計算科学研究センターにおける CPU 能力の変遷
| 年 | 機種 | 理論総演算性能 (MFLOPS) |
|---|---|---|
| 1979 | HITACHI M-180 (2台) | 36 |
| 1980 | HITACHI M-180 | 18 |
| HITACHI M-200H | 48 | |
| 合計 | 66 | |
| 1982 | HITACHI M-200H (2台) | 52 |
| 1986 | HITACHI M-680H | 16 |
| HITACHI S-810/10 | 315 | |
| 合計 | 331 | |
| 1988 | HITACHI M-680H | 16 |
| HITACHI S-820/80 | 2,000 | |
| 合計 | 2,016 | |
| 1991 | HITACHI M-680H(+) | 32 |
| HITACHI S-820/80 | 2,000 | |
| 合計 | 2,032 | |
| 1994 | HITACHI M-680H(+) | 32 |
| NEC SX-3/34R (3 CPU) | 19,200 | |
| 合計 | 19,232 | |
| 1995 | IBM SP2 (Wide24 台) | 288.0×24 |
| IBM SP2 (Thin24 台) | 118.0×24 | |
| NEC HSP | 300 | |
| NEC SX-3/34R (3 CPU) | 19,200 | |
| 合計 | 29,244 | |
| 1999 | IBM SP2 (Wide24 台) | 288.0×24 |
| IBM SP2 (Thin24 台) | 118.0×24 | |
| NEC SX-5 (8 CPU) | 64,000 | |
| NEC SX-3/34R (3 CPU) | 19,200 | |
| 合計 | 92,944 | |
| 2000 | IBM SP2 (Wide24 台) | 288.0×24 |
| IBM SP2 (Thin24 台) | 118.0×24 | |
| NEC SX-5 (8 CPU) | 64,000 | |
| Fujitsu VPP5000 (30 PE) | 288,000 | |
| SGI SGI2800 (256 CPU) | 153,000 | |
| 合計 | 514,744 | |
| 2001 | IBM SP2 (Wide24 台) | 288.0×24 |
| IBM SP2 (Thin24 台) | 118.0×24 | |
| NEC SX-5 (8 CPU) | 64,000 | |
| Fujitsu VPP5000 (30 PE) | 288,000 | |
| SGI SGI2800 (192 CPU) | 115,200 | |
| SGI Origin3800 (128 CPU) | 102,400 | |
| 合計 | 579,344 | |
| 2003 | NEC SX-7 (32 CPU) | 282,560 |
| NEC TX7 (64 CPU) | 256,000 | |
| Fujitsu VPP5000 (30 PE) | 288,000 | |
| SGI SGI2800 (192 CPU) | 115,200 | |
| SGI Origin3800 (128 CPU) | 102,400 | |
| 合計 | 1,044,160 | |
| 2006 | NEC SX-7 (32 CPU) | 282,560 |
| NEC TX7 (64 CPU) | 256,000 | |
| Fujitsu PRIMEQUEST (640 core) | 4,096,000 | |
| SGI Altix4700 (640 core) | 4,096,000 | |
| 合計 | 8,730,560 | |
| 2008 | Fujitsu PRIMEQUEST (640 core) | 4,096,000 |
| SGI Altix4700 (640 core) | 4,096,000 | |
| Hitachi SR16000 (288 way) | 5,414,400 | |
| 合計 | 13,606,400 | |
| 2011 | Fujitsu PRIMERGY RX300S7 (5,472 core) | 126,950,400 |
| (+ NVIDIA Tesla M2090 32台) | 21,280,000 | |
| Fujitsu PRIMEHPC FX10 (1,536 core) | 20,152,320 | |
| SGI UV1000 (576 core) | 6,128,640 | |
| Hitachi SR16000 (288 way) | 5,414,400 | |
| 合計 | 179,925,760 | |
| 2012 | Fujitsu PRIMERGY RX300S7 (5,472 core) | 126,950,400 |
| (+ NVIDIA Tesla M2090 32台) | 21,280,000 | |
| Fujitsu PRIMEHPC FX10 (1,536 core) | 20,152,320 | |
| SGI UV2000(1024core) | 21,299,200 | |
| Fujitsu PRIMERGY CX250S1 (5,888 core) | 136,601,600 | |
| 合計 | 326,283,520 | |
| 2015 | Fujitsu PRIMERGY RX300S7 (5,472 core) | 126,950,400 |
| (+ NVIDIA Tesla M2090 32台) | 21,280,000 | |
| Fujitsu PRIMEHPC FX10 (1,536core) | 20,152,320 | |
| SGI UV2000 (1,024core) | 21,299,200 | |
| Fujitsu PRIMERGY CX2550M1 (7,280 core) | 302,848,000 | |
| 合計 | 492,529,920 | |
| 2017 | NEC LX LX 2U-Twin2サーバ 406Rh-2 (31,760 core) | 2,439,000,000 |
| NEC LX 1Uサーバ 110Rh-1 (800 core) | 61,400,000 | |
| NEC LX 1Uサーバ 110Rh-1 (5,724 core) | 550,000,000 | |
| NEC LX 4U-GPUサーバ 108Th-4G (2,304 core) | 221,000,000 | |
| (+NVIDIA Tesla P100 2枚×96台) | 806,000,000 | |
| Fujitsu PRIMEHPC FX10 (1,536 core) | 20,152,320 | |
| 合計 | 3,097,552,320 | |
| 2022 | HPE Apollo2000 Gen 10 Plus (102,912 core) | 4,034,000,000 |
| HPE Apollo2000 Gen 10 Plus (1,792 core) | 70,000,000 | |
| HPE Apollo6500 Gen 10 Plus (2,048 core) | 80,000,000 | |
| (+NVIDIA A100 NVLink 128個) | 2,496,000,000 | |
合計 | 6,680,000,000 |
詳細版
| 添付 | サイズ |
|---|---|
| スパコンマシン変移図2015.png | 623.28 KB |
| スパコン演算性能変移図2015.png | 63.64 KB |
| スパコン演算性能変移図2015-2.png | 21.51 KB |
