Where's my 10GHz CPU that I was promised in 2005?
>>107583448Not far fetched from any perspective of Eternity, though for anyone with less temporal acuity than myself and apparently OP, kind of is a long run if you try to take it raw.
>>107583448>Where's my 10GHz CPUrunning too hot for it to be physically feasiblebut no worries, now a process only uses 1 of your 24 cores so the whole computer doesn't hang up when one program hogs resources
>>107583448If we ever figure out photonics it might be eventually possible.
4 core x 2.5 GHz
>>107583448you cannot block roentgen via atx case can you
>>107584459/thread/
>>107583448It never materialized because silicon has unforseen problems at that high frequencies. We will need new materials or even photonics to get above 5-8GHz at reasonable power levels.
>>107584922Make a thread on /sci/, I can consult.
>>107583448moore's law is dead
>>1075853382005 does not know if it dies fast or slow
>>107583448dennard/frequency scaling died in the early 2000s, I think because the operating voltages started to get small enough that transistors stopped behaving as nicelySuppose you can make your IC 50% smaller. The small metal viaducts a) consume less power due to being smaller and b) charge up faster because of a reduced a RC constant. b) allows higher frequencies to be used. I think a) allows lower voltages to be used so your now smaller die is still able to dissipate that heat
>>107583448pretty soon anon. Computer power is improving crazy fast by 2015 we will be able to simulate entire universes on our PCs. this is the worst it will ever be
>>107583448"Running Microsoft Word can only take so much processing power"Today advancements in software engineering have brought us significant CPU usage rendering the blinking cursor.
>>107585363The heat is primarily caused by current.A 300 Watt GPU running at 1.1 volts is sucking in over 250 amps which is why it’s surrounded by 12 V to 1.1 V DC/DC converters.The silicon junction needs .7 to turn on, if you started making them with, say, germanium, you could probably achieve chip supply voltages of around 0.4 V.I’ve seen other devices using something like GaAs junctions that go into 2-300 GHz range based on other properties (such as electron mobility)
>>107583448Cpus stopped getting faster because we maxed out the simulation. It was unexpected because physics was modeled after the real world, but only to the point that simulation could support it. That is why scaling hit a wall, the wall should not be there. And in the real world, scaling just kept going and going.
>>107585574I maxed out stimulation
bleeding edge "AI" companies are currently investing in phonotic computers.So eventually they'll trickle down into consumer level.
>>107585909photonic*
this aged like someone sayingdude. imagine core scaling. imagine what we could do with 512 coresthe worst thing about multicore is a lot of software doesn't urilise it
>>107585553holy moly
>running MS Word can only take so much processing powerHow prescient
