Video files don't run, they're viewed or manipulated (i.e. editing).That's a file size, not the amount of memory required to run it.
Video files don't run, they're viewed or manipulated (i.e. editing).That's a file size, not the amount of memory required to run it.
Well that may be your thinking but the M1 Unified Memory behaves alot different than your standard DDR3/4 RAM.Yes, they are. A 2GB video file consumes 2GB regardless of whether it's being processed by an Intel or ARM based system.
You appear to be confusing performance with memory requirements. The unified memory architecture is useless when the system needs to pull memory from secondary storage (i.e. the SSD). The speed of the M1 and its unified memory becomes irrelevant at that point. You 16GB i7 Mini Server would outperform the M1 on tasks which require 16GB of memory.Well that may be your thinking but the M1 Unified Memory behaves alot different than your standard DDR3/4 RAM.
My new base 8GB M1 is amazing - it runs circles around the 16GB quad core i7 Mini Server!
Plus no more heat pumping out the back, no more fan spin ups, no more lag and the best part: - it runs cool all day long!
Now I can get rid of the laptop cooling fan underneath it and sell my old Intel i7 Mini!
Hello M1 !
Hi, I’ve just tested Unity with a current project on an 8gb mbp, it does swap out quite a chunk (~2-3gb for my medium size project) however... I haven’t actually noticed any obvious performance drop caused by this, it’s very fast- e.g. it’s running my game 60fps 1440p full screen, where my old 2015 13” mbp I could barely run it at 30fps 640x480!Hello,
I'm looking at buying my son a Mac book Air for homework and he has started to learn coding and uses unity .
I'm from the pc world ( ducks down ) Is the 8gig base Mac book Air going to last him a 5 years or so ? Or would i be best putting 16Gig in the machine ?
Pc wise it's always more ram the better but reading here i;m unsure wit the Mac book Air . And also is the 7‑Core GPU compared to the 8‑Core GPU noticeable for standard things ?
Thank you
I don’t know if you’ve been paying attention to the real world usage reviews or not....No, it's not. ARM does not magically reduce memory requirements by 33%
Right... they are files with a file size... such as 2GBs. That is not the amount of RAM necessary to view, edit, or store it.Video files don't run, they're viewed or manipulated (i.e. editing).
I am paying attention to my many years using computers, my degree in computer science (i.e. programming), many years in IT (and still currently so), and my enjoyment of all different types of computers.I don’t know if you’ve been paying attention to the real world usage reviews or not....
If you're going to edit a 2GB file there are multiple ways it can be done:Right... they are files with a file size... such as 2GBs. That is not the amount of RAM necessary to view, edit, or store it.
The amount of RAM required to run anything is not equal to the size of the file. You suggested that a 2GB file requires 2GBs of RAM to run (run means view, edit or otherwise manipulate it in any way).
What people are trying to say is that the SoC allows memory to be accessed more efficiently than it does on a non-SoC, thus it *probably* doesn't require the same amount of RAM for the same performance that a non-SoC does.If you're going to edit a 2GB file there are multiple ways it can be done:
So yes, it's possible to edit a 2GB file on a 1GB system. However doing so slows down the process. Why hobble a fast processor by having it wait on, relatively speaking, slow secondary storage?
- The file is loaded entirely into memory. Where memory is available this is usually the preferred method.
- Parts of the file is copied from secondary storage into memory as the parts are needed. The overhead of copying the file from secondary storage to memory is incurred throughout the editing session.
That said none of this means the M1 requires less memory than an x64 system.
The manner in which the processor interacts with memory is not relevant when there's not enough memory to store the information the processor needs therefore causing the memory system to retrieve information from secondary storage. The Unified Memory configuration of the SoC applies to the CPU and GPU having access to the same memory so that there is no need to copy memory for use by the GPU. It has little, if anything, to do with secondary storage (i.e. the SSD). The fact that the SSD is on the same silicon as RAM likely reduces latency but not sufficiently so to eliminate the need for RAM. Otherwise Apple would just ship an 8GB system and call it a day.What people are trying to say is that the SoC allows memory to be accessed more efficiently than it does on a non-SoC, thus it *probably* doesn't require the same amount of RAM for the same performance that a non-SoC does.
So if you can edit a 2GB file with a 1GB RAM system slowly on a non-SoC system, then it stands that you *may* be able to run it fast on an SoC system with the same amount of RAM. People bump up the RAM to improve performance just as much to meet a hard minimum requirement. Thus you can probably say that you don't need as much RAM on a highly optimized architecture versus a less optimized architecture... which means do more with less.
That said I suggest if you are going to discuss this with me that you first obtain an understanding of the topic at hand.
Memory transfers are significantly fast than a modern SSD. DDR4 memory is approximately 20GB/sec whereas the fastest SSDs are approximately 3GB/sec. SSDs have narrowed the gap but not sufficiently enough.I wonder how much faster RAM is than secondary storage these days. Both have made gains, but I get the impression that SSD speeds have increased much more quickly than RAM. Is it a 2x difference? 10x difference? 100x? I wonder if I'd even notice that my computer is swapping.
That's fine, they're free to remain ignorant and avoid having a discussion with me. It won't make their comments any less wrong but it might make them feel better.I think you’re doing a pretty great job of ensuring nobody will want to discuss it with you.
Not even 8 - Apple SOC wouldn't even need RAM at all.The fact that the SSD is on the same silicon as RAM likely reduces latency but not sufficiently so to eliminate the need for RAM. Otherwise Apple would just ship an 8GB system and call it a day.
Was this the case all the way up until OS 9? I just don't remember people adjusting their RAM per application, neither on Windows 95. I thought all operating systems manage RAM just as MacOS does today.
I will be surprised if there is anything magic about M1's memory usage. The only architectural difference I can think of is that ARM machine code typically results in about a 5% larger code size, because x86 has better instruction density. But actual machine code is a very small portion of RAM usage.
Unified memory is also not magic. An Intel processor with integrated graphics also uses the main system RAM for GPU resources.
I would be surprised if Big Sur running on Apple Silicon has massively different memory usage to the same OS and same apps running on Intel, but I could be wrong.
The one advantage Apple does have is its extreme vertical integration. With absolute control over both the chip and the software running on it, they can write their drivers and OS to be as efficient as possible. Metal running on Apple Silicon may well avoid some copies, because M1's GPU doesn't need to pretend to be a discrete GPU and can expose the unified memory to the app.
Saying all that, 8GB still sounds a lot of memory to me. I remember when computers had a megabyte, or less!
If apps chew through that much RAM (unless they are actually dealing with massive datasets) then they need to be optimised. It really sucks how every time we get hardware improvements basic everyday apps (and websites) become less efficient in turn and expand to suck up the available resources.
I gave in and ordered 16GB because I always over-buy and am thinking I might want to run a VM or two, but honestly my current 5 year old Mac with 8GB is still fast and capable as anything. People worrying about 8GB not being enough for word processing or web browsing or photo editing are worrying for nothing, IMO.
This is not necessarily true. I have 5 hour 1080p h.264 video files that are over 200 GB in size. I do not have 200GB of RAM to load it all up.I am paying attention to my many years using computers, my degree in computer science (i.e. programming), many years in IT (and still currently so), and my enjoyment of all different types of computers.
A 2GB file consumes 2GB regardless of which architecture it is on. Been that way since the 68K days, been that way since the PPC days, and it's still the same today with x64 and M1.
Yes, they are. A 2GB video file consumes 2GB regardless of whether it's being processed by an Intel or ARM based system.
Yes, I understand basic math. I’m talking about *performance*. These machines are flying with less RAM compared to the Intel offerings.I am paying attention to my many years using computers, my degree in computer science (i.e. programming), many years in IT (and still currently so), and my enjoyment of all different types of computers.
A 2GB file consumes 2GB regardless of which architecture it is on. Been that way since the 68K days, been that way since the PPC days, and it's still the same today with x64 and M1.
^^^Yes, I understand basic math. I’m talking about *performance*. These machines are flying with less RAM compared to the Intel offerings.
OK, your point being?This is not necessarily true. I have 5 hour 1080p h.264 video files that are over 200 GB in size. I do not have 200GB of RAM to load it all up.