Are these HD speeds OK?

[Terry Pinnell]

On this PC (Quad Core Q9450 2.66 GHz, 4 GB DDR2 667 MHz) I have 2
identical Samsung HD753LJ 750 GB hard drives. The spec at
http://www.testfreaks.co.uk/internal-hard-drives/samsung-hd753lj/
says:

"The Samsung HD753LJ has 512 bytes per sector and a rotational speed
of 7200 RPM. In terms of its performance, it has an average latency of
4.17 ms, and an average seek time of 8.9 ms. Media transfer rate is at
a maximum of 175 MB/ second while interface transfer rate is at a
maximum of 300 MB/ second."

But the HD Tune benchmarks I just ran gives results well below those:

HD Tune: SAMSUNG HD753LJ Benchmark (My main HD, C

Transfer Rate Minimum : 44.8 MB/sec
Transfer Rate Maximum : 89.9 MB/sec
Transfer Rate Average : 73.2 MB/sec
Access Time : 13.9 ms
Burst Rate : 89.2 MB/sec
CPU Usage : 5.5%

HD Tune: SAMSUNG HD753LJ Benchmark (Mostly for backup, I

Transfer Rate Minimum : 65.1 MB/sec
Transfer Rate Maximum : 102.3 MB/sec
Transfer Rate Average : 91.8 MB/sec
Access Time : 13.8 ms
Burst Rate : 147.7 MB/sec
CPU Usage : 6.3%

In not-too-technical terms, can one of the experts advise why this can
be so please? Also, I'm puzzled why my backup drive is some 25%
faster?

 

[Jaimie Vandenbergh]

On this PC (Quad Core Q9450 2.66 GHz, 4 GB DDR2 667 MHz) I have 2
identical Samsung HD753LJ 750 GB hard drives. The spec at
http://www.testfreaks.co.uk/internal-hard-drives/samsung-hd753lj/
says:

"The Samsung HD753LJ has 512 bytes per sector and a rotational speed
of 7200 RPM. In terms of its performance, it has an average latency of
4.17 ms, and an average seek time of 8.9 ms. Media transfer rate is at
a maximum of 175 MB/ second while interface transfer rate is at a
maximum of 300 MB/ second."

But the HD Tune benchmarks I just ran gives results well below those:

"Access time" in your scores is the same as latency plus seek time
above. They come out even.

The "media transfer rate" quoted above is not believable, frankly. It
sounds like a read totally out of the HD's cache. "Burst rate" in
yours is similarly useless. The "interface transfer rate" is of course
irrelevant, it's just SATA-300.

Yours is believable - my Samsung 1gig gets between 80 and
120meg/second using an OSX benchmark program, for comparison.

HD Tune: SAMSUNG HD753LJ Benchmark (My main HD, C

Transfer Rate Minimum : 44.8 MB/sec
Transfer Rate Maximum : 89.9 MB/sec
Transfer Rate Average : 73.2 MB/sec
Access Time : 13.9 ms
Burst Rate : 89.2 MB/sec
CPU Usage : 5.5%

HD Tune: SAMSUNG HD753LJ Benchmark (Mostly for backup, I

Transfer Rate Minimum : 65.1 MB/sec
Transfer Rate Maximum : 102.3 MB/sec
Transfer Rate Average : 91.8 MB/sec
Access Time : 13.8 ms
Burst Rate : 147.7 MB/sec
CPU Usage : 6.3%

In not-too-technical terms, can one of the experts advise why this can
be so please? Also, I'm puzzled why my backup drive is some 25%
faster?

Two things. First, the C: drive is also in use by Windows while the HD
Tune benchmark is happening. This can easily massively disrupt the
test with just a couple of extra seeks.

Second, have you tried rerunning the tests? Synthetic benchmarks like
this are generally pretty irregular. I wouldn't trust anything that
took less than a minute to generate an HD score, and even then I'd run
it three times.

And a general thing - benchmarking software scores are only comparable
with the *same* benchmark on the *same* hardware (except the device
being tested). There's no way to make accurate comparisons with
different benchmarks, and/or across different host hardware. You can
get a feel, but nothing accurate.

Benchmarks are not direct indicators of real life performance.

Cheers - Jaimie

 

[Arno]

On this PC (Quad Core Q9450 2.66 GHz, 4 GB DDR2 667 MHz) I have 2
identical Samsung HD753LJ 750 GB hard drives. The spec at
http://www.testfreaks.co.uk/internal-hard-drives/samsung-hd753lj/
says:

"The Samsung HD753LJ has 512 bytes per sector and a rotational speed
of 7200 RPM. In terms of its performance, it has an average latency of
4.17 ms, and an average seek time of 8.9 ms. Media transfer rate is at
a maximum of 175 MB/ second while interface transfer rate is at a
maximum of 300 MB/ second."

.... at a (not seen in practice) maximum ...

But the HD Tune benchmarks I just ran gives results well below those:

HD Tune: SAMSUNG HD753LJ Benchmark (My main HD, C

Transfer Rate Minimum : 44.8 MB/sec
Transfer Rate Maximum : 89.9 MB/sec
Transfer Rate Average : 73.2 MB/sec
Access Time : 13.9 ms
Burst Rate : 89.2 MB/sec
CPU Usage : 5.5%
Fine.

HD Tune: SAMSUNG HD753LJ Benchmark (Mostly for backup, I

Transfer Rate Minimum : 65.1 MB/sec
Transfer Rate Maximum : 102.3 MB/sec
Transfer Rate Average : 91.8 MB/sec
Access Time : 13.8 ms
Burst Rate : 147.7 MB/sec
CPU Usage : 6.3%

Fine as well.

In not-too-technical terms, can one of the experts advise why this can
be so please?

Simple: You fell for marketing speech. Maximum xyz rates are all
only theoretical values or values only seen under very special
circumstances, e.g. long linear reads at the very start of the disk.

Also, I'm puzzled why my backup drive is some 25%
faster?

It is on a different controller? Or does it have different
settings?

Arno

 

[Arno]

On Tue, 12 May 2009 14:33:25 +0100, Terry Pinnell

"Access time" in your scores is the same as latency plus seek time
above. They come out even.

The "media transfer rate" quoted above is not believable, frankly. It
sounds like a read totally out of the HD's cache. "Burst rate" in
yours is similarly useless. The "interface transfer rate" is of course
irrelevant, it's just SATA-300.

Yours is believable - my Samsung 1gig gets between 80 and
120meg/second using an OSX benchmark program, for comparison.

Two things. First, the C: drive is also in use by Windows while the HD
Tune benchmark is happening. This can easily massively disrupt the
test with just a couple of extra seeks.

Second, have you tried rerunning the tests? Synthetic benchmarks like
this are generally pretty irregular. I wouldn't trust anything that
took less than a minute to generate an HD score, and even then I'd run
it three times.

And a general thing - benchmarking software scores are only comparable
with the *same* benchmark on the *same* hardware (except the device
being tested). There's no way to make accurate comparisons with
different benchmarks, and/or across different host hardware. You can
get a feel, but nothing accurate.

Benchmarks are not direct indicators of real life performance.

Very true.

Arno

 

[Terry Pinnell]

"Access time" in your scores is the same as latency plus seek time
above. They come out even.

The "media transfer rate" quoted above is not believable, frankly. It
sounds like a read totally out of the HD's cache. "Burst rate" in
yours is similarly useless. The "interface transfer rate" is of course
irrelevant, it's just SATA-300.

Yours is believable - my Samsung 1gig gets between 80 and
120meg/second using an OSX benchmark program, for comparison.


Two things. First, the C: drive is also in use by Windows while the HD
Tune benchmark is happening. This can easily massively disrupt the
test with just a couple of extra seeks.

Second, have you tried rerunning the tests? Synthetic benchmarks like
this are generally pretty irregular. I wouldn't trust anything that
took less than a minute to generate an HD score, and even then I'd run
it three times.

And a general thing - benchmarking software scores are only comparable
with the *same* benchmark on the *same* hardware (except the device
being tested). There's no way to make accurate comparisons with
different benchmarks, and/or across different host hardware. You can
get a feel, but nothing accurate.

Benchmarks are not direct indicators of real life performance.

Cheers - Jaimie

Thanks both, understood and reassured.

 

[Eric Gisin]

The 175MB/s is during sector I/O, there are gaps and ECC that add overhead.
Multiply that by 0.8 and you get 120MB/s on the outer zone (first several GB).
The drive or channel with lower burst rate may be configured as SATA150.