SATA II theoretical maximum speed

[Hans Gruber]

I have a desktop with SATA III. I have watched several youtube videos where they show the SATA III being maxed out by around 3.5 SATA III SSD drives. It would take 4 average SSD SATA III drives to max out the 6gbps max throughput of the SATA III bus.

The T410 is SATA II. I have a known good Toshiba 2.5" SSD that I benchmark in a T410. It seems the max throughput on the SATA II port on these thinkpads is 250mb read and 235mb write. Is there something wrong with the drivers? Those numbers reflect a SATA I bandwidth of 1.5gbps vs. the 3gbps of the SATA II bandwidth.

Essentially half the bandwidth of a SATA III bus is not less than one SATA III SSD drive that could push 500MB read and 450MB write.

A SATA III SSD drive is backward compatible with an SATA II bus and drive. There is no decrease in performance if you have a SATA III drive running on an SATA II bus or system.

Perhaps new drivers or modified drivers would help reach the maximum performance of the SATA II performance on the T410 series.
Anybody want to comment on my findings?

[ajkula66]

The T410 is SATA II. I have a known good Toshiba 2.5" SSD that I benchmark in a T410. It seems the max throughput on the SATA II port on these thinkpads is 250mb read and 235mb write. Is there something wrong with the drivers? Those numbers reflect a SATA I bandwidth of 1.5gbps vs. the 3gbps of the SATA II bandwidth.

Nope.

What you're getting is appropriate for a SATA II system. A SATA I system such as T60 would return 150 or less.

Essentially half the bandwidth of a SATA III bus is not less than one SATA III SSD drive that could push 500MB read and 450MB write.

Even on a pure SATA III machine - presuming no RAID, of course - you'll be stuck at under 600 depending on the drive used. So, in real life it goes like this

SATA I < 150

SATA II < 300

SATA III < 600

[theterminator93]

That sounds about right. In my T420, my mSATA SSD benchmarks in the (IIRC) low 500 MiBps range on a SATA-III bus - but since mSATA on the T420 is SATA-II, I see actual speeds in the mid-high 200s. Remember we're talking gigabits per second, not gigabytes per second.

3 Gbps (gigabits per second) is the equivalent to 375 MBps (megabytes per second) - or roughly 358 MiBps (Mebibytes per second). Add in transmission overhead due to 8b/10b encoding and we arrive at a maximum theoretical data throughput of 286 MiBps. SATA-I would be half that (143 MiBps), and SATA-III would be twice that (572 MiBps).

[Hans Gruber]

There are 8mbps for each 1 megabytes. The theoretical max is 375mb Sata 3.0gbps if we go by those numbers. I am stuck at 250mb. I am curious what those with T420 series are getting. The T420 has a SATA III bus.

250 is not close to 375mb.

[theterminator93]

250 is not close to 375mb.

Remember we're talking gigabits per second, not gigabytes per second.

3 Gbps (gigabits per second) is the equivalent to 375 MBps (megabytes per second) - or roughly 358 MiBps (Mebibytes per second). Add in transmission overhead due to 8b/10b encoding and we arrive at a maximum theoretical data throughput of 286 MiBps.

See above. I'm not sure you understood my entire post. Let me explain a little more below.

By definition a megabyte is 1,000,000 bytes. Due to 2 raised to powers of ten being the computer OS's way of formatting kilo/mega/giga etc, the syllable "bi" was used to replace "ga" to differentiate between 1,000 (kilo) and the computer's 1,024 (kibi) bytes. Hence the terms kibibyte, mebibyte, gibibyte, etc... and the need to reduce that 375 MBps figure by 1024 (twice) to come to the equivalent mebibyte figure of 357.6 MiB. Reduce that again by 20% due to encoding... voila, 286 MiB theoretical throughput.

My mSATA SSD (yes - the main drive bay on the T420 is SATA-III, but the mSATA port is SATA-II.), benchmarked just now, tops out at 251 MiBps sequential read. That's close enough in my book to the theoretical maximum of 286 MiBps to not be preoccupied with or concerned about the missing 35 MiBps in theoretically attainable throughput gain - which in all reality is additional overhead being used by the chipset/controller.