Confusing standards

[DANK_AS_gay]

NVMe drives can achieve ridiculous speeds, but often I see SATA drives not meeting anywhere near the same performance. This is despite both SSDs being slower than the 6GB/s cap of SATA3. I see SATA SSDs at 500MB/s, but NVMe drives are reaching 3 or 4 times that. So why does the additional bandwidth of NVMe help speeds? That doesn't make sense to me if the sequential read/write speeds of either drive are below 6GB/s. 

[SpookyCitrus]

They're different types of storage, one goes through sata and the other uses the PCIE lanes. Really just think of them as two completely different types of drives. If you really want to get into the nitty gritty of the differences and why the speeds are faster with NVMe than sata you should just google it and do some research. 

[BondiBlue]

5 minutes ago, DANK_AS_gay said:

6GB/s cap of SATA3

SATA III is 6 gigabits per second (Gb/s (lowercase b)), not 6 gigabytes per second (GB/s (uppercase B)). With overhead most decent SATA SSDs are close to the theoretical limitations of SATA III. 

 

Just a note: 6 GB/s is 48 Gb/s. That's many times faster than current NVMe drives. 

[LIGISTX]

5 minutes ago, DANK_AS_gay said:

NVMe drives can achieve ridiculous speeds, but often I see SATA drives not meeting anywhere near the same performance. This is despite both SSDs being slower than the 6GB/s cap of SATA3. I see SATA SSDs at 500MB/s, but NVMe drives are reaching 3 or 4 times that. So why does the additional bandwidth of NVMe help speeds? That doesn't make sense to me if the sequential read/write speeds of either drive are below 6GB/s. 

 

Just now, BondiBlue said:

SATA III is 6 gigabits per second (Gb/s (lowercase b)), not 6 gigabytes per second (GB/s (uppercase B)). With overhead most decent SATA SSDs are close to the theoretical limitations of SATA III. 

This.

 

6 Gbps is 6000 bit per second, 8 bits per byte = 6000/8 = 750 MB per second, MB as in MegaBytes (Big B). With the overhead and losses, ~525 is about as fast as any SATA drive can go. 

 

PCIe is MUCH faster, but more importantly NVMe is much much better for a plethora of reasons.

[DANK_AS_gay]

Just now, BondiBlue said:

SATA III is 6 gigabits per second (Gb/s (lowercase b)), not 6 gigabytes per second (GB/s (uppercase B)). With overhead most decent SATA SSDs are close to the theoretical limitations of SATA III. 

Curse the bits/bytes thing. Why does there need to be a difference between standards. It makes the advertising deceptive. Gigabit ethernet (actually 125 MB/s), but Gigabytes of storage. When I hear "Gigabit" obviously I start thinking 1 Gigabyte per second (which is absurd, I know, but still...) Brands advertise significantly faster speeds through these deceptive practices.

[BondiBlue]

Just now, LIGISTX said:

PCIe is MUCH faster, but more importantly NVMe is much much better for a plethora of reasons.

Exactly. It's also (somewhat) important to note that AHCI isn't limited to the theoretical 6 Gb/s that SATA III itself is. There were some systems that used AHCI drives over PCIe, and those were capable of breaking the 1 GB/s mark without using NVMe. 

[Oshino Shinobu]

1 minute ago, DANK_AS_gay said:

Curse the bits/bytes thing. Why does there need to be a difference between standards. It makes the advertising deceptive. Gigabit ethernet (actually 125 MB/s), but Gigabytes of storage. When I hear "Gigabit" obviously I start thinking 1 Gigabyte per second (which is absurd, I know, but still...) Brands advertise significantly faster speeds through these deceptive practices.

I hesitate to tell you that even the GB numbers you see in Windows are wrong. It uses yet another different unit, the Gibibyte (binary/base 2 units), but uses the wrong acronym (should be using GiB not GB).

[DANK_AS_gay]

I wish I could mark 2 answers, but @LIGISTXgave a little more detail in his response, so for future posterity, I will mark his.

[BondiBlue]

Just now, DANK_AS_gay said:

Curse the bits/bytes thing. Why does there need to be a difference between standards. It makes the advertising deceptive. Gigabit ethernet (actually 125 MB/s), but Gigabytes of storage. When I hear "Gigabit" obviously I start thinking 1 Gigabyte per second (which is absurd, I know, but still...) Brands advertise significantly faster speeds through these deceptive practices.

It's not deceptive advertising though. Different units are used to measure different things. Would you want to say 0.125 GB/s (1Gb/s) ethernet or that you have an 8192 Gb (1TB) hard drive? Or would you rather keep it simple and say gigabit ethernet and a 1TB hard drive? 

[DANK_AS_gay]

Just now, Oshino Shinobu said:

I hesitate to tell you that even the GB numbers you see in Windows are wrong. It uses yet another different unit, the Gibibyte (binary/base 2 units), but uses the wrong acronym (should be using GiB not GB).

I knew that much... Also weird, but the difference is pretty small, comparatively.

[DANK_AS_gay]

4 minutes ago, BondiBlue said:

It's not deceptive advertising though. Different units are used to measure different things. Would you want to say 0.125 GB/s (1Gb/s) ethernet or that you have an 8192 Gb (1TB) hard drive? Or would you rather keep it simple and say gigabit ethernet and a 1TB hard drive? 

Based on the fact that we deal with GB and MB more often than the others, it would make sense for advertisements to say those things. I guarantee fewer people would buy Gigabit ethernet if the name was changed to 125MB/s. And drives and files are listed in Bytes, both things that are used far more often, download speeds are also often displayed in Bytes (tired of typing prefixes). But advertising for network speeds are in bits. That smells suspiciously of deceptive advertising practices.

[LIGISTX]

6 minutes ago, DANK_AS_gay said:

Based on the fact that we deal with GB and MB more often than the others, it would make sense for advertisements to say those things. I guarantee fewer people would buy Gigabit ethernet if the name was changed to 125MB/s. And drives and files are listed in Bytes, both things that are used far more often, download speeds are also often displayed in Bytes (tired of typing prefixes). But advertising for network speeds are in bits. That smells suspiciously of deceptive advertising practices.

There is just so much more to it than that... All of this originated in a time when the computer was a vastly different machine. We used to have kilobit (kbps), and I am sure there was some sort of reason (even if a bad reason) why they chose bit vs bytes. Harddrives are a bit of a false advertising thing seeing as I forget who it was, but they decided to change to counting 1000 instead of 1024, and then the others obviously had to copy that so their drives don't look to be inferior.

 

So there is a bit of marketing, a bit of trying to game the system, and a bit of a lot of this was developed in such a different age and time, the people who originally decided internet would be bit (small b) instead of byte (big B) had no idea what the future would hold. And as long as you understand this, it's not deceptive at all.

[Glenwing]

35 minutes ago, LIGISTX said:

Harddrives are a bit of a false advertising thing seeing as I forget who it was, but they decided to change to counting 1000 instead of 1024, and then the others obviously had to copy that so their drives don't look to be inferior.

That never actually happened, hard drives have always used 1000 since the first hard drive was invented in 1956. But that hurts the argument of certain "1024-byte kilobyte purist" people on the internet, so they make up stories like that to try to leverage people's emotions with the good old "you're being lied to" tactic.

[LIGISTX]

5 minutes ago, Glenwing said:

That never actually happened, hard drives have always used 1000 since the first hard drive was invented in 1956. But that hurts the argument of certain "1024-byte kilobyte purist" people on the internet, so they make up stories like that to try to leverage people's emotions with the good old "you're being lied to" tactic.

Really? I could have sworn I saw evidence that (i had thought it was WD) changed from 1024 to 1000 via some marketing or spec sheet at one point. Hmm. Im not part of the "you're being lied to" scene, I just genuinely thought I had seen actual proof that it happened. Again tho, it doesn't really matter since they all advertise it the same and its what consumers are accustomed to.

[Glenwing]

22 minutes ago, LIGISTX said:

Really? I could have sworn I saw evidence that (i had thought it was WD) changed from 1024 to 1000 via some marketing or spec sheet at one point. Hmm. Im not part of the "you're being lied to" scene, I just genuinely thought I had seen actual proof that it happened. Again tho, it doesn't really matter since they all advertise it the same and its what consumers are accustomed to.

Well, if you ever find it, let me know

[kokosnh]

From my knowledge, the whole problem with capacity and GiB vs GB is the Windows fault, as Oshino Shinobu already explained.
Linux uses GiB and it shows GiB, Windows uses GiB, and shows false unit of GB instead of GiB.    

Like, where's the logic of windows, that are two different unit

As for the overhead of the SATA, it's mainly the 8/10b coding. 
On PCIe up to PCIe 2.0 we also used the 8/10b coding, and we had huge losses on bandwidth.