Ivy Bridge E alot like Sandy Bridge E

[deconstruct60]

Although it is doubtful Apple will use a -E processor in a Mac Pro, the -E models tend to very closely track what is now the Xeon E5 1600 series (formerly the Xeon 3x00 class series ). (Basically the same design with some features flipped on/off to differentiate the two on the market. )

The lasted leaked news is likely just primarily a clock speed bump with minor architecture improvements.

http://www.cpu-world.com/news_2012/2012110102_Features_of_Ivy_Bridge-E_processors.html

Number of cores -- the same ( Yes, there will be 4 core workstations. )

L3 cache -- the same ( side effect of number of cores since cache/cores come in same layer cake increments. )

Technologies -- the same

Instructions -- mostly the same some minor short float bump.

Memory -- mostly the same ( clock bump just like the cores )

PCI-e -- different ( but Xeon E5 are already at PCI-e v3.0, nothing new here for Mac Pro )

TDP -- the same


So all the hand waving that Apple is waiting on Ivy Bridge Xeon E5 for new Mac Pro because it is going to be revolutionarily different ....... not evidenced at all.

A Sandy Bridge Xeon E5 Mac Pro in 2013 is going to perform alot like a Ivy Bridge Xeon E5 Mac Pro in 2013-2014. Incrementally faster, but not a large jump. Especially the single package models. The only large gap will be in the mid-to-upper end of the dual package models and only on software that scales linearly.

There is a chance the Xeon E5 1600 line might top out at 8 instead of 6 but it doesn't look likely at the price points that Apple would likely use (given Intel's track record on pricing).

 

[xgman]

The big question still remains when?

 

[deconstruct60]

The big question still remains when?

Answered in other threads previously. But Q3-Q4 2013

http://www.engadget.com/2012/10/17/intel-roadmap-reveals-10-core-xeon-e5-2600-v2-cpu/

The significant part suggested by graph that it appears Intel will slow roll the E5 1600 rollout. Presumably so that can devote the runs for that variaton on the die-mask to i7 -E parts. I imagine the core i7 Extreme fans will be howling at the moon at that point due to the mainstream Haswell options having already been deployed and systems on the market by start of Q3.



So the full product mix that Apple would need for Mac Pro wouldn't be available at the start of the E5 1600's delayed window, but more closer to start of Q4 than end of Q2.

 

[seanm9]

A Sandy Bridge Xeon E5 Mac Pro in 2013 is going to perform alot like a Ivy Bridge Xeon E5 Mac Pro in 2013-2014.

The real question is why release a SB Mac Pro, when IB will be available when they are about ready to upgrade the MP....

 

[Wild-Bill]

The real question is why release a SB Mac Pro, when IB will be available when they are about ready to upgrade the MP....

Wouldn't be the first time Apple shoved obsolete parts into a Mac Pro box and tried to pass it off as "New".

 

[deconstruct60]

The real question is why release a SB Mac Pro, when IB will be available when they are about ready to upgrade the MP....

Because they could release a new Mac Pro up to 6 months earlier than if they wait for IB. The motherboard for the SB and IB Mac Pro is likely going to be exactly the same. The current motherboard in the current Mac Pro was probably designed in Q2-Q4 2008. The real question is why wait and continue sell the 2012 update for at least another 12 months that very limited update.

The 2012 update was to buy Apple more time. If it was "buying time to Ivy Bridge" they'd be trying to buy more than another year on an already an approximately 4 year old design effort. That doesn't seem like a particularly insightful strategy.

Alternatively, that could have been a move to buy time to Q1 2013. If Apple decided to upgrade the Mac Pro in late 2011/early 2012 they'd just need to a placeholder for the rest of 2012.

There are Ivy Bridge engineering samples out now. Apple could do a quick validation that the board is fundamentally OK with those now (Q4 2012) and release a SB model in Q1 or very early Q2 2013. That would be followed up with an update either late Q4 2013 or early Q1 2014.

Sounds un-Apple-like? Let's see iPad 3rd generation late Q1 2012 ... iPad 4th generation Q4 2012. Take basically the same model and bump the CPU package.

There is zero strategic advantage in waiting longer than necessary to upgrade the Mac Pro. Holding off on purpose for IB doesn't buy anything good.

P.S. the 2006 & 2007 Mac Pro are very likely going on the Obsolete Systems list in Q1 2013. The 2008 is rather long in tooth also even though it has a bit longer into early 2014 till it joins them.

 

[seanm9]

We aren't talking about an iToy, we are talking about a real computer... that averages over 400 days between refreshes... and by this sites buyers guide a machine that is about 145 days old (or about 830 days if you don't count the june update) and the last 2 refreshes were 511 and 685 days... its not a stretch to think the current version goes over 400 days to the ivies are ready... arent the sandies out... why wait on a launch... of course the june 2012 release was a stop gap to get you to ivy.

 

[deconstruct60]

We aren't talking about an iToy, we are talking about a real computer... that averages over 400 days between refreshes...

If, for example, Apple released a SB Mac Pro on Feb 5, 2013 and an IB Mac Pro on Feb 18, 2014 then the refresh period would be 378 days. That pretty much in the ballpark of 400 days. Quibbling over 22 days (or some +/- 1-2 month delta) is just noise.

The hand waving about 'iToy' is just pure misdirection.

... of course the june 2012 release was a stop gap to get you to ivy.

Not necessarily.

 

[xgman]

Answered in other threads previously. But Q3-Q4 2013

http://www.engadget.com/2012/10/17/intel-roadmap-reveals-10-core-xeon-e5-2600-v2-cpu/

The significant part suggested by graph that it appears Intel will slow roll the E5 1600 rollout. Presumably so that can devote the runs for that variaton on the die-mask to i7 -E parts. I imagine the core i7 Extreme fans will be howling at the moon at that point due to the mainstream Haswell options having already been deployed and systems on the market by start of Q3.



So the full product mix that Apple would need for Mac Pro wouldn't be available at the start of the E5 1600's delayed window, but more closer to start of Q4 than end of Q2.

Wow, that's really late in the year. So what, that would be 3-4 years between MP updates? Almost pointless to continue a line of computers with that sort of schedule. Letting myself get worked up all over again...)

 

[wallysb01]

We aren't talking about an iToy, we are talking about a real computer... that averages over 400 days between refreshes... and by this sites buyers guide a machine that is about 145 days old (or about 830 days if you don't count the june update) and the last 2 refreshes were 511 and 685 days... its not a stretch to think the current version goes over 400 days to the ivies are ready... arent the sandies out... why wait on a launch... of course the june 2012 release was a stop gap to get you to ivy.

Think of this from making money perspective. Why would you skip an opportunity to call something "new" by just dropping in a new processor? That's what they tried to do in June, but failed because we all knew the upgrade was mostly in name only. But with a SB to IB update, it will be a substantial upgrade. It might be light relative to other architecture changes, but it will be more than a slight spec bump in the same architecture, and it may come with core count increases depending on the pricing structure Apple uses for SB and IB. Particularly, the mid range CPU Apple uses may go from 6-core (likely the 2640) to an 8-core with SB to IB, since the general trend is to keep offering more cores for less.

So if Apple is actually going to continue the Mac Pro, there just isn't any reason to skip SB and wait another 9 months or more. You're putting way too much stock in these time between updates. The range has been from 240 to 685 with 5 data points. Its meaningless. Apple does what it feels will make the most money, that's it. They aren't waiting a specific number of days.

 

[seanm9]

If, for example, Apple released a SB Mac Pro on Feb 5, 2013 and an IB Mac Pro on Feb 18, 2014 then the refresh period would be 378 days. That pretty much in the ballpark of 400 days. Quibbling over 22 days (or some +/- 1-2 month delta) is just noise.

The hand waving about 'iToy' is just pure misdirection.



Not necessarily.

I agree that those dates would fit into the averages... but What Tim Cook said was "Later in 2013" which to me means Q3/Q4... http://articles.latimes.com/2012/jun/12/business/la-fi-tn-apple-tim-cook-email-mac-pro-20120612 ... Later is such an ambigious term tho so it could just mean any time in 2013...

which is also the time the IB's will be ready... it is possible that apple in skipping the SB Mac Pro, may have a deal worked out with Intel where they have ordered most of the early IB CPU's for use in the Mac Pro... didn't they have early access to the first round of Xeons used in the MP... and have the Mac Pro out for a few months before Dell, HP and the Rest could even get those CPU's....

that part is the fuzzy part of my memory... but I hope you're right... I need a new Mac and I dont think a fully loaded iMac is the one for me...

 

[xgman]

.. Later is such an ambigious term tho so it could just mean any time in 2013...
.

Exactly. Like "later" (not this year but in 2013 at any point) or "later in 2013" (as in late 2013). I'm still not entirely convinced it will come at all.

----------

.. I need a new Mac and I dont think a fully loaded iMac is the one for me...

At some point we might not have many other choices. There is only so much "waiting" we can do.

 

[Umbongo]

didn't they have early access to the first round of Xeons used in the MP... and have the Mac Pro out for a few months before Dell, HP and the Rest could even get those CPU's....

They have had Xeons a few weeks before others in the past. 2007, 2008 and 2009. Month or so later in 2006 and almost 6 months later in 2010.

I agree on your view about Cook's statement, but things change. I don't think they were waiting for IB. I think they were jsut undecided what to do with the Mac Pro. We heard rumours of internal conflict over it, which are probably the same sort of thing we see on here. Some understand the need for it and don't want to let of the pro-market and others are highly consumer/margin orientated. What we do know is we can't rely on Apple for high-end hardware and any professional user should take that in to account when they purchase.

 

[xgman]

I don't think they were waiting for IB. I think they were just undecided what to do with the Mac Pro. What we do know is we can't rely on Apple for high-end hardware and any professional user should take that in to account when they purchase.

same . . . . I'm trying very hard to get my head into a new mindset about what exactly I need and can effectively work with. Some compromises will inevitably need to be made.

 

[deconstruct60]

What we do know is we can't rely on Apple for high-end hardware and any professional user should take that in to account when they purchase.

There is nothing Apple can really "say" at this point to rebuild trust. They have to 'walk the walk' rather than just 'talking the talking'.

If they drop three successive Mac Pros, Q1 2013 , Q1 2014 , Q1 2015. And if Intel can release the Haswell Xeon E5 in Q3-Q4 2014 then that is a relatively low risk path back to redemption.

Well, that's if the customers show up in significant numbers. The faction that kills off products with declining sales is still there. They will regain the upper hand if by mid Ivy Bridge era of a revised product flounders. If the numbers sold are still going down or are stagnant there probably won't be another appeal granted from "death row".

Restarting on Ivy Bridge is much higher risk because the gap between the "placeholder" and a solution is around a year. That's another whole year of evaporating trust. The other significant risk factor is if Haswell Xeon hiccups and slides out. Then they are stuck with perhaps another irregular, unpredictable delay in product update. Again they would have dug the hole even deeper or at best made no progress.

The other factor is that if OS X is going to update every July-August then "around 6 months later" is an extremely likely far more stable place to be in terms of OS X. Shipping Mac Pros with a more rock solid version of the OS will also build trust faster. The primarily market are people using these machines in "production" settings. Very few are going to deploy an "unknown" OS into that kind of context. Typically they wait until the smoke clears and initial adopters take the hits before adopting.

The Mac Pro doesn't need some WWDC or other Apple large "dog and pony" show. It needs to consistently and doggedly make progress. It needs to take on what Apple labels 'hobby' status. It does well enough to keep doing it but it isn't a break-out, strategic product.

 

[deconstruct60]

Later is such an ambigious term tho so it could just mean any time in 2013...

'Later' is not an ambiguous term. It is a relative term. Just like smaller, taller, greater , earlier , faster , slower , etc. Or just like any pronoun (his , her, etc. ). Taking the phrase out of context is what renders it ambiguous if you remove what it is in relation to.

So " ... don't have anything today, but ... later in 2013 " is not ambiguous in the slightest bit. 'Later' is relative to 'today'. Similarly if the questioner asks about the lack of updates in the June event and Cook says "later in 2013". The answer is relative to the time context set by the questioner.



Much of the FUD was created by people converting the 'later' into 'late'. Two different words with different semantics. The next biggest FUD contributor is taking the "later" of out context and implicitly attaching "January 2013" as an implicit relation frame of reference.

If Apple internally debated about the Mac Pro well into late Q1 and early Q2 2012 then they may show up in late Q3-Q4 2013. If starting from a complete dead stop it could take them that long. However, that wouldn't be a "IB is great" plan.

My suspicion is that the debate earlier than that: somewhere around Q4 2011 or in Q1 2012 and that Q1 2013 was about as early as they could get it out the door from a dead stop and not really sure around June 2012 how far into 2013 it would be. Show stopper issues pop up even for Apple occasionally ( the iMac slide into Nov-Dec. ).

may have a deal worked out with Intel where they have ordered most of the early IB CPU's for use in the Mac Pro... didn't they have early access to the first round of Xeons used in the MP... and have the Mac Pro out for a few months before Dell, HP and the Rest could even get those CPU's....

I don't think Apple had any special deals in those earlier context ( except for when Intel was doing the motherboard work. If validated early they can go early). Apple didn't have much inertia on x86 or relative size so they could just ship as product what the larger vendors were shipping as "system validation" units to key customers and building inventories for launch. They also were an Intel darling and could midly step on their toes at the time ( ship before Intel's release event.)

At this point things are different. AMD is behind, not ahead with more scalable computing with a fused memory controller and core architecture. Apple is selling vastly more non Intel personal computers than Intel based ones. Apple is just another Intel system vendor now and a minor CPU package competitive developer now. A good customer, but they aren't going to sit by and idling let Apple step on their events.

The "early" folks getting Ivy Bridge are going to be the supercomputer vendors. They got Sandy Bridge Xeon E5's about 3-5 months before everyone else back in Nov-Dec 2011 and that will likely be the case this time around too. The top500 supercomputer list comes out in June and November each year. IB Xeons arriving in July 2013 are late to influence the June 2013 results and marginally in time to influence the November numbers if have to reconfigure deployed racks (with limited periodic downtime to incrementally do it.)

If trying to make a statement about Xeon performance that is a much better customer segment to get the product into early.

 

[seanm9]

If I am reading all the websites correctly weren't suitable Sandy based Xeons released in march of this year... I mean if you go from the westmere-EP e5645 in the current 12 core isn't the successor a sandybridge-ep like the e5-2630...or am i missing something?

 

[goMac]

The real question is why release a SB Mac Pro, when IB will be available when they are about ready to upgrade the MP....

This makes the assumptions that:
1) The holdup is IB.
2) Apple isn't going to use SB earlier next year.

There isn't enough information to put together a complete picture yet. Not saying we won't see IB in Mac Pros, but Apple could be waiting/working on something else and IB simply happens to track around the same time.

 

[monokakata]

I can't resist calling attention to the difference a comma makes, between:

"later in 2013"

and

"later, in 2013"

Was Cook's statement written or oral? I can't remember.