Nvidia Thinks about Multi-Chip GPUs for Future Designs
- Nvidia Thinks about Multi-Chip GPUs for Future Styles
Historically, GPUs have been designed as monolithic dies with all their functionality under one ‘roofing.’ This hasn’t constantly held true– the earliest GPUs sometimes utilized separate chips for specific performance. Both AMD and Nvidia have, at various times, used different cores to offer assistance for additional screens or to bridge connections in between PCI Express and AGP.As far as the core parts of the GPU itself, however, those have been single-die affairs for a long time. That’s why it’s a little surprising to see Nvidia is now evaluating the possibility of a multi-chip GPU that would interact with other parts of the core, in something like an MCM (Multi-Chip-Module).
Monolithic GPU styles suffer from a number of problems. They can be reticle-busters, pushing the limits of exactly what TSMC, GlobalFoundries, or Samsung can develop into a single core. Harvesting good GPUs from bad ones can cause problems when the manufacturer tries to collect good die, as taken place with Nvidia and the GTX 970. (Long story short– the method NV utilized to recover parts for the GTX 970 likewise had an effect on the GPU’s memory bandwidth when accessing its last 512MB of RAM.) If GPUs were built in modules, then connected together on a common package, the resulting chip could theoretically be bigger and more effective than any single card.In the authors’ research study The paper has more details on this, but the big-picture takeaway is that by providing each GPU block enough bandwidth, keeping latency low, and correctly designating cache resources, you can hit some considerable efficiency targets. The trick, of course, is keeping all those resources correctly balanced.For those who would argue that this is just the same two-GPUs-on-one-piece-of-silicon that we have actually seen from both AMD and Nvidia for several years, no, it really isn’t. Nvidia isn’t pondering sharing information across a silicon PCIe bridge; they’re discussing designing a GPU that’s developed, from the ground up, to share works throughout multiple GPU modules. Better or worse, it ‘d be profoundly various from anything we’ve seen before.More Articles China Wants to Construct Panda-Shaped Solar Grids All Over the World Jul 7 The IBM Design F Keyboard Returns After a 30-Year Hiatus Jul 6 MIT Establishes 3D Chip That Integrates CPU, Memory Jul
ET offers: Dell Inspiron 15 5000 15.6-inch laptop for $ 579 Jul 6