In the span of a single human lifetime, our capacity to calculate has soared from the room-filling ENIAC's hundred-thousand operations per second to the exascale frontier where today's supercomputers perform over a quintillion calculations in that same blink of an eye.
Key Takeaways
- The ENIAC computer, completed in 1945, had a peak performance of approximately 0.0000001 gigaFLOPS (100 kiloFLOPS)
- The Manchester Mark 1, operational in 1949, performed about 1.2 kiloFLOPS in floating-point operations
- The UNIVAC I, delivered in 1951, achieved around 0.000001 gigaFLOPS (1 kiloFLOPS) peak performance
- Frontier supercomputer holds the current TOP500 #1 at 1.194 exaFLOPS Rmax as of June 2023
- Aurora supercomputer ranks #2 at 1.012 exaFLOPS Rmax in June 2023 TOP500 list
- Eagle supercomputer at 561.2 petaFLOPS Rmax, #3 on June 2023 TOP500
- AMD Ryzen Threadripper PRO 5995WX scores 100 GFLOPS peak single CPU FP64
- Intel Core i9-13900K achieves 1.7 TFLOPS FP32 peak with AVX-512
- NVIDIA H100 SXM GPU delivers 67 TFLOPS FP64 Tensor Core performance
- Frontier supercomputer efficiency is 52.72 gigaFLOPS/W Green500 #1 June 2023
- Aurora at 49.03 gigaFLOPS/W #2 on Green500 June 2023
- Eagle achieves 46.18 gigaFLOPS/W efficiency #3 Green500 June 2023
- Moore's Law predicts doubling of transistors every 2 years, implying ~1.86x computing power
- Exascale computing achieved 2022, zettascale targeted by 2030 at 10^21 FLOPS
- Quantum supremacy demonstrated by Google Sycamore at 53 qubits, 200s vs classical 10k years
Computing power has exponentially increased from primitive kiloFLOPS to modern exaflop supercomputers.
CPU and GPU Performance
1AMD Ryzen Threadripper PRO 5995WX scores 100 GFLOPS peak single CPU FP64
Verified2Intel Core i9-13900K achieves 1.7 TFLOPS FP32 peak with AVX-512
Verified3NVIDIA H100 SXM GPU delivers 67 TFLOPS FP64 Tensor Core performance
Verified4AMD Instinct MI300X GPU reaches 163.4 TFLOPS FP16 peak
Directional5Apple M2 Ultra chip peaks at 31.6 TFLOPS FP32 GPU performance
Single source6Intel Xeon Platinum 8592+ offers 2.9 TFLOPS FP64 per socket peak
Verified7NVIDIA A100 80GB GPU achieves 19.5 TFLOPS FP64 with Tensor Cores
Verified8AMD EPYC 9754 (Genoa) peaks at 3.2 TFLOPS FP64 dual socket
Verified9Qualcomm Snapdragon 8 Gen 2 GPU at 3.2 TFLOPS FP32 peak for mobile
Directional10IBM Power10 processor delivers 5 TFLOPS FP64 per chip
Single source11NVIDIA RTX 4090 GPU reaches 82.6 TFLOPS FP16 peak shader
Verified12AMD Radeon RX 7900 XTX at 61 TFLOPS FP32 peak performance
Verified13Intel Arc A770 GPU delivers 17.2 TFLOPS FP16 peak
Verified14ARM Neoverse V2 core in AWS Graviton3 peaks at 0.4 TFLOPS FP32 per core
Directional15Google Tensor Processing Unit v4 (Trillium) achieves 2.7 petaFLOPS FP16 per pod
Single source16Cerebras Wafer-Scale Engine 2 (WSE-2) delivers 20 petaFLOPS FP16 AI performance
Verified17Graphcore IPU Colossus MK2 GC200 at 350 TOPS INT8 per chip
Verified18SambaNova SN40L chip reaches 2 petaFLOPS FP16 per card
Verified19Tenstorrent Grayskull at 114 TOPS INT8 peak performance
Directional20SiPearl Rhea CPU for HPC peaks at 1.7 TFLOPS FP64 per socket
Single source21Frontier's HPE Slingshot-11 interconnect enables 200 Gb/s per node for calculating power
Verified22NVIDIA DGX H100 system with 8 H100 GPUs reaches 32 petaFLOPS FP8 AI
Verified23AMD MI250X dual-GPU die at 47.9 TFLOPS FP64 peak
Verified24Intel Ponte Vecchio (Max 1550) GPU at 56 TFLOPS FP64 Tensor
DirectionalCPU and GPU Performance Interpretation
This dizzying array of silicon bragging rights reveals a computational arms race where the only universal truth is that your laptop is now officially a glorified abacus compared to these number-crunching behemoths.
Current Supercomputers
1Frontier supercomputer holds the current TOP500 #1 at 1.194 exaFLOPS Rmax as of June 2023
Verified2Aurora supercomputer ranks #2 at 1.012 exaFLOPS Rmax in June 2023 TOP500 list
Verified3Eagle supercomputer at 561.2 petaFLOPS Rmax, #3 on June 2023 TOP500
Verified4Fugaku at 442.0 petaFLOPS Rmax, #4 in June 2023
Directional5LUMI at 381.0 petaFLOPS Rmax, #5 June 2023 TOP500
Single source6Frontier's Rp peak is 1.707 exaFLOPS in June 2023
Verified7El Capitan projected at over 2 exaFLOPS, but current Leonardo at 233.3 petaFLOPS #6 June 2023
Verified8Alps supercomputer at 204.8 petaFLOPS #7 June 2023 TOP500
Verified9MareNostrum 5 at 175.5 petaFLOPS #9 June 2023
Directional10Frontier uses 37,888 AMD Instinct MI250X GPUs contributing to its exaFLOPS performance
Single source11Aurora employs Intel Xeon Max CPUs and Data Center GPU Max for 1.012 exaFLOPS
Verified12Summit at Oak Ridge has 27,648 NVIDIA V100 GPUs for 148.6 petaFLOPS Rmax current rank #13
Verified13Perlmutter at NERSC delivers 64.6 petaFLOPS Rmax with AMD GPUs, rank #20 June 2023
Verified14Frontier consumes 20.99 MW power for 1.194 exaFLOPS, efficiency 56.9 gigaFLOPS/W
Directional15Japan's ABCI-Q at 95.2 petaFLOPS Rmax for quantum simulation, rank #26 June 2023
Single source16China's OceanLite at 1.3 exaFLOPS AI performance but 125.4 petaFLOPS HPL #27
Verified17Microsoft Azure Eagle at 561 petaFLOPS but HPL 561.2 petaFLOPS #3
Verified18Nvidia-powered Isambard-AI at 132.0 petaFLOPS #18 June 2023 TOP500
Verified19HPC6 at 110.4 petaFLOPS Rmax #24 June 2023
Directional20AMD EPYC 7763 CPU in Frontier nodes contributes to overall calculating power
Single source21HPE Cray EX architecture in Frontier enables 9.2 million cores
Verified22Japan's Fugaku with A64FX processors at 442 petaFLOPS sustained
Verified23European LUMI uses AMD MI250X GPUs for 381 petaFLOPS
Verified24Selene at 63.5 petaFLOPS Rmax with NVIDIA A100 GPUs #33 June 2023
DirectionalCurrent Supercomputers Interpretation
The global supercomputing race has now reached the exascale frontier, where power has become a public competition of precision engineering, national pride, and enormous electricity bills, all to make really, really difficult math look easy.
Energy Efficiency and Power Consumption
1Frontier supercomputer efficiency is 52.72 gigaFLOPS/W Green500 #1 June 2023
Verified2Aurora at 49.03 gigaFLOPS/W #2 on Green500 June 2023
Verified3Eagle achieves 46.18 gigaFLOPS/W efficiency #3 Green500 June 2023
Verified4Alps at 40.60 gigaFLOPS/W #4 Green500, consumes less power per FLOP
Directional5LUMI supercomputer at 38.99 gigaFLOPS/W #5 Green500 June 2023
Single source6NVIDIA H100 GPU efficiency up to 1.98 TFLOPS/W FP64 Tensor
Verified7AMD MI300X at 81.7 TFLOPS/W FP16 efficiency claimed
Verified8Google TPU v5e at 2.5x better efficiency than v4, ~400 TFLOPS/W INT8
Verified9Cerebras CS-3 wafer-scale at 1200 TFLOPS/W sparsity FP16
Directional10Graphcore Bow IPU efficiency 500+ TOPS/W INT8
Single source11Frontier total power draw 21 MW for 1.194 exaFLOPS
Verified12Fugaku consumes 29.9 MW at 442 petaFLOPS, 14.8 gigaFLOPS/W
Verified13Summit power 10.1 MW for 148.6 petaFLOPS, 14.7 gigaFLOPS/W
Verified14Sunway TaihuLight used 15.37 MW for 93 petaFLOPS, 6.05 gigaFLOPS/W historical
Directional15NVIDIA A100 SXM4 400W TDP for 19.5 TFLOPS FP64, ~48.75 GFLOPS/W
Single source16AMD EPYC 9754 400W TDP dual socket ~8 GFLOPS/W FP64
Verified17Intel Xeon 8592+ 350W TDP ~8.3 GFLOPS/W FP64 per socket
Verified18Apple M1 Max 60W for 10.4 TFLOPS FP32, 173 GFLOPS/W GPU
Verified19Qualcomm Snapdragon 8 Gen 2 5nm process 4nm effective, ~640 GFLOPS/W mobile GPU
Directional20IBM Power10 at 20.6 gigaFLOPS/W in TOP500 systems
Single source21SiPearl Rhea 2.0 nm process target 50+ GFLOPS/W FP64
Verified22El Capitan projected 2 exaFLOPS at under 30 MW, ~66 gigaFLOPS/W target
VerifiedEnergy Efficiency and Power Consumption Interpretation
In the relentless pursuit of computational might, the supercomputing arena reveals a stark hierarchy of efficiency, where the Frontier system's crown for doing the most with each watt is being challenged by specialized accelerators claiming efficiency numbers so high they seem to belong to a different league entirely.
Future Projections and Theoretical Limits
1Moore's Law predicts doubling of transistors every 2 years, implying ~1.86x computing power
Verified2Exascale computing achieved 2022, zettascale targeted by 2030 at 10^21 FLOPS
Verified3Quantum supremacy demonstrated by Google Sycamore at 53 qubits, 200s vs classical 10k years
Verified4IBM roadmap to 100k+ logical qubits by 2033 for fault-tolerant quantum computing
Directional5Landauer limit theoretical minimum energy 2.8 kT ln2 per bit erasure ~3 zJ/op at room temp
Single source6Dennard scaling ended 2006, but 3D stacking to continue power efficiency gains
Verified7Optical computing could reach 10^15 FLOPS/W vs electronic 10^12
Verified8Neuromorphic computing like Intel Loihi 2 at 10^12 ops/W synaptic
Verified9Frontier to El Capitan 2x performance at same power by 2025
Directional10AMD roadmap MI400 series 5x AI performance over MI300 by 2026
Single source11NVIDIA Rubin platform R100 GPU 30x inference perf over Hopper by 2026
Verified12Intel 18A process 1.8nm for Xeon 6 by 2025, 20% perf/W gain
Verified13TSMC A16 1.6nm node 10% speed 15-20% power reduction 2026
Verified14Quantum annealers like D-Wave Advantage 5000+ qubits solve optimization 10^6x faster
Directional15Photonic chips Lightmatter Passage 36 petaFLOPS FP16 at 10 kW
Single source16Global supercomputing capacity to hit 10 exaFLOPS aggregate by 2025
Verified17AI training FLOPS doubling every 6 months, 10x every 2 years per OpenAI
Verified18Bekenstein bound limits info density 10^69 bits/m^3 black hole, theoretical compute limit
Verified19Reversible computing could approach Landauer limit, 10^42 ops/J theoretical
Directional20Planetary computing limit Bremermann's 10^50 FLOPS/kg matter
Single source21Margolus-Levitin theorem 6×10^33 ops/J energy-time limit per op
Verified22ExaEnergy project targets 60 gigaFLOPS/W sustainable by 2030
Verified23Post-Moore photonics-neuromorphic hybrid 1000x efficiency by 2040
VerifiedFuture Projections and Theoretical Limits Interpretation
We are in the breathtakingly clever phase where our computing ambitions have outpaced even our best metaphors, simultaneously chasing the ghostly potential of quantum supremacy and the sobering physical limits of thermodynamics, all while patching the fading legacy of Moore's Law with a dazzling quilt of quantum, photonic, and neuromorphic architectures.
Historical Milestones
1The ENIAC computer, completed in 1945, had a peak performance of approximately 0.0000001 gigaFLOPS (100 kiloFLOPS)
Verified2The Manchester Mark 1, operational in 1949, performed about 1.2 kiloFLOPS in floating-point operations
Verified3The UNIVAC I, delivered in 1951, achieved around 0.000001 gigaFLOPS (1 kiloFLOPS) peak performance
Verified4The IBM 701, introduced in 1953, delivered approximately 0.000016 gigaFLOPS (16 kiloFLOPS)
Directional5The CDC 6600, launched in 1964, reached 3 megaFLOPS peak performance
Single source6The Cray-1 supercomputer, released in 1976, had a peak speed of 160 megaFLOPS
Verified7The Cray X-MP, introduced in 1982, achieved up to 940 megaFLOPS in multi-processor configuration
Verified8The Connection Machine CM-5, deployed in 1991, scaled to 1.056 teraFLOPS with 1024 processors
Verified9ASCI Red, completed in 1997, became the first teraFLOPS supercomputer at 1.338 teraFLOPS
Directional10ASCI White, operational in 2000, peaked at 7.226 teraFLOPS
Single source11Earth Simulator, launched in 2002, achieved 35.86 teraFLOPS on TOP500 list
Verified12Blue Gene/L, reached 280.6 teraFLOPS in 2006
Verified13Roadrunner supercomputer hit 1.026 petaFLOPS in 2008
Verified14Tianhe-1A achieved 2.566 petaFLOPS in 2010
Directional15Fujitsu K computer reached 10.51 petaFLOPS in 2011
Single source16Titan supercomputer delivered 17.59 petaFLOPS in 2013
Verified17Tianhe-2 peaked at 33.86 petaFLOPS in 2014
Verified18Sunway TaihuLight achieved 93.01 petaFLOPS in 2016
Verified19Summit supercomputer reached 122.3 petaFLOPS in 2018
Directional20IBM Power9-based Sierra hit 94.64 petaFLOPS in 2018
Single source21Frontier became the first exaFLOPS machine at 1.102 exaFLOPS in 2022
Verified22The first TOP500 list in June 1993 was topped by TMC CM-5/1024 at 59.7 gigaFLOPS
Verified23Intel Paragon XP/S 140 at 143.4 gigaFLOPS topped November 1993 list
Verified24Numerical Wind Tunnel at 170.0 gigaFLOPS in June 1994
Directional25Intel Paragon at 281.0 gigaFLOPS in November 1996
Single source26ASCI Red at 1.340 teraFLOPS in June 1997
Verified27ASCI Red sustained 1.064 teraFLOPS in November 1997
Verified28ASCI Red at 2.382 teraFLOPS in June 1999
Verified29ASCI Q reached 4.944 teraFLOPS projected in November 2001
Directional30Earth Simulator at 35.860 teraFLOPS in June 2002
Single sourceHistorical Milestones Interpretation
The breathtaking speed at which we've rocketed from needing an entire room to calculate a single artillery trajectory to casually simulating supernovas on a desktop proves that humanity's appetite for computational power is the ultimate exponential curve—our ambitions outrace our machines almost as soon as we build them.
Sources & References
- Reference 1ENen.wikipedia.orgVisit source
- Reference 2TOP500top500.orgVisit source
- Reference 3CPUBENCHMARKcpubenchmark.netVisit source
- Reference 4TECHPOWERUPtechpowerup.comVisit source
- Reference 5NVIDIAnvidia.comVisit source
- Reference 6AMDamd.comVisit source
- Reference 7APPLEapple.comVisit source
- Reference 8INTELintel.comVisit source
- Reference 9QUALCOMMqualcomm.comVisit source
- Reference 10IBMibm.comVisit source
- Reference 11AWSaws.amazon.comVisit source
- Reference 12CLOUDcloud.google.comVisit source
- Reference 13CEREBRAScerebras.netVisit source
- Reference 14GRAPHCOREgraphcore.aiVisit source
- Reference 15SAMBANOVAsambanova.aiVisit source
- Reference 16TENSTORRENTtenstorrent.comVisit source
- Reference 17SIPEARLsiPearl.comVisit source
- Reference 18HPEhpe.comVisit source
- Reference 19ARKark.intel.comVisit source
- Reference 20ANANDTECHanandtech.comVisit source
- Reference 21SIPEARLsipearl.comVisit source
- Reference 22EXASCALECOMPUTINGPROJECTexascalecomputingproject.orgVisit source
- Reference 23NATUREnature.comVisit source
- Reference 24ENERGYenergy.govVisit source
- Reference 25NVIDIANEWSnvidianews.nvidia.comVisit source
- Reference 26TSMCtsmc.comVisit source
- Reference 27DWAVESYSdwavesys.comVisit source
- Reference 28LIGHTMATTERlightmatter.coVisit source
- Reference 29SEMIANALYSISsemianalysis.comVisit source
- Reference 30ARXIVarxiv.orgVisit source
- Reference 31EUROHPC-JUeurohpc-ju.europa.euVisit source
- Reference 32IEEEieee.orgVisit source