bochica :
I'm sure once more and more of bootable PCI-E SSDs become available, everyone will jump ship off the current SSDs. The problem is that some of these PCI-E variants are pretty expensive. Expect to buy your computer build two or three(+) times over to get one of these.
I really doubt everyone will jump ship. The problem is that MB/sec is the inverse of how humans perceive speed. We perceive things in terms of sec/MB. If something takes a fraction of a second, it's fast. If something takes several seconds, it's slow.
How does that matter? When you invert the metric to sec/MB, the vast majority of your speed gain comes from upgrading from a HDD to a regular SSD. The speed gain from switching to a 3 GB/sec PCIe SSD is marginal. e.g. Say the drive needs to read 1 GB of data to boot.
HDD (100 MB/sec) = 10 seconds
SSD (500 MB/sec) = 2 seconds (8 sec improvement from HDD)
PCIe (3000 MB/sec) = 0.3 seconds (9.7 sec improvement from HDD, 1.7 second improvement from SSD)
So the 400 MB/sec increase going from the HDD to the SSD gives an 8 sec improvement. But the 2500 MB/sec increase going from the SSD to PCIe SSD only gives a 1.7 sec improvement. That's what happens when you invert the metric from human perception - the bigger the numbers get, the smaller the perceived improvement. The bigger the number, the less it matters. Both Toms Hardware and Anandtech have published articles pretty much admitting this. That the only reason they post benchmarks in MB/sec is because if they do it in sec/MB, all the SSDs look pretty much the same. They're all very fast and there's very little difference between them. They use MB/sec to exaggerate the difference and keep readers interested and coming back to their reviews.
So if you look at the improvement range from HDD to PCIe SSD, about 83% of the improvement can be gained by buying a SSD. Going to a PCIe SSD only gets you an extra 17%. So unless PCIe drives drop down in price to where they're at most about 15%-20% more expensive than a SATA SSD, they're not going to really gain marketshare. They're going to be relegated to niche applications where you need super-fast access to large amounts of data (e.g. real-time video editing).
It'll happen eventually because of the inevitable march of technology (the SSDs will have 16- or 32- banks internally instead of 8 so cost shouldn't really be a factor). But there won't really by a compelling reason for most people to ditch their current SSDs for the new ones because the improvement in perceived speed is marginal. Incidentally, RAM disks score about 6000-10000 MB/sec on these disk benchmarks, so we are getting close to topping out drive speed. It's already hard for me to tell when the computer is swapping on a 500 MB/s SATA 3 SSD.
Facebook
Twitter
Instagram