I looked through some zfs blogs and mail lists for information about nvdimms and learned that the slog device doesn't cache data, but the information about the "transactions" so that in case of a powerfailure the changes can be undone.
With other words the performance (iops, latency & throughput) of the slog does not necessarily improve the performance of the storage pool.
If you're not clear what a SLOG device does then research it, it's been explained numerous times on this forum and
@gea has done a great write-up in various places on it as well. I never claimed it was a "cache device" and to say a high-performance SLOG device does not improve pool write performance is 100% wrong. Additionally it's been shown that using an identical drive for a SLOG device as the POOL itself uses will still yield > performance because it's not writing the log to the pool and the data.
I'm not debating or getting into this especially not in this thread. I suggest you research then make your own thread where confusion can be clarified for you as this has been rehashed over and over. It's not a cache, it improves performance, it needs tweaked\adjusted just like pools do, etc...
If you're not doing sync writes then that's different. But when someone says they were going to use an OPtane I don't immediately think they have 0 understanding of the benefits of a SLOG, I assume they understand what it is, why they need it, and are looking at options. As I mentioned, we need more info to provide more suggestions