15" MBP with 2.6 ghz and radeon pro 460 or 2.7 ghz with 455?

[moran18guerra]

Hello, I'd like to be given some advice about whether to buy the 2.6 ghz 15" 2016 Macbook Pro with the Radeon pro 460 graphics, or 2.7 ghz 15" 2016 Macbook Pro with Radeon pro 455 graphics. Both models would have 512 gb of storage.

The use I would give to the machine would be software development (especially iOS apps), 2D and 3D design, semi-professional video editing in Final Cut Pro X, Photo editing in photoshop and content creation in Illustrator. Also I have thought of connecting at least an ultra wide monitor or two 27" 5k LG Ultrafine.

Thank you

 

[New_Mac_Smell]

This question has come up a fair few times already...

Generally there isn't much difference between them, as the benefits offset each other. If you're doing anything that requires more CPU, the 2.7 makes more sense, any GPU, the 460 makes sense. Simple as that really.

As you said you would be doing 3D design, that's the only thing of compromise. 2.7 for ever so slightly faster renders, or 460 for ever so slightly faster viewport. Although the difference again is negligible.

I went with the 460 as graphically it offers a greater increase than the CPU, so I'd recommend that one. But again, you likely wouldn't notice any difference so if it's between waiting 4-5 weeks for a 460, or getting a 2.7 in store tomorrow, consider the 2.7.

 

[moran18guerra]

This question has come up a fair few times already...

Generally there isn't much difference between them, as the benefits offset each other. If you're doing anything that requires more CPU, the 2.7 makes more sense, any GPU, the 460 makes sense. Simple as that really.

As you said you would be doing 3D design, that's the only thing of compromise. 2.7 for ever so slightly faster renders, or 460 for ever so slightly faster viewport. Although the difference again is negligible.

I went with the 460 as graphically it offers a greater increase than the CPU, so I'd recommend that one. But again, you likely wouldn't notice any difference so if it's between waiting 4-5 weeks for a 460, or getting a 2.7 in store tomorrow, consider the 2.7.

Time doesn't matter to me, I could wait.

 

[New_Mac_Smell]

Time doesn't matter to me, I could wait.

Then as I said either, you'll probably get more '460' votes as it's probably considered a better value machine, personally I feel that I don't need that GPU, but I might need it for one thing in 3 years time. Whereas I am not going to get into a situation where a 2.7 is necessary over a 2.6. So there you go.

 

[cubalibre]

I do 3D and motion design and ordered the 2.6 with 460, rendering speed with the 2.7 will only be very few percents faster - whilst the difference between the 455 and 460 and the double GPU memory would be more noticeable in viewport I think. Would love to see some real-world comparisons with After Effects though, and I can't seem to find cinebench scores for all three cards.

 

[BiteSizeThumb]

I got whatever I could get faster without waiting past Jan... So I got a 2.7 with 455.

 

[Marshall73]

2.7 with 460.

 

[duervo]

You'll get more bang for your buck by getting the 460 option.

 

[vipergts2207]

This question has come up a fair few times already...

Generally there isn't much difference between them, as the benefits offset each other. If you're doing anything that requires more CPU, the 2.7 makes more sense, any GPU, the 460 makes sense. Simple as that really.

As you said you would be doing 3D design, that's the only thing of compromise. 2.7 for ever so slightly faster renders, or 460 for ever so slightly faster viewport. Although the difference again is negligible.

I went with the 460 as graphically it offers a greater increase than the CPU, so I'd recommend that one. But again, you likely wouldn't notice any difference so if it's between waiting 4-5 weeks for a 460, or getting a 2.7 in store tomorrow, consider the 2.7.

Somewhat disagree with this assessment. I agree the 2.6 vs 2.7 is a pretty marginal difference. The difference between the 455 and 460 is fairly significant though. The question there is, does one need the extra performance? The price increase is so small though when considering the price of the computer as a whole, that I'd recommend almost everyone get it.

 

[New_Mac_Smell]

Somewhat disagree with this assessment. I agree the 2.6 vs 2.7 is a pretty marginal difference. The difference between the 455 and 460 is fairly significant though. The question there is, does one need the extra performance? The price increase is so small though when considering the price of the computer as a whole, that I'd recommend almost everyone get it.

Yup, although you missed my point. A 2.7/455 compared to a 2.6/460. The benefits of both make them very similar, comparing just the 2.6/2.7, or the 455/460 then yes the 460 is more powerful.

 

[BiteSizeThumb]

I think majority of those will agree that the 460 over the 455 is a bigger upgrade than the 2.7 vs 2.6. But, what will more people notice on a daily use (non graphical)? Probably the 2.7 over 2.6 processing speed (although again, not substantial at all...)

If you do anything graphics related (video games, photo/video editing) then get the better graphics card.

 

[maratus]

460 over 455 is an upgrade, 2.7 over 2.6 is a downgrade. That's a no-brainer.

 

[xraydoc]

If you're doing heavy FCPX work, then then 460 would be better than the 2.7GHz upgrade.

For me, I went with the 2.7GHz/455 since I don't do video editing and for the occasional 3D work that I do (Osirix MD), the 455 is more than sufficient -- and even then that app lets me render on the 3D card or on the processor. The results look a bit different but on the 2016 quad core MBP, both perform quite well.

 

[vipergts2207]

460 over 455 is an upgrade, 2.7 over 2.6 is a downgrade. That's a no-brainer.

How do you reckon?

 

[maratus]

How do you reckon?

2.7 and 2.9GHz CPUs will reach higher temps and may throttle in certain conditions under 100% load in which 2.6 model specifically will not. Aside from potential throttling, higher power CPUs will result in higher fan speed, higher power consumption (less headroom before power throttling begins) and under extreme situations may result in similar or even lower performance (lower turboboost due to thermal limits). You may see more than 10C increase in CPU temperature and several degrees higher internal temps with other things being equal. Keep in mind that dGPU temps are also heavily dependant on CPU temps because they share the heatsink. I think it's known what kind of issues Apple is known for when it comes to dGPU reliability due to heat.

Overall it's a downgrade just like it was 2.7Ghz vs. 2.4Ghz for the early-2013 rMBP. A downgrade for which you pay extra! I had to return my original 2.6GHz mid-2012 rMBP and replace it with a lowest-end CPU version (otherwise maxed out!)

 

[vipergts2207]

2.7 and 2.9GHz CPUs will reach higher temps and may throttle in certain conditions under 100% load in which 2.6 model specifically will not. Aside from potential throttling, higher power CPUs will result in higher fan speed, higher power consumption (less headroom before power throttling begins) and under extreme situations may result in similar or even lower performance (lower turboboost due to thermal limits). You may see more than 10C increase in CPU temperature and several degrees higher internal temps with other things being equal. Keep in mind that dGPU temps are also heavily dependant on CPU temps because they share the heatsink. I think it's known what kind of issues Apple is known for when it comes to dGPU reliability due to heat.

Overall it's a downgrade just like it was 2.7Ghz vs. 2.4Ghz for the early-2013 rMBP. A downgrade for which you pay extra! I had to return my original 2.6GHz mid-2012 rMBP and replace it with a lowest-end CPU version (otherwise maxed out!)

I think you're way off base here. First, assuming it will begin to throttle after getting too hot from running at 100%, it will not throttle to a lower clock than the 2.6 would. Worst case scenario it will run at the same clock speed as the 2.6. Then factor in the 2 MB of increased cache and the 2.7 wins.

And explain why the 2.6 will specifically not throttle under extended heavy loads.

 

[maratus]

it will not throttle to a lower clock than the 2.6 would.

Yes, it will. Maybe not in the previous models (where heat was my main concern), but it will in 2016 rMBP.

Worst case scenario it will run at the same clock speed as the 2.6.

No, it could be much worse. With the older gens its effective frequency could drop down and stay on par with the base model under the same conditions (it happened with the 2.4 vs 2.7) – which means having the same speed for higher price and more heat.

The new 2016 rMBP implements a different algorithm – TDP is dropped to 35W after reaching throttling temperatures and stays there for a few minutes. This affects performance much more than before.

Then factor in the 2 MB of increased cache and the 2.7 wins.

Cache effect is very application specific and hardly would be enough to compensate.

And explain why the 2.6 will specifically not throttle under extended heavy loads.

Because Intel specifies lower base voltages and lover boost voltages for its lower-end CPUs. Lower power means lower temperatures. 2.6Ghz Skylake stays at 90C and 2.7+ reach 100C+ under the same conditions (iGPU, 20C ambient, max fan RPM, same 100% cpu load and same OS)

 

[vipergts2207]

The new 2016 rMBP implements a different algorithm – TDP is dropped to 35W after reaching throttling temperatures and stays there for a few minutes. This affects performance much more than before.

Since this seems to be the basis for much of what you're saying, where is this information coming from?

 

[maratus]

Also, regarding L3 cache:

1. More L3 cache increases total power consumption as well. It's not much under 100% load, but hey, every bits count and that's 33% more cache.

2. L3 cache isn't gated while cores / igp are. That means it's power contribution is higher when CPU idles, so more L3 cache means lower battery life as well.
[doublepost=1481695962][/doublepost]

Since this seems to be the basis for much of what you're saying, where is this information coming from?

There were several reports from 2.7 and/or 2.9 users on this forum

 

[vipergts2207]

Also, regarding L3 cache:

1. More L3 cache increases total power consumption as well. It's not much under 100% load, but hey, every bits count and that's 33% more cache.

2. L3 cache isn't gated while cores / igp are. That means it's power contribution is higher when CPU idles, so more L3 cache means lower battery life as well.
[doublepost=1481695962][/doublepost]
There were several reports from 2.7 and/or 2.9 users on this forum

Let's be honest, that the extra 2MB of cache is unlikely to make a real impact on battery life. Tried to find hard numbers for cache power consumption, but could not.

That's a rather amorphous, non-specific source as well.

 

[duervo]

Intel allows OEMs (ie: Apple) to configure the TDP down value. On the Skylake processors in these 2016 MBPs, the configurable TDP down is 35W. Apple has chosen to take advantage of this, and have set this 35W TDP value for the CPU's they are using (most likely somewhere within the EFI.)

You can confirm this by running Intel Power Gadget on a 2016 MBP, and then running enough workload to peg the CPU at 100%. Shortly after it hits 100%, the power consumption of the CPU drops from 45W, down to 35W, and processor speed drops accordingly.

At any rate, for the cost being charged to bump the CPU up, if you are doing so for pure performance gains, such as the OP is intending for video workload, money would be better put towards the 460 GPU.

If there is a desire to also perform tasks that are enabled with Intel vPro, then the 2.7 and 2.9 CPU options would be warranted. However, that's irrelevant to the OP's needs. It would offer more value for corporate environments (ie: I bet IBM may probably opt for the 2.7 or 2.9 CPU.)

There's also the TSX-NI feature that's finally enabled (albeit on the 2.7 and 2.9's, not the 2.6) but it's arguable if such a thing would see huger gains than using the 460 GPU for video workload.

 

[vipergts2207]

Intel allows OEMs (ie: Apple) to configure the TDP down value. On the Skylake processors in these 2016 MBPs, the configurable TDP down is 35W. Apple has chosen to take advantage of this, and have set this 35W TDP value for the CPU's they are using (most likely somewhere within the EFI.)

You can confirm this by running Intel Power Gadget on a 2016 MBP, and then running enough workload to peg the CPU at 100%. Shortly after it hits 100%, the power consumption of the CPU drops from 45W, down to 35W, and processor speed drops accordingly.

My MBP has not yet arrived to try this, but thanks for this info anyway.

 

[duervo]

My MBP has not yet arrived to try this, but thanks for this info anyway.

It's not surprising at all to me that Apple did this. I was expecting it, since it is a laptop after all. I would actually be surprised if Apple were the only OEM doing it. I suspect that all the laptop manufacturers are doing it on at least some of their products.

Took me forever to come across a video showing it. I just happened to watch that Louis Rossmann guy streaming one day, and he was checking out a 15" MBP that somebody let him take a look at. He had Intel Power Gadget running while running an ffmpeg job from CLI. It didn't happen immediately, but after a little while, you could see the power usage of the CPU in Power Gadget drop down to 35W.

Start at ~57 minutes into the video:

Watch the Intel Power Gadget window as he starts up an ffmpeg job.

Honestly, I think this is a better approach to the traditional method of throttling, where people would see noticeable impacts. I believe this method, while still essentially a form of throttling, is less noticeable.

 

[maratus]

Let's be honest, that the extra 2MB of cache is unlikely to make a real impact on battery life. Tried to find hard numbers for cache power consumption, but could not.

That's a rather amorphous, non-specific source as well.

w/e as long as you're happy with yours.

All I'm trying to do is to help others, if you prefer to not listen – it's fine.

http://i.stack.imgur.com/syJ25.jpg

Cache always impacts CPU power consumption. How much? That's hard to test, I don't have that info. I think it'll be as small as +2MB on real-world performance tbh

 

[vipergts2207]

w/e as long as you're happy with yours.

All I'm trying to do is to help others, if you prefer to not listen – it's fine.

http://i.stack.imgur.com/syJ25.jpg

Cache always impacts CPU power consumption. How much? That's hard to test, I don't have that info. I think it'll be as small as +2MB on real-world performance tbh

I don't have mine yet unfortunately. And it's not that I 'prefer not to listen,' I just don't blindly believe anything someone says. I like hard evidence. That's why I attempted to find cache power consumption data.