[ashtonsix]

I just did a mini-ablation study for this (prefix sum). By getting rid of the cross-block carry (16 values), you can increase perf from 19.85 to 23.45 GB/s: the gain is modest as most performance is lost on accumulator carry within the block.

An absolute value every 16 deltas would undermine compression: a greater interval would lose even the modest performance gain, while a smaller interval would completely lose the compressibility benefits of delta coding.

It's a different matter, although there is definitely plausible motivation for absolute values every X deltas: query/update locality (mini-partition-level). You wouldn't want to transcode a huge number of values to access/modify a small subset.

[zamadatix]

I think what GP is saying is you can drop an absolute value every e.g. 8192 elements (or even orders of magnitude more if you're actually processing GBs of element) and this frees you to compute the blocks in parallel threads in a dependency free manor. The latency for a block would still bound by the single core rate, but the throughput of a stream is likely memory bound after 2-3 cores. It still hurts the point of doing delta compression, but not nearly as bad as every 16 values would.

Even if one is willing to adopt such an encoding scheme, you'd still want to optimize what you have here anyways though. It also doesn't help, as mentioned, if the goal is actually latency of small streams rather than throughput of large ones.

[ashtonsix]

Oh right. That's sensible enough. Makes total sense to parallelise across multiple cores.

I wouldn't expect a strictly linear speed-up due to contention on the memory bus, but it's not as bad as flat-lining after engaging 2-3 cores. On most AWS Graviton instances you should be able to pull ~5.5 GB/s per-core even with all cores active, and that becomes less of a bottleneck when considering you'll typically run a sequence of compute operations between memory round-trips (not just delta).