New Mac Pro arrived, speed concerns

greyhoundgames said:

I just got my new MacPro(basic configuration) and I am concerned because the various performance tests I have run on it show little improvement compared to the 2012 macbook pro I upgraded from. I was expecting a monster desktop mac to handily beat a 2012 laptop. Here is the tests I ran and the results

#1 Blackmagic hard disk speed test app from the App Store
#2 Geekbench downloaded from the web
#3 Real life test building my Xcode project which takes 1 minute on the macbook 2012 laptop

Results
#1 As expected, the Mac Pro hard disk is 2x as fast as the macbook hard disk.
#2 Geekbench test results had only a 8% gain on single core tasks and even less gain on multi core tasks
#3 The build time of 1 minute only changed by a few seconds for the mac pro.

These results are pretty pathetic. Does anyone know what to think of it? The only software difference is my laptop is not running mavericks.

Click to expand...

Why did you even think there would be a big speed increase compared to a 4 core MBP? The single core speed is more or less the same on both chips.

What were your MBP config specs?

Indeed, it is a bit of a disappointment. I played with the nMP for the first time today (there's one in at Yoobee in Riccarton Mall, chch NZ). Lovely looking. Ordinary performance. Not a significant step up from a recent MBP or iMac in most respects, unless (key indicator) you're doing heavy video processing with an app that can use the GPUs. Needless to say, no apps were loaded that were written for this. So the machine itself is deeply underwhelming, particularly for the price.

If you want a sexy video processing beast for certain apps that can do it, fine. If you want a much faster computer, this isn't it.

Maybe the lesson here is that upgrading a computer that is less than 2 or 3 years old probably isn't going to gain much in performance on "ordinary" tasks...this, I believe, speaks highly of the 2 or 3 year old computer than poorly of the new machines. That build time of 1 minute on your MBP probably would have taken 10 or 20 times that 5 or 6 years ago.

Where you might see the biggest gain with the xeon chip and nMP in general is when you start doing stuff that takes HOURS to do. You shouldn't notice the throttling that might take place on the "lesser" MBP.

I remember several years ago (6 or 7) when I was in charge of a project that involved a video server. We had been using simple file sharing and a consumer Pentium based PC. We decided to spend the money and purchase an actual server....complete with dual xeon chips, Hardware raid with hotswap backplane, dual NIC, dual PS, the whole nine yards. Not sure what I expected, but when it came in, I was SO UNDERWHELMED by what it "could do." But I set it up, and put it online. Within hours I figured out what it really could do! What it could do was pump out 10 plus streams of HD video over ethernet at once, without missing a beat...and still asking for more! The thing is a workhorse. Will it open solitaire any faster than a dual core pentium machine? NOPE. Does that matter? NOPE.

Edit to add this: You have two awesome computers! Both of which BLOW away anything that as available a few years ago. They each have their purpose, and you WILL find things you enjoy doing more on each...and each will have strengths and weaknesses. I can't wait to get mine!

The single core score on the 3.7 4-core nMP is 3610. On the late 2013 2.6 rMBP running at 2.6 ghz its 3710 but keep in mind Im pretty sure that geekbench is measuring the full speed boost potential of the CPU which is 3.8 GHZ so those results make sense.

If you really want overall CPU performance thats significantly higher then the latest quad iMacs and 15" rMBP the 6 core scores about 20500 on multicore which is pretty good..

What I do find interesting is the single core 3ghz 8 core nMP is a few points higher then the single core score 3.5 ghhz on the 6 core model which makes no sense.

With the nMP your paying for an industrial workstation design thats more upgradable then an iMac or macbook.

I used this to get a good idea of the raw CPU speeds.

http://browser.primatelabs.com/mac-benchmarks

If the i7's turbo boost is 3.8ghz this is expected. But it won't be able to provide that performance for long before it has to throttle.

nMP is for heavy lifting. If ones demands are 5 minute or less tasks expect to see little if any speed improvement (quad nMP vs any modern mac quad). Do 50% load for hours on end - big difference in thermal design. 100% load for an hour or a day - huge difference.

just an example of why the nmp has dual gpu/single cpu..

it's not like apple could have just put a faster cpu in there which blows away the laptop cpus.. the laptops have caught up and the cpu speed advancements have moreORless topped out (or drastically slowed in their progression)..

adding more cores was a possibility but it's just way too expensive for most people and not very many apps (relatively speaking) can take advantage of it anyway..

so instead, the actual design/arrangement of components has been refined (finally after 30 years or whatever) and the computer is equipped with lots of potential power in the gpus.. that power just needs unlocked by the software developers now.

early adopters aren't going to be able to immediately utilize all that potential power but they will within the lifespan of the computer.. until then, yes- don't expect a nmp (or any 'pro' machine for that matter) to preform leaps&bounds above more modest hardware/laptops/imacs in a big percentage of tasks (in certain more specialized tasks however, the 6+ core machines will definitely be outperforming the mbp/imac/etc.. even the quad mp will best your mbp, by far, in things like cpu based raytracing)

The CPU on the Mac Pro isn't expected to be that much faster for Xcode.

All other things even (which they mostly are), core count is the primary attribute that would speed up XCode. Because your Mac Pro is a four core machine, I wouldn't expect to see much of a speed up from your MacBook Pro.

If you want to see Xcode speed gains from a Mac Pro the 6, 8, and 12 core config are where it's at.

Mine arrived today - 8 core, 3GHz, 32GB RAM, 512GB SSD, Dual 700's.

I am just playing around while I get all my latest software versions downloaded and installed. I have a 2011 MBP. This is the first "desktop" Mac Pro I've purchased since the PowerMac G5 Quad that I bought years ago, mainly because I have access to the 2010/2012 Mac Pro's at work.

I'm rendering a Mandelbulber animation in the background at 1920x1080, and it's ripping through the frames insanely fast, likely due to the enormous amount of GPU memory available. Can't give an exact spec, but basically the MBPro probably would have just given up and gone home.

Got a new Lacie Thunderbolt to eSATA adapter so I could put my old OWC QX2's into play. I'm getting 215MB/sec writes and 230 reads, very respectable. I'm a little bummed that the Lacie adapter doesn't support port multiplication. I may take it back and try to find something else, but it's not the end of the world as I have a 24TB Promise2 en route.

Write speeds to the internal SSD with AJA System test were about 1GB/sec write and 1.15GB/sec read. That's pretty darn fast.

Once I get all the Adobe, FCP, Avid stuff installed, as well as Resolve (don't even know if Resolve works with these GPU's yet?), I'll start getting a feel for it.

I expected this machine to feel about the same as a 2012 Mac Pro in normal daily tasks. When I get into multi-threaded tasks that use a lot of GPU memory, then I expect this little guy to kick some serious a$$.

When did we ever get to own a Mac with TWELVE GB of GPU memory? That's a huge leap on the Mac side. All my 4K rendering issues are going to go away. If it wasn't for this machine, us video people would all be switching to Windows right now. Many already have, thanks to Apple's poor treatment of the professional market.

Here's the biggie to me though -- and this is not a small thing -- the MP2013 is sitting on my coffee table, taking up about as much space as a big mug of coffee (ok, a really big mug of coffee), completely obscured behind a 24" monitor, making no noise, while CPU usage is at 1450%, just a gentle warm breeze rising from the top. I'm browsing the web, download software in Safari, rendering in Mandelbulber, hitting the disk with AJA speed test -- feels like the computer isn't even in use.

It's so quiet it's hard to believe it's on. And the lights in the house aren't flickering.

I think I'm really happy!

It's just benchmarks ... try all-day use and you'll see the difference. It's been said all over this forums before, benchmarks are just a mere quick full load on cpu/gpu/ssd. The real difference is when you'll have projects you work on for hours and hours a day, performance will be stable whereas non-workstations will drop.

And you're not blown away with diskspeeds 2x as fast?? wauw.

greyhoundgames said:

...#1 As expected, the Mac Pro hard disk is 2x as fast as the macbook hard disk...
...These results are pretty pathetic. Does anyone know what to think of it?...

Click to expand...

I can understand your other points, but this one's a bit strange. Are you saying that since the improvement is expected, it's not enough because you weren't surprised by it - as in secretly hoping that it would perform better than all those units already reviewed?

I replaced a massively upgraded (CPUs, RAM, and SSD) but tired MP 1,1 with a stock 6 core nMP. The differences are more evident.

Like the previous posters have said, the single core performance is on par with other recent Mac models. The real difference is in using multi-core or GPU aware applications.

The old MP 1,1 would take 2 days to render a complex scene at large image sizes in LuxRender to a satisfactory level. The nMP crunched a far more complex scene using the GPU's in 12 hours. The nMP was quite warm, but still inaudible.

For 3D rendering work the nMP is quite impressive.

propower said:

nMP is for heavy lifting. If ones demands are 5 minute or less tasks expect to see little if any speed improvement (quad nMP vs any modern mac quad). Do 50% load for hours on end - big difference in thermal design. 100% load for an hour or a day - huge difference.

Click to expand...

Indeed. The performance gains single core have been successively smallish since the big boost the Sandy Bridge chipped Macs had in 2011 from 2010 in the other models bar the Mac Pro towers.

greyhoundgames said:

I just got my new MacPro(basic configuration) and I am concerned because the various performance tests I have run on it show little improvement compared to the 2012 macbook pro I upgraded from. I was expecting a monster desktop mac to handily beat a 2012 laptop. Here is the tests I ran and the results

#1 Blackmagic hard disk speed test app from the App Store
#2 Geekbench downloaded from the web
#3 Real life test building my Xcode project which takes 1 minute on the macbook 2012 laptop

Results
#1 As expected, the Mac Pro hard disk is 2x as fast as the macbook hard disk.
#2 Geekbench test results had only a 8% gain on single core tasks and even less gain on multi core tasks
#3 The build time of 1 minute only changed by a few seconds for the mac pro.

These results are pretty pathetic. Does anyone know what to think of it? The only software difference is my laptop is not running mavericks.

Click to expand...

You would need to go > 4 core to get some advantage from the mac pro for xcode. Xcode scales with cores. The good news is you can always put in a better cpu yourself.

greyhoundgames said:

Perhaps I don't know enough about the differences between xeon and whats in the macbook pro. The cpu ghz difference is quite huge but maybe thats just a marketing trick. If so shame on me for not doing my homework. I also thought the memory speed increase would help some too as well as the hard disk which I am thinking now only helps when copying massive files and not day to day project activities.

Click to expand...

Clock speeds can make a difference. You just have to realize they list base clocks, and the listed data may not reflect real world performance. As you mention, they use marketing tricks. I really wish people would stop being so hung up over the term Xeon. The base mac pro uses an E5-1620 v2. There is an equivalent i7 cpu that costs roughly the same amount. Both are around $300 list, just like the cpu in your notebook. Small increases in memory speed don't make a noticeable difference in most tasks. In my experience the amount of memory is far more significant. I have an SSD, and being able to hold things in ram is still much faster.

For one, a benchmark like Geekbench does not take in account the GPU. In software that can use multicore...multiple GPU's is where your going to see the most performance.

Not sure if xcode can make the most use of this type of computer or not.

The only way you could really be disappointed is if you had unrealistic expectations and a workflow that doesn't need a Mac Pro.

greyhoundgames said:

The disappointment was despite the speed increase, actual real life tasks didn't get any better.

Click to expand...

I think this is your misconception right here. Aside from the GPUs, that config of Mac Pro is only a small speed increase over your Macbook Pro. So the corresponding speedup you're seeing in things like Xcode is small.

To me, the speed increase matches exactly the real world speed increase here. There just isn't much of either.