Great article, I’ve been looking for examples of Apple’s 1st-party hardware performance beyond redlining 3rd-party CPU ISA. They really need to push these points of difference now that every PC looks like a Mac (AIO & Notebook).
Shame I can’t see the details in an iPhone. The graphs/tables don’t show in the main article and tapping the thumbnail in the comments just shows the same thumbnail image expanded.
Did you measure power (watts) or energy used during all these runs?
The advantage of the T2 could be factors of 3x to 5x on a WHr basis, which is also really nice, above and beyond the 2x in time?
Is QuickSync dependent on clock rate and number of cores? Ie, is the QS hardware in the uncore or in the cores themselves?
We didn't measure power during the runs, but I'll put it on the list of things to do for a follow-up. Interesting thought.
Regarding Quicksync, there is a (slight) increase in what you get from it as the number of cores increase, but it is not linear.
Hopefully you can get yourself both a Kill-A-Watt and in-line USB power meter to measure power at the wall and power used by USB devices (iOS devices) along with whatever Intel’s power gadget tool does.
I’m maybe surprised how little Intel QuickSync reduces transcode time (the Mac mini with no T2 and no QS vs that iMac 4K). Wonder how the transcode is split up and goes to what hardware.
Curious about the massive discrepancy between the "2018 Mac Mini i3 CPU only" at 8m 51s and the "5,1 Mac Pro" at 51m 14s, both without T2/QS.
Is it possible to break down the differences between T2 and QuickSync performances and which settings are required for each?
The 5,1 Mac Pro is a cheese grater model that is likely using 10 year old Xeon processors. A comparative system today should be about 2^3 times faster, without any fancy dedicated onchip ASICs even.
It’s upsetting to know that my 2012 Mac Pro, with two 6 core CPU’s, running at 3.47GHz, is slower than my iPhone Max, core for core, or my 2018 iPad Pro 12.9”. Such is progress.
No it’s not upsetting. It’s exciting!
Hoping Apple opens the spigot for iPad Pro so that we can do all this type of stuff a lot more easily. The A12X actually would be in the middle of the pack in a lot of these tests, and could be at the top of these Handbrake tests if it has a T2 in it. Wonder if the Neural Engine can run 8-bit compression/decompression math... probably not. But the GPU might.
The 2018 MacBook Air also has the T2 chip, which is not mentioned in this article. It would be cool to compare the encoding times with the 2018 MacBook Air.
For some reason I am not clear on which preset uses the Apple Video Toolkit, is this any preset that specifies an Apple Device?
No, when you pull down the Video Encoder menu in the Video tab of Handbrake, there's an option for "H.265 (VideoToolbox)"
Thanks Mike, however I would update the article for the uninitiated to some of the settings as to where to find them.
I also noted that if you have an older Mac, you do not see the H.265 Video Toolbox. I see it on my 2017 iMac, but its not visible on my 2012 Mini. Only H.264.
Great article. The new Mac Pro will be tricked out for the fastest encoding I am quite sure ... 🤞🏻
Yes, for sure... great article. Yeah, I'm sure the new Mac Pro will have the T2 chip. Maybe even a bunch of them? I wonder if they could run them in parallel somehow.
... The article should clarify it is comparing the use of the T2's h265 HEVC 8-bit only hardware encoder. ... If you weren't encoding h265 HEVC 8-bit then there would be no difference than having a T2 or not.
Yeah, but I think that was the point of the article. Most of us know the T2 does lots of other things too, from previous articles when they started implementing it. (Though, I guess doesn't hurt to mention.)
I'm curious about the 8-bit thing, though. Are you saying this is a more limited kind of h.265 than most people would typically export? (I'm genuinely curious, as the quality looks OK to me, but I'm no video professional.)
Did you measure power (watts) or energy used during all these runs? The advantage of the T2 could be factors of 3x to 5x on a WHr basis, which is also really nice, above and beyond the 2x in time?
Yeah, it reduces the power used... though I haven't measured by how much. My UPS has a watt display on it, so I could figure it out (assuming it is accurate). But, it certainly doesn't run as hard as it does if it's doing h.264 (and taking longer).
That said, it does run the CPU harder than idle, so the CPUs are doing something (it isn't only using the T2).
Since I came from an older machine (probably more akin to the Mac Pro 5,1 experiment), I was totally blown away by the speed. (ie: 51 min vs 2 min, lol) I was used to setting up a Screenflow project or Handbrake and letting it run overnight. Now, I can just hit export and wait 5 min, or 15 min, etc. Even a Blu-Ray rip isn't a big deal anymore.
The article seems to confirm what I've been suspecting -- that hardware acceleration only works if one accepts having certain parameter settings fixed.
Years ago I did a mountain of tests with Handbrake, adjusting various parameters to evaluate their impact on the compressed result. I eventually settled on a combination of settings that yield the best (for my purposes) balance between file size and compression artifacts. After all the effort that went into that, I'd be reluctant to surrender any of that control to a chip that may prioritize differently than I would. I'll have to wait until I can compare myself, but I suspect I may choose to live with longer encode times to get a better end result.
Yes, you have to use that video setting... though I'm happy with it so far. I wasn't aware (as the article pointed out) that you have to get other settings within proper ranges. I just used the settings that I was using for h.264 and changed the video to the VideoToolbox, and then tweaked the bitrate a bit. I guess I got lucky.
Curious about the massive discrepancy between the "2018 Mac Mini i3 CPU only" at 8m 51s and the "5,1 Mac Pro" at 51m 14s, both without T2/QS.
Is it possible to break down the differences between T2 and QuickSync performances and which settings are required for each?
The mini (2018) has the T2. I wonder if there is an error in the article there, as I wouldn't expect the difference to be quite that narrow. Hmm...
For some reason I am not clear on which preset uses the Apple Video Toolkit, is this any preset that specifies an Apple Device?
Yeah, I see Mike answered... but it IS NOT one of the presets. I think the word is getting out, but I had to find it digging on some forums when I first got my mini (and it has only been in a fairly recent build of Handbrake... like a couple months, I think). You can pick a preset as a starting point, and then have to go to the video tab and change the 'Video Encoder" setting to the h.265 VideoToolbox one. Then, you might also need to tweak the bitrate.
For other apps, it is done differently. For example, in Screenflow, you just switch to manual and pick the HEVC preset. I figured it out there before I did with Handbrake. (And, of course, some apps might not support it, I suppose.)
tht said: The 5,1 Mac Pro is a cheese grater model that is likely using 10 year old Xeon processors. A comparative system today should be about 2^3 times faster, without any fancy dedicated onchip ASICs even.
I suppose, but the key here is the T2. Even with the i7 6-core in my mini, the time goes way up if I don't use HEVC/h.265 with those settings.
melgross said: It’s upsetting to know that my 2012 Mac Pro, with two 6 core CPU’s, running at 3.47GHz, is slower than my iPhone Max, core for core, or my 2018 iPad Pro 12.9”. Such is progress.
Only for certain things, though. Your iPhone Max probably wouldn't do a very good job of doing some structural analysis in a CAD app or solving some protein folding solutions (or, doing what I currently am, running Autodesk Revit on Windows 10).
Any word on how this T2 chip helps encoding using Apple Compressor?
I haven't noticed any significant increase in render speed on my MBP 15" 2018 2.6 GHz Intel Core i7 with 32 GB 2400 MHz DDR4 compared to my 2013 MBP (using Final Cut Pro and Compressor). Then again, I haven't tweaked any of the settings in Compressor like you have with Handbrake.
Yeah, you'd have to find the settings to see the difference, as there probably wouldn't be a big difference otherwise.
That’s cool, reminds me those x87 co-processors in the early days. I thought about that T2 can be used for this purpose since it got Fusion cores in them, and hopefully they can makeup the loss from the throttling of an i9 processors.
Oh, if you were using that codec, it WAY more than makes up for that (and the machine runs cooler). But, as Mike has indicated.... I'm not sure what the limitations are. Like, if you're using video software that does effects and such, at what point does it fall-back to not using the T2? It's probably just for the actual encoding, I'm guessing the rest has to use the CPU/GPU... but I don't know if there could be some mix.
Like, when I do Screenflow projects, I'm guessing it has to do more than a straight re-encode, but my times dramatically decreased (though I wasn't coming from a speed-demon machine previously. But, way more difference than clock-speed, CPU generation).
The 2018 MacBook Air also has the T2 chip, which is not mentioned in this article. It would be cool to compare the encoding times with the 2018 MacBook Air.
Hmm, I'm not sure T2 chips are different from one another, or how. I'd expect the times to be somewhat similar, as I think the T2 is doing most of the work. Though, the CPUs (as I noted above) are working as well... so that might factor in a bit.
normang said: I also noted that if you have an older Mac, you do not see the H.265 Video Toolbox. I see it on my 2017 iMac, but its not visible on my 2012 Mini. Only H.264.
Hmm, that's odd, as the 2017 iMac wouldn't have the T2 either. Are you sure you have the latest Handbrake on both? They just added that capability in the most recent version (a couple months back).
Speed is important, but what about quality? QuickSync is known to produce worse results than good software encoders, how does T2 stack up on quality? Does nobody care about quality anymore?
Curious about the massive discrepancy between the "2018 Mac Mini i3 CPU only" at 8m 51s and the "5,1 Mac Pro" at 51m 14s, both without T2/QS.
Is it possible to break down the differences between T2 and QuickSync performances and which settings are required for each?
The 5,1 Mac Pro is a cheese grater model that is likely using 10 year old Xeon processors. A comparative system today should be about 2^3 times faster, without any fancy dedicated onchip ASICs even.
It’s upsetting to know that my 2012 Mac Pro, with two 6 core CPU’s, running at 3.47GHz, is slower than my iPhone Max, core for core, or my 2018 iPad Pro 12.9”. Such is progress.
You'd think given you have internal slots some sort of add on card would be feasible or external gizmo via the card in the bus. With the new trash can Mac Pro I have with its obsolete I/O I am shit out of luck. It transcodes slower than molasses. I await the new Mac Pro with bated breath.
Speed is important, but what about quality? QuickSync is known to produce worse results than good software encoders, how does T2 stack up on quality? Does nobody care about quality anymore?
That's what I'm after. I have a new 2018 Mac mini and the T2 is very fast compared to cpu encoding.... but tuning is limited thru these apps - I use ff-works/ffmpeg and handbrake. But... the result is not always better. I re-encoded an h264 the other day and the result was poor... the video jumped every few minutes. I tried encoding a rip of a Blu-ray that caused ff-works to crash and handbrake generates very large files.... so I basically try encoding with T2 first and if the result is good I use it. If not, I encode with the CPU. Perhaps this will improve over time so I can use the T2 exclusively.
Your mileage may vary based on driving conditions.
Curious about the massive discrepancy between the "2018 Mac Mini i3 CPU only" at 8m 51s and the "5,1 Mac Pro" at 51m 14s, both without T2/QS.
Is it possible to break down the differences between T2 and QuickSync performances and which settings are required for each?
The 5,1 Mac Pro is a cheese grater model that is likely using 10 year old Xeon processors. A comparative system today should be about 2^3 times faster, without any fancy dedicated onchip ASICs even.
It’s upsetting to know that my 2012 Mac Pro, with two 6 core CPU’s, running at 3.47GHz, is slower than my iPhone Max, core for core, or my 2018 iPad Pro 12.9”. Such is progress.
No it’s not upsetting. It’s exciting!
Hoping Apple opens the spigot for iPad Pro so that we can do all this type of stuff a lot more easily. The A12X actually would be in the middle of the pack in a lot of these tests, and could be at the top of these Handbrake tests if it has a T2 in it. Wonder if the Neural Engine can run 8-bit compression/decompression math... probably not. But the GPU might.
I didn’t really mean that it was upsetting. It was just a figure of speech. I’m waiting, more impatiently every day, for the new Mac Pro.
yeah, I’m also hoping for that big iOS change for the iPad that was supposedly delayed from last year. Hopefully, that year delay allowed them to continue work on it for this year, adding changes and functionality that would have come in the second year. I hope they didn’t basically shelve it, and just tweak it for this year. I want to see real progress.
i’m excited over the possibilities. A different UI, or so we’re led to believe. If so, I hope it’s really major, not just a shuffling around of what’s there now. The iPad Pro in particular, needs to move from the big iPhone UI configuration. I think we can handle it. ;~) The 12.9” screen, and even the 11”, is big enough for windowing, not this sliding arrangement, which while useful, is annoying and ridged.
so three things would totally change the way we look at, and use it. I think we all know what that is.
trackpad support a true mass drive capability through USB C and wifi hierarchical filing system.
something some people get worried about with all of that is that they think they will have to use it, and that’s not true. That is, Apple can keep things mainly the same for those who want that. The way they expose the filing system now, while not nearly enough, is a good example. You don’t have to use it if you don’t want to, but it’s easy to access if you do. If you don’t want a trackpad, you won’t see it with no keyboard, and possibly, some keyboards won’t have it. Though, if one is using a keyboard, I don’t understand why they wouldn’t want one.
if you don’t connect a drive, then that doesn’t matter either. Windowing can be a selection in settings. A new section dealing with all the new aspects could be added. Choose which of them you want to see, and use, if any. None could be default, at least for the first few years until people get used to it, which I bet they would.
if Apple does this, I would bet sales would leap, though sales of MacBooks might slump. Hard to tell.
Great article. The new Mac Pro will be tricked out for the fastest encoding I am quite sure ... 🤞🏻
Yes, for sure... great article. Yeah, I'm sure the new Mac Pro will have the T2 chip. Maybe even a bunch of them? I wonder if they could run them in parallel somehow.
... The article should clarify it is comparing the use of the T2's h265 HEVC 8-bit only hardware encoder. ... If you weren't encoding h265 HEVC 8-bit then there would be no difference than having a T2 or not.
Yeah, but I think that was the point of the article. Most of us know the T2 does lots of other things too, from previous articles when they started implementing it. (Though, I guess doesn't hurt to mention.)
I'm curious about the 8-bit thing, though. Are you saying this is a more limited kind of h.265 than most people would typically export? (I'm genuinely curious, as the quality looks OK to me, but I'm no video professional.)
Did you measure power (watts) or energy used during all these runs? The advantage of the T2 could be factors of 3x to 5x on a WHr basis, which is also really nice, above and beyond the 2x in time?
Yeah, it reduces the power used... though I haven't measured by how much. My UPS has a watt display on it, so I could figure it out (assuming it is accurate). But, it certainly doesn't run as hard as it does if it's doing h.264 (and taking longer).
That said, it does run the CPU harder than idle, so the CPUs are doing something (it isn't only using the T2).
Since I came from an older machine (probably more akin to the Mac Pro 5,1 experiment), I was totally blown away by the speed. (ie: 51 min vs 2 min, lol) I was used to setting up a Screenflow project or Handbrake and letting it run overnight. Now, I can just hit export and wait 5 min, or 15 min, etc. Even a Blu-Ray rip isn't a big deal anymore.
The article seems to confirm what I've been suspecting -- that hardware acceleration only works if one accepts having certain parameter settings fixed.
Years ago I did a mountain of tests with Handbrake, adjusting various parameters to evaluate their impact on the compressed result. I eventually settled on a combination of settings that yield the best (for my purposes) balance between file size and compression artifacts. After all the effort that went into that, I'd be reluctant to surrender any of that control to a chip that may prioritize differently than I would. I'll have to wait until I can compare myself, but I suspect I may choose to live with longer encode times to get a better end result.
Yes, you have to use that video setting... though I'm happy with it so far. I wasn't aware (as the article pointed out) that you have to get other settings within proper ranges. I just used the settings that I was using for h.264 and changed the video to the VideoToolbox, and then tweaked the bitrate a bit. I guess I got lucky.
Curious about the massive discrepancy between the "2018 Mac Mini i3 CPU only" at 8m 51s and the "5,1 Mac Pro" at 51m 14s, both without T2/QS.
Is it possible to break down the differences between T2 and QuickSync performances and which settings are required for each?
The mini (2018) has the T2. I wonder if there is an error in the article there, as I wouldn't expect the difference to be quite that narrow. Hmm...
For some reason I am not clear on which preset uses the Apple Video Toolkit, is this any preset that specifies an Apple Device?
Yeah, I see Mike answered... but it IS NOT one of the presets. I think the word is getting out, but I had to find it digging on some forums when I first got my mini (and it has only been in a fairly recent build of Handbrake... like a couple months, I think). You can pick a preset as a starting point, and then have to go to the video tab and change the 'Video Encoder" setting to the h.265 VideoToolbox one. Then, you might also need to tweak the bitrate.
For other apps, it is done differently. For example, in Screenflow, you just switch to manual and pick the HEVC preset. I figured it out there before I did with Handbrake. (And, of course, some apps might not support it, I suppose.)
tht said: The 5,1 Mac Pro is a cheese grater model that is likely using 10 year old Xeon processors. A comparative system today should be about 2^3 times faster, without any fancy dedicated onchip ASICs even.
I suppose, but the key here is the T2. Even with the i7 6-core in my mini, the time goes way up if I don't use HEVC/h.265 with those settings.
melgross said: It’s upsetting to know that my 2012 Mac Pro, with two 6 core CPU’s, running at 3.47GHz, is slower than my iPhone Max, core for core, or my 2018 iPad Pro 12.9”. Such is progress.
Only for certain things, though. Your iPhone Max probably wouldn't do a very good job of doing some structural analysis in a CAD app or solving some protein folding solutions (or, doing what I currently am, running Autodesk Revit on Windows 10).
Any word on how this T2 chip helps encoding using Apple Compressor?
I haven't noticed any significant increase in render speed on my MBP 15" 2018 2.6 GHz Intel Core i7 with 32 GB 2400 MHz DDR4 compared to my 2013 MBP (using Final Cut Pro and Compressor). Then again, I haven't tweaked any of the settings in Compressor like you have with Handbrake.
Yeah, you'd have to find the settings to see the difference, as there probably wouldn't be a big difference otherwise.
That’s cool, reminds me those x87 co-processors in the early days. I thought about that T2 can be used for this purpose since it got Fusion cores in them, and hopefully they can makeup the loss from the throttling of an i9 processors.
Oh, if you were using that codec, it WAY more than makes up for that (and the machine runs cooler). But, as Mike has indicated.... I'm not sure what the limitations are. Like, if you're using video software that does effects and such, at what point does it fall-back to not using the T2? It's probably just for the actual encoding, I'm guessing the rest has to use the CPU/GPU... but I don't know if there could be some mix.
Like, when I do Screenflow projects, I'm guessing it has to do more than a straight re-encode, but my times dramatically decreased (though I wasn't coming from a speed-demon machine previously. But, way more difference than clock-speed, CPU generation).
I’m just commenting on core strengths. You might remember a time when it was said that there was no possibility that ARM could ever hope to even come close to x86 in any way. What I see is that if Apple decided to change the parameters a bit, we could have two 4 core A12x chips in a machine. If there was adequate cooling and power, the efficiency cores could be removed and the room used to power up the other components. Considering that the SoC is it’s own support chip package, and contains a good GPU, with everything doubled, at least, including the ISP and the AI section, that could really be something.
apple could also decouple the GPU from the chip and have a separate one with its own graphics memory. That would enable the possibility of competing with more powerful versions from AMD and Nvidia.
Speed is important, but what about quality? QuickSync is known to produce worse results than good software encoders, how does T2 stack up on quality? Does nobody care about quality anymore?
Yeah, that's a question (and possible fear) I have as well. I'd love someone with some more video knowledge than myself to weigh in on such things. I haven't been disappointed with the output so far, but I haven't compared it in any good way with alternatives. If there is some problem, I guess that would be good to know before I fill my video library with T2-encoded HEVC stuff. Heh.
On the other hand, it's plenty fine for the YouTube stuff I've been doing... and saves a TON of time.
MacPro said: You'd think given you have internal slots some sort of add on card would be feasible or external gizmo via the card in the bus. With the new trash can Mac Pro I have with its obsolete I/O I am shit out of luck. It transcodes slower than molasses. I await the new Mac Pro with bated breath.
I think there were little hardware encoders for h.264, so I'd imagine some exist or will be coming. There's no reason something like this shouldn't be able to hang off a USB or TB port. Maybe there just isn't enough demand? (I've never even looked though.)
melgross said: so three things would totally change the way we look at, and use it. I think we all know what that is.
trackpad support a true mass drive capability through USB C and wifi hierarchical filing system.
Great list. Yes, what is holding iOS from becoming a pro solution (at least for me) is like 90% on the OS side of things, and maybe the other 10% lack of apps. I need to do a better re-evaluation one of these days, but just from what I hear/read, it doesn't seem things are that much different than when I wrote my review of the iPad 2 for such things. Most of the same workflow contortions seem to be the case.
melgross said: I’m just commenting on core strengths. You might remember a time when it was said that there was no possibility that ARM could ever hope to even come close to x86 in any way. What I see is that if Apple decided to change the parameters a bit, we could have two 4 core A12x chips in a machine. If there was adequate cooling and power, the efficiency cores could be removed and the room used to power up the other components. Considering that the SoC is it’s own support chip package, and contains a good GPU, with everything doubled, at least, including the ISP and the AI section, that could really be something.
apple could also decouple the GPU from the chip and have a separate one with its own graphics memory. That would enable the possibility of competing with more powerful versions from AMD and Nvidia.
the possibilities are endless.
Oh yeah, there are some big possibilities. But, I think the Windows thing is hard to solve. Without direct hardware, it would have to be many times faster to emulate.
cgWerks said: Yeah, that's a question (and possible fear) I have as well. I'd love someone with some more video knowledge than myself to weigh in on such things. I haven't been disappointed with the output so far, but I haven't compared it in any good way with alternatives. If there is some problem, I guess that would be good to know before I fill my video library with T2-encoded HEVC stuff. Heh.
Evaluating quality is a thorny issue, for a couple reasons.
Through lots of training and experience, I'll hear things in a recording most people won't notice. That means what I consider "good quality" will differ from what's good to someone else. Does it matter that a particular flaw exists if one doesn't notice it, or at least isn't bothered by it?
On the other hand, you may become more critical with time. I now notice flaws in transcodes I did five or six years ago that apparently didn't bother me at the time I did them. Maybe that means one should err on the side of higher quality as a hedge against more sophisticated perception in the future.
The bottom line: if in doubt, increase the bit rate!
Great article. The new Mac Pro will be tricked out for the fastest encoding I am quite sure ... 🤞🏻
Yes, for sure... great article. Yeah, I'm sure the new Mac Pro will have the T2 chip. Maybe even a bunch of them? I wonder if they could run them in parallel somehow.
... The article should clarify it is comparing the use of the T2's h265 HEVC 8-bit only hardware encoder. ... If you weren't encoding h265 HEVC 8-bit then there would be no difference than having a T2 or not.
Yeah, but I think that was the point of the article. Most of us know the T2 does lots of other things too, from previous articles when they started implementing it. (Though, I guess doesn't hurt to mention.)
I'm curious about the 8-bit thing, though. Are you saying this is a more limited kind of h.265 than most people would typically export? (I'm genuinely curious, as the quality looks OK to me, but I'm no video professional.)
Did you measure power (watts) or energy used during all these runs? The advantage of the T2 could be factors of 3x to 5x on a WHr basis, which is also really nice, above and beyond the 2x in time?
Yeah, it reduces the power used... though I haven't measured by how much. My UPS has a watt display on it, so I could figure it out (assuming it is accurate). But, it certainly doesn't run as hard as it does if it's doing h.264 (and taking longer).
That said, it does run the CPU harder than idle, so the CPUs are doing something (it isn't only using the T2).
Since I came from an older machine (probably more akin to the Mac Pro 5,1 experiment), I was totally blown away by the speed. (ie: 51 min vs 2 min, lol) I was used to setting up a Screenflow project or Handbrake and letting it run overnight. Now, I can just hit export and wait 5 min, or 15 min, etc. Even a Blu-Ray rip isn't a big deal anymore.
The article seems to confirm what I've been suspecting -- that hardware acceleration only works if one accepts having certain parameter settings fixed.
Years ago I did a mountain of tests with Handbrake, adjusting various parameters to evaluate their impact on the compressed result. I eventually settled on a combination of settings that yield the best (for my purposes) balance between file size and compression artifacts. After all the effort that went into that, I'd be reluctant to surrender any of that control to a chip that may prioritize differently than I would. I'll have to wait until I can compare myself, but I suspect I may choose to live with longer encode times to get a better end result.
Yes, you have to use that video setting... though I'm happy with it so far. I wasn't aware (as the article pointed out) that you have to get other settings within proper ranges. I just used the settings that I was using for h.264 and changed the video to the VideoToolbox, and then tweaked the bitrate a bit. I guess I got lucky.
Curious about the massive discrepancy between the "2018 Mac Mini i3 CPU only" at 8m 51s and the "5,1 Mac Pro" at 51m 14s, both without T2/QS.
Is it possible to break down the differences between T2 and QuickSync performances and which settings are required for each?
The mini (2018) has the T2. I wonder if there is an error in the article there, as I wouldn't expect the difference to be quite that narrow. Hmm...
For some reason I am not clear on which preset uses the Apple Video Toolkit, is this any preset that specifies an Apple Device?
Yeah, I see Mike answered... but it IS NOT one of the presets. I think the word is getting out, but I had to find it digging on some forums when I first got my mini (and it has only been in a fairly recent build of Handbrake... like a couple months, I think). You can pick a preset as a starting point, and then have to go to the video tab and change the 'Video Encoder" setting to the h.265 VideoToolbox one. Then, you might also need to tweak the bitrate.
For other apps, it is done differently. For example, in Screenflow, you just switch to manual and pick the HEVC preset. I figured it out there before I did with Handbrake. (And, of course, some apps might not support it, I suppose.)
tht said: The 5,1 Mac Pro is a cheese grater model that is likely using 10 year old Xeon processors. A comparative system today should be about 2^3 times faster, without any fancy dedicated onchip ASICs even.
I suppose, but the key here is the T2. Even with the i7 6-core in my mini, the time goes way up if I don't use HEVC/h.265 with those settings.
melgross said: It’s upsetting to know that my 2012 Mac Pro, with two 6 core CPU’s, running at 3.47GHz, is slower than my iPhone Max, core for core, or my 2018 iPad Pro 12.9”. Such is progress.
Only for certain things, though. Your iPhone Max probably wouldn't do a very good job of doing some structural analysis in a CAD app or solving some protein folding solutions (or, doing what I currently am, running Autodesk Revit on Windows 10).
Any word on how this T2 chip helps encoding using Apple Compressor?
I haven't noticed any significant increase in render speed on my MBP 15" 2018 2.6 GHz Intel Core i7 with 32 GB 2400 MHz DDR4 compared to my 2013 MBP (using Final Cut Pro and Compressor). Then again, I haven't tweaked any of the settings in Compressor like you have with Handbrake.
Yeah, you'd have to find the settings to see the difference, as there probably wouldn't be a big difference otherwise.
That’s cool, reminds me those x87 co-processors in the early days. I thought about that T2 can be used for this purpose since it got Fusion cores in them, and hopefully they can makeup the loss from the throttling of an i9 processors.
Oh, if you were using that codec, it WAY more than makes up for that (and the machine runs cooler). But, as Mike has indicated.... I'm not sure what the limitations are. Like, if you're using video software that does effects and such, at what point does it fall-back to not using the T2? It's probably just for the actual encoding, I'm guessing the rest has to use the CPU/GPU... but I don't know if there could be some mix.
Like, when I do Screenflow projects, I'm guessing it has to do more than a straight re-encode, but my times dramatically decreased (though I wasn't coming from a speed-demon machine previously. But, way more difference than clock-speed, CPU generation).
I’m just commenting on core strengths. You might remember a time when it was said that there was no possibility that ARM could ever hope to even come close to x86 in any way. What I see is that if Apple decided to change the parameters a bit, we could have two 4 core A12x chips in a machine. If there was adequate cooling and power, the efficiency cores could be removed and the room used to power up the other components. Considering that the SoC is it’s own support chip package, and contains a good GPU, with everything doubled, at least, including the ISP and the AI section, that could really be something.
apple could also decouple the GPU from the chip and have a separate one with its own graphics memory. That would enable the possibility of competing with more powerful versions from AMD and Nvidia.
the possibilities are endless.
Why not going fanless with consumer laptops? A 12” iPad Pro actually cools well, it passed the stress test without throttling. Something what Intel and AMD still couldn’t do even with two fans constantly running in full speed.
I had this discussion before, from what I know, x86 processors we have today has been held back by its legacy. Backwards compatibility are the biggest strength since it retains so much even from 8086, but that hurts their efficiency bad. Hence why A11 outperforms x86 rivals with higher frequencies in single core tests.
normang said: I also noted that if you have an older Mac, you do not see the H.265 Video Toolbox. I see it on my 2017 iMac, but its not visible on my 2012 Mini. Only H.264.
Hmm, that's odd, as the 2017 iMac wouldn't have the T2 either. Are you sure you have the latest Handbrake on both? They just added that capability in the most recent version (a couple months back).
I am not sure if the access to the H.265 version of the Toolkit is T2 Specific, I suspect its related to something in the CPU in the iMac that is not in the older Mini. PLys Handbrake usually always prompts on launch if there are updates, which there were in my case, had not run it in a little while. So when I noted this, it was the latest version of Handbrake.
lorin schultz said: Evaluating quality is a thorny issue, for a couple reasons.
Through lots of training and experience, I'll hear things in a recording most people won't notice. That means what I consider "good quality" will differ from what's good to someone else. Does it matter that a particular flaw exists if one doesn't notice it, or at least isn't bothered by it?
On the other hand, you may become more critical with time. I now notice flaws in transcodes I did five or six years ago that apparently didn't bother me at the time I did them. Maybe that means one should err on the side of higher quality as a hedge against more sophisticated perception in the future.
The bottom line: if in doubt, increase the bit rate!
Yeah, I guess I was thinking more along the lines of some glaring flaw or limitation, such that most pros wouldn't use it. And, if so, how I might want to take that into consideration.
normang said: I am not sure if the access to the H.265 version of the Toolkit is T2 Specific, I suspect its related to something in the CPU in the iMac that is not in the older Mini. PLys Handbrake usually always prompts on launch if there are updates, which there were in my case, had not run it in a little while. So when I noted this, it was the latest version of Handbrake.
Hmm, yeah, you're probably right. I guess I do recall people in the Screenflow forum wondering what the new HEVC option was about, and why it took so long on their machines (pre-T2 machines), and I was a bit shocked because it was so quick on mine (T2-equipped). So, the option to use it and maybe use some kind of specialized hardware has probably been there for a while (even pre-T2?). Maybe I'm more surprised the option isn't there at all for the older machine... like they couldn't do h.265 at all, or just not use that Toolkit because the toolkit is hardware oriented?
Yeah, I guess I was thinking more along the lines of some glaring flaw or limitation, such that most pros wouldn't use it. And, if so, how I might want to take that into consideration.
When you say "such that pros wouldn't use it" are you using the "pro" reference as a quality standard ("good enough to satisfy a pro"), or do you mean another "pro" may want to use your video? You probably don't want to be encoding to HEVC or H.264 if you plan to let anyone do anything else to the video later. Both are delivery formats, not intermediates, and are not good sources for later processing.
If you were recording audio, you'd probably capture and edit a wav file, and only convert to AAC or MP3 as the final step. If you send the finished product to someone else for additional processing, you send them the wav, not the MP3. Passing on an MP3 both interferes with later processing and results in a double-compressed final, as it will have to be converted to MP3 or AAC again for final delivery.
If the expectation is that others may work on the material after you're done with it, even if it's just editing portions of your material into their own, you're better off sticking to an all-ProRes workflow. That format strikes a decent balance between quality and file size, and doesn't cause the problems inter-frame compression imposes in H.264 and HEVC.
lorin schultz said: When you say "such that pros wouldn't use it" are you using the "pro" reference as a quality standard ("good enough to satisfy a pro"), or do you mean another "pro" may want to use your video?
The former. And, I suppose I didn't mean pro in that sense either... just more someone who knows the upsides/downsides of such things who can advise me.
I'm worried that I'll love the speed gain so much I'll start putting everything into the format, via this method, and then discover something like.... oh, well, h.265 is OK, but the way the T2 encodes it does xyz which impacts compatibility, or messes up xyz in some quality/capability way that I haven't caught with my untrained eyes/ears, etc.
It seems fine, but I don't want it to be a situation of sacrificing something important because this is some quick & dirty method.
lorin schultz said: When you say "such that pros wouldn't use it" are you using the "pro" reference as a quality standard ("good enough to satisfy a pro"), or do you mean another "pro" may want to use your video?
The former. And, I suppose I didn't mean pro in that sense either... just more someone who knows the upsides/downsides of such things who can advise me.
I'm worried that I'll love the speed gain so much I'll start putting everything into the format, via this method, and then discover something like.... oh, well, h.265 is OK, but the way the T2 encodes it does xyz which impacts compatibility, or messes up xyz in some quality/capability way that I haven't caught with my untrained eyes/ears, etc.
It seems fine, but I don't want it to be a situation of sacrificing something important because this is some quick & dirty method.
It's not impossible that could happen, but the odds weigh in your favour by using the T2 preset. You stand a MUCH higher chance of causing compatibility problems by messing around with encoding settings in Handbrake.
Comments
Shame I can’t see the details in an iPhone. The graphs/tables don’t show in the main article and tapping the thumbnail in the comments just shows the same thumbnail image expanded.
Hopefully you can get yourself both a Kill-A-Watt and in-line USB power meter to measure power at the wall and power used by USB devices (iOS devices) along with whatever Intel’s power gadget tool does.
I’m maybe surprised how little Intel QuickSync reduces transcode time (the Mac mini with no T2 and no QS vs that iMac 4K). Wonder how the transcode is split up and goes to what hardware.
Hoping Apple opens the spigot for iPad Pro so that we can do all this type of stuff a lot more easily. The A12X actually would be in the middle of the pack in a lot of these tests, and could be at the top of these Handbrake tests if it has a T2 in it. Wonder if the Neural Engine can run 8-bit compression/decompression math... probably not. But the GPU might.
I also noted that if you have an older Mac, you do not see the H.265 Video Toolbox. I see it on my 2017 iMac, but its not visible on my 2012 Mini. Only H.264.
Yeah, but I think that was the point of the article. Most of us know the T2 does lots of other things too, from previous articles when they started implementing it. (Though, I guess doesn't hurt to mention.)
I'm curious about the 8-bit thing, though. Are you saying this is a more limited kind of h.265 than most people would typically export? (I'm genuinely curious, as the quality looks OK to me, but I'm no video professional.)
Yeah, it reduces the power used... though I haven't measured by how much. My UPS has a watt display on it, so I could figure it out (assuming it is accurate). But, it certainly doesn't run as hard as it does if it's doing h.264 (and taking longer).
That said, it does run the CPU harder than idle, so the CPUs are doing something (it isn't only using the T2).
Since I came from an older machine (probably more akin to the Mac Pro 5,1 experiment), I was totally blown away by the speed. (ie: 51 min vs 2 min, lol) I was used to setting up a Screenflow project or Handbrake and letting it run overnight. Now, I can just hit export and wait 5 min, or 15 min, etc. Even a Blu-Ray rip isn't a big deal anymore.
Yes, you have to use that video setting... though I'm happy with it so far. I wasn't aware (as the article pointed out) that you have to get other settings within proper ranges. I just used the settings that I was using for h.264 and changed the video to the VideoToolbox, and then tweaked the bitrate a bit. I guess I got lucky.
The mini (2018) has the T2. I wonder if there is an error in the article there, as I wouldn't expect the difference to be quite that narrow. Hmm...
Yeah, I see Mike answered... but it IS NOT one of the presets. I think the word is getting out, but I had to find it digging on some forums when I first got my mini (and it has only been in a fairly recent build of Handbrake... like a couple months, I think). You can pick a preset as a starting point, and then have to go to the video tab and change the 'Video Encoder" setting to the h.265 VideoToolbox one. Then, you might also need to tweak the bitrate.
For other apps, it is done differently. For example, in Screenflow, you just switch to manual and pick the HEVC preset. I figured it out there before I did with Handbrake. (And, of course, some apps might not support it, I suppose.)
I suppose, but the key here is the T2. Even with the i7 6-core in my mini, the time goes way up if I don't use HEVC/h.265 with those settings.
Only for certain things, though. Your iPhone Max probably wouldn't do a very good job of doing some structural analysis in a CAD app or solving some protein folding solutions (or, doing what I currently am, running Autodesk Revit on Windows 10).
Yeah, you'd have to find the settings to see the difference, as there probably wouldn't be a big difference otherwise.
Oh, if you were using that codec, it WAY more than makes up for that (and the machine runs cooler). But, as Mike has indicated.... I'm not sure what the limitations are. Like, if you're using video software that does effects and such, at what point does it fall-back to not using the T2? It's probably just for the actual encoding, I'm guessing the rest has to use the CPU/GPU... but I don't know if there could be some mix.
Like, when I do Screenflow projects, I'm guessing it has to do more than a straight re-encode, but my times dramatically decreased (though I wasn't coming from a speed-demon machine previously. But, way more difference than clock-speed, CPU generation).
Hmm, that's odd, as the 2017 iMac wouldn't have the T2 either. Are you sure you have the latest Handbrake on both? They just added that capability in the most recent version (a couple months back).
Your mileage may vary based on driving conditions.
yeah, I’m also hoping for that big iOS change for the iPad that was supposedly delayed from last year. Hopefully, that year delay allowed them to continue work on it for this year, adding changes and functionality that would have come in the second year. I hope they didn’t basically shelve it, and just tweak it for this year. I want to see real progress.
i’m excited over the possibilities. A different UI, or so we’re led to believe. If so, I hope it’s really major, not just a shuffling around of what’s there now. The iPad Pro in particular, needs to move from the big iPhone UI configuration. I think we can handle it. ;~) The 12.9” screen, and even the 11”, is big enough for windowing, not this sliding arrangement, which while useful, is annoying and ridged.
so three things would totally change the way we look at, and use it. I think we all know what that is.
trackpad support
a true mass drive capability through USB C and wifi
hierarchical filing system.
something some people get worried about with all of that is that they think they will have to use it, and that’s not true. That is, Apple can keep things mainly the same for those who want that. The way they expose the filing system now, while not nearly enough, is a good example. You don’t have to use it if you don’t want to, but it’s easy to access if you do. If you don’t want a trackpad, you won’t see it with no keyboard, and possibly, some keyboards won’t have it. Though, if one is using a keyboard, I don’t understand why they wouldn’t want one.
if you don’t connect a drive, then that doesn’t matter either. Windowing can be a selection in settings. A new section dealing with all the new aspects could be added. Choose which of them you want to see, and use, if any. None could be default, at least for the first few years until people get used to it, which I bet they would.
if Apple does this, I would bet sales would leap, though sales of MacBooks might slump. Hard to tell.
I’m just commenting on core strengths. You might remember a time when it was said that there was no possibility that ARM could ever hope to even come close to x86 in any way. What I see is that if Apple decided to change the parameters a bit, we could have two 4 core A12x chips in a machine. If there was adequate cooling and power, the efficiency cores could be removed and the room used to power up the other components. Considering that the SoC is it’s own support chip package, and contains a good GPU, with everything doubled, at least, including the ISP and the AI section, that could really be something.
apple could also decouple the GPU from the chip and have a separate one with its own graphics memory. That would enable the possibility of competing with more powerful versions from AMD and Nvidia.
the possibilities are endless.
On the other hand, it's plenty fine for the YouTube stuff I've been doing... and saves a TON of time.
I think there were little hardware encoders for h.264, so I'd imagine some exist or will be coming. There's no reason something like this shouldn't be able to hang off a USB or TB port. Maybe there just isn't enough demand? (I've never even looked though.)
Great list. Yes, what is holding iOS from becoming a pro solution (at least for me) is like 90% on the OS side of things, and maybe the other 10% lack of apps. I need to do a better re-evaluation one of these days, but just from what I hear/read, it doesn't seem things are that much different than when I wrote my review of the iPad 2 for such things. Most of the same workflow contortions seem to be the case.
Oh yeah, there are some big possibilities. But, I think the Windows thing is hard to solve. Without direct hardware, it would have to be many times faster to emulate.
Through lots of training and experience, I'll hear things in a recording most people won't notice. That means what I consider "good quality" will differ from what's good to someone else. Does it matter that a particular flaw exists if one doesn't notice it, or at least isn't bothered by it?
On the other hand, you may become more critical with time. I now notice flaws in transcodes I did five or six years ago that apparently didn't bother me at the time I did them. Maybe that means one should err on the side of higher quality as a hedge against more sophisticated perception in the future.
The bottom line: if in doubt, increase the bit rate!
I had this discussion before, from what I know, x86 processors we have today has been held back by its legacy. Backwards compatibility are the biggest strength since it retains so much even from 8086, but that hurts their efficiency bad. Hence why A11 outperforms x86 rivals with higher frequencies in single core tests.
Hmm, yeah, you're probably right. I guess I do recall people in the Screenflow forum wondering what the new HEVC option was about, and why it took so long on their machines (pre-T2 machines), and I was a bit shocked because it was so quick on mine (T2-equipped). So, the option to use it and maybe use some kind of specialized hardware has probably been there for a while (even pre-T2?). Maybe I'm more surprised the option isn't there at all for the older machine... like they couldn't do h.265 at all, or just not use that Toolkit because the toolkit is hardware oriented?
If you were recording audio, you'd probably capture and edit a wav file, and only convert to AAC or MP3 as the final step. If you send the finished product to someone else for additional processing, you send them the wav, not the MP3. Passing on an MP3 both interferes with later processing and results in a double-compressed final, as it will have to be converted to MP3 or AAC again for final delivery.
If the expectation is that others may work on the material after you're done with it, even if it's just editing portions of your material into their own, you're better off sticking to an all-ProRes workflow. That format strikes a decent balance between quality and file size, and doesn't cause the problems inter-frame compression imposes in H.264 and HEVC.
I'm worried that I'll love the speed gain so much I'll start putting everything into the format, via this method, and then discover something like.... oh, well, h.265 is OK, but the way the T2 encodes it does xyz which impacts compatibility, or messes up xyz in some quality/capability way that I haven't caught with my untrained eyes/ears, etc.
It seems fine, but I don't want it to be a situation of sacrificing something important because this is some quick & dirty method.