Double the performance than similar chips from 5 years ago? Shouldn't it be doubling every 18 months sided with Moore's Law. I wonder if this is due to removing many prediction elements to the CPU. The decrease in performance is so bad after microcode updates Intel tried to prevent reviewers from benchmarking CPUs ( https://www.extremetech.com/computi...urity-patch-benchmark-ban-after-public-outcry )
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 Announces 8th-Generation Whiskey Lake and Amber Lake Processors Suitable for MacBook and MacBook Air
- Thread starter MacRumors
- Start date
- Sort by reaction score
Pair it with a Razer Core X eGPU and its perfect!
My biggest takeaway from this is the Whiskey Lake 15w variant coming with 16 PCIe lanes but this launch could have come too late for the 2018 Macs.
The i5-8265U would be great for any entry level MBA replacement if it wasn't for the previous stories of Apple getting a big discount on the Kaby Lake Refresh i5-8250U. The thing that the MBA has in its favour is lots of ports. Losing them for Thunderbolt ports (or USB-C ports) would be bad but only offering 2 would have been very bad.
As it was, the MBA replacement could have gone for 4 Thunderbolt ports but 2 half speed like the 2017 13" MacBook Pro with Touchbar. If they are building it down to a cost Apple could go with 4 USB-C ports without any Thunderbolt capability on a 13" MacBook Product.
The i5-8200Y gets a small speed bump with no additional PCIe lanes (it only gets 10). There should be sufficient for 1 Thunderbolt port but Apple choose not to give the MacBook such a port - again there could be financial reasons as well as .
The i5-8265U would be great for any entry level MBA replacement if it wasn't for the previous stories of Apple getting a big discount on the Kaby Lake Refresh i5-8250U. The thing that the MBA has in its favour is lots of ports. Losing them for Thunderbolt ports (or USB-C ports) would be bad but only offering 2 would have been very bad.
As it was, the MBA replacement could have gone for 4 Thunderbolt ports but 2 half speed like the 2017 13" MacBook Pro with Touchbar. If they are building it down to a cost Apple could go with 4 USB-C ports without any Thunderbolt capability on a 13" MacBook Product.
The i5-8200Y gets a small speed bump with no additional PCIe lanes (it only gets 10). There should be sufficient for 1 Thunderbolt port but Apple choose not to give the MacBook such a port - again there could be financial reasons as well as .
So my guess is the MacBook Air gets replaced with a Whiskey Lake-U MacBook Escape for $1,099-ish. It'll have worse graphics than the current model, but twice the CPU cores. Possibly an extra-weird $999 model with 4 GB RAM. Leave the 256 GB SSD a $200 upgrade. Makes the graphics on those machines worse, but the CPU otherwise much better, and cuts overall pricing by $200. And call it the 13-inch MacBook.
Still leaves open what the heck they want to do with the 12-inch MacBook (which is basically a completely different product), though. It also leaves quite a gap from now $1,299 to the $1,799 13-inch MacBook Pro, but maybe that's OK.
Still leaves open what the heck they want to do with the 12-inch MacBook (which is basically a completely different product), though. It also leaves quite a gap from now $1,299 to the $1,799 13-inch MacBook Pro, but maybe that's OK.
So, Amber Lake Y for the 12" MacBook. Looks like it is still stuck with HD Graphics 615, which the rumors stated a few months earlier.
Whiskey Lake U for the 13" MacBook (Air), unless Intel has a special trick up its sleeve called the Core i5-8365U w/ Intel Iris Plus Graphics 645 and the Core i7-8665U w/ Intel Iris Plus Graphics 645 that will not be announced until the rumored October hardware event, which wouldn't surprise me in the least.
Also, buried in the Ark are two brand new Kaby Lake-G CPUs sporting Radeon Pro WX Vega M GL Graphics that might end up in the Mac mini successor/update.
Let's get on with it Intel. Next up, the Core 9000-Series and Core X-Series 22-core whatever (my guess, Core i9-7990XE Extreme Edition Processor).
Whiskey Lake U for the 13" MacBook (Air), unless Intel has a special trick up its sleeve called the Core i5-8365U w/ Intel Iris Plus Graphics 645 and the Core i7-8665U w/ Intel Iris Plus Graphics 645 that will not be announced until the rumored October hardware event, which wouldn't surprise me in the least.
Also, buried in the Ark are two brand new Kaby Lake-G CPUs sporting Radeon Pro WX Vega M GL Graphics that might end up in the Mac mini successor/update.
Let's get on with it Intel. Next up, the Core 9000-Series and Core X-Series 22-core whatever (my guess, Core i9-7990XE Extreme Edition Processor).
I'm tempted to bring back my 13" MBP while I still can just to see what they come up with. A quad-core 15W 13" with 2 TB3 no touch bar for a few hundred $ less than the MBP would be interesting.
Can someone please educate me on the die size issue Intel are having.
There MUST be more to it than just how easy one company can make one type of chip with one size of track and another company can't made a different chip with another track size.
By that I mean, ok. Everyone is saying how hopeless Intel are with their 14nm tech, whilst others are at 9 or 7 or looking to go even lower.
But it's can't just be that easy,
What I mean is. Say Samsung or TSM are doing 7nm chips.
Intel can't just send their cpu design over to either of these and say "hey guys, you can do 7nm, make us some please"
I presume they would struggle/fail just as much as Intel is if they tried it.
Can anyone educate me and everyone else here, by explaining this simply?
There MUST be more to it than just how easy one company can make one type of chip with one size of track and another company can't made a different chip with another track size.
By that I mean, ok. Everyone is saying how hopeless Intel are with their 14nm tech, whilst others are at 9 or 7 or looking to go even lower.
But it's can't just be that easy,
What I mean is. Say Samsung or TSM are doing 7nm chips.
Intel can't just send their cpu design over to either of these and say "hey guys, you can do 7nm, make us some please"
I presume they would struggle/fail just as much as Intel is if they tried it.
Can anyone educate me and everyone else here, by explaining this simply?
You don't understand: this replaces the CPUs in the 2012 MacMini.
So, it's a perfectly adequate comparison.
Doubt it. Rumors are it is going away. It's already outdated.
I thought Intel's recent microcode license agreement prohibited performance benchmarking? Oh right, Intel reversed course after Debian balked.
So are those performance comparisons with or without Spectre and Meltdown mitigations, and HT disabled?
Or should I wait another 5 years before all the vulnerabilities are fixed and performance recovered in a die change?
So are those performance comparisons with or without Spectre and Meltdown mitigations, and HT disabled?
Or should I wait another 5 years before all the vulnerabilities are fixed and performance recovered in a die change?
This is the best question I've seen in months. My very uneducated laymen's guess would be too much electron leakage due to lots of legacy archetecture that's so complex, they haven't figured a way to smooth it out enough to keep volts from jumping off the rails at the electrical pressure needed to just push everything along.
With billions of pathways, you can't just go in and redraw everything by hand, it's probably the most complicated maze in history.
And they don't have the luxury of restarting from scratch, there's probably a ton of legacy in there only a bunch of long retired guys barely even remember, that no one even knows how it works.
From the Whiskey Lake presentation deck -
"Improved Voice Services via dedicated quad-core audio DSP"
I wonder if this DSP could play any part in implementing "Hey Siri" on the 13" MacBook. Crazy, crazy theory...I know.
"Improved Voice Services via dedicated quad-core audio DSP"
I wonder if this DSP could play any part in implementing "Hey Siri" on the 13" MacBook. Crazy, crazy theory...I know.
On the 13" MacBook (Air)? Yes, I would think so, if they are going to actually offer the same level of value the current MacBook Air has with the Thunderbolt 2 port it currently sports. I would expect two TB3 ports and and an audio jack similar to the nTB 13" MBP but utilizing the wedge shape of the 13" MacBook Air, maybe and SD Card slot, but not much else.
Moore's Law doesn't hold up anymore, We have literally reached a point where we can't make CPU speeds any faster while reducing power consumption and heat generation. This is why we saw the move to multi-core CPUs. Many cores running at a slower speed increases performance than 1 core running at 4Ghz. There are other ways to increase performance by also optimizing the CPU architecture.
https://www.technologyreview.com/s/601441/moores-law-is-dead-now-what/
Just ordered a 13” today with 16GB RAM, delivery due 11th September... thinking the same thing before I’ve even received mine!
Edit: Cancelled the order, going to wait and see what Apple do in Sept/Oct.
Last edited:
.....
According to Intel, its new Whiskey Lake and Amber Lake chips "raise the bar for connectivity, performance, entertainment, and productivity." The U-series chips introduce support for integrated Gigabit Wi-Fi for up to 12-times faster connectivity speeds, support for USB 3.1 Gen 2 transfer speeds, and built-in support for voice services like Alexa and Cortana.
....
This diagram is a bit overstated. There is no "built in", full implementation of Thunderbolt in the chipset.
"...
Intel also puts Thunderbolt 3 on this diagram, but it isn’t native. Intel only stated that you still need the chip after they were explicitly asked about this – we’re waiting on Intel to integrate TB3 onto the chipset for a few generations now, and to put it in the diagram just because it can be connected is a little disingenuous. ..."
https://www.anandtech.com/show/13275/intel-launches-whiskey-lake-amber-lake
It is 'built-in' a bit like Ethernet is built in. You still needs another chip near the socket(s) to implement completely implement it. The Thunderbolt controller has placement requirements of being within around 2 inches of the port. For most designs you can't pragmatically 'drag" the CPU package to the edge of the device.
At some point Intel might split some of the Thunderbolt controller implementing. Part of it in the PCH and part near the outward facing port. The PCH part would deal more so has to do with blending USB/PCI-e data traffic from the distance from the PCH to the PHYS (Physical) switch. The other part more to do with Switching and merging of DisplayPort , Power , and other ALT-mode duties. The PHYS chip would get cheaper and smaller. I don't think they are there yet.
Pragmatically, I think they have just made it easier to dedicate x4 link to the TB controller by default. ( And it probably get some priority switch routing to the DMI link ) .
That poor lonely chip, just sitting out there like some dude in a life raft after his boat sank.
I love people like this... After years of process improvements as a means to an end, the market now sees the process as an end in itself. "I don't care if it's faster, cheaper and lower power, I won't be seen in public with gates that wide! It would be embarrassing!"
No one knows the real reason because Intel wouldn't share information like that. Here are some articles about why companies shrink dies and why it's beneficial.
https://www.geek.com/chips/dont-expect-10nm-intel-processors-anytime-soon-1627212/
https://www.tomshardware.com/news/intel-cpu-10nm-earnings-amd,36967.html
https://en.wikipedia.org/wiki/Die_shrink