Supercomputing in India

[NutCracker]

Based on this, we should be able to leapfrog our way into no. 6 position and hopefully break into the top 3 s India's economy get bigger.

That list doesn't show full picture.

Real relevant stat would be how many of ours 38 are in top 100 or 200, And same data for other countries.

 

[Chinmoy]

That list doesn't show full picture.

Real relevant stat would be how many of ours 38 are in top 100 or 200, And same data for other countries.

Airawat . Position 75
Param Siddhi. Position 131
Pratyush. Position 169
Mihir. Position 316

 

[Azaad]

Avalon Technologies enters into agreement with C-DAC for indigenously manufactured HPC servers ‘RUDRA’ - Industry News | The Financial Express - https://www.financialexpress.com/bu...ously-manufactured-hpc-servers-rudra-3362865/

Perhaps this should be converted into a sticky @ezsasa ; @Indx TechStyle

 

[Indx TechStyle]

Avalon Technologies enters into agreement with C-DAC for indigenously manufactured HPC servers ‘RUDRA’ - Industry News | The Financial Express - https://www.financialexpress.com/bu...ously-manufactured-hpc-servers-rudra-3362865/

Perhaps this should be converted into a sticky @ezsasa ; @Indx TechStyle

Already a sticky thread

 

[Azaad]

 

[Indx TechStyle]

India plans 10,000-GPU sovereign AI supercomputer
Puts $1.2 billion on the table for AI skills and local LLMs, tells private enterprise it expects help

Simon Sharwood
Fri 8 Mar 2024 // 02:30 UTC

India's government has approved a ₹10,300 Crore ($1.24 billion) funding package to bolster the nation's AI infrastructure.

The cornerstone of the effort is a planned supercomputer that will have at least 10,000 GPUs. The government has released no other details of the machine – which will be part of the "IndiaAI Compute Capacity" – but has said it expects a public-private-partnership will be needed to build the machine.

Another initiative will see the creation of a new academic institution: the "IndiaAI Innovation Centre," tasked with leading the development and deployment of foundational models. It is expected to have a specific emphasis on indigenous Large Multimodal Models (LMMs) and domain-specific models. The Centre will focus on "leveraging edge and distributed computing for optimal efficiency."

Funds will also flow to three other initiatives:

• The IndiaAI Startup Financing mechanism, which will streamline funding to accelerate commercialization for both startups and industry-led AI projects;
• The IndiaAI Datasets Platform, which will get more cash to improve public sector datasets so that local AI outfits – and government – have the data needed to build appropriate AI apps;
• The IndiaAI FutureSkills program, which will improve access to graduate and postgraduate AI programs, and establish Data and AI Labs that that run foundational AI courses in data and AI across India – especially beyond major cities;

Two goals of the funding package are to "foster technological self-reliance" and "democratize the benefits of AI across all strata of society."

• India demands beta AIs secure government permission before going public
• Indian tech minister vows to stop Google removing local apps from Play Store
• India approves its first full wafer fab – a 28nm affair from Tata and Powerchip
• EU sanctions Indian tech outfit that has partnered with New Delhi's IT Ministry

It’s unclear if the planned supercomputer will meet those goals by using homegrown tech. While India has set itself a goal of developing server-grade CPUs based on the RISC-V architecture, The Register is yet to see any evidence of such devices having been developed. And India is nowhere on GPUs.
The push for indigenous LLMs will deliver, though, because India has 22 scheduled languages that the nation is required to foster by law. While some of those languages – like Bengali, Marathi, and Telugu – have over 80 million speakers, others are spoken by many fewer. The giants of AI may not prioritize LLM development for the ~35 million speakers of Malayalam or Punjabi.

India clearly intends to take on that sort of job itself.

Another omission in India's announcement is the kind of private partners sought to accelerate local AI development. India has a difficult relationship with Big Tech – lauding its local investments while also regulating it fiercely, and creating public digital goods unashamedly aimed at making it harder for tech companies to build monopolies.

 

[mokoman]

India plans 10,000-GPU sovereign AI supercomputer
Puts $1.2 billion on the table for AI skills and local LLMs, tells private enterprise it expects help

Simon Sharwood
Fri 8 Mar 2024 // 02:30 UTC

Funds will also flow to three other initiatives:

• The IndiaAI Startup Financing mechanism, which will streamline funding to accelerate commercialization for both startups and industry-led AI projects;
• The IndiaAI Datasets Platform, which will get more cash to improve public sector datasets so that local AI outfits – and government – have the data needed to build appropriate AI apps;
• The IndiaAI FutureSkills program, which will improve access to graduate and postgraduate AI programs, and establish Data and AI Labs that that run foundational AI courses in data and AI across India – especially beyond major cities;

• India demands beta AIs secure government permission before going public
• Indian tech minister vows to stop Google removing local apps from Play Store
• India approves its first full wafer fab – a 28nm affair from Tata and Powerchip
• EU sanctions Indian tech outfit that has partnered with New Delhi's IT Ministry

maybe nvidia h100 , an indian company brought 16,000 units from them .

it took openai 10,000 nvidia gpus to train chatgtp

 

[FalconSlayers]

maybe nvidia h100 , an indian company brought 16,000 units from them .

it took openai 10,000 nvidia gpus to train chatgtp

Yes, Yotta bought them from Nvidia.

https://www.reuters.com/technology/india-data-centre-firm-yottas-nvidia-ai-chip-orders-reach-1-bln-ceo-2024-01-11/

They've purchased more than 32,700 Nvidia H100 and GH200 chips in total so far for more than $1 billion to be deployed at their Navi Mumbai and Greater Noida data centres.