Sidebar Sidebar
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

Xeon Sandy Bridge Delay on top of Delay

xgman

xgman

macrumors 603
Original poster
Aug 6, 2007
5,705
1,434
according to cpu world:

It was announced a week ago that Intel had discovered a bug in the Sandy Bridge-E processors, which made us worry about a possible delay for the Xeon Sandy Bridge processors, which are very close in their conception.
This delay seems to be confirmed according to CPU-World. The new Xeon E5 processors, absolutely necessary to a new model of Mac Pro, which were scheduled to ship during Q3 2011 will now ship in Q1 2012.

At that moment, Apple will have everything, including the X79 controller, to make an entirely new Mac Pro. There will be a lot of choice to make a low end model with a quad-core clocked at 3.6 GHz and a high end one with 2x 8 cores clocked at 2.9 GHz. The only missing type will be 6 cores processors, which will ship even later.


Sounds like anyone hoping for a new 6 core box will have to wait a long time. I would assume that the difference between the speed of a 16 core 2.9 to a 4 core 3.6 would only be realized on high intensity cpu calls?
 
Some apps like Cinema 4d actually scale better with access to more cores.

There was a real world rendering test done in cinema 4d on a 8 core (16virtual) vs 12 core(24virtual). The 12 core was sometimes upto 700% faster at completing the same render.
 
True, but I wonder about stuff like Logic, photoshop, the OS itself, moving lots of data. 2.9 to 3.6 is a pretty big jump in raw power,
 
xgman said:
True, but I wonder about stuff like Logic, photoshop, the OS itself, moving lots of data. 2.9 to 3.6 is a pretty big jump in raw power,
Click to expand...

Logic uses all cores and scales very well. Photoshop does not use many cores and still likes clock speed over "more cores". It all depends. The OS likes more cores but really, the kernel runs fine on Core 2 Duo.
Both processors 8-core and 4-core will no doubt have similar top turbo speeds @ around 3.9GHz. So iTunes will rip the same, Photoshop will be a toss up depending on what exactly you are doing with it (plug ins, image size, etc) and logic will be almost 70% faster on the 8-core.
 
xgman said:
according to cpu world:
The only missing type will be 6 cores processors, which will ship even later.
Click to expand...

The new Pros are already being delayed, the last thing we want is Apple to delay release even more just because there are no 6 core machines available.
 
CPU is probably the main thing I look for. I use my 12-core MP for Maya and I love the CPU power, especially coming from a Core 2 Duo MBP! Of course, render times could still be a little faster ;)
 
Does a longer delay on Sandy Bridge mean a shorter wait till Ivy Bridge afterwards? :D
 
I guess if there is any good news in this it is that we now don't have to spend any upgrade money for another 5-6 months. I wish Intel could get all the cpus for each base platform out simultaneously, both consumer and pro/server models. They seem to lag worse and worse every year on the Xeon side. At this point I would almost prefer Apple made a high end non-Xeon based Mac Pro just to keep pace with other parts of the technology. Either way, it's frustrating.
 
JesterJJZ said:
Does a longer delay on Sandy Bridge mean a shorter wait till Ivy Bridge afterwards? :D
Click to expand...
You can dream I guess... :eek: :D :p

Blue Sun said:
So, after all this, when can we realistically expect the new Mac Pro?

January, February, March or April?
Click to expand...
I'd expect the announcement in Jan - Feb, and hopefully, systems will ship in March or April (could actually go as long as June at this point before LGA2011 Xeons reach consumers however :().

Apparently, Intel has a solution to the VT-d bug already, but they've not implemented it in the fabrication process (change the steppings from C1 to C2, which contains the fix). I suspect they're madly testing it to verify that in fact C2 does fix the issue before changing the production line with the new masks, and it's possible other bugs (aka errata) have surfaced, but not been made public (this is what makes me uncertain about the delay being only an additional month or two).
 
Zwhaler said:
Link in case anyone is interested.
Click to expand...

That's kind of a lame QA testing process where a server processor gets to C1 stepping before someone figures out that the virtualization is broken.

nanofrog said:
(could actually go as long as June at this point before LGA2011 Xeons reach consumers however ).
Click to expand...

Doubtful since the Ivy Bridge versions of the E3-1200 series is due in March-April.

http://www.cpu-world.com/news_2011/2011101001_Launch_schedule_of_Intel_Xeon_E5_microprocessors.html

Given the PCI-e v3.0 is a bit behind the curve it is OK slide to Jan-Feb. , but by April going Intel would have a riot on their hands. The time-to-market would contract with a late delay ( since all of the other components are likely finished.) in the CPU going live. There is nothing else to delay for... and probably buckets of C1 samples to shake out manufacturing processes with.
 
deconstruct60 said:
That's kind of a lame QA testing process where a server processor gets to C1 stepping before someone figures out that the virtualization is broken.
Click to expand...
I agree, especially considering how important virtualization is in the server market.

The technical issues due to ever increasing complexity are difficult enough, but if there are managerial issues as well, I fear that this may become a recurring theme rather than an occasional hiccup.

In any case, it doesn't exactly inspire confidence. :(

deconstruct60 said:
Doubtful since the Ivy Bridge versions of the E3-1200 series is due in March-April.

http://www.cpu-world.com/news_2011/2011101001_Launch_schedule_of_Intel_Xeon_E5_microprocessors.html
Click to expand...
E3 is on a different socket though.

What I can't see happening, is Intel releasing the Ivy Bridge variants of LGA2011 in under a year. Not enough money in it for them.

deconstruct60 said:
Given the PCI-e v3.0 is a bit behind the curve it is OK slide to Jan-Feb. , but by April going Intel would have a riot on their hands. The time-to-market would contract with a late delay ( since all of the other components are likely finished.) in the CPU going live. There is nothing else to delay for... and probably buckets of C1 samples to shake out manufacturing processes with.
Click to expand...
If the delay allows for PCIe 3.0 validation on the initial release date, that would be a good thing, as it would help generate initial sales (vs. others "fence sitting" as long as possible to see if it shakes out or not, just in case there are technical issues that ultimately prevent that certification from being issued).

As per a riot, perhaps. I see more contractual issues for Intel however, as it's late enough in the process that delivery dates have likely been inked based on previous release estimates, so they could incur penalties for late delivery of initial shipments.

But virtualization issues do need to be sorted before they ship any final parts, as it's critical to servers (I see this as more critical than PCIe 3.0).
 
I stopped waiting for SB two months ago and I'm glad I did. I'm happy with my '09 which will surely continue to serve me well until IB is here. Looks like that is gonna be a much bigger leap (22nm, tri-gate) anyway.

If Apple hasn't EOL'd the MacPro by then, that is.
 
ugh this money is burning a hole in my pocket. I need a new mac and an imac with laptop parts isnt really an option.
 
Here is some stuff from the earnings announcement

"The fact that PC chip sales held up meant that Intel did not have to rush to get its "Sandy Bridge-EP" Xeon E5 server processors to market and could instead just ship the ones it could make to HPC and Internet data center customers during Q3."


"Otellini reiterated that the Xeon E5 processors and their related "Romley" platforms would launch in early 2012, but provided no additional insight on when this might be. He made the same comparison to the "Nehalem-EP" Xeon 5500 ramp of three years ago that Intel's top brass was chanting in unison at IDF in September: that the Xeon E5s have 400 design wins already, twice that of the Xeon 5500s at the same point in their launch schedules, and that the company was expecting for a factor of 20 more Xeon E5s to ship in the wake of the launch as the Xeon 5500s did at the bottom of the Great Recession in early 2009."
 
I hate the fact that the Mac Pro is always waiting for Intel's Xeons. I wish there was a alternative way to get the same performance and reliability. Personally I'd sacrifice the ECC ram, server based boards if I could get a 6 core sandy, lots of ram and everything else equal.
 
xgman said:
I hate the fact that the Mac Pro is always waiting for Intel's Xeons. I wish there was a alternative way to get the same performance and reliability. Personally I'd sacrifice the ECC ram, server based boards if I could get a 6 core sandy, lots of ram and everything else equal.
Click to expand...

"always waiting" is a bit harsh. This is the first time the high-end has come last since the Intel switch.
 
Umbongo said:
"always waiting" is a bit harsh. This is the first time the high-end has come last since the Intel switch.
Click to expand...

Weren't the 2010 Imacs out many months before the 2010 MP's? I recall being frustrated then about the same thing.
 
nanofrog said:
There are serious implications here, as Intel could easily become complacent again... :(
Click to expand...

Exactly. Competition breeds innovation and excellence. If AMD doesn't have a good high end Desktop/Server chip Intel can take it easy on that front and work at catching up with AMD's Fusion APUs.
 
You must log in or register to reply here.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Top Bottom
Back