Xeon coming to laptops (and Mac Mini?)

[paulrbeers]

You know that just enabling ECC firmware support and support for 64 GB (which is basically all these new Xeons are) is likely just a few feature tick-boxes intel need to enable right?

These CPUs may be called Xeon, but you're not going to be seeing other Xeon features like 20 cores and 20 MB of L2 cache here.

They will be largely the same as a mobile i7 with ECC support and a couple of minor feature cripple tweaks removed.

Intel has made low power Xeons for a few years now.


edit:
It would not surprise me in the slightest to see this Xeon as an option or maybe standard CPU in the 15" Macbook Pro. ECC is a "Pro" feature and for people who require ECC support, the current Macbook Pros (and any other portable) are immediately ruled out until now.

I don't think some people understand just how important ECC is for fields such as scientific research, etc. - even if the user doesn't need a million cores and terabytes of memory. Regular machine with no ECC = maybe your machine crashes, you get a wierd glitch in something, etc. Scientific research for example, a memory glitch can affect processing of experiment test results.

Yeah, but who's doing scientific research on a Mac mini or Macbook Pro? You want maximum power and/or maximum RAM. You aren't going to find 4+ slots for memory on a Mini or Macbook Pro anytime soon and you are going to be heat constrained on both platforms.

It isn't that people do not understand how necessary ECC memory is, it's that in a portable and/or Mini desktop, there are too many other constraints to be used in Scientific research. Great so you have ECC, but the whole thing throttles because of the heat generated or you have ECC memory, but you can't get one with more than 16GB of RAM (largest DD4 ECC SO-dimms are currently capped at 8GB) which throttles your work flow.

Just saying....

 

[throAU]

Yeah, but who's doing scientific research on a Mac mini or Macbook Pro? You want maximum power and/or maximum RAM. You aren't going to find 4+ slots for memory on a Mini or Macbook Pro anytime soon and you are going to be heat constrained on both platforms.

It isn't that people do not understand how necessary ECC memory is, it's that in a portable and/or Mini desktop, there are too many other constraints to be used in Scientific research. Great so you have ECC, but the whole thing throttles because of the heat generated or you have ECC memory, but you can't get one with more than 16GB of RAM (largest DD4 ECC SO-dimms are currently capped at 8GB) which throttles your work flow.

Just saying....

Not all scientific research needs massive CPU. It just needs results that aren't at risk of calculation or memory error.

It could be a bunch of numbers in excel....

 

[shaunp]

You know that just enabling ECC firmware support and support for 64 GB (which is basically all these new Xeons are) is likely just a few feature tick-boxes intel need to enable right?

These CPUs may be called Xeon, but you're not going to be seeing other Xeon features like 20 cores and 20 MB of L2 cache here.

They will be largely the same as a mobile i7 with ECC support and a couple of minor feature cripple tweaks removed.

Intel has made low power Xeons for a few years now.


edit:
It would not surprise me in the slightest to see this Xeon as an option or maybe standard CPU in the 15" Macbook Pro. ECC is a "Pro" feature and for people who require ECC support, the current Macbook Pros (and any other portable) are immediately ruled out until now.

I don't think some people understand just how important ECC is for fields such as scientific research, etc. - even if the user doesn't need a million cores and terabytes of memory. Regular machine with no ECC = maybe your machine crashes, you get a wierd glitch in something, etc. Scientific research for example, a memory glitch can affect processing of experiment test results.

I really don't see Apple putting ECC RAM into a laptop. Yes it's important for some applications, but lets face it Apple are pushing style over substance and even on their high-end laptops remove useful stuff like ethernet ports (not everywhere has WiFi and adaptors are crap) and there's no matte screen option. The problem I find with Apple kit these days is it's been made to appeal to the consumer market too much and the 'Pro' models don't really scale high enough. They are nice to use but it's easy to hit the limits and then I start looking back at PC workstations so I can get my work done properly.

 

[ixxx69]

Not all scientific research needs massive CPU. It just needs results that aren't at risk of calculation or memory error.

It could be a bunch of numbers in excel....

Regarding the Excel thing, that doesn't happen in real life. The benefit of ECC RAM in modern systems is uptime... servers that can't afford to crash because of a memory error, or workstations crunching numbers or recording data for days on end where a crash could mean losing days of processing time, or an interruption could mean a loss of irreplaceable data, etc.

The reality is those kind of tasks are not the MacBook's market. Practically speaking, those kind of tasks are almost exclusively reserved for servers and desktop workstations, and the small niche market for laptops doing those kind of tasks is going to be even smaller for OS X (most of that market is still Linux and Windows). Add to that, current non-ECC RAM is much more reliable than it used to be.

It's just not nearly a large enough market for Apple to spend the resources. For any given product, Apple's looking for high millions to tens of millions per year in sales. There's probably a 10 times larger audience for a 17" MacBook than there is for a Xeon MacBook, and obviously the 17" laptop wasn't selling well enough to keep its spot in Apple's lineup.

And finally, Apple can barely keep its attention on the Xeon-based Mac Pro which is a niche product (so forget about a Xeon Mini - that's just not happening), but at least it looks different from anything else they sell and offers a unique place in the Mac lineup. People will buy it even if they don't technically need a "Xeon" workstation. A Xeon MacBook would look just like any other MacBook and would function (to the user's eyes) just like any other MacBook... there's just not enough to distinguish it... no one's going to buy it unless the specifically need that functionality... there's no spill-over potential like with the Mac Pro.

That's all to say that it is very unlikely Apple would offer such a product. But while that's an educated guess, you never know with Apple. Intel makes a ton of money on Xeon (servers are the really big market), so Xeon is an important brand for them. And Apple draws a lot of attention to anything it touches. So I wouldn't say it's out of the realm of possibility either.

 

[Moonjumper]

I really don't see Apple putting ECC RAM into a laptop. Yes it's important for some applications, but lets face it Apple are pushing style over substance and even on their high-end laptops remove useful stuff like ethernet ports (not everywhere has WiFi and adaptors are crap) and there's no matte screen option. The problem I find with Apple kit these days is it's been made to appeal to the consumer market too much and the 'Pro' models don't really scale high enough. They are nice to use but it's easy to hit the limits and then I start looking back at PC workstations so I can get my work done properly.

I understand why they got rid of the ethernet port (because it was massive), but the adaptor is clunkier than it needs to be, and won't wake from sleep on my 15" rMBP. So I mostly agree there.

But they did push for things to improve their screens. I had an early unibody aluminium MacBook. The reflective screen was terrible. But they have done a lot of work to improve that. The current models are much better. Mine is even better in strong light than the external monitor attached to it that is matte and has an anti-reflective coating (HP ZR24w). So it is not always style over substance.

 

[throAU]

Regarding the Excel thing, that doesn't happen in real life.

Maybe you should tell CERN that, who did a study on data corruption and noted that actually yes it does happen in real life.

I'm not saying that Apple is going to put out a Xeon laptop (they might, but that wasn't the point I was arguing). Merely commenting that the mobile Xeons are being made for whoever because of ECC. Some workloads demand it. ECC is a Xeon only feature in intel land.

These mobile xeon are essentially just going to be i7s without the ECC and max RAM features being crippled like they normally are on non Xeon CPUs.

They're not going to be big iron 14 core Xeon E7s or anything that draws 160 watts or whatever. They'll be basically the same as a mobile i7.

 

[ixxx69]

Maybe you should tell CERN that, who did a study on data corruption and noted that actually yes it does happen in real life.

Did you read the CERN paper? No mention of Excel or spreadsheets. The "real life" expression was maybe a little flippant, but it was meant to suggest that this isn't anything anyone outside of the target markets needs to worry about. The point was that something like the CERN study looks at data corruption on a number of fronts when dealing with petabytes of data with an average sustained I/O rate of 800MB/s over several months. That slowly morphs into macrumor posts suggesting numbers in your Excel spreadsheet might start randomly changing unless you have ECC memory.

I wasn't trying to pick on you or anything, just offering some perspective on why it doesn't make much sense for Apple to pursue this in their MacBook lines.

 

[DesertSurfer]

I heard a rumor that Apple is putting eons and petabytes of storage in the new MacBooks.

 

[paulrbeers]

Did you read the CERN paper? No mention of Excel or spreadsheets. The "real life" expression was maybe a little flippant, but it was meant to suggest that this isn't anything anyone outside of the target markets needs to worry about. The point was that something like the CERN study looks at data corruption on a number of fronts when dealing with petabytes of data with an average sustained I/O rate of 800MB/s over several months. That slowly morphs into macrumor posts suggesting numbers in your Excel spreadsheet might start randomly changing unless you have ECC memory.

I wasn't trying to pick on you or anything, just offering some perspective on why it doesn't make much sense for Apple to pursue this in their MacBook lines.

This is exactly it. Several studies have been done over the last 5-7 year on how often memory errors occur.... A few notable items:

1. Most memory errors are due to a small percentage of bad memory (one study had it that 90% of memory error from google servers came from 8% of their memory sticks).
2. The older the memory, the more likely that there is an error (generally memory over 20 months old was more prone to throwing errors).
3. Most memory errors were hardware errors (see the above two issues if you want to cover yourself) and not due to environmental errors
4. To actually see a quantifiable amount of memory errors, you have to be running 24/7 with massive amounts of memory being used. If you are only running something for a half hour, the chances of errors (especially if mitigated by #1 and #2), are virtually nill.

So the end result is, unless you are running an excel spreadsheet that would take days to process (at which point, I would point out that there would be a million other ways to accomplish this faster outside of excel), then the end is moot. No one using excel to drive their calculations are going to run into a memory error.