Carding 4 Carders
Professional
- Messages
- 2,730
- Reaction score
- 1,467
- Points
- 113
Half a million chips and power that is hard to imagine.
Tachyum said that one of its customers plans to purchase hundreds of thousands of Prodigy processors to create a powerful supercomputer with a capacity of 50 exaflops. Typically, such metrics are associated with AI calculations using 8-or 16-bit floating-point math, rather than with 64-bit calculations, which are usually used in high-performance computers. However, Tachyum claims that the system will be 25 times more powerful than "the fastest supercomputer built this year."
This is probably a reference to the recently commissioned Aurora supercomputer from Argonne National Laboratory with a capacity of over two exaflops. Tachyum also states that their future system will be able to achieve performance of up to eight zettaFLOPS for processing large language models and have hundreds of petabytes of DDR5 memory by 2025.
To better understand the scope of Tachyum's plans, it is important to pay attention to their Prodigy chip, which the company has been working on in recent years. Prodigy is presented as a universal processor that can handle a variety of tasks. To support modern standards such as x86, Arm, and RISC-V, the QEMU emulator was adapted to the Prodigy architecture.
Tachyum CEO Radoslav Danilak said that most of the 600 mm2 Prodigy crystal is dedicated to its 192 64-bit processor cores. These cores have their own instruction set architecture and will execute four out-of-order instructions on each clock cycle at frequencies above 5 GHz.
According to Danilak, a supercomputer with 50 exaflops will be absolutely gigantic. "It's a huge, huge car," he said. According to Tachyum, it will take about half a million chips to build such a machine.
It is worth noting that despite Tachyum's ambitious plans, the Prodigy chip has not yet been put into production. Nevertheless, the company remains optimistic and hopes for a successful completion of the project in the coming years.
Tachyum said that one of its customers plans to purchase hundreds of thousands of Prodigy processors to create a powerful supercomputer with a capacity of 50 exaflops. Typically, such metrics are associated with AI calculations using 8-or 16-bit floating-point math, rather than with 64-bit calculations, which are usually used in high-performance computers. However, Tachyum claims that the system will be 25 times more powerful than "the fastest supercomputer built this year."
This is probably a reference to the recently commissioned Aurora supercomputer from Argonne National Laboratory with a capacity of over two exaflops. Tachyum also states that their future system will be able to achieve performance of up to eight zettaFLOPS for processing large language models and have hundreds of petabytes of DDR5 memory by 2025.
To better understand the scope of Tachyum's plans, it is important to pay attention to their Prodigy chip, which the company has been working on in recent years. Prodigy is presented as a universal processor that can handle a variety of tasks. To support modern standards such as x86, Arm, and RISC-V, the QEMU emulator was adapted to the Prodigy architecture.
Tachyum CEO Radoslav Danilak said that most of the 600 mm2 Prodigy crystal is dedicated to its 192 64-bit processor cores. These cores have their own instruction set architecture and will execute four out-of-order instructions on each clock cycle at frequencies above 5 GHz.
According to Danilak, a supercomputer with 50 exaflops will be absolutely gigantic. "It's a huge, huge car," he said. According to Tachyum, it will take about half a million chips to build such a machine.
It is worth noting that despite Tachyum's ambitious plans, the Prodigy chip has not yet been put into production. Nevertheless, the company remains optimistic and hopes for a successful completion of the project in the coming years.
