4 ms·
That sounds great... until you get a defect on your networking block. Then what? You've got a heterogenous network with tons of "this core to this core is not
by ethbr1 3y ago
That sounds great... until you get a defect on your networking block.
Then what? You've got a heterogenous network with tons of "this core to this core is not like the others" exceptions (latency, bandwidth, etc).
I know chip-to-chip/memory interconnects burn a ton of power, but fabbing discrete "biggest chip we can get with decent yield" still seems a solid tradeoff in the reality of < 100% yields.
Does anyone have a link or search phrases on how this is currently handled for high-chiplet counts? E.g. interconnection routing architectures that are still reasonable with random manufacturing-time failing links
- arcticbull 3y agoI assume it's probably an 1800 core wafer and they just cut fuse off the 200 defective cores. Some redundancy is likely just built-in based on the expectations of process reliability. Probably multiple networking blocks, too, and you'd use less demanding process features on the things that can't be duplicated. In fact you could probably even have FPGA-style soft programmable fabric interconnects to work around process failures.
- ethbr1 3y agoThat's great if it's just hitting compute cores. But when your wafer networking flows through cores (because it's cores-all-the-way-down), a defective core starts to impact network performance. Which cascades into cache, memory, locality, etc. Which starts to make a very unpredictable hardware system for software to reason about. See also dragontamer's comment down below.
- fspeech 3y agoNot all transistors have to yield the same. You can use more forgiving design rules to ensure that critical network elements don't fail. There is also post fabrication repair. It all comes down to economics. There is no fundamental problem engineering wise.
- adgjlsfhk1 3y agoalso the networking is a small fraction of the area so you naturally expect it to have fewer defects
- nerpderp82 3y agoCerebras already solved this problem. So we have that existence proof. The redundancy overhead in v1 was 1-1.5%, and reportedly half that for v2 WSE. https://www.youtube.com/watch?v=8i1_Ru5siXc https://www.youtube.com/watch?v=8i1_Ru5siXc
- gregmac 3y agoThis is on a 22nm process, which is what Intel Haswell [1] (4th gen core) was using 10 years ago. The latest gen chips are now 7 and 5nm [2], and it seems a lot of the innovation in chip manufacturing is about shrinking this size. How much is being done to improve yields of these older process sizes, maybe using the improvements done for smaller sizes? Logically it must be possible to have 100% yield on wafers at a certain process size -- but what size is that? [1] https://en.wikipedia.org/wiki/Haswell_(microarchitecture) https://en.wikipedia.org/wiki/Haswell_(microarchitecture) [2] https://en.wikipedia.org/wiki/Microprocessor_chronology#2020s https://en.wikipedia.org/wiki/Microprocessor_chronology#2020...
- SnazzyJeff 3y ago[dead]
- Dylan16807 3y agoCore to core latency is already very heterogeneous on existing CPUs. Fusing off a few links isn't the end of the world.
- ethbr1 3y agoHeterogeneous but predictable from distance, no? Routing chip-to-chip in a 2x2 or 3x2 with a fused link is less complicated than around a 1,600-core layout with multiple fused links. See nerpderp82's link above.
- Dylan16807 3y agoPredictable in theory, but not much software actually tries to predict.