January 29, 2009 at 13:47Life's Test of Moore's Law
I am sure that all readers of the Habré have heard a lot about Moore’s law, and there has been talk for a long time that he should stop working soon, because the physical limits of development (atomic sizes and speed of light) are almost reached. However, the events of the past year (the netbook boom) make us think about a completely different reason for the slowdown in progress.
Intel has released Atom for cheap and lightweight netbooks. And what happened? For most users, its power, which is about 15% of the top processors, is enough for all tasks (except for games that can be played on specialized consoles): the Internet - no problem, casual games - easy, high-resolution video - excellent, and in netbooks with normal graphics cards - also with hardware acceleration.
It turned out that manufacturers are going to make Atom not only cheap netbooks, but also more serious models - with a metal case, a 12-14 'screen, relatively normal video cards (ASUS 10J), desktops (ASUS Eee Box).
You can imagine the shock in Intel and AMD: Intel immediately began to say that “there is no point in putting Atom in laptops with a screen larger than 10 ',” went along the beaten track - increasing the frequency, the number of cores, which few people need in this segment. AMD also said that it is "not interested in netbooks and other dead systems." Well, tell me, who needs a dual-core Atom 330 in the netbook (instead of the weakest Core2Duo)? It can hardly be expected that users will also rush for an “upgrade” if everything works fine now.
Thus, we have already reached the point where current technical capabilities are redundant for most users. If now both AMD and Intel start massively releasing a time-tested solution without including R&D costs in the price (since such a volume of development as is no longer required before), the price of the processor can drop to ~ $ 10.
Quite a bit of effort - and a processor for consumer systems with an integrated chipset, a video accelerator and one memory channel such as GDDR5 (I know that it is “only for graphics” so far) can have only about 100 pins (compared to the current 500-1000) - that's all would allow to create much more economical, compact and cheap devices for consumers (both laptops and desktops). Soldered on the board 2GB of RAM will allow you to place the chip 1cm from the processor, and not 10cm, which will greatly reduce latency (and do with smaller caches).
Now there is already the opportunity to solve the issue with PC upgrades once and for all, but companies are unlikely to want to lose a continuous stream of money, and will try to warm up the corpse of a continuous upgrade for as long as possible, and will stimulate the development of slow-moving software (for example, even slower antiviruses), without which it will be difficult to motivate users to upgrade. It is hoped that the global crisis will kick the industry in the direction of consolidating its current positions :-)
And what do the habra people think about the current situation with the development of iron?
Intel has released Atom for cheap and lightweight netbooks. And what happened? For most users, its power, which is about 15% of the top processors, is enough for all tasks (except for games that can be played on specialized consoles): the Internet - no problem, casual games - easy, high-resolution video - excellent, and in netbooks with normal graphics cards - also with hardware acceleration.It turned out that manufacturers are going to make Atom not only cheap netbooks, but also more serious models - with a metal case, a 12-14 'screen, relatively normal video cards (ASUS 10J), desktops (ASUS Eee Box).
You can imagine the shock in Intel and AMD: Intel immediately began to say that “there is no point in putting Atom in laptops with a screen larger than 10 ',” went along the beaten track - increasing the frequency, the number of cores, which few people need in this segment. AMD also said that it is "not interested in netbooks and other dead systems." Well, tell me, who needs a dual-core Atom 330 in the netbook (instead of the weakest Core2Duo)? It can hardly be expected that users will also rush for an “upgrade” if everything works fine now.
Thus, we have already reached the point where current technical capabilities are redundant for most users. If now both AMD and Intel start massively releasing a time-tested solution without including R&D costs in the price (since such a volume of development as is no longer required before), the price of the processor can drop to ~ $ 10.
Quite a bit of effort - and a processor for consumer systems with an integrated chipset, a video accelerator and one memory channel such as GDDR5 (I know that it is “only for graphics” so far) can have only about 100 pins (compared to the current 500-1000) - that's all would allow to create much more economical, compact and cheap devices for consumers (both laptops and desktops). Soldered on the board 2GB of RAM will allow you to place the chip 1cm from the processor, and not 10cm, which will greatly reduce latency (and do with smaller caches).
Now there is already the opportunity to solve the issue with PC upgrades once and for all, but companies are unlikely to want to lose a continuous stream of money, and will try to warm up the corpse of a continuous upgrade for as long as possible, and will stimulate the development of slow-moving software (for example, even slower antiviruses), without which it will be difficult to motivate users to upgrade. It is hoped that the global crisis will kick the industry in the direction of consolidating its current positions :-)
And what do the habra people think about the current situation with the development of iron?