E5-2400 was a crippled 2600, but good enough for some configs.
Not sure they still have those.
Not sure they still have those.
Got a tip for us?
Let us know
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?
- Thread starter Anto38x
- Start date
- Sort by reaction score
- Status
Skylake Xeons will probably only be available in 2017, so it's still a long wait.
Intel will also change the setup for SKL. The platform will be divided, according to a roadmap, where the 1600 and 2600/4600 will have different routes. I guess 1600 will get more mainstream, and even use desktop PCH instead of current C600 series. this could be good or bad. You get the desktop features which is great, but will it be at the expense of robustness? Time will tell.
Intel will also change the setup for SKL. The platform will be divided, according to a roadmap, where the 1600 and 2600/4600 will have different routes. I guess 1600 will get more mainstream, and even use desktop PCH instead of current C600 series. this could be good or bad. You get the desktop features which is great, but will it be at the expense of robustness? Time will tell.
There is no Broadwell-E coming to the Xeon product. EP but not E. -E is used for the derivation used for the Core i7 x9xx products that actually part of the mainstream dekstop/laptop Core i implementations only in label marketing; not in implementation. Implementation wise, they are on a different track which is synched up with Xeon E5 products and chipsets.
In general, the general micro-architecture code names do not exclusively offer same features over all of Intel product lines. For Macs, there are two basic groups of Intel processor that Apple uses. One is the general desktop/laptop group marketed under Core i label. The Mac Pro is on the Xeon E5 label. Those two get features at different rates. Some things come to Core i in a given generation and others will come to Xeon E5 first and then to Core i later. [ Core M, used in the MacBook, is a coupled to the desktop/laptop Core i implementations. It is has a few more optimizations for lower power, but on the same micro-archtecture feature grouping. As mentioned above the Core i7 x9xx are really not the same as the rest of what is labeled "Core i". ]
Intel has a consistent naming standards per product lines, but not across product lines.
Xeon ---> Xeon Ex XXXX v n [ where n is the generation and the count started at 1 ]
Core i ---> Core ix nxxxx [ where n is the generation and started counting one generation step earlier than the above. ]
Xeon E5 xxxx v4 is "Broadwell-EP" and Core i7 5xxxx is Gen 5 "Broadwell" desktop/laptop.
Skylake exclusively offers none of these across the whole of Intel's product line up.
DDR4.
DisportPort 1.3
Thunderbolt v3
Xeon E5 v3 (Haswell) has DDR4 now. The desktop/laptop line up is just getting it with Gen 6 (Skylake ). None of them have DisplayPort 1.3. Gen 6 will have embedded DisplayPort, but that doesn't go to external ports. Xeon E5 has no embedded GPU so has DisplayPort nothing in and of itself. The chipset associated with Gen 6 ( SKylake) better supports having Thunderbolt v3 implementations but Gen 6 is not necessary to deploy a TB v3 system. Thunderbolt is not "inside" of any Intel CPU package or I/O chipset package. So it is not a property of those. Discrete TB v3 controllers be done with a Xeon E5 v3 or v4 (Broadwell ) system. There is zero requirement for a "Skylake" specifically it is more so a better synergy effect for the desktop/laptop versions.
Part of the confusion here is generated because people are still treating these as simple x86 core only CPU. In other words as just a Central Process Unit. These are more so packages that contain the CPU plus other stuff. Also since Intel is defacto the only I/O chipset vendor each CPU generation comes with a designated I/O chipset. So two sources of sloppiness.
1. the 'codename' usage over a several product lines is really a similar x86 implementation baseline design. It doesn't necessarily have to do with the other stuff bundled (or not) inside the CPU package that is not what was classicly named as being the CPU. "Broadwell" Core i Gen 5 and Xeon E5 v4 share design objectives but not necessarily complete feature groupings or implementations.
2. Since the I/O chipset is coupled to the CPU packages sometimes folks label the CPU "codeword" onto the chipset features. The chipset actually have their own code words.
Frankly all of this codeword stuff is become highly bogus over time. Intel has had fairly uniform product naming guidelines for several generations now. If folks would just use the generation version numbers instead of the confusing and hard to track codewords then things would be more clear. Tracking the codewords is becoming more of the "secret handshake" than being necessary.
The additional implicit confusion added in several points of the thread is that since Xeon E5 v4 and v5 are coming later in time then other chipsets and GPUs will be available at those later dates too. That's where the DisplayPort v1.3 comes from. There are no GPU from anyone with that now. Probably none until 2016. ( embedded DisplayPort (eDP) is on a faster generation tick rate than the external port version .)
Because they can. To a large extent this is due to not have a creditable competitors. Intel can take the E5 2600 design and cripple it slightly to sell another product line into a different market segment. The price of 2600's is high so there is a viable set of folks who don't quite need all of the horsepower but want something cheaper. It is not hard for Intel to tweak the design to turn off stuff and then use that to offer at a lower price point.
I suppose it keeps some jr. design teams busy by tweaking designs with all the major problem removed.
It keeps some set of customers happy because they "demanded' a better solution for their design and got something for it.
If AMD was heavily breathing on Intel's neck they wouldn't be devoting resources to these "cripple and sell cheaper" products.
Making every product SKU possible isn't really helping the PC industry (shotgun shots at the side of the barn hoping to see what will stick ). It isn't going to help the server one either as competition goes up. Same thing for the workstation market.
The problem right now with names/versions of Intel CPUs is, in my opinion, that they messed up so much that no one can keep track or even knows which is which.
The generational gap, the different version numbers , the models - nothing matches up.
Be it Broadwell being ix-5xxx, Xeon E5-x600 v4, Core M or whatever not is so damn not logical that makes you want to scream.
Despite release cycles, couldn't they at least keep families within the same numbering scheme?
Take also the ix Core family for instance. the differences between Core i7 and i5 are not always the same, which makes it impossible for someone to make an educated decision. If you knew i7 would always have HT enabled (for example) or other features, then you'd know the difference. But when you realize that, considering desktop as well as mobile, this is not always true - some i7 may or may not have HT enabled, or have larger caches - it's a jungle really. And with each new generation it gets worse, the features are not always indicative of the family CPUs belong to.
With the iGPUs it gets even worse, the naming schemes are changing constantly and with no apparent sense. Now it went from 6xxx to 5xx, which can indicate a power end part, which is not correct.
Intel, get your act together...
The generational gap, the different version numbers , the models - nothing matches up.
Be it Broadwell being ix-5xxx, Xeon E5-x600 v4, Core M or whatever not is so damn not logical that makes you want to scream.
Despite release cycles, couldn't they at least keep families within the same numbering scheme?
Take also the ix Core family for instance. the differences between Core i7 and i5 are not always the same, which makes it impossible for someone to make an educated decision. If you knew i7 would always have HT enabled (for example) or other features, then you'd know the difference. But when you realize that, considering desktop as well as mobile, this is not always true - some i7 may or may not have HT enabled, or have larger caches - it's a jungle really. And with each new generation it gets worse, the features are not always indicative of the family CPUs belong to.
With the iGPUs it gets even worse, the naming schemes are changing constantly and with no apparent sense. Now it went from 6xxx to 5xx, which can indicate a power end part, which is not correct.
Intel, get your act together...
Indeed. Having options is great and all, but when it starts to be too much, and those options are not absolutely clear, it gets confusing.
Intel is even in a position (no competition) where they really don't need to diversify that much, fewer parts means less SKUs to handle. The good thing is they get to use even the less performing parts and powers SKUs, but it gets so confusing at times.
Intel is even in a position (no competition) where they really don't need to diversify that much, fewer parts means less SKUs to handle. The good thing is they get to use even the less performing parts and powers SKUs, but it gets so confusing at times.
I love this thread. I pop in to read every now and then to make my mouth water.
Right, I want DDR4, HDMI2 and dual independent SSDs in case you’re listening Timmy.
Right, I want DDR4, HDMI2 and dual independent SSDs in case you’re listening Timmy.
I really hope that there will be a happy end to this, and the new iteration of nMP will be up to our wishes.
You may be right. I admit it, I did not look at what exactly Computerbase.de wrote, and they plainly said that Broadwell-EP is due for Q1 2016. wccftech must've took the word about RAM that is due for Q4 2015.
Of course, the current nMP is still a great machine but the people in this thread are wondering and speculating about its 2nd edition.I'm going to hold you to that when I can't get my 22-core Mac Pro 7,1 for 4 months after announcement!
Perfectly reasoned and explained as always.
So, high probability of a Broadwell-EP / Fiji / TB3 nnMP in Q1 2016, not before. And no sneak peak in October.
I'm just a bit depressed now: koyoot gets my hopes up and then you come along with vulcan logic and demolish them! Next you'll be proving there is no Santa Claus...
brad, don't get your hopes up. 22 core seems unlikely for Apple although, if the CPU is still user replaceable, you can change it later I guess.
I'm not sure it will be a quiet upgrade. They won't make a big fuss out of it but I'd say at least some form of announcement with focus on TB3 and maybe DDR4 and faster SSDs as major updates. They will have to cash on that for sure.
If a new TBD comes out too (and let's hope it does), then for sure they'll have to show it off.
1Q'16 seems reasonable considering all the (expected) parts should be available by then. Only drawback would be that newer GPUs will be available in a couple of months (not from now, from release date) and that will make the included GPUs seem even more dated, depending on which will be used of course.
Maybe Santa will bring good news at least by Xmas
I'm not sure it will be a quiet upgrade. They won't make a big fuss out of it but I'd say at least some form of announcement with focus on TB3 and maybe DDR4 and faster SSDs as major updates. They will have to cash on that for sure.
If a new TBD comes out too (and let's hope it does), then for sure they'll have to show it off.
1Q'16 seems reasonable considering all the (expected) parts should be available by then. Only drawback would be that newer GPUs will be available in a couple of months (not from now, from release date) and that will make the included GPUs seem even more dated, depending on which will be used of course.
Maybe Santa will bring good news at least by Xmas
Check out this Intel roadmap from Cisco:
http://www.cisco.com/assets/global/AZ/events/2015/forum-baku/pdf/Cisco-15H1_SRV_5.pdf
Cool info there.
If you know Russian...
http://www.cisco.com/assets/global/AZ/events/2015/forum-baku/pdf/Cisco-15H1_SRV_5.pdf
Cool info there.
If you know Russian...
koyoot, there was another leaked Intel roadmap in a Polish conference, but was removed:
http://konferencjakdm.pcss.pl/2015/...ey–kolejna_generacja_platformy_serwerowej.pdf
Lots of details regarding Purley apparently, but Intel must have removed it.
:-(
http://konferencjakdm.pcss.pl/2015/...ey–kolejna_generacja_platformy_serwerowej.pdf
Lots of details regarding Purley apparently, but Intel must have removed it.
:-(
May be we are too technical here.
Since Apple won't upgrade the Mac Pro every year, may be they are simply waiting until next year to release the next gen MP because they want to celebrate the 10 years birth of Mac Pro line.
Nothing to do with Intel, nothing to do with the GPU availability, and nothing to do with any computer technology
Since Apple won't upgrade the Mac Pro every year, may be they are simply waiting until next year to release the next gen MP because they want to celebrate the 10 years birth of Mac Pro line.
Nothing to do with Intel, nothing to do with the GPU availability, and nothing to do with any computer technology
If they update the iMac in the fall I may consider just getting a decked out version of that. Or a maxed out i7 windows 10 machine. Really don't want to go windows. but I have been pricing them out since the options here for a powerful machine are limited.
I can't take this early2008 mac pro any longer. New audio plugins and software bring it to its knees.
I can't take this early2008 mac pro any longer. New audio plugins and software bring it to its knees.
They can always release and additional model, more cMP like, to remind how it was, for greater celebrations.May be we are too technical here.
Since Apple won't upgrade the Mac Pro every year, may be they are simply waiting until next year to release the next gen MP because they want to celebrate the 10 years birth of Mac Pro line.
Nothing to do with Intel, nothing to do with the GPU availability, and nothing to do with any computer technology
I made a thread about the upcoming products...like why can't apple just give use general information about upcoming products...who cares about secrecy...mac pro was one of things i'm curious about.
This. Also, there is reason to believe that the tech industry is colluding with the major studios and labels to shut out project studio owners from stepping into their quality spec. I'm glad I got my Mini, but I'm kicking myself for not jumping at the refurb hex-core cMP. That ECC memory -does- make difference when running Pro Tools.
Update: I almost pulled the trigger on one of these: https://www.blackmagicdesign.com/products/blackmagicmultidock but I thought, "What if it doesn't work with whatever USB-C cable/adapter comes our way?" And so that little bit of commerce came to a halt.
Last edited:
So, in essence you're saying that "Steve Jobs would never allow this to happen." ? -Sorry, I couldn't resist.
Apple might go for the solution of having one of the GPUs (whichever gets less performance penalty) in x8 while the other will keep the x16 link, having another x16 for the 3 TB3 controllers and the SSD.
Would be a compromise solution, not ideal of course.
I'd rather have full x16 on both GPUs I guess, it might come useful in the future.
Those 48 lanes in Skylake server parts would be sweet here though, no compromise.
Actually, for me only one TB3 controller and a large SSD would suffice, so the 40 lanes would be enough, but others will need more.
Would be a compromise solution, not ideal of course.
I'd rather have full x16 on both GPUs I guess, it might come useful in the future.
Those 48 lanes in Skylake server parts would be sweet here though, no compromise.
Actually, for me only one TB3 controller and a large SSD would suffice, so the 40 lanes would be enough, but others will need more.
- Status