Tachyum^® today announced details about how its TDIMM™ technology is bringing the future of AI and computing, enabling AI models with parameters many orders of magnitude greater than those of any existing solution at a fraction of the cost. TDIMM is key to reducing the estimated cost of OpenAI data center $3 trillion and 250,000 megawatts of power to $27 billion and 540 megawatts.

Prodigy Ultimate provides up to 21x higher AI rack performance than NVIDIA Rubin Ultra NVL576. Tachyum DDR5 DIMM (TDIMM) bandwidth increase is 5.5 times, from 51 GB/s to 281 GB/s. Prodigy Ultimate’s 24 channels provide 6.7 TB/s, 11x increase over 12-channel CPUs. The TDIMM supports 256 GB in standard, 512 GB in tall, and 1 TB in extra tall height, and TSV increases capacity up to 8 times to 3 PB per Prodigy node.

Existing DDR5 RDIMM provides 64-bit data, 16-bit ECC, 40 DQS, 14 C/A, and 16 control signals for a total of 146 signals in 288-pin connector. Tachyum TDIMM provides 128-bit data, 16-bit ECC, 36 DQS, 14 C/A, and 16 control signals for a total of 206 signals in 484-pin connector, only 38% signals increase for doubling the memory bandwidth. From a cost perspective, TDIMM needs 10% fewer DRAMs and is expected to be 10% lower cost.

The physical dimensions of connectors and DIMMs are the same as for DDR5 RDIMM, and for tall DDR5 MRDIMM. The extra tall TDIMM further doubles capacity when the server enclosure is 3U instead of 2U, which is the typical server height today. The contact pins have 0.5 mm pitch, the same as DDR5 SODIMM, so no new contacts need to be developed. Only the plastic mold needs to be modified, which is a low-intensity engineering task.

The DDR5 x4 RDIMM activates 20 DRAM chips and TDIMM only 18 DRAM chips per access, saving power. The DDR5 RDIMM switches 124 pins (64+16+40+2+2) at a high rate and TDIMM switches 184 pins (128+16+36+2+2). The TDIMM power consumption is expected to be 30% higher for 2x bandwidth. Using newer DRAM chips will put TDIMM power consumption at about the same level as older DDR5 RDIMM.

The DDR5 TDIMM doubles DDR5 RDIMM or MRDIMM bandwidth and capacity without waiting years for expensive DDR6 memory. The TDIMM quadruples bandwidth in comparison to DDR5 RDIMM. The existing DDR5 controller and PHY IP changes are simple by increasing the data path from 80 bits to 144 bits. The command/address and data buffers chips do not need to change, making adoption less than a year from now.

Minor changes to DDR6 controller, PHY, and MRDIMM chips will double bandwidth from 6.7 TB/s to 13.5 TB/s in 2027, exceeding Nvidia Rubin’s 13 TB/s. The TAI reduces bandwidth up to 4x, making TAI inference like with 54 TB/s of bandwidth. Evolutionary changes in 2028 would double TDIMM based AI chips bandwidth to 27 TB/s.

Making TDIMM open source and royalty-free to companies and the whole world, including JEDEC, will drive mainstream volume adoption and low cost. China was late with DDR5, but they can leapfrog to performance and density as DDR6 next year, making low-cost volume manufactured TDIMM available before Prodigy launch. Interested parties should contact Tachyum now to get access to TDIMM technology and agreements.

“The TDIMM is key in reducing the cost of AI systems trained on all the knowledge from $8 trillion and 276 gigawatts to $78 billion and 1 gigawatt in 2028,” said Dr. Radoslav Danilak, founder and CEO of Tachyum. “The TDIMM ushers in the era of affordable AI trained on all written knowledge produced by humanity, accessible to many companies and nations.”

Tachyum believes that its technology should be democratized and accessible to serve all mankind. Opening of TDIMM technology will be followed this year by making the Instruction Set Architecture (ISA) open and releasing software to open source. In 2023, Tachyum announced that TAI TPU™ technology is available for licensing to ensure wide proliferation of Tachyum technology outside of data centers to edge and IoT devices.

The Prodigy Universal Processor delivers orders of magnitude higher AI performance, 3x the performance of the best x86 processors, and 6x HPC performance of the fastest GPGPU. Eliminating the need for expensive dedicated AI hardware and dramatically increasing server utilization, Prodigy reduces data center CAPEX and OPEX significantly while delivering unprecedented performance, power and economics.

Those interested in reading the technical details for TDIMM architecture can download the Tachyum TDIMM Module Product Brief at https://www.tachyum.com/products/tdimm/datasheets/tdimm-module-product-brief.pdf.

Follow Tachyum

https://x.com/tachyum

https://www.linkedin.com/company/tachyum

https://www.facebook.com/Tachyum/

About Tachyum

Tachyum is transforming the economics of AI, HPC, public and private cloud with the world’s first Universal Processor Prodigy unifying the functionality of a CPU, an HPC GPGPU, and AI accelerators to deliver industry-leading performance, cost and power efficiency. Tachyum has offices in the United States, Slovakia, Taiwan and the Czech Republic. For more information, visit https://www.tachyum.com/.

View source version on businesswire.com: https://www.businesswire.com/news/home/20251125487955/en/

Tachyum Open Sources 281GB/s TDIMM™ for the Future of AI and Computing