The "real" reason a notebook with an SSD gets better battery life than with a HDD

[Zap]

Zap's deep thoughts for the day, the PG edition.

So I've thought about this. Notebook HDDs are actually quite power efficient these days, and the high performance SSDs probably draw around the same power.

I started thinking of CPUs. My former Vostro V131 with a "normal" wattage CPU had great battery life and my Samsung Series 3 with an ULV CPU has abysmal battery life. Most of that can be attributed to the actual battery of course, but with the Samsung I'm often waiting for it to finish "thinking." I'm not really accustomed to that anymore. Yeah, yeah, first world problem. :sneaky: Oh yeah, both had SSDs, and both are dual core Sandy Bridge with Hyperthreading. One is around 1.3GHz and the other can Turbo to 3GHz.

I then thought of the SPCR review of low wattage desktop CPUs, and how the normal wattage ones drew minimally more power overall (maybe 5W for a dual core) for the higher performance because it "raced to idle" faster.

Maybe that can apply to SSDs? Assuming no sleep mode, a notebook with SSD can be usable sometimes a half minute sooner than with a HDD (while the screen is lit and the entire system powered on) or even more if startup is cluttered, and everything you do with it that runs off local storage finishes faster. Thus, if you aren't tied to waiting for web pages to load, the few seconds here and there might add up to many minutes before Windows warns you of low battery.

Couple that with getting an SSD that has super low idle power draw, and maybe can get a few percentage points added to the battery life.

Look for Zap's deep thoughts of the day, the NC-17 edition, on the other forum.

 

[dave_the_nerd]

Makes sense.

 

[Charles Kozierok]

I think that's part of it, but idle power consumption during intermittent use is an even greater difference. SSDs basically use nearly nothing when not active, and HDDs have to keep those platters spinning.

 

[hhhd1]

The reason is "Random access time"

you see allot of people saying booting a harddrive in 4 minutes, while an SSD boots in 30 seconds ?

in case of HDD, in those 4 minutes it is the bottleneck, meaning that it is being utilized 100%, and using maximum power of random read.

in case of SSD, in those 20 seconds is has been utilized like <10% and the cpu is the bottleneck in most cases.

so to do the same operation, a hdd has used 4 minutes of battery, while drawing allot of power, while the ssd only used 20 seconds of battery, while drawing allot less.

The same applies for any page file access or any program loading or saving files.

 

[Schmide]

What if we look at this on an iop based system?

Light 5 iops/s 18k operations an hour
Medium 10 iops/s 36k operations an hour
Heavy 20 iops/s 72k operations an hour

Typical seek times

SSD 0.1ms
HD 5ms

Total time seeking per hour

Light
SSD 0.1ms * 18k = 1.8s
HD 5ms * 18k = 90s = 1.5min

Medium
SSD 0.1ms * 36k = 3.6s
HD 5ms * 36k = 180s = 3min

Heavy
SSD 0.1ms * 72k = 7.2s
HD 5ms * 72k = 360s = 6min

So pinching ms can really add up; let alone, imagine adding transfer rates into the picture.

 

[IntelUser2000]

I then thought of the SPCR review of low wattage desktop CPUs, and how the normal wattage ones drew minimally more power overall (maybe 5W for a dual core) for the higher performance because it "raced to idle" faster.

It's little more involved than that. Back then when most of the common applications(like web browsing, and office applications) were considered demanding enough, CPUs ran closer to the TDP more of the time. So back in those days the TDP made greater difference in battery life.

(Also the fact that power management technologies were considerably less advanced further magnified TDP's role in battery life, since CPUs ran at TDP more of the time)

But nowadays most common usage do not demand the CPU much at all. So to save power, the developers employ few tricks of their own. Like running the CPU at LFM frequency when running HD Video but offloading the load to the GPU.

Advanced power management also means the CPU power usage is much more "bursty" than before. When you open a browser, or open up a link to a different page in the website, the CPU peaks for a very short time and backs down. Idle modes dominate more than 90% of the time in those scenarios making TDP difference not very significant.

(TDP is of course relevant when running intensive codes, like when gaming or running a benchmark like Linpack)

Despite the theory of calculating faster to idle faster saving power, in the real world the ULV chips are specced to have lower power usage than standard chips at all operating points, including the idle C states.

Back to SSDs: Only few SSD controllers offer better battery life than notebook HDDs. Samsung, Crucial, and Intel controllers to name a few(not the Sandforce controller Intel SSDs) give better battery life than SSDs.

 

[Emulex]

Antivirus scan + backup = 3-4 hours (full daily, i'm paranoid).

with SSD = 40 minutes.

multiply by 100 machines = lots of energy saved.

Plus I tend to sleep macbook ssd after 1 minute since there is very little lag to resume from idle mode.

Originally SSD's (50nm) ate more power than 1.8" hard drives, so you would lose power runtime - i'm sure they've fixed that by now but it was a problem back in the days when you went from 5hours hard drive to 4 hours ssd. Most people didn't care since life was so much faster.