Chiroyce wrote:

b1048546 wrote:

An ISA itself does NOT neccerscarily provide any performance or efficiency gains!

x86 can take multiple clock cycles to complete an instruction - whereas ARM64 always takes one. isn't this a efficiency gain?

They both have their advantages. The “intel and AMD need to switch to ARM quick” thing is a myth. Meet Alder Lake.

MagicCrayon9342 wrote:

Apple needs to work on MacOS's terminal… a lot.

MacOS's terminal is great! Of course, third party terminals like Iterm2 are to some extent, but terminal works very well. You get Bash, ZSH, fairly customizable options, and it has enough features so you can use it instead of the GUI. It's faster too. I'd much rather use terminal because of the speed. It isn't perfect, but it's built on Unix which means it's very similar to the Unix/Linux terminals. (IIRC, 99% of the commands in Linux is the same in MacOS.)

b1048546 wrote:

MagicCrayon9342 wrote:

Apple needs to work on MacOS's terminal… a lot.

MacOS's terminal is great! Of course, third party terminals like Iterm2 are to some extent, but terminal works very well. You get Bash, ZSH, fairly customizable options, and it has enough features so you can use it instead of the GUI. It's faster too. I'd much rather use terminal because of the speed. It isn't perfect, but it's built on Unix which means it's very similar to the Unix/Linux terminals. (IIRC, 99% of the commands in Linux is the same in MacOS.)

you can also grab busybox and use all of the linux utilities… right?
https://busybox.net/

MagicCrayon9342 wrote:

b1048546 wrote:

MagicCrayon9342 wrote:

Apple needs to work on MacOS's terminal… a lot.

MacOS's terminal is great! Of course, third party terminals like Iterm2 are to some extent, but terminal works very well. You get Bash, ZSH, fairly customizable options, and it has enough features so you can use it instead of the GUI. It's faster too. I'd much rather use terminal because of the speed. It isn't perfect, but it's built on Unix which means it's very similar to the Unix/Linux terminals. (IIRC, 99% of the commands in Linux is the same in MacOS.)

you can also grab busybox and use all of the linux utilities… right?
https://busybox.net/

OSX version?

b1048546 wrote:

(IIRC, 99% of the commands in Linux is the same in MacOS.)

This. This is how I learnt 99% of the Linux commands I know today, by using it on macOS. The 1% is probably this

Chiroyce wrote:

b1048546 wrote:

(IIRC, 99% of the commands in Linux is the same in MacOS.)

This. This is how I learnt 99% of the Linux commands I know today, by using it on macOS. The 1% is probably this

sudo rm -rf /

That would be hilarious!

Chiroyce wrote:

b1048546 wrote:

(IIRC, 99% of the commands in Linux is the same in MacOS.)

This. This is how I learnt 99% of the Linux commands I know today, by using it on macOS. The 1% is probably this

sudo rm -rf /

i learned linux commands via google, guides, manuals, and –help args

i learned commands via i cant remember

MagicCrayon9342 wrote:

Chiroyce wrote:

b1048546 wrote:

(IIRC, 99% of the commands in Linux is the same in MacOS.)

This. This is how I learnt 99% of the Linux commands I know today, by using it on macOS. The 1% is probably this

sudo rm -rf /

i learned linux commands via google, guides, manuals, and –help args

Google is my favorite way to learn about terminal.

MagicCrayon9342 wrote:

i learned linux commands via google, guides, manuals, and –help args

I learnt basic commands from YouTube, an d more advanced ones using -h or –help until I found out about
replace curl with command name

Also I used Windows for 5 years and I don't know a lot of commands in CMD, but I know so much more about bash and zsh after using macOS only for 7 months. Best shell scripting language ever.

So here's the best advantage of curl and the macOS terminal along with bash/zsh.

I have a camera, it uses an SD card slot. My Mac doesn't have one, and the dongle isn't with me at the moment. The camera can host a http server in it's own WiFi network, so I connected to that, and got the URL structure for the website. It was <IP_ADDRESS>/DCIM/<camera_name>/<img/video-ID/

So I made a note of the Video/Image ID on the camera, connected my Mac to it's WiFi, and used curl to download images and videos (I used > to save the output). It was slower than SD card speeds, but hey, something is better than nothing. And I couldn't have done this on Windows (a windows laptop without an SD card slot) .. going to the website and downloading didn't work since browsers would get the videos in parts, so I can't download it as one file. But curl made it possible!

bump

dude someone remade the old itunes and macOS gadgets, as well as iOS 4 in swift.

{what method did you use::control hat
answer on profile ::motion
} ::operators

Apple M, future Apple chipsets, and future Macs deep-dive. (M1 Max x 4 and M1 Max x 2.) I won't go into leaks too much, but I will mention them as they are from quite accurate sources.

Let's start with the basics. M1 has 8-CPU cores. 4 high-performance and 4 high-efficiency. This is referred to as big.little. Big cores are performance cores and little cores are efficiency cores. Efficiency cores are used until their performance is not enough for a task and the task is switched to a high-performance core. How does Apple do this? They have a controller that picks the correct core for each task in the chip. M1 uses LPDDR4X memory. LPDDR is low power DDR memory. It runs at 3.2GHZ clock speed, which is not important. M1 has a media engine that makes specific media workloads faster.

M1 has up to 8-GPU cores. This is currently the best integrated graphics (besides themselves) in the world. It is fast and speedy. It has good performance compared with Intel integrated graphics, but the one in M1 does not compare with high-end graphics cards. M1 does go into a thin and light notebook, so this is to be expected.

There is up to 16 gigabytes of unified memory. What is unified memory? It's memory that can be shared with the CPU and GPU. It's very good for reducing latency as the CPU and GPU can share the memory much faster. The latency is reduced because they get direct access to the memory pool! Unified memory does not make the RAM itself magically faster though. Its only use is for reducing latency. Instead of taking a long time copying the memory, you just look at the memory block and share it. The CPU and GPU can work on it at the same time.

M1 has 68GB/s memory bandwidth and 8-channel memory. Memory bandwidth is basically how fast data can be stored or extrapolated from the chip. 68GB/s is unheard of in the industry. Imagine memory channels as a a RAM driving to a place called CPU. If there's only two lanes, (two being memory channels and two is common) it takes longer to get to the destination because of traffic. 8 channels is like having 8 dedicated lanes to each chunk of memory. 8-channel memory is equivalent to AMD EPYC chips! Those are fast server chips and M1 is equivalent to them! A desktop chip, such as the 2019 Mac Pro chip has 6-channel memory. This is very apparent in tasks such as Photo Editing.

Not much to explain 16-core Neural Engine. This is the only part of the chip that has not been upgraded from the A14 IP. The Neural Engine is used for machine learning. Machine learning is used in many Apple applications such as Final Cut Pro. They use it to make the overall experience of FCP better. Some benchmarks don't show the full potential of M1 because they run tasks on the GPU when Apple has optimized those tasks for use on the Neural Engine. It shows how much proper optimization matters. The speed will bring developers though. I've seen lots of developers get the new M1 Pro/Max MacBook Pros.

M1 has enough IO bandwidth for even the Mac Mini, however the Mac mini only has 2 Thunderbolt ports instead of 4 because of the lack of memory bandwidth on M1. Hopefully, later generations solve this problem by adding just enough memory bandwidth for 4 Thunderbolt ports. The M1 Mac Mini has 2 Thunderbolt ports, 2 USB-A ports, HDMI 2.0, up to 10-gigabit Ethernet, and a headphone jack. For the Mac Mini, M1 has support for up to two external displays. For the M1 Air and Pro, that goes down to one.

Are M-Series chips based off the iPhone chips or are iPhone chips based on M-Series chips? Technically speaking, the Mac chips are not based on the iPhone chips. Rather, they're based on the IP of the iPhone chips and they can add anything they want. This is apparent for M1 Pro and Max. (More on these chips later.) What do I mean? The Mac and iPhone chips are based on each other. However, an iPhone chip will get iPhone features built in and a Mac chip will have Mac features built in. You start off with one chip, then scale it up or down and build on it. A14 has 6 cores. (2P and 4E) That was scaled up to M1 Pro/Max which have tons of features that A14 didn't have.

ARM vs x86. x86 is the architecture used for Intel and AMD. ARM is the architecture used for Apple chips and phone chips in general. A common misconception is that one architecture makes a chip better than the other. Not true, the chip architecture matters more than the ISA. ARM has two licenses, an ISA and a license for their cores. An ISA license is where you get to build off ARM, which is what Apple does. A license for ARM's cores is a license which gives you access to the cores ARM designed themselves. Fun fact, NVIDA owns ARM.

M1, M1 Pro, and M1 Max are built on TSMC's 5 nanometer process node. What is a process node? A simplified explanation is that when the process node shrinks, the chip either gets more efficient with the same performance, or faster with the same efficiency. Generally, Apple uses a little bit of both but they typically want the chip to get faster. A process node is generally just marketing, but Apple has the best process node on the market. Apple usually redesigns their processors after a node shrink. By the way, just because AMD/Intel aren't on the same node, it doesn't mean we should stop comparing them. If they couldn't get their processors on the same node, then that's on them. Apple is usually ahead of everyone else on this front.

Performance per watt. For every single watt (energy) that is used in an Apple Silicon device, the performance output is is better than any chip out there. Think of it this way. Imagine that you're trying to flood a place as fast as possible. One of the ways is one garden hose and the other is five garden hoses. Clearly, five garden hoses is many more than one, so the place will be flooded quicker. It's efficient because every minute or so goes by, the 5 garden hoses are much closer to flooding the place to the one. This is why M1 is cool, fast, and has good battery life. Does the Air without a fan thermal throttle? Yes, but rarely. The chip is so efficient and fast, so the CPU/GPU can process the task quickly enough so it won't be running long enough for thermal throttling. The cooling system in M1 MacBook Pro and M1 Mini can keep up too! Apple has better performance watt on their CPU and GPU than AMD/Intel/NVIDIA. What does this mean? Once Apple scales up their processors to Mac Pro level (128-GPU cores and 40-CPU cores), they will be very fast and use much less power than competitors. A 128-core GPU Mac Pro would be faster than an RTX 3090 since GPUs scale well.

What is chip binning? Chip binning is where you take some cores and flat-out disable them. This way, companies can save money by a defective core due to manufacturing not being perfect. Not only this, but companies can make an upgrade path without having to make many processors. Apple's processors are very cheap and this saves them even more money. Keep in mind, Apple spends billions on chip R&D. The M1 MacBook Air is binned to 7-core GPU and can be upgraded to 8.

Let's move onto M1 Pro and Max. I'm mixing them as one because some things are the same in these chips. M1 Pro and M1 Max are based on M1 or A14 IP and not the A15 IP. 10-core CPU for the M1 Pro/Max (binned to 8 for base model MacBook Pro. 6P and 2E cores.) 8-high performance cores and two efficiency cores. Why only two efficiency cores? There's a lot of special features that take up space so they decided to use 2-efficiency cores. The battery life is still really good. How? Since tasks are finished faster and there's an amazing cooling system (more on that later), you're using more power for less time, bring similar or even better battery life.

Watching video can go over 24 hours on even the 14-inch MacBook Pro with a smaller battery. The cores are really fast but they don't scale perfectly. Why? It would ruin Apple's focus on performance-per-watt and the power to get it there is not worth it. They also run at 3.2GHZ clock speed like M1 which is surprising since more cores usually means smaller clock speeds. M1 Pro and Max add Pro-Res accelerators which speed up that type of media by a lot. Even a maxed out Mac Pro with an afterburner cannot beat an M1 Pro/Max at this workload. M1 Pro and Max have bandwidth for 3 Thunderbolt 4 controllers, 1 HDMI 2.0 controller, and 1 SD card controller.

There's a 16-core GPU (binned to 14 for base model MacBook Pro.) for the M1 Pro. The graphics performance is good and competes with high-end graphics cards. There's a 32-core GPU (binned to 24) for the M1 Max. This is the fastest iGPU in the world and competes with very high-end mobile graphics cards while using very little power. M1 Pro and Max keep the same 16-core neural engine from the A14 IP. The neural engine can perform up to 11 trillion operations while the A15 can perform 15 trillion.

M1, M1 Pro, and M1 Max have a Secure Enclave for security which is new. It's not much new, it's just slightly enhanced security. All the chips support PCIE 4 which is interesting for the Mac Pro. (More on that later.) M1 Pro supports up to two external displays and up to 32 GB unified memory. M1 Pro has an insane 200GB/s memory bandwidth (More than double M1.) and an absurd 16 channel memory. That's 16 lanes for the 32 GB to travel from! This also means a laptop GPU has access to 32 GB memory. All chips have an ISP which makes the image look better. The new MacBook Pros actually have a good webcam unlike the MacBook Air/Pro. M1 Pro and Max also have machine learning accelerators to make the neural engine much faster even with 16-cores.

There are 33.7 billion and 57 billion transistors respectively for the M1 Pro and M1 Max. What are transistors? They're not too important, but they make tasks more efficient and faster. 57 billion transistors is unheard of! Even 16 billion on the M1 is impressive. Each side of the M1 Pro has two blocks of memory on each side and its own dedicated memory bus on each side. That means M1 Pro has a 256-bit memory bus! That's why we have 200GB/s memory bandwidth. M1 has 24 MB L2 cache which is a lot of cache and is faster than the L3 cache typically used. The cache on M1 Pro/Max is about 48MB and 96MB respectively. That's a lot of cache for a device so small. This cache is helpful for storing memory and each block of cache is used for different things. The 16 memory channels on the M1 Pro is split into 8-channels on each side. All of this applies to the base M1 Pro chip.

M1 Max starts at 10-CPU cores, 24 GPU cores, and 32 gigabytes of unified memory. It is 3.5x bigger than the original M1 chip. It has 96MB L2 cache which is a lot! It goes up to 32 GPU cores and 64 gigabytes of unified memory. This means the GPU can have up to 64 gigabytes of memory without any copying! That is unheard of for even desktop graphics cards. M1 Max has a 512-bit bus which is double the M1 Pro. This allows an interface for up to 400GB/s which has never been possible before for even server chips. Even the base M1 Max comes with 32 channels of memory. Double M1 Pro. Don't know how crazy that is? AMD's best server chips have 8 memory channels which even M1 matches. That's 4 times the memory channels! That's 32 lanes for the memory to go through.

M1, M1 Pro, and M1 Max are so snappy because they have one of the fastest single-core performance in the world. For M1 Pro, 30 watts maximum for CPU and 30 watts maximum for the GPU. Only 60 watts max for these MacBook Pros with really good performance! Simply put, the cooling system can handle at minimum 106 degrees Celsius without thermal throttling while still being able to put it on your lap and still have great battery life. The new MacBook Pros are so fast with limited amount of ram because they have super fast SSD's. Like really fast SSD's. 1TB storage is the sweet spot for this in my opinion. MacBooks don't lose their performance while they're plugged in either. In my opinion, this is great because I don't like being near a wall all the time. The GPU is capable of gaming but MacOS isn't very good for that. For GPU comparisons, I don't like teraflops, so I'll stick with GeekBench Compute. Geekbench Compute isn't optimized for M1 Max so it gets only a score of about 68,000. If it was optimized, M1 Max could get a score of 80,000+ while using much less than 30 watts. Why is it not optimized? The frequency doesn't ramp up properly so it's not really testing the full M1 Max GPU. CPU gets about 12,500 which is really good.

Alder Lake. I'm not going to spend too much time here. Alder Lake is a really good improvement from Intel compared to previous generations and it competes with the M1 Pro/Max. However, Intel has seemingly ramped up the voltage and wattage of these processors by a lot. Not to mention the M1 Max X2 and X4 that I'll talk about later. Intel also brings in big.little! That is for sure an improvement but the efficiency cores aren't really efficient and use a lot of power. I am skeptical with mobile Alder Lake though. This shows that while mobile Alder Lake competes with M1 in single-core, M1 still beats it out in multi-core. This is the mobile i9. I'll have to wait for more Alder Lake mobile benchmarks to form an opinion though. If Alder Lake beats M1 Pro/Max, I wouldn't be too surprised unless it's not at higher power.

What are chiplets? The manufacturing process isn't perfect, so sometimes chips are not that great. This is especially apparent with high-end chips. Chiplets solve the problem. Chiplets are basically tiny parts of a Chip that are put together to make a giant chip. This means it's easy to not only scale up small chips to big ones (with threadripper), but it saves money because you're not wasting huge chips to very small defects. Use the parts that work, and either throw away the ones that don't work, or if your product includes binning, just add the chip to the binned product without any extra chiplets.

Apple Silicon 27“ iMac/iMac Pro and the inclusion of possibly M1 Max X2. Reminder, everything below is speculation on my part but we will be getting an M1 Max X2. I believe that the new iMac will start with M1 Pro, upgradable to M1 Max, and maybe even a third M1 Max X2. What is M1 Max X2? It is two M1 Max chips combined using chiplet technology. That means double everything, and Apple can even add custom things to suit the products that it will go into. Up to 20-CPU cores, up to 64-GPU cores with up to 128 gigabytes of unified memory, double the media engine, 32 neural engine cores, and more.

Why do I think we'll get M1 Max X2? The Intel iMac has up to 128 gigabytes of storage while M1 Max gets only 64, M1 Max isn't that much faster than the highest-end GPU in the iMac (it is faster, not that much though), Cinebench R23 scores are better in the highest-end (expensive), and MacOS is already built for M1 Max X2, but not M1 Max X4. How is that helpful? If the iMac comes before June (when new MacOS is typically released), then it will be ready for 2 chips if Apple wants to do that. How will they fit the chip in the chin? Simple, they won't. They can still have heatpipes in the chin, but the chip will be in around the middle of the iMac.

My prediction for the GeekBench and Cinebench score is about 22,500 which should compete with desktop Alder Lake. The 64-core GPU would be nearly twice as fast if scaled well. It will have 800GB/s memory bandwidth, four AMX multiplication blocks, eight Thunderbolt 4 controllers, 32-core neural engine that can be used as two 16 cores at the same time, double media engine, and likely double the memory channels for a whopping 64. Will we have upgradable unified memory? No, but we could get something else. We could be able to upgrade regular RAM which would act as super fast swap, but never as fast as unified memory. This is what I expect in the Mac Pro.

I expect the iMac to start at $1999 for the same configuration as the base 14” MacBook Pro or $2299 for the non-binned version. I expect the fully speced out iMac will be $3499. Upgrading to M1 Max X2 will require you to upgrade to 64 gigabytes of unified memory. I expect the CPU to be binned to 16 cores and GPU 48 cores. 16 cores will be 12 performance cores and 4 efficiency. 8x2 of 14“ MacBook Pro and 24x2 for the M1 Max binned version. Then you can upgrade to 20 CPU cores, 64 GPU cores, and the full 128 gigabytes of RAM which will be expensive. (LPDDR5 is expensive, the RAM prices on the MacBook Pros are very reasonable.) Lifuka is a codename for a discrete GPU that Apple was/is working on. If it is used, I expect it to be used as an add-on card for the GPU in the iMac/or/and the Mac Pro. If we don't get M1 Max X2, I expect Apple will do something about the RAM and GPU. I expect the iMac to have thin black bezels without a notch, 1080p webcam, 27”, Pro-Motion, Mini-LED, and all the ports of the MacBook Pro.

The Apple Silicon Mac Pro. I believe that the Mac Pro will start with M1 Max X2 binned. 48-core GPU, 16-core CPU, 64 gigabytes of RAM, and all the other features of M1 Max X2 I mentioned with the iMac. You would be able to upgrade to 20 CPU cores, then 64 GPU cores, and 128 gigabytes of unified memory so the GPU has up to 128 gigabytes of memory with only M1 Max X2. What else? M1 Max X4. Up to 40-core CPU, 128-core GPU, 256 gigabytes of unified memory/VRAM, and double everything in M1 Max X2. M1 Max X2 will use at worst 184 watts and M1 Max X4 at worst will use 368 watts. At max power but not max load, the power would be 120 watts and 240 watts respectively. That matches/is the same as desktop Alder Lake and will be much more powerful. Not just the CPU, the entirety of that Mac Pro and iMac package would use similar power to Alder Lake.

Apple's binning strategy is genius. The M1 Max has two neural engines just for a defective one to save money! It's on top of the first one. This means if Apple really wanted to, M1 Max X4 would have a 128-core neural engine. The media engine will be multiplied by 4 from the original M1 Max. An M1 Max X4 would get a GeekBench/Cinebench score of about 44,000 which is much better than Alder Lake. The M1 Max X4 would have 8 AMX chips. for matrix multiplication, 128-channel memory, 16 Thunderbolt 4 controllers, 12 6K displays, and 4 4K displays via HDM The 128-core GPU in M1 Max X4 would beat the RTX 3090. However, I would still get a gaming PC for gaming. All of this at much lower power than the competitors. How? Performance-per-watt. If Lifuka is used, I expect it to be used as an add-on card for the GPU in the Mac Pro and iMac.

A couple of my crazy ideas. Apple could have upgradeable RAM by making it so that it's like super fast swap and not unified memory. Unified memory stays the same, but you can sort of upgrade dedicated swap that's very fast. That would be possible. Apple could also offer SOC upgrades, but probably not. I do expect them to do something though. I think Apple will use MPX or PCIE 4 lanes to let you add binned/defective M1 Max dies. This would improve many things and this is more reasonable for them than SOC upgrades. I still hope for SOC upgrades though. (I mean being able to buy an SOC and replace it.) They can also make custom GPU chips without CPU cores or the other way around. This would make it really powerful for a long time.

Pricing, configurations, and final thoughts on RAM. 256 gigabytes of unified memory isn't the same as 1.5TB RAM. I'd say there's two options. Either my super fast swap idea is true, or this will not replace the Intel Mac Pro. There was a leak about there being a new Mac Pro to replace the Intel one and that might use an Intel processor. That was an old leak so things may have changed though. The Apple Silicon Mac Pro will be out as early as June. How do I know? 4 chiplet support isn't built into MacOS yet! WWDC is typically when a new version of MacOS is released. I expect the price to be $3000 with the money Apple will be saving. Base model will start with 1TB and up to 16TB. 48-core GPU, 16-core CPU, and 64 gigabytes of unified memory. I expect either Apple will switch to HBM2E memory for bandwidth or stick to LPDDR5. Perhaps, an optional 1TB super fast swap too. I expect the maximum configuration will be at least $8000 which is crazy for he performance. It'll be the fastest video and photo editing computer in the world for sure. Perhaps you can buy the custom GPU cards along with the Mac Pro. Will the Mac Pro be upgradable? I think yes, to an extent.

M2 MacBook Air. I expect a redesign with white bezels, white keyboard, and colors like the iMac. I wouldn't be surprised if there's a white notch with thin bezels. I also expect a 1080p webcam. M2 could be based on either the A15 or A16 IP. I'm hoping for A16. 8 CPU cores and 9/10 GPU cores. If M2 is based on the A15, with 10 core I expect the GPU performance to be around the Radeon Pro 570 or RX 570. That's very impressive for an iGPU! I'm not sure about the CPU improvements though, but I expect an efficiency improvement because even A15 had that. I still expect the M2 Air to be fanless, have MagSafe, have a 30 watt power adapter, Mini-LED, thinner than M1, and 2 or more displays. Based on leaks, it will be more expensive and a higher-end model of the M1 Air.

This is the Mac, Apple Silicon, and the future of Apple Silicon. If you already know a chunk of this, skim or skip the part, I tried to make it easy to be able to skim through parts you already know about. Correct me if I'm wrong in a certain area too, thanks. You can ask me a question too.

The craziness of Apple Silicon starts now. We have so many more chips to go. Just know that this all started from an iPhone chip.

b1048546 wrote:

-huge snip-

WOAH, how long did it take for you to write that all? Cool!!!

god286 wrote:

b1048546 wrote:

-huge snip-

WOAH, how long did it take for you to write that all? Cool!!!

It wouldn't have taken that long if I remembered everything off the top of my head. About 2 hours. I'm pretty sure there's some things I forgot or that I didn't explain thoroughly enough but I'm willing to edit it. 21,720 characters. The scratch forum limit is 200,000. Thanks by the way!

M2 and M2X When

b1048546 wrote:

8-high performance cores and two efficiency cores. Why only two efficiency cores? There's a lot of special features that take up space so they decided to use 2-efficiency cores. The battery life is still really good. How?

Battery is bigger as well, right?

b1048546 wrote:

M1, M1 Pro, and M1 Max have a Secure Enclave for security which is new. It's not much new, it's just slightly enhanced security.

Replacement for the T2 Chip, right? (found in Intel Macs)

b1048546 wrote:

The frequency doesn't ramp up properly so it's not really testing the full M1 Max GPU.

I've heard that it's because the benchmark is a very short one - so the chip thinks it's a light workload and may reduce the frequency in order to preserve battery life. Making it longer should help, right?

b1048546 wrote:

This shows that while mobile Alder Lake competes with M1 in single-core, M1 still beats it out in multi-core.

wait what - that seems like a fake score. Why would a Intel Core i9-12900HK only get 6438 in multi core?
I guess it's because it's on a laptop and it's thermally limited. What a shame. I thought Alder Lake for laptops would beat out M1 and compete with M1 Pro/Max but it doesn't seem optimized for it yet. Desktop version will be epic though. Much larger scores. Beats the M1 Max, right?

b1048546 wrote:

The M1 Max X4 would have 8 AMX chips. for matrix multiplication, 128-channel memory

I can't even imagine a highway with so many lanes

b1048546 wrote:

M2 Air to be fanless, have MagSafe, have a 30 watt power adapter, Mini-LED, thinner than M1

I really would've loved MagSafe on the 2020 M1 MacBook Air - but I guess you can't have everything

MagicCrayon9342 wrote:

M2 and M2X When

uh Magic 2 and Magic 2 X
Mac series
??

{what method did you use::control hat
answer on profile ::motion
} ::operators

MagicCrayon9342 wrote:

M2 and M2X When

Next year for M2. M2X probably not in 2022.

Chiroyce wrote:

b1048546 wrote:

8-high performance cores and two efficiency cores. Why only two efficiency cores? There's a lot of special features that take up space so they decided to use 2-efficiency cores. The battery life is still really good. How?

Battery is bigger as well, right?

Yeah, I mentioned that.

Chiroyce wrote:

b1048546 wrote:

M1, M1 Pro, and M1 Max have a Secure Enclave for security which is new. It's not much new, it's just slightly enhanced security.

Replacement for the T2 Chip, right? (found in Intel Macs)

I'm pretty sure that the Secure Enclave is just the T2 chip (which is basically an A10 fusion) inside the SOC.

Chiroyce wrote:

b1048546 wrote:

The frequency doesn't ramp up properly so it's not really testing the full M1 Max GPU.

I've heard that it's because the benchmark is a very short one - so the chip thinks it's a light workload and may reduce the frequency in order to preserve battery life. Making it longer should help, right?

Correct.

Chiroyce wrote:

b1048546 wrote:

This shows that while mobile Alder Lake competes with M1 in single-core, M1 still beats it out in multi-core.

wait what - that seems like a fake score. Why would a Intel Core i9-12900HK only get 6438 in multi core?
I guess it's because it's on a laptop and it's thermally limited. What a shame. I thought Alder Lake for laptops would beat out M1 and compete with M1 Pro/Max but it doesn't seem optimized for it yet. Desktop version will be epic though. Much larger scores. Beats the M1 Max, right?

Yep, I mentioned that I'm skeptical with Alder Lake as well as the score.

Chiroyce wrote:

b1048546 wrote:

The M1 Max X4 would have 8 AMX chips. for matrix multiplication, 128-channel memory

I can't even imagine a highway with so many lanes

In a perfect world, we would have a highway with lanes for each person driving.

Chiroyce wrote:

b1048546 wrote:

M2 Air to be fanless, have MagSafe, have a 30 watt power adapter, Mini-LED, thinner than M1

I really would've loved MagSafe on the 2020 M1 MacBook Air - but I guess you can't have everything

There's magnetic adapters but I'd rather use the USB-C. (I know, I'm crazy.)