AMD plans to bring RAM closer to CPU cores - but don’t expect it on PCs soon
AMD CEO sheds some light on what the company plans to do reduce data transfer energy
When you purchase through links on our site, we may earn an affiliate commission.Here’s how it works.
AMDwants to reduce the cost of communication and is on a crusade to bring down the cost of transferring bits between memory and compute significantly by puttingRAMon top of theCPU/GPU.
Company CEO Dr. Lisa Su recently delivered a high level presentation at the International Solid-State Circuits Conference (ISSCC) 2023, speaking extensively about the need to cut down on the amount of energy (expressed in Joules) per computation operations (FLOPS).
Otherwise(as she puts it) thenext Zettaflop-capable supercomputer will need a nuclear power stationto keep running - and that’s not something realistic or sustainable.
Distance
Instead, the biggest improvements in performance-per-watt, Su believes, will be achieved by reducing the physical distance between the memory and where computation takes place (either on the CPU or the GPU). She used the example of theMI300 acceleratorwhich uses a next-generation AMD InstinctAPUwith unified HBM (High bandwidth memory) to deliver some significant power savings.
Concurrently, AMD has already integrated processing-in-memory to reduce the energy required to access data.
Su presentation mentioned, “Key algorithmic kernels can be executed directly in memory, saving precious communication energy” - and for that AMD is collaborating withSamsungElectronics, whose expertise in DRAM is undeniable.
Closer is better
Memory-on-chip is already mainstream: AMD packs it in itsAMD Ryzen 9 7950X3Dand before that on itsRyzen 7 5800X3D(note that this memory is the faster and more expensive SRAM rather than DRAM). HBM is present in AMD’s Instinct MI accelerators and inNvidia’s popularA100accelerator, the brains behindChatGPT. Apple’s M-series usesHBMconnected to the processor but on the package rather than on the chip die.
Are you a pro? Subscribe to our newsletter
Sign up to the TechRadar Pro newsletter to get all the top news, opinion, features and guidance your business needs to succeed!
Eventually, HPC will move towards memory-on-chip full scale as this is the most straight-forward low hanging fruit as workloads that demand extremely large amounts of high bandwidth push tackling power requirements (and associated costs) up the priority list.
Fujitsu’s A64FX processor,launched in 2019, is a true trailblazer and pioneer, merging dozens ofArmcores with 32GB of HBM2 memory sitting atop and offering a whopping 1TBps of bandwidth and with HBM3 already available onNvidia’s Hopper H100enterprise GPU, things will get even more interesting. Rambus plans to go beyond the HBM3 specs and hinted,last April, at up to 1.05TBps of bandwidth.
Increased interest in HBM, the cloud of the 1-ton gorilla that isAppleand the never-ending quest for bandwidth without needing an exotic power supply (and equally exotic cooling system) means that HBM - in the long run - is likely to supplant DIMM (and GDDR) as the main memory format: Blame it on Apple.
Dr. Su expects the first Zettascale supercomputer to be unveiled before 2035: that leaves us with 12 years to find the perfect solution unlessAIgets there first.
Désiré has been musing and writing about technology during a career spanning four decades. He dabbled inwebsite buildersandweb hostingwhen DHTML and frames were in vogue and started narrating about the impact of technology on society just before the start of the Y2K hysteria at the turn of the last millennium.
HPE reveals critical security bug affecting networking access points
Cybersecurity is business survival and CISOs need to act now
Ireland vs New Zealand live stream: how to watch 2024 rugby union Autumn International online from anywhere
