New Mac Pros slower than rMBP, new iMacs plus all 5.1 mac pros

[MattDSLR]

New Mac Pros slower than rMBP, new iMacs plus all 5.1 mac pros at equivalent speeds running on Sata2
Lets results speak for themselves :
http://www.macworld.com/article/208...ter-weve-been-waiting-for-finally.html?page=2

Finder tests
Mac model Copy 6GB Folder------ Compress6GB Folder -----Decompress 6GB Folder
Mac Pro 8-Core/3.0GHz (Late 2013) 23.46 324.8 39.10
Mac Pro 6-Core/2.4GHz (Mid 2012) 113.22 396.43 96.52
Mac Pro quad-core/3.2GHz (Mid 2012) 147.35 329.67 129.44
27-inch iMac quad-core/3.5GHz CTO (Late 2013) 40.89 241.25 33.62
15-inch Retina MacBook Pro quad-core/2.3GHz
(Late 2013) 21.42 272.67 33.01
Results are times in seconds. Lower results are better. Best result in bold. Reference models in italics.

iMovie 10.0.1 enhance/export
Mac model Result (seconds)
Mac Pro 8-Core/3.0GHz (Late 2013) 170
Mac Pro 6-Core/2.4GHz (Mid 2012) 225.00
Mac Pro quad-core/3.2GHz (Mid 2012) 290.33
27-inch iMac quad-core/3.5GHz CTO (Late 2013) 47.33
15-inch Retina MacBook Pro quad-core/2.3GHz (Late 2013) 63.00
Lower results are better. Best result in bold. Reference models in italics.


Aperture 3.5.1 import/process
Mac model Results (seconds)
Mac Pro 8-Core/3.0GHz (Late 2013) 72
Mac Pro 6-Core/2.4GHz (Mid 2012) 127.00
Mac Pro quad-core/3.2GHz (Mid 2012) 112.00
27-inch iMac quad-core/3.5GHz CTO (Late 2013) 70.67
15-inch Retina MacBook Pro quad-core/2.3GHz (Late 2013) 75.67
Lower results are better. Best result in bold. Reference models in italics.

 

[Bazu]

In 4k video or video in overal nMP shines. Also in engineering apps.
Problem is. Not everyone in the market for Mac Pro is engineer nor video editor.

 

[MattDSLR]

if you notice they use the slowest possible 5.1 mac pros to show how well it performs but since iMac and rMBP can beat it we know equivalent speed and core 5.1 with out SSD can crush it
I love early adopters of nMP mini.

Here comes CUBE fiasco again.

 

[d-m-a-x]

i am going to ignore the i movie test, who knows what that program is optimized for

 

[MattDSLR]

In 4k video or video in overal nMP shines. Also in engineering apps.
Problem is. Not everyone in the market for Mac Pro is engineer nor video editor.

It's slower in photo processing and iTunes conversion and iMovie conversion
well its not much fast in Geekbench

and some of the 4k adopters will find huge limitation of external drives and them being limited to just below sate3 running raid
it will run good as long you do not have to export to external drive and than will be much slower than the computers with larger internal capacity

 

[ZnU]

There's no reason to expect the Mac Pro to be vastly faster at single-core tasks... or at I/O-bound tasks when up against other machines with PCIe flash (which the 2013 iMacs have). Look at tasks that run well across arbitrarily large numbers of cores and/or support OpenCL and dual GPUs, and you'll see very different results.

From that article, for instance, FCP X import (which involves a fair bit of processing as FCP X does analysis of imported clips by default) is 2.9x as fast as the iMac and export is 1.8x as fast. Cinebench is 1.6x as fast, MathematicaMark is 2.2x as fast, and overall Geekbench score is 1.8x as fast.

The above are all falling in the expected range for well-threaded CPU bound tasks. I haven't seen benchmarks posted yet, but for heavy OpenCL compute as per the spec sheets you're looking at 2.3 teraflops on the 780M BTO iMac option vs. a total of 7 teraflops across two D700 GPUs, 3x the performance. And three times the graphics memory, which quickly becomes important at higher resolutions (e.g. 4K).

 

[MattDSLR]

i am going to ignore the i movie test, who knows what that program is optimized for

I can give you that one

I want to see a 4hrs 4k video compression with running another app in the back ground like photoshop and burning it to BR

I do not think that cheese shredder on the top (i think they call it a new cooling tech) keep up with all the heat dissipation

 

[Joshua M]

My favorite part is where you cherry-picked 3 tests out of the 20 they ran, the vast majority of which the Mac Pro dominated. The remainder of which were fractionally slower, and likely only using one core, which just happens to run 500MHz slower than the other machine.

EDIT: I mean, the article is literally titled: "Lab Tested: New Mac Pro is the speedster we've been waiting for (finally)"

 

[Bazu]

The above are all falling in the expected range for well-threaded CPU bound tasks. I haven't seen benchmarks posted yet, but for heavy OpenCL compute as per the spec sheets you're looking at 2.3 teraflops on the 780M BTO iMac option vs. a total of 7 teraflops across two D700 GPUs, 3x the performance. And three times the graphics memory, which quickly becomes important at higher resolutions (e.g. 4K).

Everyone mention how fast it potentially is in OpenCL but as for now fact is in OpenGL it is really disappointing.

 

[VirtualRain]

New Mac Pros slower than rMBP, new iMacs

Those are clearly lightly-threaded or single-core tasks where Haswell is expected to perform better than Ivy Bridge. It says more about the strength of Haswell than any weakness in IB.

However, the moral of this story is probably that if your primary apps are iLife or iWork, the nMP is not your best bang-for-the-buck. But I doubt there's too many people here who are surprised by that.

plus all 5.1 mac pros

Huh? I didn't see a single test where the nMP was slower than the 5,1. In fact, the nMP spanks the 5,1 in most/all of these benchmarks. You should probably change the title of your thread, it's extremely misleading.

 

[mrsavage1]

My favorite part is where you cherry-picked 3 tests out of the 20 they ran, the vast majority of which the Mac Pro dominated. The remainder of which were fractionally slower, and likely only using one core, which just happens to run 500MHz slower than the other machine.

Yeah i loved that too. I was like loling

 

[ybz90]

if you notice they use the slowest possible 5.1 mac pros to show how well it performs but since iMac and rMBP can beat it we know equivalent speed and core 5.1 with out SSD can crush it
I love early adopters of nMP mini.

Here comes CUBE fiasco again.

As I always say, numbers never lie, but interpretations do. Just another example of people getting the incorrect conclusions from the data, specifically assigning significance without context to the meaningfulness of the test.

 

[goMac]

Really? C'mon, this isn't rocket science here.

These are single core tests.

The 8 core Mac Pro is 3.0 ghz.

The 4 core iMac is 3.5 ghz.

Which do you think is going to win in a single core test?

Which do you think is going to win in a multi core test?

 

[MattDSLR]

There's no reason to expect the Mac Pro to be vastly faster at single-core tasks... or at I/O-bound tasks when up against other machines with PCIe flash (which the 2013 iMacs have). Look at tasks that run well across arbitrarily large numbers of cores and/or support OpenCL and dual GPUs, and you'll see very different results.

From that article, for instance, FCP X import (which involves a fairly bit of processing as FCP X does analysis of imported clips by default) is 2.9x as fast as the iMac and export is 1.8x as fast. Cinebench is 1.6x as fast, MathematicaMark is 2.2x as fast, and overall Geekbench score is 1.8x as fast.

The above are all falling in the expected range for well-threaded CPU bound tasks. I haven't seen benchmarks posted yet, but for heavy OpenCL compute as per the spec sheets you're looking at 2.3 teraflops on the 780M BTO iMac option vs. a total of 7 teraflops across two D700 GPUs, 3x the performance. And three times the graphics memory, which quickly becomes important at higher resolutions (e.g. 4K).

You compering 8 cores to 4
or 6 cores at 25% less speed to 8 core
cmon
where is the true test
add 25% more processor speed and 2 extra cores and 5.1 would be 20% faster
lets compare apples to apples
and Like I sad before 2 minutes import is faster to any solid state drive but import 1 TB of 4K video in that new mac pro and lets see what will be faster the processor speed itself its slower than 5.1 and the limitation or the lack of of internal storage for large pro file processing will be the end of nMP mini

 

[Bazu]

Really? C'mon, this isn't rocket science here.
These are single core tests.
The 8 core Mac Pro is 3.0 ghz.
The 4 core iMac is 3.5 ghz.
Which do you think is going to win in a single core test?
Which do you think is going to win in a multi core test?

Leave the cores aside.
What you say about their OpenGL tests?
Considering they've tested the highest and of nMP GPU

 

[MattDSLR]

Really? C'mon, this isn't rocket science here.

These are single core tests.

The 8 core Mac Pro is 3.0 ghz.

The 4 core iMac is 3.5 ghz.

Which do you think is going to win in a single core test?

Which do you think is going to win in a multi core test?

Since when Aperture is single core?????

Also nMP mini is using:
state of the art SSD (that is 3 times faster than the drives in iMacs
DDR memory that is twice as fast

so if you buy one don't look and think why this computer is slower handling large amounts of data

I'm just opening all your eyes to the true facts and limitations of the new Mini

 

[goMac]

Leave the cores aside.
What you say about their OpenGL tests?
Considering they've tested the highest and of nMP GPU

They need to run the tests in Boot Camp.

The 79X0 drivers have issues in Mavericks. OS problems need to be eliminated from the benchmarks.

Even then, with two cards available, it's still not a great comparison.

 

[VirtualRain]

Leave the cores aside.
What you say about their OpenGL tests?
Considering they've tested the highest and of nMP GPU

There's definitely something strange going on with the Cinebench R15 OpenGL test... the nMP stomps the other platforms on the same benchmark using the CPU only, but loses on GPU? Makes no sense whatsoever. I suspect Cinebench or OS X needs an update to fix something there.

 

[goMac]

Since when Aperture is single core?????

File copies are single threaded. How exactly do you plan to multithread a file copy?

The other thing left out is how big the SSD is. The larger SSDs on the Mac Pro are faster. If the late 2013 iMac and the Mac Pro have the same SSD, you'd expect the file copies to be the same speed, which looks like what's going on.

 

[Bazu]

There's definitely something strange going on with the Cinebench R15 OpenGL test... the nMP stomps the other platforms on the same benchmark using the CPU only, but loses on GPU? Makes no sense whatsoever. I suspect Cinebench or OS X needs an update to fix something there.

But it's not only Cinebench. Unigine isn't shining also.
CNET test for games gave results below iMac and still those test used the highest end GPU.

W9000 (card which D700 suppose came from) tramps consumer cards in 90% of tests by TomsHardware. So argument that pro cards can't run games is quite invalid. All 3 W9000, W8000, W7000 performed accordingly to their consumer equivalent and on some tests even better.

But then you see single W9000 cost over 2000$ and you wonder how apple squished two of them for price of one. It simply doesn't add up.

 

[MattDSLR]

Leave the cores aside.
What you say about their OpenGL tests?
Considering they've tested the highest and of nMP GPU

Cinebench R15 OpenGL
Mac model Results (score)
Mac Pro 8-Core/3.0GHz (Late 2013) 87.4
Mac Pro 6-Core/2.4GHz (Mid 2012) 53.45
Mac Pro quad-core/3.2GHz (Mid 2012) 58.58
27-inch iMac quad-core/3.5GHz CTO (Late 2013) 90.17
15-inch Retina MacBook Pro quad-core/2.3GHz
(Late 2013) 53.80
Results are frames per second. Higher results are better. Best result in bold. Reference models in italics.
Old mac pros have 4 year old video cards in them
just add $500.00 HD 7950 Mac Edition video card and lets see how close both computers will be

oh here for all those 4K video editors thats how your desk will look if you work with footage longer than 2 hrs

 

[goMac]

just add $500.00 HD 7950 Mac Edition video card and lets see how close both computers will be

As I mentioned upthread, the 7950 on a 2012 Mac Pro is having the same problems. It's a driver thing.

http://www.tonymacx86.com/graphics/113286-bad-graphics-performance-mavericks-sapphire-7950-cf.html

 

[MattDSLR]

File copies are single threaded. How exactly do you plan to multithread a file copy?

The other thing left out is how big the SSD is. The larger SSDs on the Mac Pro are faster. If the late 2013 iMac and the Mac Pro have the same SSD, you'd expect the file copies to be the same speed, which looks like what's going on.

Did you ever import more than one card or copy information from more than one drive??

That will eat up your cores

 

[Bazu]

Old mac pros have 4 year old video cards in them
just add $500.00 HD 7950 Mac Edition video card and lets see how close both computers will be

Exactly my point. For the price those nMP GPU's are really disappointing. Especially when compared fairly to their supposed equivalents from other FirePro's which performs very well.

 

[ZnU]

oh here for all those 4K video editors thats how your desk will look if you work with footage longer than 2 hrs

My desk (or, well, the desk in our color grading suite) will contain the Mac Pro, three video monitors (up-close video reference monitor for spotting stuff too hard to see on the projector, video scopes display, GUI monitor), a keyboard, mouse, and my Tangent Element color grading panel... and some Thunderbolt cables running to the equipment rack under the desk.