Directory
Parts taxonomy
The component layer the systems graph draws its edges into. Every part page lists the systems using it, so a question like “what is installed underneath Slingshot” has an answer you can click to rather than assemble by hand.
Accelerators
| Part | Supplier | Generation | Released | Systems | Rmax underneath |
|---|---|---|---|---|---|
| AWS Trainium2 | Amazon | Trainium | 2024 | — | — |
| NVIDIA GB200 NVL72 | NVIDIA | Blackwell | 2024 | — | — |
| AMD Instinct MI300A | AMD | CDNA 3 | 2023 | 1 | 1.74 EFlop/s |
| Google TPU v5p | Alphabet | TPU v5 | 2023 | — | — |
| Intel Data Center GPU Max 1550 | Intel | Ponte Vecchio | 2023 | 1 | 1.01 EFlop/s |
| NVIDIA GH200 Grace Hopper Superchip | NVIDIA | Grace Hopper | 2023 | 1 | 434.9 PFlop/s |
| NVIDIA H100 SXM5 | NVIDIA | Hopper | 2022 | 1 | 561.2 PFlop/s |
| AMD Instinct MI250X | AMD | CDNA 2 | 2021 | 2 | 1.73 EFlop/s |
| NVIDIA A100 SXM4 64GB | NVIDIA | Ampere | 2020 | 1 | 241.2 PFlop/s |
| NUDT Matrix-2000 | NUDT | Matrix | 2017 | 1 | 61.44 PFlop/s |
| NVIDIA Tesla V100 | NVIDIA | Volta | 2017 | 2 | 243.2 PFlop/s |
| NVIDIA Tesla P100 | NVIDIA | Pascal | 2016 | — | — |
| Intel Xeon Phi 31S1P | Intel | Knights Corner | 2013 | 1 | 61.44 PFlop/s |
| NVIDIA Tesla K20X | NVIDIA | Kepler | 2012 | 1 | 17.59 PFlop/s |
| NVIDIA Tesla M2050 | NVIDIA | Fermi | 2010 | 1 | 2.57 PFlop/s |
| IBM PowerXCell 8i | IBM | Cell BE | 2008 | 1 | 1.03 PFlop/s |
CPUs
| Part | Supplier | Generation | Released | Systems | Rmax underneath |
|---|---|---|---|---|---|
| AMD 4th Gen EPYC 24C (in MI300A) | AMD | Zen 4 | 2023 | 1 | 1.74 EFlop/s |
| Intel Xeon CPU Max 9470 | Intel | Sapphire Rapids HBM | 2023 | 1 | 1.01 EFlop/s |
| Intel Xeon Platinum 8480C | Intel | Sapphire Rapids | 2023 | 1 | 561.2 PFlop/s |
| NVIDIA Grace | NVIDIA | Grace | 2023 | 1 | 434.9 PFlop/s |
| AMD EPYC 9004 (Genoa) | AMD | Zen 4 | 2022 | — | — |
| AMD EPYC 7A53 (Trento) | AMD | Zen 3 | 2021 | 2 | 1.73 EFlop/s |
| Intel Xeon Platinum 8358 | Intel | Ice Lake-SP | 2021 | 1 | 241.2 PFlop/s |
| Fujitsu A64FX | Fujitsu | Armv8.2-A SVE | 2019 | 1 | 442.0 PFlop/s |
| IBM POWER9 22C | IBM | POWER9 | 2017 | 2 | 243.2 PFlop/s |
| Sunway SW26010 | NRCPC | Sunway | 2016 | 1 | 93.01 PFlop/s |
| Intel Xeon E5-2690 v3 | Intel | Haswell-EP | 2014 | — | — |
| Intel Xeon E5-2692 v2 | Intel | Ivy Bridge-EP | 2013 | 1 | 61.44 PFlop/s |
| AMD Opteron 6274 (Interlagos) | AMD | Bulldozer | 2011 | 1 | 17.59 PFlop/s |
| Fujitsu SPARC64 VIIIfx | Fujitsu | SPARC64 | 2010 | 1 | 10.51 PFlop/s |
| Intel Xeon X5670 | Intel | Westmere-EP | 2010 | 1 | 2.57 PFlop/s |
| AMD Opteron 2435 (Istanbul) | AMD | K10 | 2009 | 1 | 1.76 PFlop/s |
| AMD Opteron 2210 (Santa Rosa) | AMD | K8 | 2006 | 1 | 1.03 PFlop/s |
| IBM PowerPC 440 700MHz | IBM | BlueGene/L | 2004 | 1 | 478.20 TFlop/s |
| NEC SX-6 vector processor | NEC | SX | 2002 | 1 | 35.86 TFlop/s |
| Intel Pentium Pro 200 MHz | Intel | P6 | 1996 | 1 | 1.07 TFlop/s |
| Thinking Machines CM-5 vector node | Thinking Machines | Connection Machine | 1993 | 1 | 59.7 GFlop/s |
Interconnect
| Part | Supplier | Generation | Released | Systems | Rmax underneath |
|---|---|---|---|---|---|
| NVIDIA InfiniBand NDR (Quantum-2) | NVIDIA | NDR | 2022 | 1 | 561.2 PFlop/s |
| HPE Slingshot-11 | Hewlett Packard Enterprise | Slingshot | 2021 | 5 | 4.92 EFlop/s |
| Fujitsu Tofu Interconnect D | Fujitsu | Tofu D | 2019 | 1 | 442.0 PFlop/s |
| NVIDIA InfiniBand HDR | NVIDIA | HDR | 2019 | 1 | 241.2 PFlop/s |
| Eviden BXI v2 | Eviden | BXI | 2016 | — | — |
| Sunway interconnect | NRCPC | Sunway | 2016 | 1 | 93.01 PFlop/s |
| Intel Omni-Path Architecture | Intel | OPA | 2015 | — | — |
| Mellanox InfiniBand EDR | NVIDIA | EDR | 2014 | 2 | 243.2 PFlop/s |
| NUDT TH Express-2 | NUDT | TH Express | 2013 | 1 | 61.44 PFlop/s |
| Cray Aries | Cray | Aries | 2012 | — | — |
| Cray Gemini | Cray | Gemini | 2010 | 1 | 17.59 PFlop/s |
| Fujitsu Tofu Interconnect | Fujitsu | Tofu | 2010 | 1 | 10.51 PFlop/s |
| NUDT Arch interconnect | NUDT | Arch | 2010 | 1 | 2.57 PFlop/s |
| Cray SeaStar2+ | Cray | SeaStar | 2007 | 1 | 1.76 PFlop/s |
| Mellanox InfiniBand DDR | NVIDIA | DDR | 2006 | 1 | 1.03 PFlop/s |
| IBM BlueGene 3D torus | IBM | BlueGene | 2004 | 1 | 478.20 TFlop/s |
| NEC Earth Simulator crossbar | NEC | SX | 2002 | 1 | 35.86 TFlop/s |
| Intel ASCI Red mesh | Intel | Paragon | 1996 | 1 | 1.07 TFlop/s |
| Thinking Machines CM-5 fat tree | Thinking Machines | Connection Machine | 1993 | 1 | 59.7 GFlop/s |
Memory
| Part | Supplier | Generation | Released | Systems | Rmax underneath |
|---|---|---|---|---|---|
| HBM3E | open standard | HBM3E | 2024 | — | — |
| Micron HBM3E 8-High | Micron | HBM3E | 2024 | — | — |
| SK hynix HBM3E 12-High | SK hynix | HBM3E | 2024 | — | — |
| HBM3 | open standard | HBM3 | 2022 | — | — |
| DDR5 RDIMM | open standard | DDR5 | 2021 | — | — |
| HBM2E | open standard | HBM2E | 2020 | — | — |
Storage
| Part | Supplier | Generation | Released | Systems | Rmax underneath |
|---|---|---|---|---|---|
| Intel DAOS | Intel | DAOS | 2020 | 1 | 1.01 EFlop/s |
| HPE Cray ClusterStor E1000 | Hewlett Packard Enterprise | ClusterStor | 2019 | 3 | 3.47 EFlop/s |
| VAST Data Platform | VAST Data | VAST | 2019 | — | — |
| DDN EXAScaler | DDN | EXAScaler | 2018 | — | — |
| WEKA Data Platform | WEKA | WEKA | 2018 | — | — |
| IBM Spectrum Scale ESS (GPFS) | IBM | Spectrum Scale | 2017 | 1 | 148.6 PFlop/s |
Networking
| Part | Supplier | Generation | Released | Systems | Rmax underneath |
|---|---|---|---|---|---|
| Arista 7800R4 AI | Arista Networks | 7800R | 2024 | — | — |
| Broadcom Tomahawk 5 | Broadcom | Tomahawk | 2022 | — | — |
| NVIDIA Spectrum-4 | NVIDIA | Spectrum | 2022 | — | — |
Cooling
| Part | Supplier | Generation | Released | Systems | Rmax underneath |
|---|---|---|---|---|---|
| Direct-to-chip cold plate | open standard | DLC | 2018 | — | — |
| Single-phase immersion | open standard | Immersion | 2015 | — | — |
| Direct liquid cooling, warm water | open standard | DLC | 2012 | 7 | 5.60 EFlop/s |
| Rear-door heat exchanger | open standard | RDHx | 2010 | — | — |
| Air cooling (CRAC/CRAH) | open standard | Air | 1990 | 1 | 35.86 TFlop/s |
Power
| Part | Supplier | Generation | Released | Systems | Rmax underneath |
|---|---|---|---|---|---|
| Schneider Galaxy VX UPS | Schneider Electric | Galaxy | 2016 | — | — |
| Vertiv Liebert UPS | Vertiv | Liebert | 2015 | — | — |
| Eaton Power Xpert busway | Eaton | Power Xpert | 2014 | — | — |
Racks
| Part | Supplier | Generation | Released | Systems | Rmax underneath |
|---|---|---|---|---|---|
| OCP Open Rack v3 | open standard | ORv3 | 2022 | — | — |
| HPE Cray EX cabinet | Hewlett Packard Enterprise | Cray EX | 2019 | — | — |
Facility
| Part | Supplier | Generation | Released | Systems | Rmax underneath |
|---|---|---|---|---|---|
| Dry cooler | open standard | — | 2005 | — | — |
| Medium-voltage switchgear | open standard | — | 2000 | — | — |
| Evaporative cooling tower | open standard | — | 1990 | — | — |
Operating systems
| Part | Supplier | Generation | Released | Systems | Rmax underneath |
|---|---|---|---|---|---|
| HPE Cray Operating System | Hewlett Packard Enterprise | COS | 2019 | 5 | 4.92 EFlop/s |
| Sunway RaiseOS | NRCPC | RaiseOS | 2016 | 1 | 93.01 PFlop/s |
| Kylin Linux | KylinSoft | Kylin | 2010 | 2 | 64.01 PFlop/s |
| Cray Linux Environment | Cray | CLE | 2007 | 2 | 19.35 PFlop/s |
| IBM Compute Node Kernel | IBM | CNK | 2004 | 1 | 478.20 TFlop/s |
| Red Hat Enterprise Linux | Red Hat | RHEL | 2002 | 3 | 484.4 PFlop/s |
| SUSE Linux Enterprise Server | SUSE | SLES | 2000 | — | — |