Tachyum’s Prodigy CPU Upgraded To 192 Cores Thanks To New EDA Instruments

Tachyum has upgraded the specs of its Prodigy HPC CPU by 50% which was made potential with the assistance of latest EDA instruments.

Tachyum’s Prodigy CPU Now Higher Than Ever (On Paper), Upgraded To 192 Cores

In its newest presser, Tachyum confirms that by using the newest EDA instruments, the corporate achieved a serious spec bump for its Prodigy CPU. The Common Processor, which was initially speculated to function 128 cores, now will get 64 further cores which is a rise of fifty% over the unique design and rounds as much as a complete of 192 cores.

The corporate states that with the brand new EDAA instruments, the variety of SERDES additionally went up from 64 to 96 per chip. This elevated the die dimension from 500mm2 to 600mm2 however that is a marginal improve of 20%. The Prodigy CPU can even home further cores to the Prodigy chip that will nonetheless match throughout the reticle restrict of 858mm2 however then the chip would turn out to be bandwidth restrict. The chip already runs a 16-channel DDR5 reminiscence interface with speeds of 7200 MT/s and past so further cores will certainly want a a lot better reminiscence commonplace.

Picture Supply: Tachyum

Perhaps on-die HBM3 reminiscence might turn out to be the subsequent logical step for future Prodigy designs. However there are additionally different enhancements which have come from the newest instruments which are not simply restricted to the variety of cores. The L2 and L3 cache has additionally been elevated from 128 MB to 192 MB and the opposite adjustments embrace:

  • 50% Improve in core rely from 128 to 192
  • Improve of the chip L2/L3 cache from 128MB to 192MB
  • Assist of DDR5 7200 reminiscence along with DDR5 6400
  • Extra velocity with 1 DIMM per channel
  • Bigger bundle accommodates further 32 serial hyperlinks and as many as 32 DIMMs linked to a single Prodigy chip

Prodigy is a Common CPU within the sense that it could actually change from conventional HPC workloads to AI/ML workloads which shall be a key space that Tachyum shall be focusing on. The corporate already claims some actually massive numbers reminiscent of Prodigy CPUs delivering 4x the efficiency of as we speak’s highest performing x86 CPUs (for cloud workloads) and 3x the efficiency of the very best performing GPUs in HPC and 6x in AI workloads. The primary chips are stated to be on observe for supply by the top of 2023 so we are going to know quickly sufficient if Tacyum can hit its efficiency objectives.

Share this story