Theoretically being that the quoted speeds of the 980 Pro should be what I'm currently getting with the T700 which is a gen 2 PCIE5.0 SSD. In raw numbers the T700 is getting 7.1GB/s and the 980 Pro is getting 6.4GB/s. The upper ceiling for PCIE4.0 x4 is 8GB/s and the 980 Pro should be getting 7GB/s. The 980 Pro actually beats it in Random 4k Q32 both read and write but is beat everywhere else by the T700.
IDK what you're on about constantly quoting MLC vs TLC and the channels. The PS5 SSD has a 12 channel controller and without the built in decompression chip working it only gets 5.5GB/s raw and it's a gen 4 drive, typical of most first generation Gen 4 SSDs. Hell, my old 960 EVO is TLC and 8 channels lol. In fact the 980 Pro also has TLC and 8 channels sooo? Should probably double check your info; the Samsung website states MLC but 3 bits shows that's a typo and again MLC is better than TLC, like SLC is better than MLC. The 980 Pro got so much flack because of the downgrade to TLC from MLC due to endurance and write speeds.
Yes I would expect a newer device to be faster, 1+1 does = 2, but when I was currently on a drive that "theoretically" should have already maxed the bus out I didn't expect much. The fact that I'm not seeing this drive max the bus out when it's a 12GB/s device is why I kept saying theoretically. Neither device does.
Now if you're questioning why I was surprised, it's because I did not expect any better 4k random QD1 improvements. My 980 Pro has similar 4k random performance to my Sabrent Rocket 4+ 1TB, my SPCC 1TB Gen 4 drive (1st gen), and my Sabrent Rocket 4 1TB (also 1st gen). 4k random QD1 performance doesn't come from controllers, dram cache, or the type of NAND (outside of QLC which is dog slow). It comes from actual improvement in NAND and latency which we haven't seen since the minor bump up from SATA SSDs to NVMe as well as protocol tech (hence the bump from SATA AHCI to NVMe). I don't care much about full sequential numbers, when I said "beat the snot out of" I was referring to 4k random performance in QD1 and not in the way of write caching. The irony here being the T700 actually only has 4 chip enables but the actual specs of the NAND itself and topology cause it to have actual better real world 4k QD1 performance which isn't a cause of write caching. This can be seen in the 52% increase in 4k QD1 read speeds that can't be masked by write caching. The 8 chip enables to 2 chips at 8Tb/s is what causes the 980 Pro to have faster 4k QD32 speeds due to write combining being faster in parallel.
** While I was writing this I realized that Gen 5 drives are NVMe 2.0 which explains the big bump in 4k QD1 performance. Big improvements to latency.