Highlights - November 2026
This is the 68th edition of the TOP500.
Here is a summary of the systems in the Top 10:
-
LineShine is the No. 1 system on the TOP500 since June 2026.
It is installed at the National Supercomputing Centre in Shenzhen (NSCS),
China, and was built by the Shenzhen Cloud Computing Center. It achieved
2.198 Exaflop/s on the HPL benchmark using 13,789,440 cores. The system
is based on the custom LingKun platform with LX2 304C 1.55GHz processors,
the proprietary LingQi interconnect, and Kylin OS. LineShine also leads
the HPCG ranking with 22.00 Petaflop/s.
-
El Capitan at the Lawrence Livermore National Laboratory,
California, USA, is the No. 2 system on the TOP500. The HPE Cray EX255a
system was measured with 1.809 Exaflop/s on the HPL benchmark. El Capitan
has 11,340,000 cores and is based on AMD 4th generation EPYC™ processors
with 24 cores at 1.8 GHz and AMD Instinct™ MI300A accelerators. It uses
the Cray Slingshot 11 network for data transfer and achieves an energy
efficiency of 60.94 Gigaflops/watt. The system is No. 2 on the HPCG
benchmark ranking with 17.41 Petaflop/s.
-
Frontier is the No. 3 system in the TOP500. It is installed
at the Oak Ridge National Laboratory (ORNL) in Tennessee, USA, where it is
operated for the Department of Energy (DOE). It has achieved 1.353 Exaflop/s
using 9,066,176 cores. The HPE Cray EX architecture combines 3rd Gen
AMD EPYC™ CPUs optimized for HPC and AI, with AMD Instinct™ MI250X
accelerators, and a Slingshot-11 interconnect.
-
Aurora is the No. 4 system with a HPL score of
1.012 Exaflop/s. It is installed at the Argonne Leadership Computing
Facility, Illinois, USA, where it is also operated for the Department
of Energy (DOE). The Intel system is based on HPE Cray EX - Intel
Exascale Compute Blades. It uses Intel Xeon CPU Max Series processors,
Intel Data Center GPU Max Series accelerators, and a Slingshot-11
interconnect.
-
JUPITER Booster is the No. 5 system. It is installed at
EuroHPC/FZJ in Jülich, Germany, where it is operated by the Jülich
Supercomputing Centre. It is based on Eviden's BullSequana XH3000 direct
liquid-cooled architecture, which utilizes Grace Hopper Superchips.
It remains measured at 1.000 Exaflop/s, the first Exascale system
in Europe.
| Rank |
Site |
System |
Cores |
Rmax (TFlop/s) |
Rpeak (TFlop/s) |
Power (kW) |
| 1 |
National Supercomputing Centre in Shenzhen (NSCS) China
|
LineShine - LingKun, LX2 304C 1.55GHz, LingQi, Kylin OS
Shenzhen Cloud Computing Center Co., Ltd. |
13,789,440 |
2,198.40 |
2,735.82 |
42,220 |
| 2 |
DOE/NNSA/LLNL United States
|
El Capitan - HPE Cray EX255a, AMD 4th Gen EPYC 24C 1.8GHz, AMD Instinct MI300A, Slingshot-11, TOSS
HPE |
11,340,000 |
1,809.00 |
2,821.10 |
29,685 |
| 3 |
DOE/SC/Oak Ridge National Laboratory United States
|
Frontier - HPE Cray EX235a, AMD Optimized 3rd Generation EPYC 64C 2GHz, AMD Instinct MI250X, Slingshot-11, HPE Cray OS
HPE |
9,066,176 |
1,353.00 |
2,055.72 |
24,607 |
| 4 |
DOE/SC/Argonne National Laboratory United States
|
Aurora - HPE Cray EX - Intel Exascale Compute Blade, Xeon CPU Max 9470 52C 2.4GHz, Intel Data Center GPU Max, Slingshot-11
Intel |
9,264,128 |
1,012.00 |
1,980.01 |
38,698 |
| 5 |
EuroHPC/FZJ Germany
|
JUPITER Booster - BullSequana XH3000, GH Superchip 72C 3GHz, NVIDIA GH200 Superchip, Quad-Rail NVIDIA InfiniBand NDR200, RedHat Enterprise Linux
Bull |
4,801,344 |
1,000.00 |
1,226.28 |
15,794 |
| 6 |
Eni S.p.A. Italy
|
HPC7 - HPE Cray EX255a, AMD 4th Gen EPYC 24C 1.8GHz, AMD Instinct MI300A, Slingshot-11, RHEL 9
HPE |
3,461,472 |
571.50 |
861.13 |
8,735 |
| 7 |
Microsoft Azure United States
|
Eagle - Microsoft NDv5, Xeon Platinum 8480C 48C 2GHz, NVIDIA H100, NVIDIA Infiniband NDR
Microsoft Azure |
2,073,600 |
561.20 |
846.84 |
|
| 8 |
Eni S.p.A. Italy
|
HPC6 - HPE Cray EX235a, AMD Optimized 3rd Generation EPYC 64C 2GHz, AMD Instinct MI250X, Slingshot-11, RHEL 8.9
HPE |
3,143,520 |
477.90 |
606.97 |
8,461 |
| 9 |
RIKEN Center for Computational Science Japan
|
Supercomputer Fugaku - Supercomputer Fugaku, A64FX 48C 2.2GHz, Tofu interconnect D
Fujitsu |
7,630,848 |
442.01 |
537.21 |
29,899 |
| 10 |
Swiss National Supercomputing Centre (CSCS) Switzerland
|
Alps - HPE Cray EX254n, NVIDIA Grace 72C 3.1GHz, NVIDIA GH200 Superchip, Slingshot-11, HPE Cray OS
HPE |
2,121,600 |
434.90 |
574.84 |
7,124 |
-
HPC7, the No. 6 system, is installed at the Eni S.p.A.
center in Italy. It is another HPE Cray EX255a system, pairing AMD
4th Gen EPYC™ 24C 1.8GHz processors with AMD Instinct™ MI300A
accelerators over a Slingshot-11 interconnect. It achieved
571.5 Petaflop/s, making it the most powerful system in Eni's HPC fleet.
-
Eagle, the No. 7 system, is installed by Microsoft in
its Azure cloud. This Microsoft NDv5 system is based on Xeon Platinum
8480C processors and NVIDIA H100 accelerators and achieved an HPL
score of 561.2 Petaflop/s.
-
HPC6, the No. 8 system, is installed at the Eni S.p.A.
center in Ferrera Erbognone, Italy. It is an HPE Cray EX235a system
with 3rd Gen AMD EPYC™ CPUs, AMD Instinct™ MI250X accelerators,
and a Slingshot-11 interconnect. It achieved 477.9 Petaflop/s.
-
Fugaku, the No. 9 system, is installed at the RIKEN
Center for Computational Science (R-CCS) in Kobe, Japan. It has
7,630,848 cores, which allowed it to achieve an HPL benchmark score
of 442 Petaflop/s. It is now the third fastest system on the HPCG
benchmark with 16 Petaflop/s, behind LineShine and El Capitan.
-
Alps at the Swiss National Supercomputing Centre (CSCS)
in Switzerland, is at No. 10. It is an HPE Cray EX254n system with
NVIDIA Grace 72C and NVIDIA GH200 Superchip and a Slingshot-11
interconnect. It achieved 434.9 Petaflop/s.
Highlights from the List
-
A total of 294 systems on the list are using accelerator/co-processor technology,
up from 276 six months ago.
110 of these use NVIDIA Hopper
chips, 61 use NVIDIA Ampere, and
34 systems with AMD Instinct.
-
Intel continues to provide the processors for the largest share (50.20 percent)
of TOP500 systems,
down from 53.20 % six months ago.
199 (39.80 %) of the systems in the current list used
AMD processors,
up from 38.40 % six months ago.
-
The entry level to the list moved up to the
2.87 Pflop/s mark on the Linpack
benchmark.
-
The last system on the newest list was listed at position 473 in the
previous TOP500.
-
Total combined performance of all 500 exceeded the Exaflop barrier with
now 19.70 exaflop/s (Eflop/s) up from
18.73 exaflop/s (Eflop/s) 6 months ago.
-
The entry point for the TOP100 increased to
23.24 Pflop/s.
-
The average concurrency level in the TOP500 is 445,578 cores
per system up from 305,404 six months ago.
General Trends
Installations by countries/regions:
HPC manufacturer:
Interconnect Technologies:
Processor Technologies:
Green500
In the Green500 the systems of the TOP500 are ranked by how much computational
performance they deliver on the HPL benchmark per Watt of electrical power
consumed. This electrical power efficiency is measured in Gigaflops/Watt.
This ranking is not driven by the size of a system but by its technology,
and the ranking order looks therefore very different from the TOP500.
The computational efficiency of a system tends to slightly decrease with
system size, which among technologically identical systems gives smaller
systems the advantage.
This edition of the Green500 sees three systems with identical architecture
in the top 3 positions. They are all BullSequana XH3000 systems using the
Grace Hopper Superchip 72C 3GHz, NVIDIA GH200 Superchip, and Quad-Rail
NVIDIA InfiniBand NDR200. All three systems debuted on the Green500 list
twelve months ago, and the order among them is unchanged since.
Additionally the TOP10 system see no change at all since June 2026
speaking to the stagnation of the HPC segment of the high-end market.
-
KAIROS, a BullSequana XH3000 system at the CALMIP /
University of Toulouse – CNRS center, retains the No. 1 spot on the
Green500 with an energy efficiency of 73.28 Gigaflops/Watt. It achieved
3.046 Petaflop/s on the HPL benchmark. The system uses the Grace Hopper
Superchip 72C 3GHz, NVIDIA GH200 Superchip, and Quad-Rail NVIDIA
InfiniBand NDR200.
-
ROMEO-2025 holds the No. 2 spot, as it did last edition.
The system is installed at the ROMEO HPC Center - Champagne-Ardenne in
France and has 47,328 total cores. With an HPL benchmark of
9.863 Petaflop/s, it achieved an efficiency of 70.91 Gigaflops/Watt.
The architecture is identical to the No. 1 system KAIROS; ROMEO-2025
is substantially larger, which results in its energy efficiency being
slightly lower.
-
Levante GPU extension at DKRZ - Deutsches
Klimarechenzentrum in Germany again holds the No. 3 spot. It also has
an identical architecture to the No. 1 and No. 2 systems and achieved
6.747 Petaflop/s HPL performance and an efficiency of
69.43 Gigaflops/Watt.
Here are the top 10 of the Green500 ranking:
| Rank |
TOP500 Rank |
System |
Cores |
Rmax (PFlop/s) |
Power (kW) |
Energy Efficiency (GFlops/watts) |
| 1 |
472 |
KAIROS - BullSequana XH3000, GH Superchip 72C 3GHz, NVIDIA GH200 Superchip, Quad-Rail NVIDIA InfiniBand NDR200, RedHat Enterprise Linux
, Bull
CALMIP / University of Toulouse - CNRS France
|
13,056 |
3.05 |
46 |
73.282 |
| 2 |
205 |
ROMEO-2025 - BullSequana XH3000, Grace Hopper Superchip 72C 3GHz, NVIDIA GH200 Superchip, Quad-Rail NVIDIA InfiniBand NDR200, Red Hat Enterprise Linux
, Bull
ROMEO HPC Center - Champagne-Ardenne France
|
47,328 |
9.86 |
160 |
70.912 |
| 3 |
267 |
Levante GPU extension - BullSequana XH3000, GH Superchip 72C 3GHz, NVIDIA GH200 Superchip, Quad-Rail NVIDIA InfiniBand NDR200, RedHat Enterprise Linux
, Bull
DKRZ - Deutsches Klimarechenzentrum Germany
|
35,904 |
6.75 |
110 |
69.426 |
| 4 |
255 |
Isambard-AI phase 1 - HPE Cray EX254n, NVIDIA Grace 72C 3.1GHz, NVIDIA GH200 Superchip, Slingshot-11
, HPE
University of Bristol United Kingdom
|
34,272 |
7.42 |
117 |
68.835 |
| 5 |
335 |
Otus (GPU only) - ThinkSystem SD665-N V3, AMD EPYC 9655 96C 2.6GHz, NVIDIA H100 SXM5 80GB, Infiniband NDR, Rocky Linux 9.4
, Lenovo
Universitaet Paderborn - PC2 Germany
|
19,440 |
4.66 |
|
68.177 |
| 6 |
94 |
Capella - Lenovo ThinkSystem SD665-N V3, AMD EPYC 9334 32C 2.7GHz, Nvidia H100 SXM5 94Gb, Infiniband NDR200, AlmaLinux 9.4
, MEGWARE
TU Dresden, ZIH Germany
|
85,248 |
24.06 |
445 |
68.053 |
| 7 |
388 |
SSC-24 Energy Module - HPE Cray XD670, Xeon Gold 6430 32C 2.1GHz, NVIDIA H100 SXM5 80GB, Infiniband NDR400, RHEL 9.2
, HPE
Samsung Electronics South Korea
|
11,200 |
3.82 |
69 |
67.251 |
| 8 |
54 |
Helios GPU - HPE Cray EX254n, NVIDIA Grace 72C 3.1GHz, NVIDIA GH200 Superchip, Slingshot-11, Linux
, HPE
Cyfronet Poland
|
269,280 |
53.57 |
239 |
66.948 |
| 9 |
478 |
AMD Ouranos - BullSequana XH3000, AMD 4th Gen EPYC 24C 1.8GHz, AMD Instinct MI300A, Infiniband NDR200, RedHat Enterprise Linux
, Bull
Atos France
|
16,632 |
2.99 |
48 |
66.464 |
| 10 |
91 |
Portage - HPE Cray EX255a, AMD 4th Gen EPYC 24C 1.8GHz, AMD Instinct MI300A, Slingshot-11, RHEL 8.9
, HPE
Hewlett Packard Enterprise United States
|
129,024 |
24.50 |
370 |
66.277 |
HPCG Results
The TOP500 list now includes the High-Performance Conjugate Gradient (HPCG)
Benchmark results.
-
LineShine remains the leader on the HPCG ranking with
22.00 HPCG-Petaflop/s.
-
El Capitan is in second position on the HPCG benchmark
with 17.41 HPCG-Petaflop/s.
-
Supercomputer Fugaku is in third position with
16.00 HPCG-Petaflop/s.
-
Frontier at ORNL is in fourth position with
14.05 HPCG-Petaflop/s.
-
The new HANGANG system takes the eighth position
with 4.29 HPCG-Petaflop/s.
-
The JUPITER Booster system has still not submitted
an HPCG result.
| Rank |
TOP500 Rank |
System |
Cores |
Rmax (PFlop/s) |
HPCG (PFlop/s) |
| 1 |
1 |
LineShine - LingKun, LX2 304C 1.55GHz, LingQi, Kylin OS
,
National Supercomputing Centre in Shenzhen (NSCS) China
|
13,789,440 |
2,198.40 |
22.00 |
| 2 |
2 |
El Capitan - HPE Cray EX255a, AMD 4th Gen EPYC 24C 1.8GHz, AMD Instinct MI300A, Slingshot-11, TOSS
,
DOE/NNSA/LLNL United States
|
11,340,000 |
1,809.00 |
17.41 |
| 3 |
9 |
Supercomputer Fugaku - Supercomputer Fugaku, A64FX 48C 2.2GHz, Tofu interconnect D
,
RIKEN Center for Computational Science Japan
|
7,630,848 |
442.01 |
16.00 |
| 4 |
3 |
Frontier - HPE Cray EX235a, AMD Optimized 3rd Generation EPYC 64C 2GHz, AMD Instinct MI250X, Slingshot-11, HPE Cray OS
,
DOE/SC/Oak Ridge National Laboratory United States
|
9,066,176 |
1,353.00 |
14.05 |
| 5 |
6 |
HPC7 - HPE Cray EX255a, AMD 4th Gen EPYC 24C 1.8GHz, AMD Instinct MI300A, Slingshot-11, RHEL 9
,
Eni S.p.A. Italy
|
3,461,472 |
571.50 |
5.95 |
| 6 |
4 |
Aurora - HPE Cray EX - Intel Exascale Compute Blade, Xeon CPU Max 9470 52C 2.4GHz, Intel Data Center GPU Max, Slingshot-11
,
DOE/SC/Argonne National Laboratory United States
|
9,264,128 |
1,012.00 |
5.61 |
| 7 |
11 |
LUMI - HPE Cray EX235a, AMD Optimized 3rd Generation EPYC 64C 2GHz, AMD Instinct MI250X, Slingshot-11
,
EuroHPC/CSC Finland
|
2,752,704 |
379.70 |
4.59 |
| 8 |
12 |
HANGANG - HPE Cray EX254n, NVIDIA Grace 72C 3.1GHz, NVIDIA GH200 Superchip, Slingshot 400, HPE Cray OS
,
KISTI Supercomputing Center South Korea
|
1,700,544 |
342.60 |
4.29 |
| 9 |
23 |
CHIE-4 - NVIDIA DGX B200, Xeon Platinum 8570 56C 2.1GHz, NVIDIA B200 SXM 180GB, Infiniband NDR400, Ubuntu 24.04.2 LTS
,
SoftBank Corp. Japan
|
662,256 |
135.40 |
3.76 |
| 10 |
10 |
Alps - HPE Cray EX254n, NVIDIA Grace 72C 3.1GHz, NVIDIA GH200 Superchip, Slingshot-11, HPE Cray OS
,
Swiss National Supercomputing Centre (CSCS) Switzerland
|
2,121,600 |
434.90 |
3.67 |
HPL-MxP Results
On the HPL-MxP benchmark, which measures performance for mixed-precision
calculations, the El Capitan system achieved 16.7 Exaflop/s and remains
the No. 1 system.
The HPL-MxP benchmark highlights the use of mixed-precision computations.
Traditional HPC uses 64-bit floating-point computations. Today, we see
hardware with various levels of floating-point precision, 32-bit, 16-bit,
and even 8-bit. The HPL-MxP benchmark demonstrates that by using
mixed-precision during the computation, much higher performance is possible
(see the Top 5 from the HPL-MxP benchmark), and using mathematical
techniques, the same accuracy can be computed with the mixed-precision
technique when compared with straight 64-bit precision.
-
This year’s winner of the HPL-MxP category is the
El Capitan system with 16.7 Exaflop/s.
-
Aurora is in second place with an 11.6 Exaflop/s
score on the HPL-MxP benchmark.
-
Frontier remains in third place with a score
of 11.4 Exaflop/s.
-
LineShine is at fourth place with 7.92 Exaflop/s.
-
The Softbank system achieved an impressive
24.4 speedup over HPL on the HPL-MxP benchmark.
| HPL-MxP Rank |
Site |
Computer |
Cores |
HPL-MxP (Eflop/s) |
TOP500 Rank |
HPL Rmax (Eflop/s) |
Speedup of HPL-MxP over HPL |
| 1 |
DOE/SC/LLNL, USA |
El Capitan, HPE Cray 255a |
11,039,616 |
16.7 |
2 |
1.809 |
9.2 |
| 2 |
DOE/SC/ANL, USA |
Aurora, Intel/HPE Cray EX |
8,159,232 |
11.6 |
4 |
1.012 |
11.5 |
| 3 |
DOE/SC/ORNL, USA |
Frontier, HPE Cray EX235a |
8,699,904 |
11.4 |
3 |
1.353 |
8.4 |
| 4 |
NSCC Shenzhen |
LineShine |
13,698,720 |
7.92 |
1 |
2.198 |
3.6 |
| 5 |
EuroHPC/FZJ, Germany |
JUPITER Booster, BullSequana XH3000 |
4,801,344 |
6.25 |
5 |
1.0 |
6.3 |
| 6 |
Softbank, Japan |
CHIE-4, NVIDIA DGX B200 |
662,256 |
3.3 |
17 |
0.135 |
24.4 |
About the TOP500 List
The first version of what became today’s TOP500 list started as an exercise for a small conference in
Germany in June 1993. Out of curiosity, the authors decided to revisit the list in November 1993 to see how
things had changed. About that time they realized they might be onto something and decided to continue compiling
the list, which is now a much-anticipated, much-watched and much-debated twice-yearly event.