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.
Intel Rumored to Debut Basin Falls Platform in May, Launch Coffee Lake Chips in August
- Thread starter MacRumors
- Start date
- Sort by reaction score
The reason why people are talking about them isn't because they're anything special, but because Intel moved up their launch by several months just like that. Goes to show that Intel really was dragging their feet with AMD not being competitive and now that they are competitive again, we may very well see some real progress in the CPU arena again.
Many iOS developers have put off buying new Macs because of the stagnation (me included). The true cost to iOS will be felt soon once new machines become necessary for those developers.
Nothing has been done while it was not very noticeable on sales charts, but it will be too late when it can be seen.
I would say a minor spec bump might be possible for around June. Most likely we'll have to wait for the Kaby Lake refresh sometime in Q4 '17.
Higher-end doesn't mean Skylake-X chips, it can also mean Xeons - remember, the iMac is not a gaming machines and there's really no point in going with very expensive X-class 6-8-10 core chips. Also, massive heat issues and power requirements in an iMac, even a thicker one? Not happening.
I don't follow. Comparable Xeons are neither cheaper nor lower-watt than Core-X chips. They'd only differ in allowing ECC memory, and be more expensive at that.
They don't need X-class, but may still pick up core count a bit without changing thermal power limits (TDP).
With desktop versions of gen 8 (CoffeeLake), mainstream i7 versions are suppose to pick up 6 cores. That appears to be a bit of Intel swapping integrated graphics processor (iGPU) space on the die for more x86 core space. ( they will only be GT2 graphics). That actually makes sense in the context that Ryzen 5 and Ryzen 7 are devoting no die space to iGPU at all. ( AMD integrated options are coming later in the year. Not sure Intel has an answer for those any time in next 9-10 months. )
The mainstream gen 8 should have an Xeon E3 equivalent that should also pick up the 6 core. E3 12xx v7 series. If Intel rapidly moved up the time table they may be able to shrink the that schedule also. So if the desktop moved approximately 5 moths back ( January '18 to August '17) then perhaps the Xeon E3 v7 moved back the same amount to land in October '17 ( March '18 to October '17). Xeon processors though usually get a higher amount of beta testing and defect quality control so Intel may not be able to claw back almost half a year.
E3 v6 would be a safer choice for Apple if shooting for a fixed timeline. if the E3 v6 to v7 timeline has seen a major disruption then Apple's E3 plans are probably screwed up from the original projected timeline. November-December has a decent chance to being more realistic if this upper edge shift to E3 is true. ( Not sure it is when the mainstream top end is shifting to 6 core. Apple could sell just 6 core sizzle as the hype tag line. But yeah a six core E3 v7 with 32-64GB of ECC ram would + an 5K display for less than most Mac Pro prices would draw in a decent number of "pro" market customers at the lower end of the Mac Pro market. ).
There is a new PCH chipset with gen 8 ( 300 series) which technically should not throw Apple's work in a loop if they were already deep into design using v6 (and/or Gen 7 KabyLabe ). Likewise the E3 v7 has similar chipset bump but Apple could just skip some of the more advanced features if caught off guard.
All of that said Intel knocking 5 months off the release timeline. That's is kind of dubious. The approximately 3 months ( a Quarter ) for Skylake-X is far more tractable. Intel probably had some slack in the schedule to get around any last minute significant bugs that might pop up. 5 months is suggestive of throwing something out there that isn't fully baked and vetted. Yes gen 8 is just an incremental change, but there is decent amount of new stuff in the chipset ( bump to USB 3.1 gen 2 among a few other things.)
[doublepost=1492793284][/doublepost]
Skylake-X <==> Xeon E5 1xxxx v5 ~ 140-160W TDP
mainstream KabyLake <===> Xeon E3 12xx v6 ~ 70-80W ( max out four cores )
mainstream CoffeeLake <===> Xeon E3 12xx v7 ~ 70-85W ( max out at 6 cores )
Xeon doesn't necessary mean the E5 class. There are four ranges of Xeon D , E3 , E5 , and E7. Not being limited to E5 leaves plenty of other options both up and down the TDP range.
The iMac is likely to select from the last two product equivalency classes above. Even if the enclosure is modified to get a bigger TDP envelope that would probably get consumed by a more "desktop" like GPU than in pushing on the CPU core count front.
What is awkward for the 21.5" iMac is the Intel 'retreat' on iGPUs with CoffeeLake. Apple would probably have to shift to dGPUs there too.
I'm in the same boat. I have two riMacs and two cMac Pros. I also have a wad of cash set aside for a new machine.
But I never find myself saying, "I could be more productive if I just had X."
I think Intel is to blame, or more correctly, the laws of physics are to blame.
All the improvements recently in CPUs have been around efficiency, which makes sense, because what are you going to do with a massively faster chip?
Granted, there are a very few professionals out there that need it, but to risk using the dreaded car analogies, no one needs a 1000 HP engine in a car that can only be driven on city streets and a freeway with a speed limit, which is where 99.99% of us drive.
Wait whaaat? They're idiots because they make their computers thinner? I bet 99% of iMac users 1) do nothing that generate heat, and 2) want a thin sexy computer.
Given the memory sizes people are talking about wanting (32 Gb+) I'd think ECC would be desirable.
Sales spiel that says nothing and promises nothing. Until it's on the market it's all hot air and vapourware.
Your are thinking of Xeon E5, Xeon E3 are both cheaper and lower wattage than Core-X chips. They are essentially Core-i7 with ECC ECC Memory.
The question is for the iMac, what will this unannounced Xeon E3 1285 chip be. Based on the numbering convention it looks like a very similar chip to what is currently being used but with ECC RAM.
Hopefully Apple offers an iMac configuration that goes beyond that. My guess is the iMac will ship near the end of the year with the highest configuration shipping with a ~4.0 GHz quad core Xeon by default, but can be BTO configured with up to an 8-core Coffe Lake chip.
Exactly. Except for a used late-2013 iMac for my son, the most Mac I have is a 2012 cMBP. They value isn't there to upgrade to a newer model, since mine has 16 gigs of RAM and lots of SSD storage.
I've written it elsewhere, but 6% of the Apple customers create the content for the other 94%. And that 6% has been pretty much ignored by Apple, except for one product line. And even that saw a ton of backlash this time around. Is Apple too late to respond? We'll see.
You seem to be restating my point. I wasn't talking about E5. And iMacs don't currently use -X; they use -S.
You omitted:
mainstream KabyLake <==> KabyLake-X (GPU disabled, higher TDP) (max out four cores)
Incorrect
The early 2011 MBP was one of the first laptops one could actually buy w/ Sandy Bridge (on sale 2/24/11).
And during a couple month period it was the fastest laptop to run windows on.
Going back to what I said late last year:
The iMac Pro will most likely feature a 10-core skylake-X as a custom upgrade option
A 10-core iMac Pro w/ better (vega) GPU options in late 2017 will satisfy a wide range of professionals.
And give Apple time to develop the mMP (EEC + Xeon multi-socket) for late-2018 to mid-2019.
A 4-core Kaby i5/i7 based iMac in 2017 will be fine (and cheaper) for standard users.
Last edited:
From a pure performance standpoint, those are good points. I'm sure Intel will retain their places in the mainstream and "pro" lines.
However, Apple is also a company that has, more so in recent times, shown itself to also be concerned with the bottom line.
This is a company that's still shipping 5400RPM laptop drives in premium $1000+ desktops in 2017(!?). Using lame ULV laptop processors in them well. Cutting back to the bare minimum solid-state storage in the 1TB Fusion configurations, even for the better models.
"Good enough" has entered their lexicon, and practice, if not for the high end, then certainly for their low end offerings.
If AMD can score a coup and get Apple as a CPU customer, they'd probably do what they could to make Apple happy...like Intel used to.
A lineup with a mix of Intel and AMD processors isn't so far fetched to me. And recent comments suggest that the product managers are working overtime to stratify their product segments further apart again.
That options is bizzaro weird. I've seen rumor reports on it. I just don't see where it would fit with any Mac product. Perhaps it is an assumption that X-Series is the same socket. If so that is an odd duck.
For a Mac Pro it seems to suck. The graphic I saw it is not only 4 cores but it backslides all the way back to 16 PCE-e lanes too. The memory channels drop out too. So it would be stuffed into a package with almost 1000 more pins and it doesn't incrementally nothing more than the original mainstream socket.
https://www.extremetech.com/computi...h-new-high-end-desktop-parts-earlier-expected ]
If the iGPU was stripped off and the Skylake-X uncore ( memory controller , PCIe lanes , etc. ) was bolted on I could see a point. At least then the layout from the memory slots and standard physical PCI-e slots would be the same. You'd pick up more memory bandwidth and PCI-e I/O to throw at faster clocked cores.
The only thing that makes sense would be those two are heading for two different boards. The "iGPU-less' versions are really coupled to the E3 12x0 v6 variants that also have the iGPU disabled, but with ECC off. I think that is quite confusing the weave that into Skylake-X messaging because it is a vastly different underlying die design and a different socket. in that case, it is the PCH chipset being shared across to socket designs. So the main take away of Basin Falls is shared chipset over sockets ?
If it is the same mainstream socket then it is more like
mainstream KabyLake <==> Xeon E3 12x5 v6
KabyLake-X (GPU disabled, higher TDP) <====> Xeon E3 12x0 v6 ( GPU disabled, not quite as high TDP )
KabyLake-X probably being a smaller binning to select out more easily over clocked parts.
I just want to add that naming the products/projects is actually a very serious business.
Quote: Intel has changed the name of his next Itanium processor from Tanglewood to Tukwila, in order to avoid a trademark dispute with the Tanglewood Music Center.The old name is associated with the famous Tanglewood Music Center in Massachusetts, said an Intel spokesman. Many of the engineers working on the project came to Intel from Massachusetts-based Digital Equipment and live and work in the state.
Tukwila is a town in Washington about 15 minutes away from Seattle. The mayor's name is Steven Mullet.
Intel typically selects geographic place names for code names to avoid copyright and trademark disputes. Many of the code names derive from towns in the Pacific Northwest, but other names are creeping in. The Pentium M chips carry Israeli code names such as Dothan and Banias because the chips are designed in Israel.
sadly I sit in the apparent 0.01% that migrated to Macs in the early 00's because it was *the* platform if you did video/creative work, only for it to become the platform for casual users and you should go back to a big desktop with neon lighting if you need rendering horsepower.
