>>>>> "Boaz" == Boaz Harrosh <boaz@xxxxxxxxxxxxx> writes: Boaz> Does it make any sense to you? I mean with multy-queue with HW Boaz> channels for every CPU, parallel IO, and poling because even Boaz> interrupts are too slow, you want me to cram everything and Boaz> serialize it on 4 open/close streams? Boaz> Either I'm completely missing something. Or please please explain Boaz> what is going on. How does 4 stream make any sense in today's NvME Boaz> HW? How does open/close anything make any sense? For number of streams you should think hardware flash channels. Boaz> On the surface it looks like someone is smoking something really Boaz> bad. No, the hardware folks have good reasons why they want things to work a certain way. I am mostly objecting to the fact that the model of the current streams proposal is heavily biased towards a fairly narrow use case and closely tied to how current generations of flash controllers are implemented. I would like to see a bit more abstraction put into the model so it is less "controller with N flash-channels and anemic ARM core" centric. That obviously means more complexity in the controller design, hence the pushback from the vendors. However, I think it'll be worth it in the long run. -- Martin K. Petersen Oracle Linux Engineering -- To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
