Hi guys,
I've heard that SSD speed boost through the NVMe protocol mainly applies to sequential data reads/writes (e.g. 4K editing) rather than to random reads/writes (e.g. gaming, office apps). In other words the article claimed that for random reads/writes NVMe is not much faster than SATA.
And finally:
3) How do x16 GPU slot, M.2 #1 slot and M.2 #2 slot share lanes if both M.2 devices (SSDs) are connected to PCIe bus? As far as I've read, if both SSDs are connected to SATA - the drop in their performance is just not possible at all, but is PCIe connection trickier from this perspective? My guess is: if there are totally 20 lanes available, 16 will be reserved by the GPU (dedicated use) and if I had only 1 SSD, it will use the 4 lanes left (also dedicated use) which will bring the max. possible speed (8000Mb/s). With a 2nd SSD added - shall it be 4 lanes split into 2 (per each SSD) or will rather GPU lanes be "borrowed"?
Thank you!
I've heard that SSD speed boost through the NVMe protocol mainly applies to sequential data reads/writes (e.g. 4K editing) rather than to random reads/writes (e.g. gaming, office apps). In other words the article claimed that for random reads/writes NVMe is not much faster than SATA.
- Do you agree? If yes - is this statement true only when a SATA 3 bus (600Mb/s) is considered, while SATA 2 bus (300Mb/s) will be much slower compared to NVMe?
- Is there any difference in the results of the "NVMe vs. SATA" comparison when NVMe M.2 sits on PCIe 4.0 vs. PCIe 3.0 bus? In other words, among those - NVMe PCIe 4.0 vs. NVMe PCIe 3.0 vs. SATA 3 vs. SATA 2 - which ones are almost comparable when random reads/writes (e.g. gaming, office apps) are considered?
And finally:
3) How do x16 GPU slot, M.2 #1 slot and M.2 #2 slot share lanes if both M.2 devices (SSDs) are connected to PCIe bus? As far as I've read, if both SSDs are connected to SATA - the drop in their performance is just not possible at all, but is PCIe connection trickier from this perspective? My guess is: if there are totally 20 lanes available, 16 will be reserved by the GPU (dedicated use) and if I had only 1 SSD, it will use the 4 lanes left (also dedicated use) which will bring the max. possible speed (8000Mb/s). With a 2nd SSD added - shall it be 4 lanes split into 2 (per each SSD) or will rather GPU lanes be "borrowed"?
Thank you!