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

nMP 2015: where is it?

Most of people would simply answer: go back to old Mac Pro case.

I like this design more. Of course, I would not want to be left on Apples will alone, in terms of hardware, but I see future in this design. It will have purpose.

I like that it delivers great performance in this efficient design.

I dont like the price in comparison to what it offers, though :D.
 
I'm with you koyoot, the only thing I must admit is that it's in fact overly expensive for the machine it is. Still, remarkable piece of art and efficient design.

flat, way configurable indeed, but still looks like any regular PC. Powerhouse, configure as you want, no doubt. Haven't played with the configurator, but it seems it's no bargain either.
 
i was being sarcastic.. but that's basically the extent of answers you'll get.

the confusing part comes when people say "i want the most widespread design of the desktop.. you know, the same exact design that the first pc was.. components in a box.. "
"please please.. i understand there are thousands of choices for me to pick from that are exactly what i want.. except-- they don't have a picture of an apple on the outside.. please put the logo on the box for me.. k thanks"
 
2010 Mac Pro was released at WWDC
2012 Mac Pro was released at WWDC
2013 Mac Pro was announced at WWDC and released in December of that year

Macrumors buyers guide shows that as of 2010 they're on a roughly 2 year release cycle...

WWDC this year should be an announcement of an update at the very least... the 2013 gap between announcement & release would have been due to the new model and the new US manufacturing setup they built for it, so you'd have to think that this year the announcement would be almost the same time as the release.
 
koyoot said:
Most of people would simply answer: go back to old Mac Pro case.

I like this design more. Of course, I would not want to be left on Apples will alone, in terms of hardware, but I see future in this design. It will have purpose.

I like that it delivers great performance in this efficient design.

I dont like the price in comparison to what it offers, though :D.
Click to expand...

They are really two very different products, brilliant in their own way, and personally I think they could coexist.

Since we're moving our office I've been pulling apart the Dell Precisions and HP Z820/Z840s we just got... the Dell is a horrible cable-filled mess of an interior layout (with the HDD's basically stuck randomly on the very outside of the case for whatever reason), and while the HP is definitely a bit more organized, it's a strange machine that feels like its meant to be easily serviced, but not easily upgraded—sure, you can get to the PSU without using a tool, but uncovering the shrouds and fans to get at the slots, processors, and RAM is a production compared to the ease of the 4,1 and 5,1 Mac Pro towers. They are still a killer design.

I guess maybe I'd have loved it if Apple had approached the Mac Pro the same way they've approached things on the laptop side—produce the new cutting edge design but keep the old one kicking around for a while.

I think I'll be immensely happy with a nMP when I can finally justify getting one, but I do feel for the people who feel like they no longer have a desktop option that really speaks to them in the Macintosh lineup.

On one hand, you can say "Apple is the most successful company in modern history, they can afford to make the graters!" On the other, you don't become the most successful company probably by sticking in the past, and Apple has always been one to embrace the future, perhaps earlier than they should.

mustgroove said:
2010 Mac Pro was released at WWDC
2012 Mac Pro was released at WWDC
2013 Mac Pro was announced at WWDC and released in December of that year

Macrumors buyers guide shows that as of 2010 they're on a roughly 2 year release cycle...

WWDC this year should be an announcement of an update at the very least... the 2013 gap between announcement & release would have been due to the new model and the new US manufacturing setup they built for it, so you'd have to think that this year the announcement would be almost the same time as the release.
Click to expand...

At this point though I don't see much of a reason to upgrade to Haswell Xeons when theoretically we'll be seeing the next processor revision by June in just a few months. The nMP would certainly benefit from a refresh at any point, but it would be odd to be so out of sync with the processor lineup—dropping a new rev in late 2015 makes more sense to me at this point.

ManuelGomes said:
My mistake, the W7100 and W5100 are there already, not on the left side links but below the links are there.

----------

OK, I'm about to start a war here I guess :)
Who wants to comment the bad decisions Apple made in the nMP?
Click to expand...

The only "bad" design decision I'd say is as you brought up the non-identical card fittings. Everything else I don't see as "bad design", because creating a machine that isn't for you isn't a matter of design, it's a matter of scope. It might be bad from a business perspective, but that's a separate issue.
 
Last edited:
fuchsdh said:
At this point though I don't see much of a reason to upgrade to Haswell Xeons when theoretically we'll be seeing the next processor revision by June in just a few months. The nMP would certainly benefit from a refresh at any point, but it would be odd to be so out of sync with the processor lineup—dropping a new rev in late 2015 makes more sense to me at this point.
Click to expand...

Where are you seeing reports of Broadwell Xeons in June? I think that was maybe the original schedule before last year's big Broadwell delay?

The latest reports I've seen with leaked Intel roadmap slides are indicating Q1 2016 for Broadwell-E (which is the basis for the 16xx v4 Xeon used in the low-end nMPs). Broadwell-EP (26xx v4 Xeon used in the high-end model) will likely arrive around the same time. This makes sense as it would be about a year following the mainstream Broadwell launch which is just starting now and would be consistent with Haswell.

I've suggested this before, but I think the better crystal ball to be peering into is not the one owned by Intel, but the one owned by AMD. The question there is... When will the next-gen AMD GPUs arrive? (Edit: Summer for AMDs new top-end Fiji GPU)
 
Last edited:
VirtualRain said:
Where are you seeing reports of Broadwell Xeons in June? I think that was maybe the original schedule before last year's big Broadwell delay?

The latest reports I've seen with leaked Intel roadmap slides are indicating Q1 2016 for Broadwell-E (which is the basis for the 16xx v4 Xeon used in the low-end nMPs). Broadwell-EP (26xx v4 Xeon used in the high-end model) will likely arrive around the same time. This makes sense as it would be about a year following the mainstream Broadwell launch which is just starting now and would be consistent with Haswell.

I've suggested this before, but I think the better crystal ball to be peering into is not the one owned by Intel, but the one owned by AMD. The question there is... When will the next-gen AMD GPUs arrive? (Edit: Summer for AMDs new top-end Fiji GPU)
Click to expand...

Sorry my wording was unclear—I meant that if Apple waited as long as June to update, it seems odd they would update with parts that would be at best mid-cycle by that point, not that the processors themselves would be out in June.

I do agree that if there was big movement on the GPUs that would be a possible reason.
 
fuchsdh said:
Sorry my wording was unclear—I meant that if Apple waited as long as June to update, it seems odd they would update with parts that would be at best mid-cycle by that point, not that the processors themselves would be out in June.

I do agree that if there was big movement on the GPUs that would be a possible reason.
Click to expand...

Remember that the Haswell and Broadwell should use the same socket - so Apple has to update the motherboard and CPU daughtercard regardless.
 
My point exactly, since both Haswell and Broadwell use the same socket, and since Broadwell is still far off, they might as well update to Haswell in June when Fiji's out. H1'16 will see Broadwell with no changes to the motherboard.
Only in H1'17 or so when SkyLake is out there will be changes again, with the equivalent PCH of the desktop 100 series (C620 or C700?), where PCIe 4 and TB3. But then again Apple will probably only update in the next iteration, successor to the SkyLake a year or so later.

----------

Broadwell will also bring support for faster DDR4 memory, 2400 instead of 2133, but the difference should be minimal.
 
ManuelGomes said:
[/COLOR]Broadwell will also bring support for faster DDR4 memory, 2400 instead of 2133, but the difference should be minimal.
Click to expand...

Yeah it doesn't seem like DDR4 is getting pushed that heavily, especially since the on-paper percentage increase with 2133 vs. 1833 isn't necessarily what you'll get—I just realized that of the processor models HP offers, we're only getting 1833MHz DDR4 with a 4-core 3GHz v3 on the new computers in our office. In reality we're probably not getting much real-world benefit (aside from the fact that most of our editing rigs are 2008 or so so there's obvious speed advantages from that much of a leap.)
 
Right, DDR4 started out quite quickly but it seems it will go up in speed slowly.
Some manufacturers are providing fast modules, but that's normal in the desktop world. For workstation we'll be getting small bumps I guess, now 2133, 2400 later with Broadwell, probably 2666 or so with SkyLake.
But the real benefits are quite small with such increases I'd say.
Lower power is actually where it shows it's benefits.
 
mustgroove said:
2010 Mac Pro was released at WWDC
Click to expand...

Not true. https://buyersguide.macrumors.com//#Mac_Pro
July 2010 not June 2010 .

2012 Mac Pro was released at WWDC
Click to expand...

A dramatically limited speed bump. This wasn't really an upgrade as much as a "The Mac Pro isn't dead yet" announcement. 2-3 year old GPU complete non-update and minor clock speed bump on two year old CPU architecture. [ In contrast every other workstation vendor was moving to Xeon E5 v1 in the June 2012 time frame. Several were shipping by June 2012. ]. This was an appeal to wait until 2013 for anyone who needed substantive improvements. [primary upside was that it got another round of 3rd party GPU card updates going as this model was going to have to fill stopgap duty for a even longer extended period of time. ]



2013 Mac Pro was announced at WWDC and released in December of that year
Click to expand...

Again nothing released at WWDC and was at least as much an announcement that the old Mac Pro form factor and product line was dead than informative about the new product. [ The Mac Pro had been bannede from EU markets for about 4 months at this point because it was so old and out of compliance. ] In terms of a product release/update this was October/December not WWDC.


There is no actual new (as in substantive upgrade) Mac Pro correlation here at all with WWDC. None in 2009 either ( March). None in 2007 ( April ). None in 2006 ( August ).

This whole "Mac Pro in WWDC " hand waving does always show up in this forum every Spring in the Feb-April time frame. The track record for accuracy and relevancy is dismally bad.


P.S. If Apple has established anything is that WWDC might be when they announce their "excuse of the year" for not shipping any substantively new Mac Pro. Reality Distortion Field adjustments; not actually doing something.
 
Last edited:
VirtualRain said:
The latest reports I've seen with leaked Intel roadmap slides are indicating Q1 2016 for Broadwell-E (which is the basis for the 16xx v4 Xeon used in the low-end nMPs).
Click to expand...

Backwards. The -EP 16xx v4 is a baseline for the Core i7 (-E) variants with some swamps on features. It is the reverse of the Xeon E3 being variation on the baseline Core i mainstream baseline.

The associated chipset follows the (-EP) once per tick/tock cycle update frequency which is yet another indicator of which implementation is the baseline and which one is the variant.


Broadwell-EP (26xx v4 Xeon used in the high-end model) will likely arrive around the same time.
Click to expand...

Xeon E5 26xx and E5 16xx use the exact same chipset. They aren't in different product arch categories. There is a different mask (and die size) used in 16xx level, but there are different masks inside the 26xx range too. That doesn't drive the (-EP) designation.



, but I think the better crystal ball to be peering into is not the one owned by Intel, but the one owned by AMD. The question there is... When will the next-gen AMD GPUs arrive? (Edit: Summer for AMDs new top-end Fiji GPU)
Click to expand...

But is Apple going to drop FirePro drivers and alignment. If stick with FirePro then it is several months downstream after the FirePro updates. ( since Apple's drivers are trailing )

Apple hasn't enabled/rolled out "same time as Windows" drivers in quite a long time. All the more true when there are customized graphics cards involved. There are no indications they are going to drop the 'air cover' that FirePro lends to their GPU pricing strategies.

AMD is unlikely to trim availability of their brand new top end GPU to highly custom cards of Apple. That is just less GPUs available for what probably will be a red-hot market in trying to match paces with Nvidia over next 6 months.
 
ManuelGomes said:
nVidia just launched the M6000 based on the Titan X.
Poor FP64 performance, 1/32 as the Titan X.
Click to expand...

True, but it's targeted towards more graphically demanding tasks.

If you want to use it for computing, there are other options. Mostly the cards based on the GK110/GK210 chip offer better compute performance.

Nevertheless, I would *love* to see a M6000 as a BTO Option in the eagerly expected nMP 2015.
 
Seems strange that that new M6000 only has a 8pin cable. In torture tests on tomshardware it almost hit 250 watts, so it overloaded its power rating of 225 watts.
 
Surrat said:
Seems strange that that new M6000 only has a 8pin cable. In torture tests on tomshardware it almost hit 250 watts, so it overloaded its power rating of 225 watts.
Click to expand...

Evidently Nvidia is implicitly expecting that the card will only be used in systems with over provisioned power supplies.

"... . We asked NVIDIA about this, and they have told us that the card is pulling the extra power from the 8-pin connector, and though not officially in spec, the kind of systems expected to house the M6000 are expected to have no problem delivering the extra amperage necessary. ..."
http://anandtech.com/show/9096/nvidia-announces-quadro-m6000-quadro-vca-2015

Perhaps the expectation is built into their minimal power supply recommendations but is a bit of an indirect requirement.

Go cheap or tight specs on the rest of your system specs and probably have a problem. That kind of mentality though pretty much runs completely opposite of where the Mac Pro is now.
 
deconstruct60 said:
...Go cheap or tight specs on the rest of your system specs and probably have a problem. That kind of mentality though pretty much runs completely opposite of where the Mac Pro is now.
Click to expand...

Never go tight on the power supply - the failure rates go up quickly when you run them at close to the rated load. (Except for some which are very conservatively rated.)
 
AidenShaw said:
Never go tight on the power supply -...
Click to expand...

If you control the components then don't necessarily need tons of slack. If have a general erector set of highly arbitrary components then probably do need tons of slack due to the variability and interaction effects. This M6000 card is in the high end erector set zone.


(Except for some which are very conservatively rated.)
Click to expand...

IMHO that is the disconnect with what Nvidia is doing. They are basically taking the conservative capacity allocated for reliability and pushing it into running an operational "norm". Probably not going to immediately damage the power supplies, but it is a trade-off.

A trade-off that Apple probably won't sigh up for.
 
beaker7 said:
Why not? They're running the current Mac Pro right at the edge of its thermal and electrical limit, throttling when necessary.
Click to expand...

Because it is only indirectly attached to the thermal/power management feedback loop. There is no control loop telling the power supply to pump extra juice. Apparently, they just "yank" and pull the delivery ignoring the spec envelope.

There is little evidence that the current Mac Pro components ignore the parameters they are externally told to stay inside of. If the thermals build up too much on the three major producers they all three will walk back from their max allocated loads. As major sink A draws more power major sink B is thermally coupled and will fall back isn't a problem because it is coordinated.

The wider dynamic swings the individual components can unilaterally make the less coordinated they will be.
(e.g., OWC's SSD that bleeds power off the GPU is probably has consequences also. )



P.S. I don't Apple will be crying any tears either if the Mac Pro power supplies die after 4-6 years either when Apple's liability is gone (or largely gone for most users). But they will still probably have less liability for replacements then when end users to stuffing random components in and over juicing just because they could.
 
Last edited:
deconstruct60 said:
Because it is only indirectly attached to the thermal/power management feedback loop. There is no control loop telling the power supply to pump extra juice. Apparently, they just "yank" and pull the delivery ignoring the spec envelope.
Click to expand...

Perhaps true - but going 10% over the power budget on a card in the $1K range seems pretty reasonable.

I wonder how many cards out there are running with 6-pin to 8-pin adapters without any issues....

41VWBjORreL._SX425_.jpg

http://www.amazon.com/Manhattan-8-I...qid=1426886894&sr=8-2&keywords=6+pin+to+8+pin
 
As an Amazon Associate, MacRumors earns a commission from qualifying purchases made through links in this post.
All recent Nvidia GPU BIOSs allow you to set a power target. If Apple wanted to they could easily interact with BIOS and tie it all up together.

If they wanted a modern and power efficient GPU that made their machines relevant, it's there.

As I just pointed out in another thread, the drivers for "Big Maxwell" are already available in the Nvidia Web Driver for 10.10.2. So, they were there on day of release.
 
You must log in or register to reply here.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Top Bottom
Back