PARAM
India's home-grown supercomputer line, built by C-DAC in Pune since the early 1990s
PARAM is a family of supercomputers designed and put together in India by C-DAC, a government body based in Pune. The line began in the late 1980s after foreign suppliers refused to sell India such machines, and it grew into dozens of systems used at home and abroad.
- First PARAM shipped in August 1991 with 64 nodes of Inmos transputers
- PARAM 8000 matched 28 times the power of the sought Cray X-MP at the same $10m cost
- A 256-node PARAM 8000 was rated at 1 GFLOPS theoretical, 100-200 MFLOPS sustained
- 52 PARAM systems deployed by 2008, eight of them outside India
- PARAMNet-3 used C-DAC's GEMINI co-processor and ANVAY 48-port switches
What it is
PARAM is a line of parallel supercomputers built in India by the Centre for Development of Advanced Computing, an organisation headquartered in Pune. The name is a Sanskrit word meaning supreme and doubles as an acronym for PARAllel Machine. Systems in the family range from the early PARAM 8000 through the 8600, 9000 and 10000 to later one-off machines, including units built under the National Supercomputing Mission from the late 2010s. By 2008, 52 of these systems had been installed, with eight of them outside India. They serve both government and commercial users.
How it came about
C-DAC came into being in November 1987, first called the Centre for Development of Advanced Computing Technology. The trigger was difficulty buying supercomputing hardware from abroad, which pushed the Indian government towards building its own technology. The opening machine, PARAM 8000, was constructed from the ground up and shown at a 1990 event in Zurich, where it ranked behind only one American system. A 64-node version shipped in August 1991. Successive models followed roughly every few years, and manufacturing of hardware was later handed to private partners.
The compute
PARAM 8000 ran on Inmos T800 and T805 transputers in a distributed-memory MIMD layout with a rewiring interconnect; a 256-node setup was rated at 1 GFLOPS on paper and 100 to 200 MFLOPS in real use. The 8600 paired an Intel i860 with four transputers per node and kept the PARAS programming environment, so code moved between the two. PARAM 9000 mixed cluster and massively parallel work, scaling to 200 CPUs over a clos network, with SuperSPARC II, UltraSPARC and DEC Alpha variants. PARAM 10000 used Sun Enterprise 250 nodes carrying dual 400 MHz UltraSPARC II chips. Interconnects progressed through PARAMNet, PARAMNet II at 2.5 Gbit/s, and PARAMNet-3 with GEMINI co-processors, ANVAY switches and the KSHIPRA protocol stack.
Where the GPUs are
The machines are designed and assembled in Pune. Systems have been placed with operators in India and shipped to Russia, Singapore, Germany and Canada, among other destinations. Later hardware production involved outside firms: Netweb Technologies handled manufacturing and design work for PARAM Yuva II and AIRAWAT, while VVDN Technologies, Avalon Technologies and Kaynes Technology worked with C-DAC on PARAM Rudra.
Customers and deals
Users span public bodies and private firms. Early sales reached Germany, the United Kingdom and Russia, with a unit going to ICAD Moscow in 1991 under a Russian partnership. A price of $350,000 drew 14 further buyers and helped the line take the domestic market. Later installations went to countries including Tanzania, Armenia, Saudi Arabia, Ghana, Myanmar, Nepal, Kazakhstan, Uzbekistan and Vietnam.
Money and ownership
The Indian government originally sought a Cray X-MP at a quoted $10 million. PARAM 8000 delivered about 28 times that machine's capability for the same sum, and later sold at $350,000 a unit. No funding rounds, valuations or revenue figures are recorded here.
Its place in AI
PARAM gave India a domestic route into high performance computing after import barriers blocked foreign purchases. Its first machine outperformed every system bar one at the 1990 Zurich showcase. The series fed into the National Supercomputing Mission, which produced further machines from the late 2010s, and it sits alongside other Indian efforts such as ISRO's SAGA-220.
Questions and criticism
The programme began because overseas vendors would not supply supercomputers, leaving India dependent on imports. Building capability at home required sustained government backing and, more recently, reliance on private manufacturers for hardware production.
Questions people ask about PARAM
What does PARAM mean?
It is the Sanskrit word for supreme. The name also works as an acronym, standing for PARAllel Machine, which describes the design approach of the series.
Who builds these machines?
C-DAC, a government organisation based in Pune, designs them. Hardware manufacture has been outsourced in part to firms such as Netweb Technologies, VVDN Technologies, Avalon Technologies and Kaynes Technology.
Why was the series started?
India faced obstacles buying supercomputers from foreign suppliers, so the government set up C-DAC in November 1987 to build indigenous computing technology. The first machine arrived in August 1991.
How fast was the first system?
A 256-node PARAM 8000 was rated at 1 GFLOPS in theory and sustained 100 to 200 MFLOPS. That was about 28 times the capability of the Cray X-MP the government had wanted, at the same $10 million price.
Where have PARAM machines gone?
Beyond Indian users, systems were exported to Germany, the United Kingdom, Russia and Singapore, and later placed in Canada, Tanzania, Armenia, Saudi Arabia, Ghana, Myanmar, Nepal, Kazakhstan, Uzbekistan and Vietnam.
Good reading and sources
- ‘PARAM Brahma will allow scientists to address complex scientific problems’The Indian Express, Article
- electronics: C-DAC unveils India's fastest supercomputer Param Yuva IIThe Economic Times, Article
- President Droupadi Murmu inaugurates supercomputer facility at IIT GuwahatiHindustan Times, Article
The facts on this page come from the Wikipedia article "PARAM", written by its contributors and used under the CC BY-SA 4.0 licence, and from its Wikidata record (CC0). EveryNeocloud rewrote the text in its own words, with AI assistance, and compares every page with the article to make sure no passage is copied. It may be reused under the same CC BY-SA 4.0 licence with a link back.
It may contain mistakes, and figures change over time. Nothing here is investment or buying advice. If something is wrong, tell us. Last refreshed 2026-11-14.