...an updated Mac Pro. I didn’t buy one last year, because I wanted the new PCIe 4 bus, which is now becoming common enough for Apple to move to.
In an Apple Silicon Mac a PCIe bus may not do everything that we might want. For example, a PCIe video card may not work since MacOS Big Sur may not support any video cards even including Metal-based AMD video cards.
I agree that Apple will probably support PCIe v4 in an Apple Silicon Mac with expansion ports, but if it's not for video cards, maybe Apple should just re-think the whole purpose of the expansion card idea entirely. Could Thunderbolt be enough? How about Thunderbolt 4 which is due any day now from Intel (and probably from Apple too, since Apple now supports Thunderbolt without any Intel chips)? What type of cards do people want to install in a Mac that can't be done through a Thunderbolt port in some other way? Thunderbolt 4 is 26% faster than even a 16-lane PCIe v4 connector, and it's almost as fast as 16-lane PCEe v5 which won't be out for a couple of years. Why would anyone want a slow connection like a 16-lane PCIe v4 when Thunderbolt 4 is so much faster? I'll answer that - because you don't have to buy a Thunderbolt cable. So to save $50 people want to settle for a lower bus speed?
This is food for thought; I'm not sure Apple will do what I'm suggesting. I expect the usual claims from people here questioning my sanity.
Maybe not your sanity, but I think you messed up your units here.
Thunderbolt 4: PCIe data at 32 Gb/s (GigaBITS) PCIe v4 x 16: 32 GB/s (GigaBYTES) = 256 GigaBITS
"The PCIe 4.0 specification will also bring the OCuLink-2 connector, an alternative to Thunderbolt 3, that provides 8GB/s of bandwidth via four PCIe 4.0 lanes" — That's 64 Gb/s.
Someone correct me if I screwed up here, but PCIe v4 x16 is quite a bit faster than Thunderbolt 4.
Uhoh, that's embarrassing. I will believe you. I think part of the reason I was confused is that some websites talk about GT/s (gigatransfers/sec) and I was probably not discrerning between "gigatransfers of bytes" and "gigatransfers of bits."
But I won't give up so easily. Please note that Thunderbolt is currently using copper cabling. When Thunderbolt was first introduced as "Light Peak" it used fibre optic instead. Corning has bragged about having fibre-optic Thunderbolt (see the video below). I'm sure that fiber-optic is faster than copper, and it's potentially a way forward for Apple without using PCIe. Corning has been silent about this product for years, which could mean they've given up, or could mean they have an NDA with Apple for a future product.
Doubt it will be a new ATV. I think the reason Apple has waited so long to update is the A14X. iPad Pros will come out with it in early 2021, then ATV in a later quarter. Seems like if Apple was going to do an A12X or A12Z version of the ATV, it would have been already released.
...an updated Mac Pro. I didn’t buy one last year, because I wanted the new PCIe 4 bus, which is now becoming common enough for Apple to move to.
In an Apple Silicon Mac a PCIe bus may not do everything that we might want. For example, a PCIe video card may not work since MacOS Big Sur may not support any video cards even including Metal-based AMD video cards.
I agree that Apple will probably support PCIe v4 in an Apple Silicon Mac with expansion ports, but if it's not for video cards, maybe Apple should just re-think the whole purpose of the expansion card idea entirely. Could Thunderbolt be enough? How about Thunderbolt 4 which is due any day now from Intel (and probably from Apple too, since Apple now supports Thunderbolt without any Intel chips)? What type of cards do people want to install in a Mac that can't be done through a Thunderbolt port in some other way? Thunderbolt 4 is 26% faster than even a 16-lane PCIe v4 connector, and it's almost as fast as 16-lane PCEe v5 which won't be out for a couple of years. Why would anyone want a slow connection like a 16-lane PCIe v4 when Thunderbolt 4 is so much faster? I'll answer that - because you don't have to buy a Thunderbolt cable. So to save $50 people want to settle for a lower bus speed?
This is food for thought; I'm not sure Apple will do what I'm suggesting. I expect the usual claims from people here questioning my sanity.
Maybe not your sanity, but I think you messed up your units here.
Thunderbolt 4: PCIe data at 32 Gb/s (GigaBITS) PCIe v4 x 16: 32 GB/s (GigaBYTES) = 256 GigaBITS
"The PCIe 4.0 specification will also bring the OCuLink-2 connector, an alternative to Thunderbolt 3, that provides 8GB/s of bandwidth via four PCIe 4.0 lanes" — That's 64 Gb/s.
Someone correct me if I screwed up here, but PCIe v4 x16 is quite a bit faster than Thunderbolt 4.
Uhoh, that's embarrassing. I will believe you. I think part of the reason I was confused is that some websites talk about GT/s (gigatransfers/sec) and I was probably not discrerning between "gigatransfers of bytes" and "gigatransfers of bits."
But I won't give up so easily. Please note that Thunderbolt is currently using copper cabling. When Thunderbolt was first introduced as "Light Peak" it used fibre optic instead. Corning has bragged about having fibre-optic Thunderbolt (see the video below). I'm sure that fiber-optic is faster than copper, and it's potentially a way forward for Apple without using PCIe. Corning has been silent about this product for years, which could mean they've given up, or could mean they have an NDA with Apple for a future product.
Optical fiber isn't faster than copper for this. Both can get full Thunderbolt speeds.
You can also buy optical Thunderbolt cables already. Have been able to for years. The starting cost is a lot higher than that of copper cables. Corning makes cables up to 50m:
AOC-CCU6JPN005M20
AOC-CCU6JPN010M20
AOC-CCU6JPN015M20
AOC-CCU6JPN025M20
AOC-CCU6JPN050M20
The number just ahead of the "M20" is the cable's length in meters.
Optical fiber isn't faster than copper for this. Both can get full Thunderbolt speeds.
You can also buy optical Thunderbolt cables already. Have been able to for years. The starting cost is a lot higher than that of copper cables. Corning makes cables up to 50m:
AOC-CCU6JPN005M20
AOC-CCU6JPN010M20
AOC-CCU6JPN015M20
AOC-CCU6JPN025M20
AOC-CCU6JPN050M20
The number just ahead of the "M20" is the cable's length in meters.
I wasn't aware of the optical TB cables, thanks.
Sure, copper Thunderbolt is as fast as fiber Thunderbolt now, but fiber is faster than copper in general. Copper currently can't transmit faster than 40 gigabits per second but fiber has a current maximum of 200+ terabits per second. That's 5000 times faster. It's even 1000 times faster than PCIe v4 16-bit. I was simply speculating that fiber could hypothetically replace PCIe.
Optical fiber isn't faster than copper for this. Both can get full Thunderbolt speeds.
You can also buy optical Thunderbolt cables already. Have been able to for years. The starting cost is a lot higher than that of copper cables. Corning makes cables up to 50m:
AOC-CCU6JPN005M20
AOC-CCU6JPN010M20
AOC-CCU6JPN015M20
AOC-CCU6JPN025M20
AOC-CCU6JPN050M20
The number just ahead of the "M20" is the cable's length in meters.
I wasn't aware of the optical TB cables, thanks.
Sure, copper Thunderbolt is as fast as fiber Thunderbolt now, but fiber is faster than copper in general. Copper currently can't transmit faster than 40 gigabits per second but fiber has a current maximum of 200+ terabits per second. That's 5000 times faster. It's even 1000 times faster than PCIe v4 16-bit. I was simply speculating that fiber could hypothetically replace PCIe.
Ahhh... Perhaps it would be better to say that fiber has a greater bandwidth than copper. The actual "speed" really isn't much different - photon speed vs electron speed and all. I know... I know... slew rate limitations of copper due to inductance... yada-yada-yada. But is it really speed? Maybe I need to spend more time in my Zen garden.
Personally, I have become a fan of fiber over the last couple years. While gigabit to the desktop is fine and dandy, the limitations on copper at 10+ Gb speeds caused me to invest in fiber and I am not looking back!
Optical fiber isn't faster than copper for this. Both can get full Thunderbolt speeds.
You can also buy optical Thunderbolt cables already. Have been able to for years. The starting cost is a lot higher than that of copper cables. Corning makes cables up to 50m:
AOC-CCU6JPN005M20
AOC-CCU6JPN010M20
AOC-CCU6JPN015M20
AOC-CCU6JPN025M20
AOC-CCU6JPN050M20
The number just ahead of the "M20" is the cable's length in meters.
I wasn't aware of the optical TB cables, thanks.
Sure, copper Thunderbolt is as fast as fiber Thunderbolt now, but fiber is faster than copper in general. Copper currently can't transmit faster than 40 gigabits per second but fiber has a current maximum of 200+ terabits per second. That's 5000 times faster. It's even 1000 times faster than PCIe v4 16-bit. I was simply speculating that fiber could hypothetically replace PCIe.
Ahhh... Perhaps it would be better to say that fiber has a greater bandwidth than copper. The actual "speed" really isn't much different - photon speed vs electron speed and all. I know... I know... slew rate limitations of copper due to inductance... yada-yada-yada. But is it really speed? Maybe I need to spend more time in my Zen garden.
Personally, I have become a fan of fiber over the last couple years. While gigabit to the desktop is fine and dandy, the limitations on copper at 10+ Gb speeds caused me to invest in fiber and I am not looking back!
Right, I was talking about data bandwidth and not photon/electron speed. Sorry for not saying that more clearly.
But actually when it comes to physics, the physical speed of an electron through a wire is a completely different thing that the speed of the electronic signal through the wire. The actual physical speed of an electron is much slower than 1mm/second, depending on amperage and conductivity. Yet the speed of the electromagnetic pressure wave is about 90% the speed of light.
Optical fiber isn't faster than copper for this. Both can get full Thunderbolt speeds.
You can also buy optical Thunderbolt cables already. Have been able to for years. The starting cost is a lot higher than that of copper cables. Corning makes cables up to 50m:
AOC-CCU6JPN005M20
AOC-CCU6JPN010M20
AOC-CCU6JPN015M20
AOC-CCU6JPN025M20
AOC-CCU6JPN050M20
The number just ahead of the "M20" is the cable's length in meters.
I wasn't aware of the optical TB cables, thanks.
Sure, copper Thunderbolt is as fast as fiber Thunderbolt now, but fiber is faster than copper in general. Copper currently can't transmit faster than 40 gigabits per second but fiber has a current maximum of 200+ terabits per second. That's 5000 times faster. It's even 1000 times faster than PCIe v4 16-bit. I was simply speculating that fiber could hypothetically replace PCIe.
Ahhh... Perhaps it would be better to say that fiber has a greater bandwidth than copper. The actual "speed" really isn't much different - photon speed vs electron speed and all. I know... I know... slew rate limitations of copper due to inductance... yada-yada-yada. But is it really speed? Maybe I need to spend more time in my Zen garden.
Personally, I have become a fan of fiber over the last couple years. While gigabit to the desktop is fine and dandy, the limitations on copper at 10+ Gb speeds caused me to invest in fiber and I am not looking back!
Right, I was talking about data bandwidth and not photon/electron speed. Sorry for not saying that more clearly.
But actually when it comes to physics, the physical speed of an electron through a wire is a completely different thing that the speed of the electronic signal through the wire. The actual physical speed of an electron is much slower than 1mm/second, depending on amperage and conductivity. Yet the speed of the electromagnetic pressure wave is about 90% the speed of light.
Too true! Of course one must take into account the light path down a fiber cable is not straight as it bounces off the outer edge of the fiber while it traverses from transmitter to receiver... not unlike the tiger in a zoo moving from one side of the cage to the other as it tries to find a way to it's next meal - you!
The long and the short... Almost the same "speed" - although I believe fiber is a few percent faster than copper (68%c vs 64%c).
...an updated Mac Pro. I didn’t buy one last year, because I wanted the new PCIe 4 bus, which is now becoming common enough for Apple to move to.
In an Apple Silicon Mac a PCIe bus may not do everything that we might want. For example, a PCIe video card may not work since MacOS Big Sur may not support any video cards even including Metal-based AMD video cards.
I agree that Apple will probably support PCIe v4 in an Apple Silicon Mac with expansion ports, but if it's not for video cards, maybe Apple should just re-think the whole purpose of the expansion card idea entirely. Could Thunderbolt be enough? How about Thunderbolt 4 which is due any day now from Intel (and probably from Apple too, since Apple now supports Thunderbolt without any Intel chips)? What type of cards do people want to install in a Mac that can't be done through a Thunderbolt port in some other way? Thunderbolt 4 is 26% faster than even a 16-lane PCIe v4 connector, and it's almost as fast as 16-lane PCEe v5 which won't be out for a couple of years. Why would anyone want a slow connection like a 16-lane PCIe v4 when Thunderbolt 4 is so much faster? I'll answer that - because you don't have to buy a Thunderbolt cable. So to save $50 people want to settle for a lower bus speed?
This is food for thought; I'm not sure Apple will do what I'm suggesting. I expect the usual claims from people here questioning my sanity.
We really don’t know. But it’s not true that external GPUs can’t be made to work, depending on the definition of what that is. If you understand the way the M1, and subsequent chips are designed, you can see ways around what was presented with the M1. Remember that everything on the substrate, where the chip and memory is mounted, uses Apple’s “fabric”. That includes everything on the die, and the RAM, on the substrate.
bringing the lines for the RAM out of the die onto the substrate, to the two RAM packages, shows that any lines from the fabric can be brought out. That not only includes extended lines for more RAM packages, but also lines to the GPU, the Neural Engine, etc.
apple could, in a higher end chip package, take the GPU off the die, and put it onto the substrate, as they did with the RAM. That would give several advantages. One would be that Apple could increase the size of the GPU by several times. The RAM is on the right side of the package. If they made a separate GPU, they could put on the top of the package. If they did that, and made it the length of the SoC die, then it could have about four times the cores, assuming they didn’t increase the height of that chip over the current height of the current GPU that’s on the die now. Two would be that the heat would be taken off the rest of the SoC die, and would allow better cooling. Three would be that Apple would have about 20% empty space on the die that they could use for more CPU cores, or larger cores, or more cores in the Neural Engine, etc.
the interesting thing here is that these packages would be very small. No larger than say, a 14, or at most, a 24 pin IC. It could also be pluggable, if Apple decided to offer different numbers of cores. Same thing with RAM. Apple could have the two built-in packages, but by extending that substrate a bit, could have a socket for a double length RAM module for additional add-on memory.
whether Apple intends something like this in a future M series, is just guesswork on my part. But there’s nothing that wouldn’t allow this to work. In fact, something like this would be required for Apple to significantly increase GPU power in the future. Even with process shrinkage, there is only so much room on the die. Last point, they won’t be able to fit more on it without yields to drop significantly, and costs rising commensurately. It’s just not practical as far as cooling goes as well.
I think that people are misunderstanding what Apple’s doing here. The fact that they have this fabric, which is sort of a very high speed intelligent bus, connecting every item in the design together, give more flexibility, not less.
Apple said new intel macs later this year. So likely they will update the Macbook Pro 16 with 10th-gen intel processor like they did with iMac 27 2020.
Where did Apple specifically say they were releasing new Macs with Intel?
It's unclear whether Tim Cook was talking about multiple Mac models or a single model, but it sounds more like multiple models. My guess is that the iMac Pro and Mac Pro are the Intel models that Cook was talking about, although it's also possible that they will be offered simultaneously with M-chip alternatives. It's also possible that there could be hybrid machines, for example, an existing or future Intel Mac Pro could have an Mx-chip "Mac-on-a-PCI-card." (The Mx-chip on this might be a variation without a video core because the Mac Pro has its own video. This sounds too complicated to me, but it remains a possibility.)
Oh I guess I missed that. Thank you for the quote and the link. I am guess it would be the iMac Pro and the Mac Pro since they still need to figure out how to maximize the memory for those machines.
Another honest question, is Apple able to solder multiple M1 chips on one logic board and somehow allow each of them to contribute to the amount of the power per chip? So like a Duo M1 for example could in theory have 16 processor cores and up to 32 GBs of memory, etc?
I’ve proposed that for a Mac Pro. I don’t see any reason why not. Apple has lots of experience with two socket machines. But a question is whether the current M1 has the links required for such a thing. Otherwise it becomes a bit clumsy and inefficient. Remember that the biggest difference between an Intel Xeon and their other CPUs is the Quick Link, or whatever they’re calling it these days. That allows two and four socket machines. Higher end Xeons allow more sockets.
I’ve been hoping, with the rumors of a possible Mac Pro Mini, that we might see a one or two processor possibility. It could even be that a full sized machine could have four, or even six. Imagine six twelve core processors, all with a GPU, Neural Engine, and 16 to 64GB RAM! Each one having four USB 4/TB ports.
...an updated Mac Pro. I didn’t buy one last year, because I wanted the new PCIe 4 bus, which is now becoming common enough for Apple to move to.
In an Apple Silicon Mac a PCIe bus may not do everything that we might want. For example, a PCIe video card may not work since MacOS Big Sur may not support any video cards even including Metal-based AMD video cards.
I agree that Apple will probably support PCIe v4 in an Apple Silicon Mac with expansion ports, but if it's not for video cards, maybe Apple should just re-think the whole purpose of the expansion card idea entirely. Could Thunderbolt be enough? How about Thunderbolt 4 which is due any day now from Intel (and probably from Apple too, since Apple now supports Thunderbolt without any Intel chips)? What type of cards do people want to install in a Mac that can't be done through a Thunderbolt port in some other way? Thunderbolt 4 is 26% faster than even a 16-lane PCIe v4 connector, and it's almost as fast as 16-lane PCEe v5 which won't be out for a couple of years. Why would anyone want a slow connection like a 16-lane PCIe v4 when Thunderbolt 4 is so much faster? I'll answer that - because you don't have to buy a Thunderbolt cable. So to save $50 people want to settle for a lower bus speed?
This is food for thought; I'm not sure Apple will do what I'm suggesting. I expect the usual claims from people here questioning my sanity.
I’m not thinking of a full PCIe 4 bus. But remember that the M1 itself has an internal PCIe 4 bus built-in. Thunderbolt 4 is out in the USB 4 port specification. Thunderbolt is PART of the pci bus. It runs through the bus. PCIe 4 has, as you can see from the link, 2GB/s over each lane. That’s 16GB/s. A TB card can use four lanes, for a total of 64GB/s. That’s plenty of speed. Most hardware controllers for drives like this use four lanes. 16 lanes have a total of 256. I don’t know what your numbers mean.
my big post just above got delayed from being listed. If you read that you’ll see some of my thinking on this.
...an updated Mac Pro. I didn’t buy one last year, because I wanted the new PCIe 4 bus, which is now becoming common enough for Apple to move to.
In an Apple Silicon Mac a PCIe bus may not do everything that we might want. For example, a PCIe video card may not work since MacOS Big Sur may not support any video cards even including Metal-based AMD video cards.
I agree that Apple will probably support PCIe v4 in an Apple Silicon Mac with expansion ports, but if it's not for video cards, maybe Apple should just re-think the whole purpose of the expansion card idea entirely. Could Thunderbolt be enough? How about Thunderbolt 4 which is due any day now from Intel (and probably from Apple too, since Apple now supports Thunderbolt without any Intel chips)? What type of cards do people want to install in a Mac that can't be done through a Thunderbolt port in some other way? Thunderbolt 4 is 26% faster than even a 16-lane PCIe v4 connector, and it's almost as fast as 16-lane PCEe v5 which won't be out for a couple of years. Why would anyone want a slow connection like a 16-lane PCIe v4 when Thunderbolt 4 is so much faster? I'll answer that - because you don't have to buy a Thunderbolt cable. So to save $50 people want to settle for a lower bus speed?
This is food for thought; I'm not sure Apple will do what I'm suggesting. I expect the usual claims from people here questioning my sanity.
Maybe not your sanity, but I think you messed up your units here.
Thunderbolt 4: PCIe data at 32 Gb/s (GigaBITS) PCIe v4 x 16: 32 GB/s (GigaBYTES) = 256 GigaBITS
"The PCIe 4.0 specification will also bring the OCuLink-2 connector, an alternative to Thunderbolt 3, that provides 8GB/s of bandwidth via four PCIe 4.0 lanes" — That's 64 Gb/s.
Someone correct me if I screwed up here, but PCIe v4 x16 is quite a bit faster than Thunderbolt 4.
Uhoh, that's embarrassing. I will believe you. I think part of the reason I was confused is that some websites talk about GT/s (gigatransfers/sec) and I was probably not discrerning between "gigatransfers of bytes" and "gigatransfers of bits."
But I won't give up so easily. Please note that Thunderbolt is currently using copper cabling. When Thunderbolt was first introduced as "Light Peak" it used fibre optic instead. Corning has bragged about having fibre-optic Thunderbolt (see the video below). I'm sure that fiber-optic is faster than copper, and it's potentially a way forward for Apple without using PCIe. Corning has been silent about this product for years, which could mean they've given up, or could mean they have an NDA with Apple for a future product.
Glass fiber is only for longer distances. You don’t get higher speeds than 40Gb/s. What you do get, externally, is the ability to get that higher speed of 40Gb/s with longer cable lengths. Several companies have TB 3 glass fiber cables out, up to 300 feet in length. But even short ones are very expensive. Alternatively, you can buy copper cables with built-in boosters, but they are expensive too.
Optical fiber isn't faster than copper for this. Both can get full Thunderbolt speeds.
You can also buy optical Thunderbolt cables already. Have been able to for years. The starting cost is a lot higher than that of copper cables. Corning makes cables up to 50m:
AOC-CCU6JPN005M20
AOC-CCU6JPN010M20
AOC-CCU6JPN015M20
AOC-CCU6JPN025M20
AOC-CCU6JPN050M20
The number just ahead of the "M20" is the cable's length in meters.
I wasn't aware of the optical TB cables, thanks.
Sure, copper Thunderbolt is as fast as fiber Thunderbolt now, but fiber is faster than copper in general. Copper currently can't transmit faster than 40 gigabits per second but fiber has a current maximum of 200+ terabits per second. That's 5000 times faster. It's even 1000 times faster than PCIe v4 16-bit. I was simply speculating that fiber could hypothetically replace PCIe.
Apple said new intel macs later this year. So likely they will update the Macbook Pro 16 with 10th-gen intel processor like they did with iMac 27 2020.
Where did Apple specifically say they were releasing new Macs with Intel?
It's unclear whether Tim Cook was talking about multiple Mac models or a single model, but it sounds more like multiple models. My guess is that the iMac Pro and Mac Pro are the Intel models that Cook was talking about, although it's also possible that they will be offered simultaneously with M-chip alternatives. It's also possible that there could be hybrid machines, for example, an existing or future Intel Mac Pro could have an Mx-chip "Mac-on-a-PCI-card." (The Mx-chip on this might be a variation without a video core because the Mac Pro has its own video. This sounds too complicated to me, but it remains a possibility.)
I’m not so sure.
That comment was made round about June by the looks of. Since then, we had this:
When asked about future Intel Macs, Greg Joswiak shrugged it off.
"When we said we would support Intel systems for years to come, that was talking about the operating system." Apple had commented that future Intel hardware was coming, but Joswiak said "We still had Intel systems that were in the pipeline" when it was said, "That very next month, we introduced an Intel-based iMac."
Sounds to me like they’re done with Intel Macs.
That's a very good argument. Yet I'm not entirely convinced. If I had to bet even money, I'd put it on another Intel Mac.
LOL these guys don't just toss out misstatements — their entire job is crafting a narrative. He even used the past tense with "We still had..." and then the resolution of that proposition as having already happened.
...an updated Mac Pro. I didn’t buy one last year, because I wanted the new PCIe 4 bus, which is now becoming common enough for Apple to move to.
In an Apple Silicon Mac a PCIe bus may not do everything that we might want. For example, a PCIe video card may not work since MacOS Big Sur may not support any video cards even including Metal-based AMD video cards.
I agree that Apple will probably support PCIe v4 in an Apple Silicon Mac with expansion ports, but if it's not for video cards, maybe Apple should just re-think the whole purpose of the expansion card idea entirely. Could Thunderbolt be enough? How about Thunderbolt 4 which is due any day now from Intel (and probably from Apple too, since Apple now supports Thunderbolt without any Intel chips)? What type of cards do people want to install in a Mac that can't be done through a Thunderbolt port in some other way? Thunderbolt 4 is 26% faster than even a 16-lane PCIe v4 connector, and it's almost as fast as 16-lane PCEe v5 which won't be out for a couple of years. Why would anyone want a slow connection like a 16-lane PCIe v4 when Thunderbolt 4 is so much faster? I'll answer that - because you don't have to buy a Thunderbolt cable. So to save $50 people want to settle for a lower bus speed?
This is food for thought; I'm not sure Apple will do what I'm suggesting. I expect the usual claims from people here questioning my sanity.
Maybe not your sanity, but I think you messed up your units here.
Thunderbolt 4: PCIe data at 32 Gb/s (GigaBITS) PCIe v4 x 16: 32 GB/s (GigaBYTES) = 256 GigaBITS
"The PCIe 4.0 specification will also bring the OCuLink-2 connector, an alternative to Thunderbolt 3, that provides 8GB/s of bandwidth via four PCIe 4.0 lanes" — That's 64 Gb/s.
Someone correct me if I screwed up here, but PCIe v4 x16 is quite a bit faster than Thunderbolt 4.
Uhoh, that's embarrassing. I will believe you. I think part of the reason I was confused is that some websites talk about GT/s (gigatransfers/sec) and I was probably not discrerning between "gigatransfers of bytes" and "gigatransfers of bits."
But I won't give up so easily. Please note that Thunderbolt is currently using copper cabling. When Thunderbolt was first introduced as "Light Peak" it used fibre optic instead. Corning has bragged about having fibre-optic Thunderbolt (see the video below). I'm sure that fiber-optic is faster than copper, and it's potentially a way forward for Apple without using PCIe. Corning has been silent about this product for years, which could mean they've given up, or could mean they have an NDA with Apple for a future product.
Sorry, but I don't think you have any idea what you're talking about. PCIe is a bus standard, not a connector or cable. Thunderbolt literally uses PCIe transaction layers for data transfer alongside DisplayPort and USB etc. They're not competing products. If you're talking about internal connected devices vs external, external has a LONG way to go as outlined above regardless of whether it's Thunderbolt, or OCuLink, or any other interconnect, especially considering PCIe versions are doubling throughput on internal connections as well with each revision and external connects are far, far behind.
And, we already have optical Thunderbolt cables — they're terribly expensive at hundreds of dollars apiece and used for longer distances than copper can sustain. This isn't a future product, it's already here, and doesn't miraculously improve transfer speeds on existing protocols.
Apple said new intel macs later this year. So likely they will update the Macbook Pro 16 with 10th-gen intel processor like they did with iMac 27 2020.
Where did Apple specifically say they were releasing new Macs with Intel?
It's unclear whether Tim Cook was talking about multiple Mac models or a single model, but it sounds more like multiple models. My guess is that the iMac Pro and Mac Pro are the Intel models that Cook was talking about, although it's also possible that they will be offered simultaneously with M-chip alternatives. It's also possible that there could be hybrid machines, for example, an existing or future Intel Mac Pro could have an Mx-chip "Mac-on-a-PCI-card." (The Mx-chip on this might be a variation without a video core because the Mac Pro has its own video. This sounds too complicated to me, but it remains a possibility.)
I’m not so sure.
That comment was made round about June by the looks of. Since then, we had this:
When asked about future Intel Macs, Greg Joswiak shrugged it off.
"When we said we would support Intel systems for years to come, that was talking about the operating system." Apple had commented that future Intel hardware was coming, but Joswiak said "We still had Intel systems that were in the pipeline" when it was said, "That very next month, we introduced an Intel-based iMac."
Sounds to me like they’re done with Intel Macs.
That's a very good argument. Yet I'm not entirely convinced. If I had to bet even money, I'd put it on another Intel Mac.
I ordered another Intel Mac yesterday... Should be delivered tomorrow or Monday. :-)
Perhaps in a few years I will consider an M-series Mac, but not now. Even when Apple switched to Intel it took me five years to make the switch. I will let the eager beavers deal with the growing pains.
Yes of course I knew that PCIe was a physical bus and TB was a physical cable/connector. But I think you missed my point. Consider that Apple has recently allowed users to switch from using PCIe as the connection between their Mac's CPU and their video cards and instead use TB as the "bus" for eGPUs. The TB cable is now the "bus" for video cards. I don't care that it doesn't look like a "bus," it still is. And it could be the bus for other things too, perhaps for storage or even RAM. And since the bandwidth of fiber optic currently can reach 2000+ terabits per second, why is this not a good idea? All I'm suggesting is that Apple do the same thing for other devices that it has done for video cards. How is that bad?
I admit it doesn't look like a bus, but functionally, it is. It is the physical connection between eGPUs and the CPU, exactly as a PCIe bus provides. It serves the exact same purpose as a PCIe bus: connecting an external device to the CPU. You have to learn to think outside the bus.
I admit it doesn't look like a bus, but functionally, it is. It is the physical connection between eGPUs and the CPU, exactly as a PCIe bus provides. It serves the exact same purpose as a PCIe bus: connecting an external device to the CPU. You have to learn to think outside the bus.
What on earth are you talking about?
Your original proposition was that Apple drop PCIe support and just use Thunderbolt because it's faster, a very false statement. Then you shifted to saying that optical cabling could make up the difference, a false statement, since the protocol is not bottlenecked by the cable. Now you're saying that fibre cable is a bus, a fundamental misunderstanding of what a bus is?
Give up and go home dude, doubling down on your ignorance has not worked out well for you here.
Comments
Uhoh, that's embarrassing. I will believe you. I think part of the reason I was confused is that some websites talk about GT/s (gigatransfers/sec) and I was probably not discrerning between "gigatransfers of bytes" and "gigatransfers of bits."
But I won't give up so easily. Please note that Thunderbolt is currently using copper cabling. When Thunderbolt was first introduced as "Light Peak" it used fibre optic instead. Corning has bragged about having fibre-optic Thunderbolt (see the video below). I'm sure that fiber-optic is faster than copper, and it's potentially a way forward for Apple without using PCIe. Corning has been silent about this product for years, which could mean they've given up, or could mean they have an NDA with Apple for a future product.
You can also buy optical Thunderbolt cables already. Have been able to for years. The starting cost is a lot higher than that of copper cables. Corning makes cables up to 50m:
The number just ahead of the "M20" is the cable's length in meters.
Sure, copper Thunderbolt is as fast as fiber Thunderbolt now, but fiber is faster than copper in general. Copper currently can't transmit faster than 40 gigabits per second but fiber has a current maximum of 200+ terabits per second. That's 5000 times faster. It's even 1000 times faster than PCIe v4 16-bit. I was simply speculating that fiber could hypothetically replace PCIe.
Personally, I have become a fan of fiber over the last couple years. While gigabit to the desktop is fine and dandy, the limitations on copper at 10+ Gb speeds caused me to invest in fiber and I am not looking back!
But actually when it comes to physics, the physical speed of an electron through a wire is a completely different thing that the speed of the electronic signal through the wire. The actual physical speed of an electron is much slower than 1mm/second, depending on amperage and conductivity. Yet the speed of the electromagnetic pressure wave is about 90% the speed of light.
The long and the short... Almost the same "speed" - although I believe fiber is a few percent faster than copper (68%c vs 64%c).
bringing the lines for the RAM out of the die onto the substrate, to the two RAM packages, shows that any lines from the fabric can be brought out. That not only includes extended lines for more RAM packages, but also lines to the GPU, the Neural Engine, etc.
apple could, in a higher end chip package, take the GPU off the die, and put it onto the substrate, as they did with the RAM. That would give several advantages. One would be that Apple could increase the size of the GPU by several times. The RAM is on the right side of the package. If they made a separate GPU, they could put on the top of the package. If they did that, and made it the length of the SoC die, then it could have about four times the cores, assuming they didn’t increase the height of that chip over the current height of the current GPU that’s on the die now. Two would be that the heat would be taken off the rest of the SoC die, and would allow better cooling. Three would be that Apple would have about 20% empty space on the die that they could use for more CPU cores, or larger cores, or more cores in the Neural Engine, etc.
the interesting thing here is that these packages would be very small. No larger than say, a 14, or at most, a 24 pin IC. It could also be pluggable, if Apple decided to offer different numbers of cores. Same thing with RAM. Apple could have the two built-in packages, but by extending that substrate a bit, could have a socket for a double length RAM module for additional add-on memory.
whether Apple intends something like this in a future M series, is just guesswork on my part. But there’s nothing that wouldn’t allow this to work. In fact, something like this would be required for Apple to significantly increase GPU power in the future. Even with process shrinkage, there is only so much room on the die. Last point, they won’t be able to fit more on it without yields to drop significantly, and costs rising commensurately. It’s just not practical as far as cooling goes as well.
I think that people are misunderstanding what Apple’s doing here. The fact that they have this fabric, which is sort of a very high speed intelligent bus, connecting every item in the design together, give more flexibility, not less.
I’ve been hoping, with the rumors of a possible Mac Pro Mini, that we might see a one or two processor possibility. It could even be that a full sized machine could have four, or even six. Imagine six twelve core processors, all with a GPU, Neural Engine, and 16 to 64GB RAM! Each one having four USB 4/TB ports.
now that would be something.
my big post just above got delayed from being listed. If you read that you’ll see some of my thinking on this.
And, we already have optical Thunderbolt cables — they're terribly expensive at hundreds of dollars apiece and used for longer distances than copper can sustain. This isn't a future product, it's already here, and doesn't miraculously improve transfer speeds on existing protocols.
Maybe it's better if you give up so easily.
Perhaps in a few years I will consider an M-series Mac, but not now. Even when Apple switched to Intel it took me five years to make the switch. I will let the eager beavers deal with the growing pains.
I admit it doesn't look like a bus, but functionally, it is. It is the physical connection between eGPUs and the CPU, exactly as a PCIe bus provides. It serves the exact same purpose as a PCIe bus: connecting an external device to the CPU. You have to learn to think outside the bus.
Your original proposition was that Apple drop PCIe support and just use Thunderbolt because it's faster, a very false statement.
Then you shifted to saying that optical cabling could make up the difference, a false statement, since the protocol is not bottlenecked by the cable.
Now you're saying that fibre cable is a bus, a fundamental misunderstanding of what a bus is?
Give up and go home dude, doubling down on your ignorance has not worked out well for you here.