Overclock.net › Forums › Industry News › Hardware News › [custompc] IBM develops light pulse switch for multi core CPUs
New Posts  All Forums:Forum Nav:

[custompc] IBM develops light pulse switch for multi core CPUs - Page 2

post #11 of 26
Quote:
Originally Posted by max302 View Post
Wow someone needs to learn respect and stop being a lamer with a member who's more repped than you are. Keep it respectful.

What he means (or at least what I got from it) is that on such a short circuit, you don't give a woot about going really really fast or really really really fast.
Agree that the poster was a bit harsh, but let's not get side tracked. The benefits of the technology are low heat and switching speed. Remember that when overclocking you raise vcore to give transistor gates a higher threshold to drop current with a less likelihood of false on's because the current is not dropping fast enough. The light based gate may not have these issues of heat nor false on's because the current has not dropped in the gate soon enough.
Black Box 3
(13 items)
 
  
CPUMotherboardGraphicsHard Drive
i7 920 4GHz 1.248 VCore Gig EX58-UD3R 5870 2x160GB Hitachi RAID 0, 500GB WD 
Optical DriveOSPowerCase
ASUS DRW-2014L1T Windows 7 Ultimate 64bit SilverStone Zeus 850w Antec 902 
  hide details  
Reply
Black Box 3
(13 items)
 
  
CPUMotherboardGraphicsHard Drive
i7 920 4GHz 1.248 VCore Gig EX58-UD3R 5870 2x160GB Hitachi RAID 0, 500GB WD 
Optical DriveOSPowerCase
ASUS DRW-2014L1T Windows 7 Ultimate 64bit SilverStone Zeus 850w Antec 902 
  hide details  
Reply
post #12 of 26
I think the main advantage they boasted was the ability to use different wavelengths of light on a single switch at the same time, kind of like multicore wires. So while light itself may not be much faster than copper wires, try making a single copper wire transmit two packets completely simultaneously.
post #13 of 26
Quote:
Originally Posted by xdaseinx View Post
Agree that the poster was a bit harsh, but let's not get side tracked. The benefits of the technology are low heat and switching speed. Remember that when overclocking you raise vcore to give transistor gates more time to drop current with a less likelihood of false on's. The light based gate would not have these issues of heat nor false on's because the current has not dropped in the gate soon enough.
yes i mite have been a bit to harsh, but i defenetly agree with what ur saying and which is what i was trying to say. As you said light based gates wouldnt have such a problem which means that 2/3 of the speed of light that is the electricity, would be more like 2/5 in the long run because of all the thing light based technology would allow.
Insanity 2.0
(16 items)
 
Insanity 1.0
(14 items)
 
Insanity 1.5
(9 items)
 
Reply
Insanity 2.0
(16 items)
 
Insanity 1.0
(14 items)
 
Insanity 1.5
(9 items)
 
Reply
post #14 of 26
Quote:
Originally Posted by Licht View Post
Far as i know on the size of a die there isn't much of a boost in latency with switching from regular HT to light-based HT. Because electricity travels near the same speed as light.
This doesn't address latency though. This addresses bandwidth.
post #15 of 26
Quote:
Originally Posted by noldevin View Post
I think the main advantage they boasted was the ability to use different wavelengths of light on a single switch at the same time, kind of like multicore wires. So while light itself may not be much faster than copper wires, try making a single copper wire transmit two packets completely simultaneously.
Yes also i defenatly agree with u too. Even tho 2/3 of the speed of light comparing to the 3/3 is a big jump already, as u said transmiting two packets at same exect time between the cores, would increase the speeds even more boosting it to up two times the speed of light.
Insanity 2.0
(16 items)
 
Insanity 1.0
(14 items)
 
Insanity 1.5
(9 items)
 
Reply
Insanity 2.0
(16 items)
 
Insanity 1.0
(14 items)
 
Insanity 1.5
(9 items)
 
Reply
post #16 of 26
stop arguing. basically IBM found a way to transmit more data at a given time without copper THATS ALL.
    
CPUMotherboardGraphicsRAM
Q6700 Intel DP45SG Visiontek HD 4870 2x2GB DDR3 1066 
Hard DriveOptical DriveOSMonitor
320 Western Digital + 1TB Black Caviar WD Samsung Sata vista ultimate x64/Windows 7 Ultimate 64x Samsung T220 22 inch 
KeyboardPowerCaseMouse
Dynex(Temp until G15) BFG 800 Watt Antec 900 Logitech G5 Gaming mouse 
Mouse Pad
Commando(none) 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
Q6700 Intel DP45SG Visiontek HD 4870 2x2GB DDR3 1066 
Hard DriveOptical DriveOSMonitor
320 Western Digital + 1TB Black Caviar WD Samsung Sata vista ultimate x64/Windows 7 Ultimate 64x Samsung T220 22 inch 
KeyboardPowerCaseMouse
Dynex(Temp until G15) BFG 800 Watt Antec 900 Logitech G5 Gaming mouse 
Mouse Pad
Commando(none) 
  hide details  
Reply
post #17 of 26
Quote:
Originally Posted by gex80 View Post
stop arguing. basically IBM found a way to transmit more data at a given time without copper THATS ALL.
Thats ALL? lmao u do understand optics is the future, and by saying thats all? is like saying, wow she ate a coockie thats all? no lol its more then just a way to transmit mroe data at a given time, its more of a huge step in our technology. lol sometimes i wonder about u ppl..
Insanity 2.0
(16 items)
 
Insanity 1.0
(14 items)
 
Insanity 1.5
(9 items)
 
Reply
Insanity 2.0
(16 items)
 
Insanity 1.0
(14 items)
 
Insanity 1.5
(9 items)
 
Reply
post #18 of 26
And dont just think of it as two packets, this technology can send out as many packets at a time as there are shades of colors, limited only to how accurate their frequency detection is.
post #19 of 26
Quote:
Originally Posted by noldevin View Post
And dont just think of it as two packets, this technology can send out as many packets at a time as there are shades of colors, limited only to how accurate their frequency detection is.
very true, but i dont think that the begining of the technology will alow more then maybe 3 or 4 packets at a time. l8r they will advance. but as of right now, i dont think they can detect the frequencys accurate enough.
Insanity 2.0
(16 items)
 
Insanity 1.0
(14 items)
 
Insanity 1.5
(9 items)
 
Reply
Insanity 2.0
(16 items)
 
Insanity 1.0
(14 items)
 
Insanity 1.5
(9 items)
 
Reply
post #20 of 26
I can see at least the basic colors all being used at the get-go. ROYGBIV, thats 7 channels on one pipeline. And who says they have to use just one pipe?
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Hardware News
Overclock.net › Forums › Industry News › Hardware News › [custompc] IBM develops light pulse switch for multi core CPUs