3 ms·
> Graviton4 processors deliver up to 30% better compute performance, 50% more cores, and 75% more memory bandwidth than Graviton3. This seems ambiguous. Presum
by snewman 3y ago
> Graviton4 processors deliver up to 30% better compute performance, 50% more cores, and 75% more memory bandwidth than Graviton3.
This seems ambiguous. Presumably this is 50% more cores per chip. What about "30% better compute performance" and "75% more memory bandwidth": is that per core, or per chip? If the latter, then per-core compute performance would actually be lower.
Also, "up to" could be hiding almost anything. Has anyone seen a source with clearer information as to how per-core application performance compares to earlier Graviton generations?
- p1esk 3y agoWait, how could "50% more cores, and 75% more memory bandwidth" result in anything less than 50% of better compute performance?
- jagger27 3y agoClock speed
- snewman 3y agoBecause AWS doesn't rent you chips, it rents you cores. If a chip has 50% more cores, that doesn't help you as a renter of cores, just means they get to rent out more cores. Of course it might give Amazon room to offer these instances at a lower hourly rate per core, which would ultimately cash out as improved cost/performance for AWS customers.
- bluedino 3y agoIt's the best Graviton processor yet!
- kevincox 3y agoI would assume that "up to" means that for all of the workloads that they benchmarked the best result was 30% better compute performance. Not a very useful number as your workload is very unlikely to hit the right set of requirements to see that uplift.
- otterley 3y agoAWS Graviton specialist here! The performance improvement is on a per-core basis. The pending availability of 96-vCPU Graviton4 instances is icing on the cake!
- Klasiaster 3y agoDo any of the (soon-to-be) available non-metal offers support nested virt?
- stingraycharles 3y agoI read it as: each core is 30% faster, each chip has 50% more cores, and each chip has 75% more memory bandwidth. But you’re right it is ambiguous.