out, please read a few reviews from someone with credentials beyond a Youtube video.
 
  • Like
Reactions: tuxon86
The problem are the scores in Benchmarks. Somehow, after OCing the Memory, and GPU core you get higher scores on 3dMark benchmarks than before.
http://uk.hardware.info/reviews/615...-flag-ship-graphics-card-overclocking-results

The same goes for the benchmark results that guy from the film posted on YT. And no, I completely don't care about how well performing are Nvidia cards in this event. I cared only about higher scores before/after OCing Fury. Stock Fury X should get around 14k points in 3dMark FireStrike, and around 16k graphics score. AMD says the HBM memory OC is locked by them, and yet, somehow the results are better. You cannot just say its placebo effect.

The benchmarks before/after look promising. The other question is - how it will translate to relative performance. Because, right now the bottleneck is not in the AMD hardware. Which is shown by the difference in performance between Fury, and Fury X.GPU specs show that the gap between those two GPUs should be wider, and it is not. In 4K in most games its only 2-3 FPS.

From my perspective, and Mac Pro perspective, as I said higher in post - best way would be getting the core to 900 MHz, and increasing the bandwidth to 640 GB/s by running the HBM on 625 MHz.
 
fuchsdh said:
If I were in Apple's shoes, I would just put out a better v3 line with the better GPUs and slightly better processors, or just make some coin off people with some standalone GPU upgrades.
Click to expand...

(emphasis mine)

While we're discussing how potentially awesome Fury/nano based cards could be, how realistic is it that Apple would offer a GPU board upgrade path?

I seem to hazily remember there were such things in the past, but of course Apple has changed quite a bit in the modern "glue everything down" era.

I've been looking at picking up a secondhand nMP to tide me over until a (hopefully) October or new year refresh, but I just can't bring myself to pay top dollar on the eBay market to get such anaemic video cards, that I just know will be blown away by the tech they put in next.

However, it would take the sting out a purchase now (either secondhand or at Apple's jaw dropping prices for 2013 tech) if I knew that the D500 in a nMP could be replaced in future.

A justifiable hope, or a pipe dream?
 
  • Like
Reactions: JamesPDX
Status
Not open for further replies.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Top Bottom
Back