Samsung Ecogreen 1.5TB f2 better than f3?

[phshannon]

According to the Samsung site the Ecogreen 1.5TB f2 seems to have better specs than the f3, despite being older & cheaper:

• higher media-buffer transfer rate (166MB/s vs 135MB/s)
• lower latency (5.52ms vs 5.56ms)
• quieter (2.5 bel vs 2.6 bel idle; 2.8 bel vs 2.9 bel seek)
• lower power consumption (5.7W vs 7.1W seek)
• lighter (650g vs 690g).

Can anyone explain? Due to drive failure I must buy a new disk soon.

The only disadvantage with the f2 (surely a mistake) is that it apparently can't be used below an altitude of 300m; would using it here in the Netherlands then invalidate the guarantee?

 

[Stereodude]

Not sure why. I wouldn't expect any noticeable difference in real use though. I'd get the cheaper one myself.

 

[Howell]

The F2 may be cheaper at retail but more expressive to produce. I'd check for RMA rates as best as possible and make my decision based on that unless you don't need to worry about it.

 

[LunarMist]

I have a few HDS1541UI and numerous HD203WI drives. Although the latter are 2TB drives the specs are about the same for the 1.5TB version. I don't think there is much difference between the F2 and F3 in practical use. HDTach results also indicate that transfer rates are similar. The altitude is likely a typo, probably -300m rather than 300m. Consider that about 1/3 of the global population lives at 100m or less. As well, barometic pressure does not vary significantly over a few hundred meters of elevation compared to the effect of weather systems.

 

[phshannon]

The F2 may be cheaper at retail but more expressive to produce. I'd check for RMA rates as best as possible and make my decision based on that unless you don't need to worry about it.

Why more expensive? I don't know any differences between the f2 & f3 apart from the above.

How to check for RMA rates?

Thanks.

 

[time]

I've had a look through the specs for quite a few of Samsung's drives, and it does appear that there was some sort of problem going from the F2 500GB platters to the F3 666GB platters. I'll hazard a guess that the transfer rate is approximately half what it could be, so presumably Samsung had to use some kind of weird interleaving or whatever.

I notice my local supplier is sticking with the F2 1.5TB and I think you should too.

Having said that, I've also been looking at the specs for the F4 series, and they are simply colossal. They *all* come close to maxing out SATA II, yet idle at 4W and barely exceed 5W during seeks, while emitting just 2.6 bels (2.8 during seeks). And their hard error rate is an order of magnitude better than their competitors!

The 1.5TB and 2TB models are hitting 280MB/s, the only drawback being 4k blocks. There's no 1TB yet, but the 320GB 7200 rpm model is rated at 285MB/s, same incredibly low noise and power figures, and uses 512 byte blocks.

Samsung has taken the ball and run out of the stadium with it.

 

[LunarMist]

The 1.5TB and 2TB models are hitting 280MB/s, the only drawback being 4k blocks.

Where? The benchmarks from quick searches show up to 150 MB/sec. sustained on the outer tracks, dropping below 80 on the inside tracks. Of course that is excellent for a 5400 RPM drive. Results of 280 MB/sec. must be affected by cache.

 

[jtr1962]

I've had my 2TB F4 for over a week now with no problems to report. Yes, STRs are blazingly fast. Even though my M/B only has SATA 150, average STR over the entire disk is around 100 MB/sec ( the interface is bottlenecking STRs to ~115 MB/sec ). On the inner tracks STR is about 65 MB/sec. DTemp reports about 6°C less than my old 3-platter 200 GB 7200 RPM. Access times are 15.6 ms according to HDTune. This isn't that much more than most 7200 RPM drives. The 4K sectors appear to be a non-issue if the drive is partitioned correctly. I used diskpar, and started at sector 64 instead of 63.

Oh, and it's quiet also. I barely hear seeks even during benchmark testing.

 

[time]

Typo, s/b 250MB/s for the 5400rpm 2TB drive.

You're right, they're seems to be a much larger drop than I expected from the quoted raw transfer rate to measure sustained transfer rates. The 2TB seems to top out at about 137MB/s and the 7200rpm 320GB at 150MB/s. Still impressive, but not a reflection of Samsung's data sheets - presumably track-to-track time has become more significant than raw transfer rates?

From a You Tube comparison between a 1TB 7200rpm F1 and a 2TB 5400rpm F4 in HDTune: F1 F4

 

[Handruin]

I did my own few tests between the 1TB F3 HD103SJ and the 1.5TB F2 HD154UI if any of that might be useful.

 

[time]

Another typo, F1 s/b F3.

It's the same drive as yours, Doug. Results marginally better for some unknown reason.

Note that the 7200rpm F3 1TB has a quoted media transfer rate of 250MB/s. It's only the 5400rpm F3 1.5TB (and 2TB?) that has the lower specs.

 

[Handruin]

Another typo, F1 s/b F3.

It's the same drive as yours, Doug. Results marginally better for some unknown reason.

Note that the 7200rpm F3 1TB has a quoted media transfer rate of 250MB/s. It's only the 5400rpm F3 1.5TB (and 2TB?) that has the lower specs.

I kinda figured you meant the F3 since the model number was the same. It's not a big deal. The slight difference could be due to the higher revision in the benchmark software in your example or possibly a better SATA controller than mine. My tests were using an nForce 570 chipset with an AMD CPU. The two are fairly close in results though. My CPU usage percent is a lot lower than in your example.

 

[time]

To summarize sustained transfer rates compared to media transfer rates:

5400rpm F2 1.5TB: 64%
7200rpm F3 1TB: 57%
5400rpm F4 2TB: 55%
7200rpm F4 320GB: 53%

So the trend is apparently to almost two revolutions to read one track. I can sorta understand the 320GB having more problems because it only has one track per cylinder (1 head).

What's Samsung doing? Any hypotheses?

 

[LunarMist]

To summarize sustained transfer rates compared to media transfer rates:

5400rpm F2 1.5TB: 64%
7200rpm F3 1TB: 57%
5400rpm F4 2TB: 55%
7200rpm F4 320GB: 53%

So the trend is apparently to almost two revolutions to read one track. I can sorta understand the 320GB having more problems because it only has one track per cylinder (1 head).

What's Samsung doing? Any hypotheses?

It's been like that since the 1990s.

 

[time]

I'm not so sure, Lunar.

Hitachi 7K1000.B: 117/175 MB/s = 67%

Allowing for 13.5% encoding overhead brings the raw rate down to 152MB/s.

If the average head switch/cylinder switch time is 2.5mS, that equates exactly, so no mystery here.

Samsung claims an order of magnitude better hard error rate. Increasing ECC bits by say 10% takes the overall overhead to 14%. However, the newer 4kB sector drives reduce this overhead down to just 3.3% (according to ArsTechnica).

Roughly calculated average head/cylinder switch times:

5400rpm F2 1.5TB: 3.8mS
7200rpm F3 1TB: 4.2mS
5400rpm F4 2TB: 8.4mS
7200rpm F4 320GB: 5.2mS

The numbers are obviously only very approximate. I'll hazard a wild guess that head switches are about 3mS and cylinder switches 5.3mS (the 320GB doesn't do head switches). Which is slow, but still doesn't come close to explaining the 2TB F4, which by rights should be delivering 177MB/s.

I can get close if I assume ArsTechnica is flat wrong *and* Samsung decided not to bother with skewing between either tracks or platters, i.e. you have to wear average rotational latency every time you change heads or tracks. :crap:

 

[LunarMist]

I meant the general relationship between media-buffer transfer rates and sustained transfer rates.