Solid State Drives (SSD) Reduce Battery Life?

[iSee]

For the record the test performed at by TH had a critical flaw.
The test was simply ran continuously until the battery was dead. Since the SSD is significantly faster then a traditional HDD, more iterations of the test may have been completed. In terms of "work" done in by each drive, the SSD may have moved more bits of data during the same time.

It's like testing the MPG of two cars w/o considering the speed of each vehicle.

I would like to see the tests re-run with comparable work loads for both drives.

Are you sure about this?

The battery tests were conducted with MobileMark 2007 which is designed to measure battery life under real world conditions. I don't know anything about MobileMark 2007, but it would be a pretty crappy tool for measuring battery life if it penalized systems with faster I/O.

Can you point us to anything to back this up? I don't think we should assume a beachmark designed to measure battery life is fundementally flawed without some proof.

 

[macduke]

It seems like the MBA SSD is a high-quality drive since it is slightly better to even in power consumption vs. the 1.8" HD option.

Tom's Hardware points out that none of the SSDs they tested would match or beat a 1.8" "normal" hard drive, so it's impressive that Apple's does...

... or maybe the MBA's standard 1.8" hard drive is a power hog!

...or maybe OS X is a superior operating system? They were using Windoze.

I mean, pushing that data faster must require extra overhead, right? Perhaps OS X just manages it more efficiently. Beyond that, I have no idea.

 

[irun5k]

One thought- SSD's are a relative new kid on the block, and they're coming out of the block about as good as traditional mechanized drives. Gut instinct tells us they should be more efficient (no moving parts.) However, perhaps these drives need a few iterations of improvements to make them more efficient. Something tells me at the end of the day the technology has the potential to be much more efficient than it is.

Regardless, drives aren't exactly a big ticket item n energy consumption. Displays are a big draw. I think better battery life is ultimately going to depend on new battery technology.

 

[dizastor]

this is madness!

 

[daneoni]

Hopefully that changes in the future but in the meantime...

 

[ThanatosId]

It would be extremely nice if Apple offered a 15" version of the MBA, with more ram, more storage, bigger battery, and Wi-Max card.

That's a great idea!! And, They should call it a "macbook pro"...oh, wait...

 

[zif2000]

Benchmarks are misleading

First off, this may be right that in a performance situation the battery life for a HDD is better than SSD. However, the SSD gets it done faster, so less total energy is consumed and you do need to run the performance task as long.

Energy = Power * Time and is the real metric for comparing a performance task.
Battery Life is best mesured when doing very little as most of us tend to.

To understand how a faster/lower power SSD could have less battery life you need to take into account what the CPU is doing. When the CPU is waiting on HDD/SSD it consumes very little power. When is it not waiting it consumes alot of power doing the work on the data. So a faster HDD/SSD lets the CPU work more of the time and thus consumes power faster, but is more engery efficient. Let's look at an examples.

CPU_ACTIVE = 10W
CPU_IDLE = 0W (it is really higher)
CPU_TIME = 80s
HDD_ACTIVE = 2W
HDD_IDLE = 0W
HDD_TIME = 20s
SSD_ACTIVE = 1W
SSD_ACTIVE = 0W
SSD_TIME = 10s
SSD_SPEED = 2X


Average Power With a HDD
(HDD_ACTIVE * HDD_TIME + CPU_ACTIVE + CPU_TIME ) / (HDD_TIME + CPU_TIME) = (40 + 800)/100 = 8.4W (average)

Average Power With a SSD
(SSD_ACTIVE * SSD_TIME + CPU_ACTIVE + CPU_TIME) / (SSD_TIME + CPU_TIME) = (10 + 800)/90 = 9.0W (average)

What happened. The lower power faster drive is higher average power. But look at the work performed and you see the SSD is more efficient.

Work for HDD = 840 Ws
Work for SSD = 810 Ws (less it better) and provides more battery for other tasks.

 

[Mousse]

We've looked at almost a dozen different flash SSDs from seven vendors over the last few months, and measured acceptable or sometimes even disappointing power requirements with most flash SSDs. In an effort to determine the actual impact on notebook systems, we took four SSDs that we had available in our test lab, and ran a series of Mobilemark benchmark runs on a Dell Latitude D630 notebook.

But something about the way the worded the intro makes me think they're more interested in getting the facts to fit the hypothesis. They've already concluded that SSDs consume more power even before they started the experiments. From the article, their mindset is "We already KNOW SSDs drains your batteries faster than HDs, we just don't know buy how much."

All I can conclude from reading the article that HDs are more efficient than SSDs under this condition: a laptop running Windows performing a Mobilemark benchmark.

I'd like to see other test under other circumstances before making a definitive claim.

On a side note: I'll still stick with HDs until SSDs drop into the priced range of mere mortals.

 

[Bubba Satori]

There's a sucker born every minute. The tech industry has been pulling stunts like this forever and people keep falling for it.

 

[137489]

No, that's wrong.

Mac applications may appear under the Finder as being a single file, but the truth is that they are really directories containing many files. Inside this directory as well as there being the binary file that comprised the main application, there's can also be dynamic library files (the equivalent to Windows DLLs) and other applications that the main one may use. There directory structure will also contain other resources such as help files, international language support, graphic files, sound files and anything else the particular application may need.

As well as that there's other shared libraries scattered around the Mac OS X disk. Many of the core ones are in places like /usr/lib and invisible to finder. Others are hidden inside Frameworks in /System/Library/Frameworks.

I stand corrected? ok, then one more thing I like about a mac. Finder makes it look like one file, so all I have to do is drag the one file from the application folder to the trash and it gets rid of everything rather than on Windows which sometimes leaves stuff behind and I have to remove them manually. Unless mac does the same thing and I just don't know it? Mostly I was talking about installed applications like Pages, Bento, Audacity, etc.but hey, if finder makes it easier - I am all for it. I hate the way Windows uninstaller leaves stuff littered around my drive.

 

[SkyBell]

I stand corrected? ok, then one more thing I like about a mac. Finder makes it look like one file, so all I have to do is drag the one file from the application folder to the trash and it gets rid of everything rather than on Windows which sometimes leaves stuff behind and I have to remove them manually. Unless mac does the same thing and I just don't know it? Mostly I was talking about installed applications like Pages, Bento, Audacity, etc.but hey, if finder makes it easier - I am all for it. I hate the way Windows uninstaller leaves stuff littered around my drive.

Actually, it's pretty much the same. Just dragging the app to the trash doesn't completely get rid of it. There's still some leftover files.

 

[Analog Kid]

I wish people would stop saying SSDs are new. They've been around at least 20 years. It wasn't until now that they were even viable as a mainstream technology.

I think a lot of this is coming from the erase operations-- writing to flash requires a lot of erase time and power. People talk about having to shuffle bits around on HDs, but there's a tremendous amount of bit shuffling happening on flash. You can't just overwrite a byte in flash-- you have to erase a whole block after moving all the data out of it before you can write anything new to that block.

I think the reason we're seeing better battery life with the MBA is because it uses a relatively slow SSD. Manufacturers realized the reason people like SSDs is because they're fast, so they're pressing their advantage at the expense of power.

They don't mention how much work the computer did - perhaps the bettery was just used up sooner with a SSD because an IO bottle neck was removed, allowing the CPU to do more work (and use up the battery faster).

They attempted to correct for this by benchmarking performance*life.

Interesting. I wonder if there are cases where the heat of an HD makes the fans run more?

Less battery life means more power draw, more power draw means more heat, more heat means fans run more. If the tests are right, SSDs will run the fans more.

The other problem is that with an HDD spinning constantly it is reasonably efficient, the problem is with spin up, against an SSD's 'instant on'. So constant data read / write you may expect an HDD to be as or more efficient than an SSD (barring I/O bottlenecks caused during the test), but for intermittent read / writes requiring random access the SSD should absolutely trash the HDD, even without any OS level optimisation.
...
A good example was a task that reads from disk one second in every ten. Repeat this test over one minute and the SSD will go to 'read power' for a total of 6 seconds (lets give some margin and say 12, which is hugely exaggerating), an HDD will most likely be spinned up ('read power') for a full minute. This is not so unusual an example of disk usage.

Slashdot is covering this - there are some very good examples of just how bad the test methodology is, plus some interesting ways a better test may be run.

Personally - I prefer HDD's for now for other reasons (cost and better understood MTBF).

That was THs point-- hard drives give a max power rating, but spend very little time drawing that much power. Being spun up for a full minute isn't much of a problem-- the platters are already going, you're just compensating for friction.

 

[Analog Kid]

First off, this may be right that in a performance situation the battery life for a HDD is better than SSD. However, the SSD gets it done faster, so less total energy is consumed and you do need to run the performance task as long.

Energy = Power * Time and is the real metric for comparing a performance task.
Battery Life is best mesured when doing very little as most of us tend to.
...
What happened. The lower power faster drive is higher average power. But look at the work performed and you see the SSD is more efficient.

Work for HDD = 840 Ws
Work for SSD = 810 Ws (less it better) and provides more battery for other tasks.

Again, read the article-- particularly the performance*life benchmarks. They did compensate for this but people refuse to read carefully enough...

The performance*life benchmarks are doing exactly what you suggest--
(ops/sec)*(sec/charge)=ops/charge

The only real question I see in all of this is how well the benchmark models real life. Their methodology seems sound to me, and this is no more questionable than any other benchmark test.

We have another data point. That's what this means. Your mileage may vary.

 

[digitalbiker]

That's a great idea!! And, They should call it a "macbook pro"...oh, wait...

Except that a MBP is heavier, thicker, uses 2.5" drives, contains a DVD drive, doesn't allow using a SSD internal drive, is styled different, uses a different keyboard, has a heavier removable battery, doesn't have a magnetic latch. Other than that it is like a MBA.

I am talking about a ultra-thin, ultra-light, no DVD drive, portable styled like the MBA but with a larger internal battery, more ram, larger capacity SSD, and 15" display rather than a 13" display.

 

[Butthead]

Personally I am waiting for the next update to MBA. Hopefully it will use newer larger SSD drives and also improve the OS to optimize the drive for battery life.

The performance boost is the most important thing to me. Using the MBA with SSD in the Apple Store, compared to the slow HDD versions makes a big difference.

You really notice the almost instant on speed of the MBA SSD and the very quick launch of apps for the first time. Some apps take 3 or 4 times longer to load when initially launched.

It would be extremely nice if Apple offered a 15" version of the MBA, with more ram, more storage, bigger battery, and Wi-Max card.

But something about the way the worded the intro makes me think they're more interested in getting the facts to fit the hypothesis. They've already concluded that SSDs consume more power even before they started the experiments. From the article, their mindset is "We already KNOW SSDs drains your batteries faster than HDs, we just don't know buy how much."

All I can conclude from reading the article that HDs are more efficient than SSDs under this condition: a laptop running Windows performing a Mobilemark benchmark.

I'd like to see other test under other circumstances before making a definitive claim.

On a side note: I'll still stick with HDs until SSDs drop into the priced range of mere mortals.

Odd, Arn doesn't report 1st page news on latest SSD announced price reductions but somehow is inclined to post this TH story...as it applies to Apple how???

Super Talent has announced a 1.8in 120GB $679 SSD, that nearly doubles the MBA, for likely 1/2 the price of the Samsung (price to Apple)...which is also 1/2 the performance, currently in use. Hopefully with Monteviña chipset, the MBA will move to SATA compatibility.

Yesterday (actually before yesterday in EU said to be shipping within a week),

http://www.electronista.com/articles/08/07/01/ocz.sata.ii.ssds/

OCZ announced 2.5in 128GB SSD, which electronista is reporting a cost of $479...to me that is a far more an important story! This 128GB SSD is MLC, but performance is boosted again with more advanced controller, well beyond the old tech slug in the current MBA.

Despite the reduced cost, the drives are claimed to be ten times faster in seeking data than a rotating notebook hard drive and transfer that data more quickly as well, reading between 120 to 135 Mbps and writing at between 80 to 93 Mbps.

The drives are also claimed to be as reliable as the pricier drives, running for about 1.5 million hours of continuous use while still consuming half as much power as a traditional hard disk. All the Core Series drives should be available soon, with prices of $169 for a 32GB drive, $259 for a 64GB version, and a $479 for a flagship 128GB version.

Intel's chipset could be another area for power saving increases.

http://www.techfuzz.com/roadmaps/2009.aspx

# Intel Moorestown platform is expected to be launched in 2009. Moorestown is the next generation platform for MIDs and UMPCs and will replace Intel's Menlow platform. Moorestown will consist of Lincroft CPU, Langwell IOH chipset, and EvansPeak comms chipset. Some devices based on Moorestown could see up to an 100% improvement in battery life due to a total system TDP of less than 2 Watts.
# Intel's Lincroft CPU is expected to be launched in 2009 and will be part of the Moorestown platform. Lincroft will bring the graphics core and memory controller all into the processor itself. It will either be a 45 nm or 32 nm chip. It will include support for DDR3, PCIe, USB, and SATA. It will likely use Intel's QuickPath Interconnect technology thus replacing the FSB.
# Intel's Langwell IOH chipset is expected to be launched in 2009 and is a component of the Moorestown platform. Langwell will include support for solid state devices (SSD).

Better Battery Life? Not Necessarily
http://www.anandtech.com/showdoc.aspx?i=3287&p=4

From nearly $4k, to just under $0.5k, for similar high-performance (Memoright is SLC, OCZ's is MLC + more advanced controller) in just a mater of months...impressive! Front page news if you ask me, but what do I know? Can't wait to see what Samsung and/or OCZ do as far as the 200Mb sustained transfer rate 256GB model still vaporware but due out later this year or early next year, will cost...128GB is still fairly restrictive capacity compared to HD's, what with the Hitachi 7k320 Travelstar now shipping.

 

[Analog Kid]

But something about the way the worded the intro makes me think they're more interested in getting the facts to fit the hypothesis. They've already concluded that SSDs consume more power even before they started the experiments. From the article, their mindset is "We already KNOW SSDs drains your batteries faster than HDs, we just don't know buy how much."

I think they wrote the introduction after doing the tests... They didn't run the tests while you were reading.

 

[Digital Skunk]

I will have to agree with those that say the speed increase outweighs the battery life drain, and that the machine was probably worked to death during the testing. It does seem skewed indeed.

But I will be waiting to use SSDs on my main machines until the price comes down dramatically. Getting a 320 7k100 is a lot more economical and adds to the speed of my machine greatly, and increases my storage capacity at the same time, while only spending about $250 for the drive and installation.

 

[deputylove8]

SSD = Slightly slower devices? Hahaha

 

[jhsfosho]

And I thought SSD drives were supposed to increase battery life?? Seems like a step in the wrong direction.

 

[Analog Kid]

Don't know if anyone else has read the MobileMark white paper, but it does include idle time throughout the test. It also outputs two metrics: performance (how quickly it gets the work done) and battery life (by performing the same amount of work per hour regardless of performance).

This was already posted and ripped to shreds on Slashdot. Read the comments.

http://hardware.slashdot.org/article.pl?sid=08/07/02/134256

I haven't read the whole comment thread, but what I have read indicates that most people haven't actually looked at the tests.

 

[ryan200]

someone mentioned displays as being a bigger draw i wonder when they will come out with oled displays for laptops. maybe 7in first for a ultra portable.

and how much power it will consume.

 

[twoodcc]

wow. well hopefully they'll continue to improve the technology

 

[ryan200]

when they stop improving technology we will stop buying it. unless they just make everything so crappy that it falls apart and that will force you to buy a new one not a technological advance.

but as far as laptop hard drives are concerned they have a huge jump start. but when ssd becomes more aged it should replace hard drives because of speed alone.