C2D vs. ULV SB Aperture Performance

[unclet]

Let me just preface this by saying that I've read most of the discussions on these forums about whether a Sandy Bridge update would be better or worse than the current configuration of the MBA. The reason I am adding yet another thread to the debate is that I wanted to see if anyone can comment on the importance of the GPU to photo editing, specifically in Aperture.

I sold my 2009 15" MBP back in December in order to offset most of the cost of my 13" Ultimate MBA. At the time I was worried about two things, the reduction in screen size and the slower processor. Since the only work that I do that is somewhat processor intensive is photo editing, I was willing to make the sacrifice for the greater portability of the MBA. Most of my photo editing is done in Aperture using NIK software plugins like Color Efex Pro, Silver Efex Pro, and HDR Efex Pro. To my surprise, my new computer performed just as well, if not slightly better in Aperture than my MBP with a more powerful C2D processor. I had thought that the processor downgrade would significantly affect performance, but it didn't at all. I am wondering if this is mainly because of the SSD, or if the 320M comes into play here. I know that Silver Efex Pro 2.0 has been designed to make more use of the GPU, but I don't know how the other NIK plugins or Aperture itself utilize the GPU.

The reason I am asking is because with all this talk of ULV SB updates to the MBA, I am curious as to whether the much better CPU would improve my photo editing experience even more, or is there a chance that trading the 320M for the HD 3000 could actually have a negative effect. In other words, how important is the GPU to photo editing?

For me personally, it would take a pretty great leap in performance to justify giving up the amazing machine I have right now. It is hands down the best computer I have ever owned. But I thought this discussion might be useful to other people thinking about buying now or waiting for the update, since most of the discussion so far has been about the GPU's effect on gaming performance.

 

[macaddict3]

me too, I have been thinking of this if I should trade my mba for a sandy bridge for the faster processor but slower gpu intel HD3000 no idea too..

 

[nebulos]

i'm no expert on GPUs, but i did a bit of research on them recently and, as far as i can tell, GPU shouldn't matter too much for photo editing. but, let's see what others have to say.

which 15" did you have? i'm assuming you didn't have the 2.53 with the integrated only 9400M.

SSD makes things like opening big images very fast, for example. there are some videos online of the 11" opening like 20 or so big files in a fraction of the time it takes, i believe, a stock 2011 MBP. it's all drive speed.

i don't think photo editing is too demanding on the CPU either, but i mean stuff like typical photoshop work; i don't know anything about the programs you mentioned.

from what i've read, i would guess that either machine is fine for the vast majority of photo editing.

alternatively, did you have one of the 15"s with 2GB RAM? and do you have 4GB now? i do believe RAM is something that affects photo editing noticeably.

 

[bcaslis]

Actually, Aperture makes significant use of the GPU for the photo display, zooming, panning, etc... How much difference there is will for a new MBA with SB and IGP is really impossible to know without one to test.

 

[unclet]

Thanks for the replies, everyone. I thought my post was doomed to obscurity. Actually, I did have the 15" MBP with a 2.53 Ghz processor, 9400M integrated GPU, and 4GB RAM. The 320M in the MBA is a real improvement over that. I did not know how much Aperture uses the GPU. That could make a real difference in the performance of the current MBA and it's possible update. I guess we really won't know until we see what the update is, but I'm starting to think I might just hang on to the one I have. It's a beauty of a machine as it is.

 

[kcdude]

It will be nice to see this post continued when the new units get here. I am actually considering moving from a woefully underused 2008 MP setup to a MBP or MBA w/monitor&ext drive setup. It helps to see others doing Aperture work on a MBA...I just can see myself going for a MBP with the fan noise etc. For my websurfing and photowork I'm beginning to warm up to the idea of a MBA (wishing it had more RAM though)...now if I could only work in a Thunderbolt external I would be set.

 

[unclet]

Yeah, I probably jumped the gun on this. Instead of speculating about what might be it would make more sense to look at real-world results. That being said, it is useful to hear what others think.

I hadn't thought about Thunderbolt, but that would be extremely useful in working with all the photos I've got stored on external drives. Of course, we'll need Thunderbolt hard drives before that will really be beneficial.

 

[nebulos]

of course, we won't know for sure till they come out, but if you want to get an idea, the new samsung series 9 has the slowest available ULV Sandy Bridge CPU and integrated HD3000. assuming the new airs will indeed move to Sandy Bridge, this gives you a baseline to compare with, roughly.

its very common for reviewers to explicitly compare with the current gen Air(s).

check out the reviews. there should be some that do some photoshop, and maybe aperture* tests.

the GPU tests will be on games, and the current Air, with 320M, seems to obliterate the Samsung. however, this should be taken with a grain of salt, as performance differences in something like Aperture, may be much less dramatic. experts?

******************************************

* EDIT: Goops! Obviously, I did not realize Aperture was OSX only! As I had said previously: "i don't know anything about the programs you mentioned." I hate when people talk when they don't know and here I was doing it myself. Sorry.

 

[fyrefly]

the GPU tests will be on games, and the current Air, with 320M, seems to obliterate the Samsung. however, this should be taken with a grain of salt, as performance differences in something like Aperture, may be much less dramatic. experts?

Well, I'm no expert, but I think the people at Macworld seem to know at least something about what they're doing...

And everything I've read says the LV HD 3000 IGP comes about on par with the 9400m.

So when you look at Macworld's comparison of the 2010 Mac Mini with 320m vs the previous gen with 9400m, we can see a bit of what we might be looking at graphics-wise with the IGP HD 3000 in a Sandy Bridge LV Chip.

In Aperture, they "timed the import and thumbnail/preview creation time for 150 photos."

The 320m-equipped Mac Mini 2.4Ghz did the test in 3m13s.

The 9400m-equipped Mac Mini 2.53Ghz did the test in 3m28s.

It's notable that the 2.26Ghz Mac Mini with 9400m did the test in 3m55s, suggesting that the test is more processor bound, then GPU-bound - otherwise the 2.26Ghz and 2.53Ghz should have had the same scores.

If Aperture is CPU-bound, but also aided by the Graphics card - then I'd say the Aperture performance should be about on par with the current MBA in a Sandy Bridge machine.

But again, I'm no expert and I'm just extrapolating based on the data.

 

[nebulos]

good reference.

minis tested: (including RAM)

2.4GHz, 2GB, 320M

2.53GHz, 4GB, 9400M

2.26GHz, 2GB, 9400M

in the aperture import test (whatever that is), the 2.4 was 12% faster than the 2.53 and 18% faster than the 2.26.

with such a mix of specs, i don't know if its clear exactly what that means, other than that GPU seems to be doing something significant. (for example, it could be GPU >> RAM >> CPU.)

there are tests out there on the 2010 and 2011 13" MBPs that seem to say that the 3000 HD coupled with a powerful CPU is comparable or better (in OSX) to a C2D + 320M. the CPUs in the new airs will be ULV/LV, so, perhaps not enough to help the 3000 HD keep up with the 320M, but maybe enough to surpass the 9400M.

but, yeah, might as well wait and see. ... i'm just bored waiting for these things to come out.

 

[unclet]

but, yeah, might as well wait and see. ... i'm just bored waiting for these things to come out.

I'm with you there.

Thanks for the Macworld link. It seems to suggest that the GPU does play a role in Aperture, and can make a slower processor like the 2.4 outperform a 2.53. However, if you look at the iMac Benchmarks, the 3.06 iMac with the 9400M significantly outperforms the Mac Mini with the 320M in the Aperture test. The main difference there is that the iMac has 4GB RAM, while the Mac Mini only has 2, but I can't imagine that the RAM is being maxed out in these tests, so that shouldn't really make a difference in the speed, right?

As I said in my original post, my 2.13/320M MBA seems to outperform my old 2.53/9400M MBP in Aperture, although I don't have any benchmarks to back it up, and it is probably due more to the SSD than anything.

It would be interesting if there were tests that did more than test import speed. I'm more interested in responsiveness to adjustments and rendering speed, and I suspect that the GPU plays more of a role in these areas than it does in importing. Is that correct?

 

[neteng101]

I would think if the LV CPU in SB does provide a significant jump in performance, it will be very useful for Aperture when processing RAW files, but since the GPU matters too, a weak GPU could be a bad thing.

I believe the LV CPUs aren't quite as capable as the SB GPUs in the MBPs which aren't low voltage chips, so it could crimp things. Of course, we all still don't quite know what the SB MBA will look like yet either.

 

[bcaslis]

Aperture is both a memory and disk hog. I don't think the Macworld tests are very useful. Importing isn't going to use the GPU. If they had performed graphics related tests such as using different filters, zooming, and panning, etc... it would have been more relevant.

 

[nebulos]

as for RAM in the tests:

the 3.06 iMac with the 9400M significantly outperforms the Mac Mini with the 320M in the Aperture test. The main difference there is that the iMac has 4GB RAM, while the Mac Mini only has 2 ...

see above:

2.53GHz, 4GB, 9400M

as for GPU in the tests:

It would be interesting if there were tests that did more than test import speed. I'm more interested in responsiveness to adjustments and rendering speed, and I suspect that the GPU plays more of a role in these areas than it does in importing. Is that correct?

Aperture is both a memory and disk hog. I don't think the Macworld tests are very useful. Importing isn't going to use the GPU. If they had performed graphics related tests such as using different filters, zooming, and panning, etc... it would have been more relevant.

i know it sounds silly to say GPU matters when importing; i mean, like i said, i dont know much about GPUs, but i would have guessed it does NOTHING here. the GPU is just, seemingly, the only thing the 2.4 had on the 2.53 in the tests. there should be some explanation; maybe they moved to slightly better/faster HDDs that year?



... anyways, i should stop going on about stuff i don't understand.

 

[spacepower7]

IIRC the 9400M only uses 256MB of ram while the 320M uses 384MB of ram, meaning 50% more ram which will definitely improve aperture performance. I also believe that the SSD has a major impact when importing and especially scroll through albums.

The graphic card probably helps for working on a single image, but copying and applying edits (white balance etc....) from one image to multiple, probably benefits more from the processor and SSD rather than the graphics card.

After years of dramatic processor improvement, I think Apertures reliance on a graphics card is not as beneficial (necessary) in comparison to other Apple apps such as Motion and Color.

 

[bcaslis]

Their test doesn't have enough data to judge fully. It could easily be hard drive differences. Anyone who uses Aperture knows that using import speed is the silliest measurement, it's something that is easy to do though. The real bottlenecks are the image processing and adjustments. That's where the GPU would be used.

 

[fyrefly]

As I said in my original post, my 2.13/320M MBA seems to outperform my old 2.53/9400M MBP in Aperture, although I don't have any benchmarks to back it up, and it is probably due more to the SSD than anything.

I think you nailed it there. The SSD does wonders. I have a MBP equipped with an SSD, and it's just as fast as my MBA (faster, obviously on CPU-dependent tasks).

It would be interesting if there were tests that did more than test import speed. I'm more interested in responsiveness to adjustments and rendering speed, and I suspect that the GPU plays more of a role in these areas than it does in importing. Is that correct?

I'm sure that the GPU does play more of a role here, but (and I'm only using my experience in FCP with this response), the CPU will play a huge role in any sort of "rendering". That's almost purely CPU dependent, is it not? In that regard, I bet the LV i5/i7 chips will spank the C2D.

Their test doesn't have enough data to judge fully. It could easily be hard drive differences. Anyone who uses Aperture knows that using import speed is the silliest measurement, it's something that is easy to do though. The real bottlenecks are the image processing and adjustments. That's where the GPU would be used.

You're right. It was mostly meant as a jumping off point, as there's no ULV/LV Sandy Bridge Machines out right now that can run Aperture.

But I'd also argue that your "image processing" comment is similar to the one above, where I'd say it's moreso bound to the CPU to process through those changes.

Sure, modern day software like Photoshop and Aperture offload to the GPU somewhat, but when the processor is speedy like the Sandy Bridge chips (especially if they are allowed to turbo boost up to ~2.7Ghz) then the beefy CPU will surely offset most/all of the GPU losses in processing/rendering stuff?

 

[bcaslis]

...
You're right. It was mostly meant as a jumping off point, as there's no ULV/LV Sandy Bridge Machines out right now that can run Aperture.

But I'd also argue that your "image processing" comment is similar to the one above, where I'd say it's moreso bound to the CPU to process through those changes.

Sure, modern day software like Photoshop and Aperture offload to the GPU somewhat, but when the processor is speedy like the Sandy Bridge chips (especially if they are allowed to turbo boost up to ~2.7Ghz) then the beefy CPU will surely offset most/all of the GPU losses in processing/rendering stuff?

I don't disagree with you for the most part. But Aperture made a big deal of how they are utilizing the GPU beyond what you would normally expect back when the software was released. I've seen this myself on various Macs, the GPU can make a much bigger difference with Aperture than expected. However the SB CPUs could potentially overcome this speed difference compared to the C2D. But within an actual machine or even specs of the machine, it's really difficult to say.

 

[fyrefly]

I don't disagree with you for the most part. But Aperture made a big deal of how they are utilizing the GPU beyond what you would normally expect back when the software was released. I've seen this myself on various Macs, the GPU can make a much bigger difference with Aperture than expected. However the SB CPUs could potentially overcome this speed difference compared to the C2D. But within an actual machine or even specs of the machine, it's really difficult to say.

Totally True.

But without crazy speculation, where would the MR Forums (and especially the MBA forums) ever be?!

Heck, people are already trying to figure out if they should by a (not yet announced, or released) SB MBA or wait for a (1 yr+ away and theoretical again) Ivy Bridge MBA!

 

[bcaslis]

Totally True.

But without crazy speculation, where would the MR Forums (and especially the MBA forums) ever be?!

Heck, people are already trying to figure out if they should by a (not yet announced, or released) SB MBA or wait for a (1 yr+ away and theoretical again) Ivy Bridge MBA!

Yes!

Have they released the name of the CPUs after Ivy Bridge? If so we need to start a thread.

 

[fyrefly]

Yes!

Have they released the name of the CPUs after Ivy Bridge? If so we need to start a thread.

Hah, yep. It's called "Haswell". I'm sure the "Buy MBA now or wait for Haswell MBA?" threads will be here any day now...

 

[TrollToddington]

Hah, yep. It's called "Haswell". I'm sure the "Buy MBA now or wait for Haswell MBA?" threads will be here any day now...

Here we go https://forums.macrumors.com/threads/1145250/ Feel free to enter your input on Haswell