Technology: Past, Present and Future is a channel for fans of science and technology. I upload videos about the history and context of various technological breakthroughs as well as their impact on the world.
The Shuttle failed its purpose of saving money by having reusable rockets as maintenance to make reusability possible ended up making the whole thing way more expensive than if they had just continued using the expendable Saturn V rockets from Apollo.
What part of SpaceX’s journey to dominance are you most impressed by? Jan 8, 2024 Update: Vulcan has now lifted off, launching an unmanned lander towards the moon
I thought this channel had at least 100k subs when watching this. Couldn’t believe it when I saw it was just a small content creator. Absolutely amazing work!
I think relativity space has the best chance to compete with SpaceX. Mainly through application of the iteration principle. But, its space, so only time will tell.
The Cybertruck was officially delivered today with the tri-motor version taking 2.6 seconds for 0-60, having 845 horsepower and starting at $99,990! Which of the 12 cars in the video do you think showed the most impressive technological leap?
Wow. I had no clue those 8-bit CPUs had only a few thousand transistors. One thing's for certain: I definitely don't think the improvement from those CPUs in terms of real-world performance has quite been on that scale,. mainly due to bloated OSes.
I agree that real-world performance doesn't seem to have improved quite as much! Bloated software is one issue. Another is that I think we continually do more demanding things on our computers (video editing, higher-resolution gaming, etc), which means they don't get nearly as fast as we expect
I have been a chip designer since around the time of the MOS 6502 and I owned a BBC Model2 IIRC as my first 6502 unit. As a student I studied DRAM design and later worked for a DRAM company that also created a 32bit parallel cpu with the worlds first floating point on board, it wasn't Intel. If you can understand the internals of DRAMs used in those early 6502 computers which were likely Mk 4096 64x64 arrays, you can pretty much understand todays modern DRAMs. The earliest DRAMs that were clunky in design cycled in about 600ns but were quickly improved by the Mostek muxed design with advanced circuit design that sped things up to about 250ns full RAS cycle time and say CAS cycle in 60ns. As each generation quadrupled in size, 1 more address pin was needed, but other simplifications gave back a pin esp less power supply pins and the overall DRAM got 10% faster each generation over say 10 quadruplings. Eventually changes in packaging made pin count unimportant. Todays DRAM in the Gbit range still RAS cycle in around 20ns which is only 10x faster than the first Mostek muxed design. But they now have DDR5 ports that are hundreds of times faster in transfering say 256 bit blocks when the early DRAM was CAS cycling 1 bit every 10ns. But the RAS cycle is still only 10x faster. So why do we think CPUs are so much faster, they are only faster because of the 3 or more levels of huge caches with the inner L1 cache running at sub 1ns rates barely keeping up with the pipelines. If a programmer sequentially accesses a huge data array in order, performance will run at Ghz rates, but if you randomly access a Gigabyte array, performance drops all the way back to the limit of the DRAM RAS cycle, which can be hundreds of times slower. Thats called the memory wall. Most of the time PCs run well behaved code, video codecs are especially well behaved and run billions of times faster than a 68K Mac ever could. But some of the time code is badly behaved, we see this in the more mundane parts of computing such as booting up or in file searching which is really only 10x faster than 40 years ago. The sheer bloatedness of Windows 11 in comparison to a lightweight OS like Haiku demonstrates 10 fold difference in performmance because 1 OS wants to use 50GB of disk space while the other can be about 500MB and still just as functional. Its a question of making work vs just doing the job. Of course having more cores and process threads largely allow for computers to appear productive most of the time, but when things get really slow, its possible to see the entire pyramid almost collapsing and some times spectacularly. Programers text books today say the opposite of 1970s coding text book, the general rule is no matter how complex your data structure wants to be, always force it into a non random in order array to be cache friendly. Linked lists and trees are very bad for caches but are the natural way to build complex programs like compilers. End of rant, back to bed.
Apple just unveiled the M3 CPU today, which has 25 billion transistors, close to the 26 billion I'd predicted. This makes it 7 million times faster than the Apple I - an impressive accomplishment over the last few decades! While the newest computers are definitely going to be the fastest, which Mac design was your favorite?
Thanks for the Update! 3nm is pretty close to 1nm where quantum tunnelling becomes serious. When would it be likely to reach it. And what do you think will the future of Apple and computing in general will be?
Great question! I think quantum tunneling has been something chipmakers have been dealing with for a while but you're right it's going to become a bigger and bigger issue as we get close to 1nm. Over the next few years (definitely by 2030), manufacturers will probably need to spend more and more resources to mitigate it or start exploring completely different non-silicon architectures. For Apple and computing generally, my guess is that we'll see computing progress slow down a bit over the next couple of years until the industry figures out the best way to deal with quantum tunneling, etc. But that's just an educated guess from my end!
As it turned out, the same payloads and missions could have been flown less expensively over those decades by an expendible system (all-new, or improved/adapted legacy hardware). But they didn't know that going in, and besides the funding for it arrived intact only because it was to be a "national project" -- there wouldn't have been political support for all of those individual payloads. So many of the satellites and space probes would have still flown, but things like the Spacelab missions probably wouldn't have happened.
