> It's ironic OP doesn't complain about SSDs never reaching the theoretical and advertised interface speed. Which is no different from internet speeds not reaching the network interface speed as many others here call an example of being cheated.
This comparison is not exact. I haven't seen SSD manufacturers advertise transfer speeds of the interface. They say they support this interface, but the transfer numbers advertised are different. Whether those actually happen is a different story.
In the case of internet connections, they don't usually advertise the "interface speed" either. They advertise actual bandwidth[0]. Yes, nowadays, a lot of providers will advertise 1 Gbps connectivity, which coincides with the interface speeds with which people are familiar. But for example, in my case, advertised speed is 400 Mbps. Over a PON (passive optical network). My transceiver (optical interface) will support much more than that. I've never had a 400 Mbps network interface in my computers. Should I expect them to give me 1Gbps? Turns out, they usually do, but I guess that's more of an exception than the rule.
The issue is when a provider will advertise say 400 Mbps, like in my case, but in practice I'm never able to reach more than say 40 Mbps. In this case, yes, there may be some kind of law based on which this behavior is "lawful" or not.
---
[0] Which may be "wire" bandwidth, as opposed to "useful data" bandwidth, so they'll happily count encapsulation, etc, but I'd argue that's somewhat different, although still scummy.
> This comparison is not exact. I haven't seen SSD manufacturers advertise transfer speeds of the interface.
I went to the first consumer grade SSD manufacturer I could think of, Corsair, and lo and behold, the description of the SSD advertises the interface speed [0] as "up to"
1,950MB/s write and 4,700MB/s read.
I get the feeling we might understand the term interface differently. What do you take it to be? To me, the interface is PCI-Express 4.0 4x. The "1950 MB/s write and 4700 MB/s read" are the specifics of the drive's performance.
According to Wikipedia, PCIE 4.0 4x can sustain more than the spec sheet advertises [0].
The spec sheet advertises 4700 MB/s write and 1950 MB/s read. The interface can handle more than 7000 MB/s in each direction. That's more than twice the advertised bandwidth of this drive.
If what you mean is that the MP600 isn't able to handle the claimed 4700 MB/s "most of the time" or something, than that's fair (I have no idea if that's the case or not). But the "interface" doesn't really come into play, here. The drive is advertised as PCIE 4.0 4x because the 3.0 revision couldn't handle the advertised maximum read specs.
Just like it would seem strange for an internet provider to sell a 400 Mbps (advertised) connection over a 100 Mbps line.
This comparison is not exact. I haven't seen SSD manufacturers advertise transfer speeds of the interface. They say they support this interface, but the transfer numbers advertised are different. Whether those actually happen is a different story.
In the case of internet connections, they don't usually advertise the "interface speed" either. They advertise actual bandwidth[0]. Yes, nowadays, a lot of providers will advertise 1 Gbps connectivity, which coincides with the interface speeds with which people are familiar. But for example, in my case, advertised speed is 400 Mbps. Over a PON (passive optical network). My transceiver (optical interface) will support much more than that. I've never had a 400 Mbps network interface in my computers. Should I expect them to give me 1Gbps? Turns out, they usually do, but I guess that's more of an exception than the rule.
The issue is when a provider will advertise say 400 Mbps, like in my case, but in practice I'm never able to reach more than say 40 Mbps. In this case, yes, there may be some kind of law based on which this behavior is "lawful" or not.
---
[0] Which may be "wire" bandwidth, as opposed to "useful data" bandwidth, so they'll happily count encapsulation, etc, but I'd argue that's somewhat different, although still scummy.