Why no 3.33 12-Core?

[englishman]

Why is the 3.33 Hex only available in single CP configuarations?

[Hellhammer]

Because X5680 is 130W while X5670 is 95W. X5680 could be used and you should be able to upgrade to them later on but Apple likes the greenness and thus has used the lower TDP CPUs in at least 2009 and 2010 Mac Pros.

[cube]

Maybe the power supply is not beefy enough, nothing about "greenness".

[Hellhammer]

Maybe the power supply is not beefy enough, nothing about "greenness".

The PSU is 1000W AFAIK, that's more than enough for 2x130W CPUs as Apple offered dual X5482 (3.2GHz 8-core) in 2008 Mac Pro and those CPUs are 150W each...

[Capt Underpants]

Maybe the power supply is not beefy enough, nothing about "greenness".

The MP power supply could easily accomodate the extra 70 watts.

[Umbongo]

Maybe the power supply is not beefy enough.

The 2008 models were available with two X5482s, 8 FB-DIMMs, 4 SAS drives and a Quadro FX 5600 which needed more power than two X5680s, 8 DDR3 DIMMs and 4 SATA/SSD drives with the Radeon 5870 would.

[cube]

OK, but assuming the power supply was not downgraded in the new models.

[Chupa Chupa]

According to Intel's spec sheet the W3680 is only compatible in a SINGLE CPU configuration so it really wasn't Apple's choice: http://ark.intel.com/Product.aspx?id=47917

[Hellhammer]

According to Intel's spec sheet the W3680 is only compatible in a SINGLE CPU configuration so it really wasn't Apple's choice: http://ark.intel.com/Product.aspx?id=47917

There is X5680 which is basically the same as W3680 but has dual QPI and thus supports dual CPU configurations.

[VirtualRain]

Why is the 3.33 Hex only available in single CP configuarations?

It most certainly is related to the 130W TDP.

If you look at the single processor Mac Pro's you will see that the HSF (heat-sink fan) is much larger than those used in the dual processor systems. Two of the SP HSF's simply wouldn't fit in a DP system. Hence, I'm guessing that the HSF in the SP systems is suitable for 130W TDP and the HSF in the DP systems is suitable for up to 95W of TDP at acceptable noise levels.

In fact, I would say the DP systems are even marginal for 95W TDP processors as Apple had those running naked (sans heat spreader) in the 2009 DP Mac Pro's which is usually an indication that they are trying to extract every last watt of cooling potential from the HSF possible.

It will be interesting to see if the DP processors are running naked in the 2010's.

[AZREOSpecialist]

It most certainly is related to the 130W TDP.

If you look at the single processor Mac Pro's you will see that the HSF (heat-sink fan) is much larger than those used in the dual processor systems. Two of the SP HSF's simply wouldn't fit in a DP system. Hence, I'm guessing that the HSF in the SP systems is suitable for 130W TDP and the HSF in the DP systems is suitable for up to 95W of TDP at acceptable noise levels.

In fact, I would say the DP systems are even marginal for 95W TDP processors as Apple had those running naked (sans heat spreader) in the 2009 DP Mac Pro's which is usually an indication that they are trying to extract every last watt of cooling potential from the HSF possible.

It will be interesting to see if the DP processors are running naked in the 2010's.

The 2009 8-core models also did not offer the highest 3.33 GHz variant of those Nehalem CPUs either even though the 3.33 GHz variant was offered on the Quad. There must be a parallel for the 2010 models, and further confirmation that very little has changed from the 2009 models.

[CaptainChunk]

Aside from cosmetics, the power supply itself in the Mac Pro hasn't changed since the original 2006 models launched. It's always been 980W and manufactured by Delta.

I think this is more of a case of Apple looking at TDP numbers than anything else.

[Hellhammer]

The 2009 8-core models also did not offer the highest 3.33 GHz variant of those Nehalem CPUs either even though the 3.33 GHz variant was offered on the Quad. There must be a parallel for the 2010 models, and further confirmation that very little has changed from the 2009 models.

The option for W3580 was added on December 2009, nine months after the release of 2009 Mac Pros. Before that, the quad only had an option for W3570.

[Concorde Rules]

OK, but assuming the power supply was not downgraded in the new models.

It's 99% the same as it is in the 2009 Mac Pro, which is either 1000 or 1200W IIRC. Edit: 100% 1200W as the 2009 Mac Pro is!

All apple has changed is the GPU/CPU lineup with higher spec memory and a bumped HD layout.

Everything else however, is the same as the 2009 Mac Pro.

[Transporteur]

I don't think that heat is a problem.
The 2009 DP models can easily handle the heat of two 130W processors.
Tutor and gugucom have upgraded their processors with the W3380, respectively W5590 processors.

[VirtualRain]

I don't think that heat is a problem.
The 2009 DP models can easily handle the heat of two 130W processors.
Tutor and gugucom have upgraded their processors with the W3380, respectively W5590 processors.

Yeah, but they are probably exceeding the specs for the HSF and/or Apple's noise level requirements. There's no technical reason why you can't run higher clocked / higher TDP CPU's in those sockets... they will just run hotter and/or noisier... something Tutor and gugucom might not care about, but Apple didn't want to offer.

[xgman]

maybe because the cost would just be sick.

[Ryan P]

My current intention is to get the 2010 2.4 Octa and replace the processors with the 3.33 5680 Xeon or perhaps even then 3.47 5690 Xeon. I'm hoping this is possible. I wonder if putting a better cooling system in the Mac Pro's is possible, water cooling perhaps.

[Transporteur]

My current intention is to get the 2010 2.4 Octa and replace the processors with the 3.33 5680 Xeon or perhaps even then 3.47 5690 Xeon. I'm hoping this is possible. I wonder if putting a better cooling system in the Mac Pro's is possible, water cooling perhaps.

Water cooling is always possible, requires some major modifications to the case though, which is IMO not worth it since you'd lose any warranty claims whatsoever.

[Cynicalone]

My current intention is to get the 2010 2.4 Octa and replace the processors with the 3.33 5680 Xeon or perhaps even then 3.47 5690 Xeon. I'm hoping this is possible. I wonder if putting a better cooling system in the Mac Pro's is possible, water cooling perhaps.

It's been done to the 09's. So it should be possible.

[Greg Punzo]

The 3.33's bettter work in the 12 core system cuz that's the only reason why i got the 2.66 12 core in the first place was to upgrade later. Otherwise I would have just gotten the 3.33 6 core.
Btw, where do u find these Chips to buy online anyways?


Sent from my iPad

[Umbongo]

The 3.33's bettter work in the 12 core system cuz that's the only reason why i got the 2.66 12 core in the first place was to upgrade later. Otherwise I would have just gotten the 3.33 6 core.
Btw, where do u find these Chips to buy online anyways?


Sent from my iPad

Just search the part number.

[deconstruct60]

There is X5680 which is basically the same as W3680 but has dual QPI and thus supports dual CPU configurations.

Well there is the $999 (W3680) versus the $1663 difference factor too. :-)
There is a $223 difference in price between the W5680 and W5670. Multiply that by 2 : $446 . Add a 40% Apple profit margin: $624 . So, you'd have a top end Mac Pro that was $700 (gotten have price end in "99" ) more expensive. The more expensive the product, the less folks are going to buy There is a limit to what folks will pay. Additionally, the higher the core box the less likely going to buy the add-ons (memory, drives , etc. ). $700 would just about give a CTO 12 2.93 box 16GB of memory which is a reasonable config for that number and speed of cores. Apple still gets the extra $700 and many customers are happier because it is a better set up out of the box with just one purchase order.


Heat dissipation probably plays a factor too.

Did those previous model 300W dual processor packages boxes have a spotless long term record on power supply problems ?

Folks can say the power supply is the same as before (with 130W processors ), but two factors:

1. Internal case model to adjust for the effectiveness of the multiple thermal zones. Graphics cards are getting more power greedy than 3-4 years ago.

2. The power supply is a zero sum game. The same folks throwing mega dollars at the base box the receive from Apple are likely much more inclined to fill every open slot with stuff. So probably not a tendency to add more of just one subcomponent it will be more on every single dimension possible.
If have 3.33GHz probably going to stuff as much memory as possible in so that can leverage more of that full speed, more disks to leverage more of the speed, more GPU horsepower to display the results. The standard GB Ethernet has gotten be tooo slow for that kind of speed so some higher I/O card. The disk I/O is too limit so RAID card. Oh look, an empty SATA connector in the optical drive.... yeah stick something on that too and run it constantly. Not every user on the highest end box is that way, but that is one of the places in the product spectrum they tend to congregate.


Wouldn't be a very surprising result if Apple went back through previous generation Mac Pro warrantee power supply replacements trouble tickets and found a slight positive correlation between people who run the supply at higher fractions of constant power and those who needed replacements during warrantee period. Apple can't really control folks from adding new components into every nook an cranny. However, they can improve profitibility but given those folks a broader power envelope to work with by capping the dual package offering at 190. That won't remove all of the power supply failure claims but definitely helps remove the more unnecessary ones.


So yeah there is some slack in the 900-980W standard supply in the configurations as shipped by Apple, but using every single last drop of power isn't a prudent idea. It is even worse idea to ship the box from the factory with a small reserve.

[Tutor]

Clean up

[Tutor]

Clean up

[VirtualRain]

Hottest of my 8 cores at idle and under load is core no.1 (per Real Temp), coming in at scorching 37 degrees centigrade when idle, but when rendering a QT movie in Cinebench 11.5 it rises to an hellacious 63 degrees centigrade. Could it be the temps? And as for noise, maybe I forgot to reinstall my fans because I didn't hear the fans at idle or at render time. Could it be the noise? Over one year A.S.*, my 3.2 Ghz 8-core still runs cool and quietly even when rendering in Motion, Maya, and Cinema 4d 11.5 and while achieving a geekbench score of 19,474. Does a very powerful weightlifter have to strain (i.e., sweat) or cry out when doing heavy lifting that a weaker soul could achieve? I submit the answer is "no." If my temps are too hot, then maybe our favorite company needs to get out of my kitchen.


A.S. means After the Swap.

As I suspected, it's not a concern for you. :) Perhaps it wasn't a concern for Apple either, but I think that's why you're not seeing 130W CPUs in the two socket boards. However, others have raised other possible explanations such as stratospheric price points (not traditionally an issue for our favorite company), or power constraints (less likely to be the issue in my opinion with the limited peripherals that need power for such a large seemingly over spec'd PSU).

[Tutor]

Clean up

[iRobertM]

Someone just posted a picture of the 3.33 heat-sink. It takes up almost the whole bottom board so I am guessing you can't order the x2 3.33 because they could not adequately cool them.

[deconstruct60]

Someone just posted a picture of the 3.33 heat-sink. It takes up almost the whole bottom board so I am guessing you can't order the x2 3.33 because they could not adequately cool them.

That heatsink is probably being conservative. They want to be in the bounds of whatever gets done on the CPU ( or plugged in elsewhere in the box). If you center the heat sink as a target for the circular fan blowing at it then the fan 's air flow can be more efficient. Apple doesn't want the heat from the CPU to go out into the box. They want the heat to be blown out the case hole that is directly behind the CPU in the air flow.

So distributing the heat radiation along that large diameter air flow vector allows it to pick most of it up and chuck it out the box without resulting to creating mini turbo jet levels of airflow. The more efficient the transfer of heat to the moving air the less volume you need to push through the box.