then eni ac 1 9 4 6 5 khz the future is not
play

Then... ENI AC 1 9 4 6 ( 5 kHz) The Future is not 1 8 0 0 0 - PDF document

Then... ENI AC 1 9 4 6 ( 5 kHz) The Future is not 1 8 0 0 0 radiorr sladdprogram m erad w hat it used to be... 5 KHz Erik Hagersten Professor i Datorarkitektur Uppsala Universitet AVDARK ENIAC 2 0 1 1 Dept of I nform


  1. Then... ENI AC 1 9 4 6 ( ”5 kHz”) The Future is not 1 8 0 0 0 radiorör sladdprogram m erad w hat it used to be... ”5 KHz” Erik Hagersten Professor i Datorarkitektur Uppsala Universitet AVDARK ENIAC 2 0 1 1 Dept of I nform ation Technology| w w w .it.uu.se Erik Hagersten| user.it.uu.se/ ~ eh “Recently” APZ 2 1 2 , 1 9 8 3 Then ( in Sw eden) Ericsson’s Supercom puter ( “5 MHz”)  BARK (~ 1950)  8 000 relays,  80 km cables  BESK (~ 1953)  2 400 vac. tubes  ”20 kHz” (world record) AVDARK AVDARK 2 0 1 1 2 0 1 1 Dept of I nform ation Technology| w w w .it.uu.se Erik Hagersten| user.it.uu.se/ ~ eh Dept of I nform ation Technology| w w w .it.uu.se Erik Hagersten| user.it.uu.se/ ~ eh APZ 2 1 2 6 5 years of “im provem ents” m arketing brochure quotes:  Speed  ”Very compact”  Size  6 times the performance  1/ 6: th the size  Price  1/ 5 the power consumption  Price/ performance  ”A breakthrough in computer science”  Reliability  ”Why more CPU power?”  ”All the power needed for future development”  Predictability  ”… 800,000 BHCA, should that ever be needed”  Energy  ”SPC computer science at its most elegance”  Safety  ”Using 64 kbit memory chips”  ”1500W power consumption  Usability… . AVDARK AVDARK 2 0 1 1 2 0 1 1 Dept of I nform ation Technology| w w w .it.uu.se Erik Hagersten| user.it.uu.se/ ~ eh Dept of I nform ation Technology| w w w .it.uu.se Erik Hagersten| user.it.uu.se/ ~ eh

  2. ”Moore’s Law ” Exponentiell utveckling Pop: Double perform ance every 1 8 -2 4 th m onth Exam ple: Doubling every 2 nd year Perform ance How long does it it take for 1 0 0 0 x im provem ent? [ log] Multicore Exam ple: Doubling every 1 8 th m onth 1000 How long does it it take for 1 0 0 0 x im provem ent? Single-core Log 100 scale 1000 100 10 10 1 time 1 ? Linear Year scale AVDARK AVDARK 2006 2 0 1 1 2 0 1 1 Dept of I nform ation Technology| w w w .it.uu.se Erik Hagersten| user.it.uu.se/ ~ eh Dept of I nform ation Technology| w w w .it.uu.se Erik Hagersten| user.it.uu.se/ ~ eh Ray Kurzw eil pictures Linear scale 1 9 4 0  2 0 1 7 w w w .Kurzw eilAI .net/ pps/ W orldHealthCongress/ ( 2 x perform ance every 1 8 th m onth) Doubling every 18th month since 1940 4,E+15 3,E+15 Performance 3,E+15 2,E+15 2,E+15 1,E+15 5,E+14 0,E+00 40 50 60 70 80 90 0 10 Year AVDARK AVDARK 2 0 1 1 2 0 1 1 Dept of I nform ation Technology| w w w .it.uu.se Erik Hagersten| user.it.uu.se/ ~ eh Dept of I nform ation Technology| w w w .it.uu.se Erik Hagersten| user.it.uu.se/ ~ eh Ray Kurzw eil pictures Ray Kurzw eil pictures w w w .Kurzw eilAI .net/ pps/ W orldHealthCongress/ w w w .Kurzw eilAI .net/ pps/ W orldHealthCongress/ AVDARK AVDARK 2 0 1 1 2 0 1 1 Dept of I nform ation Technology| w w w .it.uu.se Erik Hagersten| user.it.uu.se/ ~ eh Dept of I nform ation Technology| w w w .it.uu.se Erik Hagersten| user.it.uu.se/ ~ eh

  3. Ray Kurzw eil pictures Exponentiell utveckling: w w w .Kurzw eilAI .net/ pps/ W orldHealthCongress/ Doublerings/ halverings-tider ( according to Kurzw eil)  Dynam ic RAM Mem ory ( bits per dollar) 1 .5 years Average Transistor Price 1 .6 years  Microprocessor Cost per Transistor Cycle 1 .1 years   Total Bits Shipped 1 .1 years  Processor Perform ance in MI PS 1 .8 years Transistors in I ntel Microprocessors 2 .0 years  Log scale 1000 100 10 1 time AVDARK AVDARK 2 0 1 1 2 0 1 1 Dept of I nform ation Technology| w w w .it.uu.se Erik Hagersten| user.it.uu.se/ ~ eh Dept of I nform ation Technology| w w w .it.uu.se Erik Hagersten| user.it.uu.se/ ~ eh HiPEAC Roadm ap -- High Perform ance Em bedded Architecture and Com pilers AVDARK 2 0 1 1 Dept of I nform ation Technology| w w w .it.uu.se Erik Hagersten| user.it.uu.se/ ~ eh

Recommend


More recommend