Mac Pro 2011 "Sandy Bridge E" everything we know

[thekev]

This TechPowerUp article has some presentation slides with further details about SandyBridge Xeons and first performance claims ("up to 80% increase" compared to Westmere 3.46GHz, with Linpack value going as high as 2,2x compared to the Westmere baseline).

There are also informations about a new technology named Data Direct I/O Technology (DDIO) which bypasses main memory by sending data directly into the CPU cache (whatever that means in detail - slide is pure marketing so far). The presentation claims DDIO to increase I/O performance by up to 2.3 times that of Westmere, reducing latency and allowing system memory to remain in low power state. Probably needed with the highest E5-2600 SB-EP topping out at a whopping 150W TDP...

Intel is definitely serving kool aid.

I hate it when they use drastically inflated numbers. The gain doesn't look that amazing when comparing a cpu with its logical successor. This takes price and tdp under consideration. I'm not sure if the current design of the mac pro is really suitable for 150W cpus considering the desire to keep it silent. Other OEMs often pass on cpus that run this hot when it comes to workstations. I wonder how hot they run overclocked .

 

[goMac]

I hate it when they use drastically inflated numbers. The gain doesn't look that amazing when comparing a cpu with its logical successor. This takes price and tdp under consideration. I'm not sure if the current design of the mac pro is really suitable for 150W cpus considering the desire to keep it silent. Other OEMs often pass on cpus that run this hot when it comes to workstations. I wonder how hot they run overclocked .

Depends entirely on the pricing. The new 8 core could be the logical successor to the 6 core if the pricing shifts correctly.

The iMac is about 2 years ahead of the MP in technology ... mostly due to the Intel roadmap. Thunderbolt ports, USB 3, Sandy/Ivy Bridge Performance... The Mac Pro should be leading the way yet here we are nearly 2 years after it's release and it's still shipping...without the latest technology surrounding the cores. Sorry, but this is not a one legged race down the tech highway...so don't go pointing at all the cores . You want to buy a MP today and add your external HDs via the USB 2 or FW800? Or maybe go buy an eSata card... how exciting... You go for it... pathetic state that MP is in.

I care about actual processing power, and in that respect, the Mac Pro blows the iMac out of the water.

Mac Pro still ships with the latest Xeons. And a six core Mac Pro will run circles around the iMac. Sorry, the iMac just isn't competitive.

If the iMac came in 6, 8, and 12 core varieties, sure, maybe it would be competitive.

 

[deconstruct60]

Depends entirely on the pricing. The new 8 core could be the logical successor to the 6 core if the pricing shifts correctly.

Pricing leaked a long time ago.

http://www.cpu-world.com/news_2011/2011102701_Prices_of_Xeon_E5-2600-series_CPUs.html

The E5 2670 ( $1552 ) or E5 2665 ( $1440) roughly matches the X5670 ($1440 ). I suspect Apple would simply tack on the extra $200 to bump the price up to cover the increased cost of the E5 2670 on the top end model if they go that route. On strictly pricing, it is 2620 (new $406 old $387) , 2650 ( $1106 vs $996 ), and 2660 ( $1440 vs. $1440 ). However, that would be a base GHz drop with Turbo may/may not make up for. Intel has put some price creep into the E5 prices. Apple will most likely pass that along. ( Until AMD gets their act together, that will continue to happen. )

Matching clock rate is flawed. Apple is very unlikely to go with E5 2600 models that weigh in at over 115W TDP. Even 115W is a stretch.


There is no "8 core" E5 1600 model. In the single package context, the 6 core is the successor of the 6 core. The single package model will still top out at with 6 cores. It won't be as I/O choked though ( more cache and additional memory controller). May see $100 bump in box price though (new $1099 vs. old $999 ).
http://www.cpu-world.com/news_2011/2011091901_Prices_of_Intel_Xeon_E5-1600_series_CPUs.html

 

[deconstruct60]

haha - they will def release the new MP before the new iMac, but with only sandy bridge architecture....it will not have USB 3, only thunderbolt. So again, when the iMac is released with it's Ivy Bridge and USB3 supported natively at the chip sets, then Mac Pro will again be playing catch up to the iMac and more... obviously, this is Intel road maps that Apple is stuck with

Eh? First, most of the aspects you are talking about here have nothing to do with the CPU package or the x86 architecture implemented. USB 3 and Thunderbolt are implemented outside of the CPU package. Second, for the Mac Pro's larger case and motherboard it isn't particularly constraining to implement both TB and USB 3.0 with discrete supporting controller chips.
Third, TB for the Mac Pro isn't a 'slam dunk'. The box may not get it.


Additionally, USB 3.0 has a software component (xHCI) which Apple has been completely radio (and leak) silent on. That's in stark contrast to their competitors in Redmond http://blogs.msdn.com/b/b8/archive/2011/08/22/building-robust-usb-3-0-support.aspx . I would have expected some aspect of the software stack to have leaked by now. I would not be surprised if Apple stiffed Mac users in 2012 by delaying the xHCI software stack until 2013 (i.e., might get some USB 3.0 Macs this year but support isn't there. Shades of OpenGL 4.0 GPUs but a Mac OS X software stack stuck in the 3.x timeframe. ) Depends if the execs at Apple were drinking TB marketing kool-aid all through 2010 and 2011.

In contrast, looking at the CPU packages, the Sandy-Bridge Xeon E5's have PCI-e v3.0. So does the upcoming Ivy Bridge mainstream Core-i series. In this aspect, the Ivy Bridge is "catching up" to the SB lead. The Xeon series typically gets more memory I/O in the generation previous to the updates coming to Core-i even though the generic overall architectures are marketed under different names. Because the Xeon class processors are larger and (so far) don't have GPUs, Intel tends to drop a few changes on those packages first before trickling it down to the mainstream series.

 

[wallysb01]

On strictly pricing, it is 2620 (new $406 old $387) , 2650 ( $1106 vs $996 ), and 2660 ( $1440 vs. $1440 ). However, that would be a base GHz drop with Turbo may/may not make up for. Intel has put some price creep into the E5 prices. Apple will most likely pass that along. ( Until AMD gets their act together, that will continue to happen. )

I hope they don't stick exactly to the prices seen in the 2010 model. That 2620 with only a 2.0 GHz clock rate scares me a little. Even though I'm closer to the core count than the core speed camp when it comes to work flow, a 2.0 clock speed has me worried. I guess single threaded stuff is likely to boost to 2.53 or maybe a little higher, but even still...I think it would be well worth an extra $400 for a pair of 2630s.

I think the 2650 and 2670 sit in good spots for the other options. Particularly if they chose the 2630 for the base DP. Then you get a little per core speed bump in the 2630 relative to the 2650 if that's important to you, or if you just want cheap-ish core count you go for the 2650, and both is in the 2670.

If they go for only the 2660 at the top end, I don't think they will use the 2650. It just isn't different enough in price or speed from the 2660. So then does the mac pro line only have one 8 core processor option? That would be a mistake in my opinion. Then picking the 2665 just seems cheap.

 

[Neodym]

But, how is FW800 a bad choice? Particularly on a Mac Pro that houses at least 4 drives internally? If you use the optical bays, and get something like this: http://www.transintl.com/store/category.cfm?Category=2799&RequestTimeOut=500

then you could have up to 10 HDs in your Mac Pro. Very few people really need external storage for more than a back up on the Mac Pro. So, why does a few extra MB/sec really matter with USB 3.0 compared to Firewire 800, if you can even get that, since FW800 usually saturates slow external drives?

People on this forum often claim the MacPro being targeted at "Pro" users. Especially in those environments the MP is not an island, but exchanges data with other computers/users. This does not always happen in form of a network, but instead in form of external drives.

"Pro" users tend to have huge amounts of data and thus are thankful for any improvement to lower transfer times.

And "Pro" users tend to not use 'slow external drives', but instead the fastest they can get, with the trend strongly moving towards SSD and RAID housings...

Eh? First, most of the aspects you are talking about here have nothing to do with the CPU package or the x86 architecture implemented. USB 3 and Thunderbolt are implemented outside of the CPU package.

Some people have claimed that USB3 and Thunderbolt can not be implemented into the MacPro until the next CPU generation (with according chipset) is available from Intel...

Second, for the Mac Pro's larger case and motherboard it isn't particularly constraining to implement both TB and USB 3.0 with discrete supporting controller chips.

I fully agree! Apple could have easily improved the MacPro's value by adding TB / USB3 independently from the next CPU revision eventually coming from Intel!

Third, TB for the Mac Pro isn't a 'slam dunk'. The box may not get it.

The Mac Pro desperately needs an interface for DAS faster than both FW800 and USB2 (see my reply to wallysb01 further above). Thunderbolt would be a good candidate...

 

[thekev]

I hope they don't stick exactly to the prices seen in the 2010 model. That 2620 with only a 2.0 GHz clock rate scares me a little. Even though I'm closer to the core count than the core speed camp when it comes to work flow, a 2.0 clock speed has me worried. I guess single threaded stuff is likely to boost to 2.53 or maybe a little higher, but even still...I think it would be well worth an extra $400 for a pair of 2630s.

This is the kind of thing that annoys me with Intel to a degree. At equivalent clock speeds, Sandy Bridge E doesn't look like too massive of a gain over what is currently available.

Additionally, USB 3.0 has a software component (xHCI) which Apple has been completely radio (and leak) silent on. That's in stark contrast to their competitors in Redmond http://blogs.msdn.com/b/b8/archive/2011/08/22/building-robust-usb-3-0-support.aspx . I would have expected some aspect of the software stack to have leaked by now. I would not be surprised if Apple stiffed Mac users in 2012 by delaying the xHCI software stack until 2013 (i.e., might get some USB 3.0 Macs this year but support isn't there. Shades of OpenGL 4.0 GPUs but a Mac OS X software stack stuck in the 3.x timeframe. ) Depends if the execs at Apple were drinking TB marketing kool-aid all through 2010 and 2011.

Apple has been lazy on drivers in general for some time. Intel never said it would kill usb anyway.

People on this forum often claim the MacPro being targeted at "Pro" users. Especially in those environments the MP is not an island, but exchanges data with other computers/users. This does not always happen in form of a network, but instead in form of external drives.

"Pro" users tend to have huge amounts of data and thus are thankful for any improvement to lower transfer times.

And "Pro" users tend to not use 'slow external drives', but instead the fastest they can get, with the trend strongly moving towards SSD and RAID housings...
.

While that is correct, thunderbolt doesn't actually outpace current technology available in pci form factors. Someone may bring up that it will likely increase in speed down the line, but no major speed increases are planned for this year.

 

[deconstruct60]

This is the kind of thing that annoys me with Intel to a degree. At equivalent clock speeds, Sandy Bridge E doesn't look like too massive of a gain over what is currently available.

Got benchmarks to back that up? In contrast,

"... due to the productivity gains it provides, Core i7-3960X looks to be a substantial upgrade as a result of its Sandy Bridge roots. ..."
http://www.tomshardware.com/reviews/core-i7-3960x-x79-performance,3026-15.html

I don't think Intel or anyone else promised "massive" gains where core count and clock speed were equal. I think some Intel material is hyping very significant gains where the software can take advantage of increased core count.


It is difficult to deliver equivalent clock speeds when the core count gets close to the double digits. Significant performance is being gained by software not living in the last century by leveraging the additional cores present. The other issue is what happens when these processors fall back to the "old" core count ( turn off 2 cores and fall back to the 4 (or 6) core equivalents.) where the clock speeds end up.

It is bit early to "write off" these 2620/2650 updates because the base clock is lower without knowing where the Turbo range is.

Apple has been lazy on drivers in general for some time. Intel never said it would kill usb anyway.

Intel as a whole perhaps, but some Intel product/marketing folks were dishing out the "eventually one port to rule them all" kool-aid early on in the Lightpeak phase. Yeah, they have put a cap on that goofiness as of late but early one they were fanning the flames.

While that is correct, thunderbolt doesn't actually outpace current technology available in pci form factors. Someone may bring up that it will likely increase in speed down the line, but no major speed increases are planned for this year.

or Next. http://www.anandtech.com/show/5405/the-first-thunderbolt-speed-bump-likely-in-2014

Not sure how TB can possibly outpace PCI-e since it pragmatically depends upon PCI-e.

Even a bit questionable at that point whether TB gets a bump to cover PCI-e v3.0 in 2014. Wouldn't be surprising to see it still regulated to the legacy v2.0 zone but with some updated bandwidth ( possibly an x8 connector instead of x4 , but suspect more flow increase along Dispaly Port traffic that PCI-e. ).

 

[deconstruct60]

...I think it would be well worth an extra $400 for a pair of 2630s.

It will be more than $400. It will be $400 + Apple's 30% mark-up; so about $520 more. I suspect that most will balk at a $1040 price increase. Note that for the dual package E5 2600 series models you are buying in groups of two. So any $100, $200, $400 increase becomes a $200 , $400, $800 user cost increase. I think most Mac Pro users are very price sensitive to amounts over $300.

If they go for only the 2660 at the top end, I don't think they will use the 2650. It just isn't different enough in price or speed from the 2660.

My listing was a bit off. Should have been the E5 2665 at $1440. So that is about a $400 difference (round up to even 100's for Apple's xx99 pricing gap formula) .

Again. there is a $400+ difference between the two. When bought in groups of two that makes for a $1000+ price difference once you tack on Apple's +30% mark up. So I think users will notice.

 

[thekev]

Got benchmarks to back that up? In contrast,

"... due to the productivity gains it provides, Core i7-3960X looks to be a substantial upgrade as a result of its Sandy Bridge roots. ..."
http://www.tomshardware.com/reviews/core-i7-3960x-x79-performance,3026-15.html

I don't think Intel or anyone else promised "massive" gains where core count and clock speed were equal. I think some Intel material is hyping very significant gains where the software can take advantage of increased core count.


It is difficult to deliver equivalent clock speeds when the core count gets close to the double digits. Significant performance is being gained by software not living in the last century by leveraging the additional cores present. The other issue is what happens when these processors fall back to the "old" core count ( turn off 2 cores and fall back to the 4 (or 6) core equivalents.) where the clock speeds end up.

It is bit early to "write off" these 2620/2650 updates because the base clock is lower without knowing where the Turbo range is.

Higher clock rate and turbo boost are what boosted the i7 2600k. It was clocked significantly above the previous generation. In the case of Xeon cpus, that's not really what we're seeing. The 3960x has a max turbo boost comparable to the i7 2600k which isn't really outpacing the W3680 on anything that can use the available cores effectively. Geekbench also supports this. On a side note, the price creeping from both Intel and Apple irritates me. It's unlikely to keep me from buying one, but it's annoying anyway. I kind of wonder if Apple will figure from the current price of the 6 core when adjusting for the next generation, even though it started around $1000. On the mac pro end, Apple's markup seems to be higher than what you suggest.

Intel as a whole perhaps, but some Intel product/marketing folks were dishing out the "eventually one port to rule them all" kool-aid early on in the Lightpeak phase. Yeah, they have put a cap on that goofiness as of late but early one they were fanning the flames.

I still read the effects of TB marketing kool-aid on Apple fan sites. I definitely didn't see mice, keyboards, etc. being remade with TB connections.

or Next. http://www.anandtech.com/show/5405/the-first-thunderbolt-speed-bump-likely-in-2014

Not sure how TB can possibly outpace PCI-e since it pragmatically depends upon PCI-e.

Even a bit questionable at that point whether TB gets a bump to cover PCI-e v3.0 in 2014. Wouldn't be surprising to see it still regulated to the legacy v2.0 zone but with some updated bandwidth ( possibly an x8 connector instead of x4 , but suspect more flow increase along Dispaly Port traffic that PCI-e. ).

You know that I meant in terms of available cards, and that I know how thunderbolt works. For comparable money to what would be spent on a thunderbolt enclosure, you can put together a very nice Raid setup with higher throughput. Not everyone needs it. For something like uncompressed video capture, it could be quite useful. Also I recall a test where Anandtech bottlenecked the TB data rates via use of displayport.

 

[wallysb01]

It will be more than $400. It will be $400 + Apple's 30% mark-up; so about $520 more. I suspect that most will balk at a $1040 price increase. Note that for the dual package E5 2600 series models you are buying in groups of two. So any $100, $200, $400 increase becomes a $200 , $400, $800 user cost increase. I think most Mac Pro users are very price sensitive to amounts over $300.



My listing was a bit off. Should have been the E5 2665 at $1440. So that is about a $400 difference (round up to even 100's for Apple's xx99 pricing gap formula) .

Again. there is a $400+ difference between the two. When bought in groups of two that makes for a $1000+ price difference once you tack on Apple's +30% mark up. So I think users will notice.

It was $206/processor for going from the 2620 to 2630. Like you mention, its not time to write these "low" clocked CPUs off until we see the turbo capabilities, but its going to have to be pretty darn aggressive for me not like a 15% faster CPU for "only" another $520-ish (including "Apple tax").

Part of my opinion on this matter stems from the desire to have a very good base model, even if the base is slightly more expensive. The jump to the 2650 would be $700/processor over the 2620. So if we stick with the 30% apple tax, that's $1800. That's another $300 over the current gap between the the base DP and the mid model. And if Apple likes the ~$1500 price difference between base and mid, this would help attain something closer to that.

And with an ever larger price gap, if having a 2.0 GHz clock rate is a hard pill to swallow on the 12 core, you just don't have a lot of options to balance core count and speed on the DP Mac Pro. That is unless they shake things up and offer 4 options, including the 2640. However, you're reaching diminishing returns. The $206 from 2620 to 2630 gains you 15% or 300 MHz, but you need $272 to go from the 2630 to 2650 and it gains you 8.7% or 200 MHz. So, unless the turbo boosts ratios are different, I just don't see a good reason to not pay ~$500 for a 15% faster CPU.

 

[wallysb01]

People on this forum often claim the MacPro being targeted at "Pro" users. Especially in those environments the MP is not an island, but exchanges data with other computers/users. This does not always happen in form of a network, but instead in form of external drives.

"Pro" users tend to have huge amounts of data and thus are thankful for any improvement to lower transfer times.

And "Pro" users tend to not use 'slow external drives', but instead the fastest they can get, with the trend strongly moving towards SSD and RAID housings...

The Mac Pro desperately needs an interface for DAS faster than both FW800 and USB2 (see my reply to wallysb01 further above). Thunderbolt would be a good candidate...

Sure, faster is better with anything. But USB 3 just sits in a funny spot. eSata is not that much slower, thunderbolt is probably on its way and will be faster. A thunderbolt only refresh would be rather impossible. I just don't really see this as the issue to be impatient about. eSata can be added on easily enough if speed to externals is that important to you, FW800 is not terribly slow, and you can absolutely load these Mac Pro's internally if you want. Is it perfect? No. Buts damn good.

 

[derbothaus]

People on this forum often claim the MacPro being targeted at "Pro" users. Especially in those environments the MP is not an island, but exchanges data with other computers/users. This does not always happen in form of a network, but instead in form of external drives.

"Pro" users tend to have huge amounts of data and thus are thankful for any improvement to lower transfer times.

True. But real "Pro's", like my users, have had FAS SAN's and 32 head RAID's for years and no Thunderbolt consumer doodad will change that. Any other Mac that can use those HBA's? No.
For pro's it DOES happen in the form of a network. GigE is everywhere and I have it to the desktop. But maybe we are not talking about the same caliber of Pro. I'm not talking home video editor but Enterprise pro.

 

[goMac]

True. But real "Pro's", like my users, have had FAS SAN's and 32 head RAID's for years and no Thunderbolt consumer doodad will change that. Any other Mac that can use those HBA's? No.

Thunderbolt isn't a huge issue for Pros, but it's hard to imagine Apple would continue shipping a Mac that didn't fit into it's general Thunderbolt strategy. Especially if we start seeing things like Thunderbolt cameras.

Thunderbolt doesn't worry me because it's a feature I would need, it worries me because the ability to add it to the Mac Pro would imply Apple's general interest in the machine.

 

[thereubster]

Finally Some Xeon E5 news

Looks like HP are getting ready to announce their Xeon E5 Gen 8 servers
http://www.channelregister.co.uk/2012/02/07/hp_server_launch_preview/
so we might see something from Apple sometime soon? I have been a PowerMac and Mac Pro User for more than 20 years, and Apple have been consistant about several things for their launches (of Macs);
- They ALWAYS hold off launching until they can ship at least one model immediately
- They have usually gotten the first shipments of the next-gen Intel processors, so although others might announce first, Apple usually ships first.
- They rarely release a new version of the Pro Mac unless it is the highest end/highest performing Workstation PC on the market AT that moment in time. I am not including Processor updates, etc. This is so they can maintain the ultra-high end pricing structure.
- They usually launch a new technology at the same time, either hardware or software, to make the new High end Mac a "must have".

With all of this in mind I am predicting
- A new Pro Mac workstation (might be redesigned) with Xeon E5 processors up to 16 cores/128Gb RAM possibly with the new Nvidia 6XX series GPU and Thunderbolt.
- a new Hi-DPI monitor maybe 30 inch or more + Hi DPI updates to OSX (already appearing so this is a no brainer)
- Updates to Apples pro software plus possibly a new product.

OK so none of this is too ground breaking, but the pieces are finally coming together......

 

[barkmonster]

CeBIT: 6-10 March, 2012Very last revision = end of lifecycle where fixed cost amortization risks are lowest. Plus the PowerMac was the last to be converted to Intel, so there would have been inventory to "clear out". Either makes it a bad comparison.

It's not the fact it was the last revision of a range, they ALWAYS had a low end Powermac G3, G4 and G5 with EVERY range they introduced from the original beige G3 desktop (which I owned) to the very last G5s.

Even before that Apple offered the high end 9600 tower systems and the entry level 7300 desktop systems.

They're literally ignoring a segment of their users who want to be able to add drives and expansion cards, use their existing screens and want a few years worth out of their system but only have the budget for the entry level tower/desktop systems they no longer offer. Fortunately, the used market fills the gap nicely with 2.66Ghz Mac Pro's available for as little as £700 after 6 years and the 8 core models around the £1,000 mark but it shouldn't be this way.

 

[Umbongo]

- They ALWAYS hold off launching until they can ship at least one model immediately

This has seemed to be the case, but they did announce the 2010 models a fair bit before shipping. By this time one would expect Apple to be just needing processors from Intel in the right volume or Intel to give the go-ahead for release.

- They have usually gotten the first shipments of the next-gen Intel processors, so although others might announce first, Apple usually ships first.

Yeah true for the 2008 and 2009 models and the 3GHz 2007 8-core update. The 2006 model was within a reasonable time frame, I would think they waited for WWDC because they could and to sell off as many PPC systems as they could. 2010 was a big wait and the 2009 3.33GHz processor addition came a fair while after they were available elsewhere.

With their previous history as it is now, I would not be surprised if Apple announced it tomorrow or in May. Too many variables we have no idea about, but my gut says they will release it a good while before they update all the others to Ivy Bridge in May/June.

They rarely release a new version of the Pro Mac unless it is the highest end/highest performing Workstation PC on the market AT that moment in time. I am not including Processor updates, etc. This is so they can maintain the ultra-high end pricing structure.
- They usually launch a new technology at the same time, either hardware or software, to make the new High end Mac a "must have".

I think these two have subsided a bit in the last few years as sales have dwindled in regard to the Mac Pro. The 2009 and 2010 models were a fair bit off the top end available from everyone else. 3.2GHz 4 and 8-core, with 6 and 12 memory slots at 1333MHz were available from all other workstation vendors in 2009 yet Apple chose to limit those things. I can't see them offering a 3.1GHz 16-core workstation with 16 memory slots either which should be available from HP, Dell, Fujitsu etc.

With all of this in mind I am predicting
- A new Pro Mac workstation (might be redesigned) with Xeon E5 processors up to 16 cores/128Gb RAM possibly with the new Nvidia 6XX series GPU and Thunderbolt.
- a new Hi-DPI monitor maybe 30 inch or more + Hi DPI updates to OSX (already appearing so this is a no brainer)
- Updates to Apples pro software plus possibly a new product.

OK so none of this is too ground breaking, but the pieces are finally coming together......

Doubt there will be an NVIDIA card as they are at least a few months away and the 7000 series AMD drivers have already been seen and those cards will have been available for sometime.

Currently monitors over 30" (4K resolution) are tens of thousands of dollars and I would think a 30" 2560x1600 LED would be too similar to the 27" for Apple to bother. Then again they have had multiple displays in the past and have introduced new panels before. So here's to hoping!

 

[xgman]

I wish someone would leak some real news about this.

 

[goMac]

I would bet on an ATI card and not an NVidia.

 

[-hh]

It's not the fact it was the last revision of a range, they ALWAYS had a low end Powermac G3, G4 and G5 with EVERY range they introduced from the original beige G3 desktop (which I owned) to the very last G5s.

Even before that Apple offered the high end 9600 tower systems and the entry level 7300 desktop systems.

Sure, but the 7100-9600 models (and Performa's too) dates from the Pre-Steve-Return-to-Apple days, when Apple rashly had too many models.

Subsequently, the proliferation count dropped pretty rapidly. In 1997/8 there was the G3 "Gossimer" and the "All In One" for EDU sales. As the iMac launched, this was superceded by the G3 Blue&White, which is when you could say that we first hit what we have today, namely one basic box that's sold with a couple of different CPU speeds (and later: number of cores) inside.

They're literally ignoring a segment of their users who want to be able to add drives and expansion cards, use their existing screens and want a few years worth out of their system but only have the budget for the entry level tower/desktop systems they no longer offer. Fortunately, the used market fills the gap nicely with 2.66Ghz Mac Pro's available for as little as £700 after 6 years and the 8 core models around the £1,000 mark but it shouldn't be this way.

Yes, but the context is while in a market which over the past 15+ years has gone from 75% desktops to 75% laptops, and also where the hardware based "Apple Tax" (at least in the USA) has dropped from easily a 30% premium to only around 10%.

And considering that what I paid for my Apple ][+ easily thirty years ago is still about the same in literal dollars as a brand new Mac Pro today, I'd love for it to be half the price, but by the same token, I'm not horribly surprised that its not. That, and a broader timeline examination of historical prices personally brings me to the opinion that the anomoly isn't today's "high" prices, but is instead a relatively narrow slice of time where they were unusually low that invariably gets used in these comparisons: if we go a little bit further back in time, we again find Apple towers to be expensive.

To whit..a post of mine from May 12, 2009:

For example, as per this online tool baed on the Consumer Price Index:

What cost $2000 in 1993 would cost $2944.06 in 2008.

And what would cost $2000 in 2008 would have cost $1369.22 back in 1993.

Thus, if we plug in to today a value of $2500 as a ballpark for a basic Mac Pro, what we find is the following comparisons:

1993 Equivalent: $1711.52
1998 Equivalent: $1942.00
2003 Equivalent: $2179.86

And if we go look for historical Mac baseline hardware for those years, we find:

1993:
Centris 610 introduced 1993.02.10 at $2,520;

That $2520 today would be $3709.52
Conclusion: 1993 was ~48% more expensive than today.

1998:
Beige G3/233 introduced 1997.11.15 at $1,999;

That $1999 today would be $2629.10
Conclusion: 1998 was slightly ~5% more expensive than today.

2003:
PowerMac G5/1.6GHz (IMO a poor machine): $1,999;
reduced on 11/03 to $1,799;
The (IMO better) 1.8 GHz sold for $2,399;

That $1999 today would be $2335.33, and
the price-reduced $1799 would be $2101.68 today.

FYI, the better 1.8GHz $2399 would today be $2802.63

And FYI, yes I am aware of the 1.8 GHz single 256/80 introduced 2004.10.19 at $1,499 (discontinued 2005.06.20). The problem was that it was a sour lemon 'Road Apple' that was actually slower than the older 2003 1.6GHz because its FSB was lobotimized to 1/3rd speed (600MHz), whereas the other PM G5's had used 1/2 speed (1.6GHz used 800MHz and the older 1.8GHz was 900MHz). Nevertheless, if you really wanted this $1499 Road Apple reject today, it would still cost you $1695.26 ... which is more than a 24" iMac.

Please note that the above is using 2008 dollars, not 2011 dollars, so it does need to be updated. The URL has been changed to a newer Inflation Calculator which has 2011 rates; for example, the aforementioned 2004 $1499 Road Apple is equivalent to $1785 in 2011 dollars...up nearly $100 since the above 2008 price point.

-hh

 

[Hellhammer]

Doubt there will be an NVIDIA card as they are at least a few months away and the 7000 series AMD drivers have already been seen and those cards will have been available for sometime.

But can't Apple get them early? They have gotten in the past! :trollface:

The first Kelper SKUs (GTX 680, 670 and 660) will be released in April if reports are to believe. Hopefully, the new Mac Pro is already out by then. Availability of new GPUs is always very low so if Apple was to use NVidia, then I would say we are looking at May-June launch. Don't really see Apple going back to NVidia though, AMD is doing pretty well nowadays.

 

[nanofrog]

But can't Apple get them early? They have gotten in the past! :trollface:

In theory, Yes.

But it has to do with what, and how much they're purchasing in order for factors other than price to be negotiable, such as an early delivery schedule. In the case between Apple and Intel, keep in mind that Intel was also supplying the boards for the 2006 - 2008 models.

BTW, there's higher margins in board manufacturing, which would have been a significant factor in why they were willing to make such negotiations at the time.

In the case of the 2009 systems, I suspect there were different dates on the CPU and board contracts, which is what allowed Apple to get early access to CPU's for the 2009's, while shifting their board manufacturing to Hon Hai Precision (Foxconn). Apple simply played the dates when they got a very attractive offer from Foxconn (i.e. free or nearly free final assembly as a means of attracting Apple's board contracts). It gave Apple a lower cost per unit, and Hon Hai Precision increased their margins. Intel was the only "loser" in the deal.

I'd imagine Intel was a bit unhappy over that one, but more importantly, when you consider the reduced CPU volume for the Xeons destined for the MP's (and ultimately profit for that segment from Apple), Intel isn't motivated to offer them the incentives they did in the past for Xeons. They're a business, and exist to make money. So if Apple is willing to purchase large enough volume on a particular product/segment and/or enter contracts for say manufacturing boards as well, they'd still be willing to deal if doing so ultimately increases their profits vs. not offering such incentives.

As per the workstation market, it's been shrinking overall as the advent of more powerful consumer variants that have allowed those that were previously stuck with Xeons to shift to less expensive consumer systems (i.e. do not require ECC or more memory than the consumer variants can handle).

 

[Hellhammer]

In theory, Yes.

I guess your sarcasm detector wasn't on

Besides, we were talking about GPUs and AFAIK, Apple has never got them early.

 

[nanofrog]

I guess your sarcasm detector wasn't on

Besides, we were talking about GPUs and AFAIK, Apple has never got them early.

Apparently not. I keep thinking of the thread's title and went with it, blindly missing the GPU aspect.

The trollface bit left me with a expression. Must be a crazy European cultural thing. Or I just haven't consumed enough caffeine yet. Either one would generate the same thing.

 

[thekev]

As per the workstation market, it's been shrinking overall as the advent of more powerful consumer variants that have allowed those that were previously stuck with Xeons to shift to less expensive consumer systems (i.e. do not require ECC or more memory than the consumer variants can handle).

There are quite a lot of issues. Even for demanding users, much of the time a lot of what they do won't run much faster on a dozen cores than four. It's possible to be in a situation where you can use more power, yet the type of power offered doesn't grant much in the way of performance. I have to wonder if Intel has tried to determine how much longer they can milk X86 and where they see themselves after that.