PDA

View Full Version : BareFeats speed comparison: 4-Core vs 8-Core




sigmadog
Apr 3, 2009, 04:48 PM
http://www.barefeats.com/nehal08.html

Interesting results.



VirtualRain
Apr 3, 2009, 04:55 PM
http://www.barefeats.com/nehal08.html

Interesting results.

I think it confirms what many non-biased folks here have been saying, that the best choice of machine depends on the applications you use, and whether you earn a paycheck with your machine to justify the cost of that added performance in some apps.

In my (now biased) mind it demonstrates that for most day-to-day apps, the 2.93 Quad will be a much better choice at around $3K than the 2.26 Octo. The main advantage the Octo has is that it's cheaper to load up with RAM but that advantage will disappear over time.

This is also interesting...

Compressor - When we encoded an HD DVD with Compressor 3, we tried various flavors of QuickClusters using from 2 to 8 instances (aka cores). Comparing the fastest times, the 8-core's "sweet spot" was 6 instances and the 4-core's sweet spot was 4 instances. Once the number of instances was "tuned," the 8-core completed the task 29% faster than the 4-core.


It indicates that you can only get max performance from the Octo core with trial and error... bizarre?!

jjahshik32
Apr 3, 2009, 05:10 PM
But look at the RAM pricings. 3x4GB costs $1458 compared to 6x2GB which costs $208.

Think about it, you can have 12GB or even 16GB of RAM for your 8 core 2.26GHz machine while being whatever % faster than the Quads and overall cheaper than the $2499 + $1458= $3957!

I would much rather pay, $3246 (2.26GHz 8 core) + $208 (12GB of RAM) = $3454!

You dont have to wait almost a year or two for the 4GB dims to drop in price + you get superior hardware for much cheaper. No brainer.

GGGUUUYYY
Apr 3, 2009, 05:14 PM
People buying a quad wouldn't necessarily need 12 gb right away..

jjahshik32
Apr 3, 2009, 05:21 PM
People buying a quad wouldn't necessarily need 12 gb right away..

True but thats the whole argument anyway right. Basically you get higher clock speeds and it does take 4GB dims so their saying that you should get the Quad. But only problem with that is the 4GB stick prices.

Plus I would be dying to put in 12GB of RAM in one of those and with a 2.26GHz octo you can buy the 12GB of RAM right away still end up costing nearly $500 cheaper.

I wouldnt want to wait almost 2 years for the 4GB sticks to come down to a reasonable price. And when it does, even better for the octo users.

Basically to me I wouldnt want to pay $1500 for RAM when I could use it to get a superior hardware in the beginning.

Plutonius
Apr 3, 2009, 05:34 PM
But look at the RAM pricings. 3x4GB costs $1458 compared to 6x2GB which costs $208.

Think about it, you can have 12GB or even 16GB of RAM for your 8 core 2.26GHz machine while being whatever % faster than the Quads and overall cheaper than the $2499 + $1458= $3957!

I would much rather pay, $3246 (2.26GHz 8 core) + $208 (12GB of RAM) = $3454!

You dont have to wait almost a year or two for the 4GB dims to drop in price + you get superior hardware for much cheaper. No brainer.

Where did you get your RAM pricing from. 3x4GB for the quad is $750.00 while 6x2GB for the octo is $219.00. If you need lots of RAM now, the octo is probably a better choice but you are paying much more for the octo in the first place.

I'm also curious why you think the octo is superior hardware ? If you need lots of RAM or run the few apps that will take advantage of the 8 cores, you should be looking at the octo. Otherwise, you will get a much faster computer for the money if you get a quad. If you say you are buying for the future, there will be much better Mac Pros out by time many of the apps are able to take advantage of the current processors.

DrawingArt
Apr 3, 2009, 05:35 PM
4-core offcourse :) now this is nice test that shows how quad is definatelly better > cheeper way then octo for users that are not hard core video or 3d users ... but for most of the adobe things and other everyday work quad is much better choice.

jjahshik32: you are comparing totally wrong 2 products. compare prices of quad 2,93 and octo 2,93. Btw. where did you get those prices for 4GB sticks? OWC 16GB (4x4GB) = $979.99 and OWC 12GB (3x4GB) = $749.00 ;)

Keniff
Apr 3, 2009, 05:36 PM
http://www.barefeats.com/nehal08.html

Interesting results.


These comparisons mean nothing until 'Snow Leopard' has been released/installed.
We're going to see a big change, imho...

sigmadog
Apr 3, 2009, 05:38 PM
But look at the RAM pricings. 3x4GB costs $1458 compared to 6x2GB which costs $208.

Current prices at OWC have 3x4GB for $750. That's still plenty more than the 2GB sticks, yet far easier to stomach than prior prices.

DrawingArt
Apr 3, 2009, 05:38 PM
I wouldn't be so sure, you might be dissapointed.

jjahshik32
Apr 3, 2009, 05:42 PM
http://farm4.static.flickr.com/3644/3409764435_32f573c90e_o.png

From Transintl.com

I dont trust OWC. Their RAM always goes bad or haywire later on. But if you do go with OWC's RAM:

2.66GHz quad ($2489 after tax and student discount) + $749 (12GB RAM) = $3238.

2.26GHz 8 core ($3246) + $208 (12GB of RAM) = $3454

Difference of $216 dollars. I'd still pick the 2.26GHz 8 core model.

My whole point is I dont feel comfortable paying $750 bucks on RAM, I'd rather use it towards a higher end model, which the 8 core is higher end.

Plutonius
Apr 3, 2009, 05:43 PM
These comparisons mean nothing until 'Snow Leopard' has been released/installed.
We're going to see a big change, imho...

I wouldn't base any buying assumptions on what you think Snow Leopard will do.

sigmadog
Apr 3, 2009, 05:43 PM
These comparisons mean nothing until 'Snow Leopard' has been released/installed.
We're going to see a big change, imho...

On the contrary. For people considering a purchase now, the promise of Snow Leopard means very little. My decision to purchase in the next few days is based more the interactions of current hardware and software configurations than the hazy speculation (which is really all we have right now) of Snow Leopard improvements.

DrawingArt
Apr 3, 2009, 05:46 PM
http://farm4.static.flickr.com/3644/3409764435_32f573c90e_o.png

From Transintl.com

I dont trust OWC. Their RAM always goes bad or haywire later on.

OWC is almost 50% cheeper and I trust them :) Btw how come those prices are cheeper for 4*4 then 3*4 :rolleyes: weird, like this its more afordable to buy 4x4 the 3x4.

jjahshik32
Apr 3, 2009, 05:48 PM
OWC 50% cheeper I trust them :) Btw how come those prices are cheeper for 4*4 then 3*4 :rolleyes:

Either way:

2.66GHz quad ($2489 after tax and student discount) + $749 (12GB RAM) = $3238.

2.26GHz 8 core ($3246 after tax and student discount) + $208 (12GB of RAM) = $3454

Difference of $216 dollars. I'd still pick the 2.26GHz 8 core model.

My whole point is I dont feel comfortable paying $750 bucks on RAM, I'd rather use it towards a higher end model, which the 8 core is higher end.

I still think its insane to pay 2k+ on a w3500 series. Basically you either pay for an overpriced X series xeon or you pay for the over priced w3500 series (QUAD) AND overpriced RAM.

I'd rather avoid the latter.

DrawingArt
Apr 3, 2009, 05:51 PM
Either way:

2.66GHz quad ($2489 after tax and student discount) + $749 (12GB RAM) = $3238.

2.26GHz 8 core ($3246 after tax and student discount) + $208 (12GB of RAM) = $3454

Difference of $216 dollars. I'd still pick the 2.26GHz 8 core model.

My whole point is I dont feel comfortable paying $750 bucks on RAM, I'd rather use it towards a higher end model, which the 8 core is higher end.

Hmhm again you are comparing 2 different CPU clocks. Compare 2,66 4-core with 2,66 8-core or 2,93 4-core with 2,93 8-core and see prices :) But still your comparison is 216$ difference and you get bigger speed. I'm telling this from single cpu hard user not from dual. And believe me a lot of the mac pro buyers use only apps that are single cpu based. They just buy slow octo to say "yeah I have 8-core" lol.

jjahshik32
Apr 3, 2009, 05:54 PM
Hmhm again you are comparing 2 different CPU clocks. Compare 2,66 4-core with 2,66 8-core or 2,93 4-core with 2,93 8-core and see prices :)

If you think about though, I'd rather have 2.26GHz 8 physical cores as to only 2.66GHz 4 physical cores. To me it feels like a step backwards.

I look at the machine as a whole, 2.26GHz(x) 8 cores = 18.08, 2.66GHz(x) 4 cores = 10.64.

jjahshik32
Apr 3, 2009, 05:58 PM
Hmhm again you are comparing 2 different CPU clocks. Compare 2,66 4-core with 2,66 8-core or 2,93 4-core with 2,93 8-core and see prices :) But still your comparison is 216$ difference and you get bigger speed. I'm telling this from single cpu hard user not from dual. And believe me a lot of the mac pro buyers use only apps that are single cpu based. They just buy slow octo to say "yeah I have 8-core" lol.

Ummm no. Most Mac Pro buyers use their machines to take advantage of as many CPUs cores as they can.

What do you think the main purpose of more cores are anyways to begin with?

DrawingArt
Apr 3, 2009, 05:58 PM
Hmhm ok ... but you are looking wrong. Check how many applications work and use 8-core and you'll see. but like I said if you need all thos 8-cores then octo is a way but if you don't need them like I don't then it is waste of money for such a slow machine. I use 75% of the time photoshop and since photoshop doesn't use 8-cores its waste of money mate. When it will use it we don't know and I don't care, when it will I'll buy anyway something new ... until then 4-core is faster in photoshop and other apps I use.

To finish this thing, each user needs to know what he will use this machine for. If one CPU is gathering dust inside then you threw your money to trash.

jjahshik32
Apr 3, 2009, 06:00 PM
Hmhm ok ... but you are looking wrong. Check how many applications work and use 8-core and you'll see. but like I said if you need all thos 8-cores then octo is a way but if you don't need them like I don't then it is waste of money for such a slow machine.

To finish this thing, each user needs to know what he will use this machine for. If one CPU is gathering dust inside then you threw your money to trash.

The whole point of Snow Leopard my friend. And its only a few months away. ;)

RAM it'll always get cheaper and you can always add more. Same goes for GPU. But the CPU is like the first thing you look to buy.

To me paying $750 for RAM is a waste thrown into the trash.

DrawingArt
Apr 3, 2009, 06:02 PM
You might be right really on this one. I agree, but still I will not be suprised if it will not be such big step forward. But looking forward to it and hopefully it does what it should :P Still I think my 2,93 quad will keep me satisfied for 2-3 years and then we'll see what we will be new on market :P But yes I plan upgrade to 16gb ram max. 1 year. For now 8 will do.

jjahshik32
Apr 3, 2009, 06:05 PM
You might be right really on this one. I agree, but still I will not be suprised if it will not be such big step forward. But looking forward to it and hopefully it does what it should :P

I guess I'm just not comfortable paying that much money for RAM. Almost a 1k for RAM just listen to how that sounds!

If anything, I would much rather have a 2.66GHz OCTO with 6GB of RAM than a 2.93GHz QUAD with 32GB of RAM. :p

sushi
Apr 3, 2009, 06:07 PM
I think it confirms what many non-biased folks here have been saying, that the best choice of machine depends on the applications you use
Maybe I can provide some insight as to why.

Taking advantage of multi-core systems requires overhead to manage the processes. The same is true with multi-CPU systems. This is why when you double the core amount you will not see a doubling of the speed across the board. Some applications are written to take advantage of multiple cores or are written where the OS can easily farm out the tasks to multiple cores.

To make an analogy, think of it in terms as the management of a small office. In this office, let's say that you have a manager (the OS) and 8 employees (the cores).

Say I have a project that requires 8 tasks (A through H). In this case, each one of these tasks is reliant upon the prior task being completed. So A must be completed before B, and B must be completed before C, and so on. In this case, you could have each person do one task, however, they would be waiting on each other to complete the prior task. So in essence one employee could complete the 8 tasks as quickly as 8 employees. And in reality, the one person could probably do it quicker since they would not need to wait on the prior task person to pass them the information. If the manager were inexperienced in this type of situation, he would probably assign it to one person and wait for that one person to be done.

Now take the example above, but you have a manager that has more experience with similar type taskings. He knows that within each task, there are subtasks that other individuals in the office could help with saving the main employee some time and effort to complete the tasks. This manager would then farm out the project to one employee, but also have other employees assist the one to complete all 8 tasks. The more experienced the manager, the more he can involved the other employees.

In the examples above the more experienced and effective the manager, the faster the 8 tasks sequential tasks can be completed. However, no matter how experienced the manager, he will not be able to drop the time it takes to 1/8 of the total time required. Through his experience, he may be able to drop it some more but will not hit that level because the tasks required are not conducive to farming them out to the 8 employees.

At the other end of the spectrum is a project that requires 8 tasks that can be accomplished at the same time. Each one is not dependent on the others but at the end the 8 results much be placed together in a package.

In this case, the manager can easily divide the tasks out among the 8 employees. Each employee completes their tasks and returns them to the manager who then compiles the results into the final package or tasks that out to one individual. Either way, the the tasks are completed in 1/8 the time plus the time to compile the results.

In the real world, there is combination of the two types of projects. The OS can only go so far. The way the application is written can greatly aide the OS in using the multiple cores that are available. Regardless, there is always overhead in both the app and OS to take advantage of multiple cores. So while it sounds neat on paper, the actual through put of multi-core systems can end up being somewhat disappointing at times.

Anyhow, my attempt at a layman's explanation. Granted it glosses over some issues but I think gives a decent explanation of why you won't see doubling of speed when you double the number of cores available.

jjahshik32
Apr 3, 2009, 06:13 PM
Maybe I can provide some insight as to why.

Taking advantage of multi-core systems requires overhead to manage the processes. The same is true with multi-CPU systems. This is why when you double the core amount you will not see a doubling of the speed across the board. Some applications are written to take advantage of multiple cores or are written where the OS can easily farm out the tasks to multiple cores.

To make an analogy, think of it in terms as the management of a small office. In this office, let's say that you have a manager (the OS) and 8 employees (the cores).

Say I have a project that requires 8 tasks (A through H). In this case, each one of these tasks is reliant upon the prior task being completed. So A must be completed before B, and B must be completed before C, and so on. In this case, you could have each person do one task, however, they would be waiting on each other to complete the prior task. So in essence one employee could complete the 8 tasks as quickly as 8 employees. And in reality, the one person could probably do it quicker since they would not need to wait on the prior task person to pass them the information. If the manager were inexperienced in this type of situation, he would probably assign it to one person and wait for that one person to be done.

Now take the example above, but you have a manager that has more experience with similar type taskings. He knows that within each task, there are subtasks that other individuals in the office could help with saving the main employee some time and effort to complete the tasks. This manager would then farm out the project to one employee, but also have other employees assist the one to complete all 8 tasks. The more experienced the manager, the more he can involved the other employees.

In the examples above the more experienced and effective the manager, the faster the 8 tasks sequential tasks can be completed. However, no matter how experienced the manager, he will not be able to drop the time it takes to 1/8 of the total time required. Through his experience, he may be able to drop it some more but will not hit that level because the tasks required are not conducive to farming them out to the 8 employees.

At the other end of the spectrum is a project that requires 8 tasks that can be accomplished at the same time. Each one is not dependent on the others but at the end the 8 results much be placed together in a package.

In this case, the manager can easily divide the tasks out among the 8 employees. Each employee completes their tasks and returns them to the manager who then compiles the results into the final package or tasks that out to one individual. Either way, the the tasks are completed in 1/8 the time plus the time to compile the results.

In the real world, there is combination of the two types of projects. The OS can only go so far. The way the application is written can greatly aide the OS in using the multiple cores that are available. Regardless, there is always overhead in both the app and OS to take advantage of multiple cores. So while it sounds neat on paper, the actual through put of multi-core systems can end up being somewhat disappointing at times.

Anyhow, my attempt at a layman's explanation. Granted it glosses over some issues but I think gives a decent explanation of why you won't see doubling of speed when you double the number of cores available.

Could you also say that on a 8 physical core machine it will have 8 virtual personal assistants (2 threads on each core) to aid each employee before putting it into a package? HEHE.

DrawingArt
Apr 3, 2009, 06:15 PM
I guess I'm just not comfortable paying that much money for RAM. Almost a 1k for RAM just listen to how that sounds!

If anything, I would much rather have a 2.66GHz OCTO with 6GB of RAM than a 2.93GHz QUAD with 32GB of RAM. :p

I agree 1k for 16gb ram is too much, absolutelly too much but I really think prices will drop in next 6 months like always do.

Hehe ... go for octo then I think you are sure you will use all 8-cores. I'll go with my small quad and I'm also sure it will be my friend for quite some time and I'm sure I won't use 100% of its power every day, maybee 10% of the whole time working on it. For that percent I'm willing to wait few seconds more.

v0n
Apr 3, 2009, 08:09 PM
Wasn't 4 core Nahalem supposed to max out at 8Gb?

Fomaphone
Apr 3, 2009, 08:18 PM
edit: baleted!

i misunderstood his question, thinking it referred to the chip itself

DrawingArt
Apr 4, 2009, 02:46 AM
Wasn't 4 core Nahalem supposed to max out at 8Gb?

No, 16gb for now, maybee even more.

VirtualRain
Apr 4, 2009, 03:19 AM
If anything, I would much rather have a 2.66GHz OCTO with 6GB of RAM than a 2.93GHz QUAD with 32GB of RAM. :p

Yeah, but the question is, would you rather have a 2.26GHz Octo with 6GB (or even 12GB) vs. a 2.93GHz Quad with 6 (or 12GB)?

I think these benchmarks prove that unless you are heavily reliant on the few professional apps that can really utilize that many cores, you are better off with the Quad.

Even if you are heavily reliant on the few pro apps that can use that many cores, it's clear from this review that you may have to spend time doing trial and error runs with your software to see what yields the biggest gains... who would have guessed the compressor runs best on the Octo with 6 threads?! That makes zero sense to me.

Finally, with the entry-level Octo do you realize that most of your day-to-day tasks are just chugging along at 2.26GHz? That's barely better than a low-end Macbook. 12GB or 8 cores aint going to help you load these forums any faster... for many things, clock speed is still king. (BTW there were at least a couple of people here who purchased the entry-Octo machines and returned them promptly because they felt sluggish for exactly this reason).

These are just a few of the main reasons, I selected a Quad. While I'm admittedly very biased, I think my reasoning is sound (at least to me LOL)... and I'm having real trouble understanding the reasoning for anyone to buy the entry level Octo for anything other than bragging rights. Anyone serious about multi-threaded performance is probably making money with their machine and is likely undaunted by the extra cost of the 2.66 or 2.93 Octo upgrades which would probably pay for themselves with just a couple of small jobs. :confused:

BTW, I wish people would stop hanging on to Snow Leopard as being the next best thing since sliced bread or multi-core processors. OS X has been SMP capable since inception... it's the applications that are lagging.

Fomaphone
Apr 4, 2009, 03:31 AM
...so i can blame my processors for the amount of time i waste here? :D

sorry, but what about SMP makes the gains coming in leopard negligible?

DrawingArt
Apr 4, 2009, 03:33 AM
Yeah, but the question is, would you rather have a 2.26GHz Octo with 6GB (or even 12GB) vs. a 2.93GHz Quad with 6 (or 12GB)?

I think these benchmarks prove that unless you are heavily reliant on the few professional apps that can really utilize that many cores, you are better off with the Quad.

Even if you are heavily reliant on the few pro apps that can use that many cores, it's clear from this review that you may have to spend time doing trial and error runs with your software to see what yields the biggest gains... who would have guessed the compressor runs best on the Octo with 6 threads?! That makes zero sense to me.

Finally, with the entry-level Octo do you realize that most of your day-to-day tasks are just chugging along at 2.26GHz? That's barely better than a low-end Macbook. 12GB or 8 cores aint going to help you load these forums any faster... for many things, clock speed is still king. (BTW there were at least a couple of people here who purchased the entry-Octo machines and returned them promptly because they felt sluggish for exactly this reason).

These are just a few of the main reasons, I selected a Quad. While I'm admittedly very biased, I think my reasoning is sound (at least to me LOL)... and I'm having real trouble understanding the reasoning for anyone to buy the entry level Octo for anything other than bragging rights. Anyone serious about multi-threaded performance is probably making money with their machine and is likely undaunted by the extra cost of the 2.66 or 2.93 Octo upgrades which would probably pay for themselves with just a couple of small jobs. :confused:

BTW, I wish people would stop hanging on to Snow Leopard as being the next best thing since sliced bread or multi-core processors. OS X has been SMP capable since inception... it's the applications that are lagging.

Nicelly said, I agree! In 14 days I'll be able to say how I'm satisfied with my 2.93 Quad.

MasterM6
Apr 4, 2009, 06:43 AM
My quad 2.93 will land before the 8th!!! :D

VirtualRain
Apr 4, 2009, 01:27 PM
sorry, but what about SMP makes the gains coming in leopard negligible?

What do you perceive as the gains coming in Snow Leopard?

From what I've seen, Snow Leopard will apparently make it easier to develop multi-threaded applications. It will also apparently include some tuning to the task schedular.

It's not going to turn single-threaded code into multi-threaded code, and it's not going to fix poorly coded multi-threaded apps.

Whatever benefits it brings, will be realized on everything from dual-cores to quad-cores to octa-cores.

Just don't expect it to unleash the power of your octo-core... that just isn't going to happen until the app developers do more to thread their apps and keep in mind that not all applications or tasks lend themselves to multi-threading.

cmaier
Apr 4, 2009, 01:32 PM
What do you perceive as the gains coming in Snow Leopard?

From what I've seen, Snow Leopard will apparently make it easier to develop multi-threaded applications. It will also apparently include some tuning to the task schedular.

It's not going to turn single-threaded code into multi-threaded code, and it's not going to fix poorly coded multi-threaded apps.

Whatever benefits it brings, will be realized on everything from dual-cores to quad-cores to octa-cores.

Just don't expect it to unleash the power of your octo-core... that just isn't going to happen until the app developers do more to thread their apps and keep in mind that not all applications or tasks lend themselves to multi-threading.

I would actually expect an across-the-board speed increase in all apps because the OS is now pretty much fully 64-bit, and thus can take advantage of extended register sets, etc. At AMD when we designed AMD64 (which Intel appropriated), we saw that even 32-bit code running under a 64-bit OS increased in speed by around 10-15%.

Plutonius
Apr 4, 2009, 03:51 PM
I would actually expect an across-the-board speed increase in all apps because the OS is now pretty much fully 64-bit, and thus can take advantage of extended register sets, etc. At AMD when we designed AMD64 (which Intel appropriated), we saw that even 32-bit code running under a 64-bit OS increased in speed by around 10-15%.

Like every new OS, the eye candy, new security, etc. will offset any of the performance increases they get.

cmaier
Apr 4, 2009, 04:14 PM
Like every new OS, the eye candy, new security, etc. will offset any of the performance increases they get.

Maybe. Apple isn't Microsoft. If anything, sl seems to be all about slinning things down and optimizing things. I'm willing to bet you see speed improvements across the board.

Tesselator
Apr 4, 2009, 04:33 PM
Yep! I think so too... But as you stated it'll be mainly due to the new 64-bit components. This will affect all Mac Pros relatively in the same percentage.

It's good news! I'm buying SL the day it hits! (like I did Leopard too.)

VirtualRain
Apr 4, 2009, 08:29 PM
Maybe. Apple isn't Microsoft. If anything, sl seems to be all about slinning things down and optimizing things. I'm willing to bet you see speed improvements across the board.

I agree... Hopefully folks don't misinterpret my comments about multi-threading and SL to imply there won't be any benefits... I'm just skeptical that Snow Leopard will single-handedly unleash the potential of the octo core platforms... IMHO it will not... so buying a 2.26 Octo in hopes that SL will suddenly transform it into a power-house machine on every-day apps, is bound to end in disappointment.

cmaier
Apr 4, 2009, 08:35 PM
I agree... Hopefully folks don't misinterpret my comments about multi-threading and SL to imply there won't be any benefits... I'm just skeptical that Snow Leopard will single-handedly unleash the potential of the octo core platforms... IMHO it will not... so buying a 2.26 Octo in hopes that SL will suddenly transform it into a power-house machine on every-day apps, is bound to end in disappointment.

I agree. I expect to see something like 10% speed improvement on existing apps, across the board. Perhaps more on some things that used to rely on 32-bit OS calls.

Multi-threaded apps I expect little additional improvement until developers deliver updates, and even then I don't expect to see massive improvements except maybe in highly parallelizable tasks like encoding, image processing, etc.

jjahshik32
Apr 4, 2009, 08:42 PM
This is the reason why I'm waiting for snow leopard to release before choosing a mac pro. I guess base it off which benefits most after some reviews. Especially to see if apple speed bumps anything in the mac pro line ups.