SSDs vs HDDs battery life - SSDs not necessarily better

[IntelUser2000]

Toms has one of the only modern SSD battery life review: http://www.tomshardware.com/reviews/notebook-battery-life-storage,5152.html

Though the platform is not ideal, battery life comparisons are far superior to sticking in a power meter and inputting that number as a comparison.

The reviews show SATA-based SSDs are generally better in battery life than NVMe ones. The top HDDs can beat lots of the NVMe SSDs in battery life.

Important note in the article:

The wake time is usually determined by the controller or the firmware. Some controller makers develop the firmware in house, but some SSD manufacturers choose to build custom firmware. Developing firmware is complicated and expensive. Not every company has the resources to dedicate to corner-case work, like making efficient wake time.

The very frugal HDDs have peak power in the range of 2W, while NVMe SSDs can reach 5W, or more. So as the article states, if the firmware is not done well, you may have a case where even with very low idle power, it realistically can't transition to idle and vice versa fast enough, rendering power management moot. The reviewers(and readers) mindset is also a problem, because they usually think in terms of Desktops. Desktops HDDs use 3-4x more power in load, while SSDs are viewed with little distinction between Desktop and Notebook usage.

SSD reviews need to add battery life as a comparison. Forget W numbers. Minutes and Hours is what should be tested.

 

[IntelUser2000]

Another case of watt numbers being misleading.

Anandtech 760p review: https://www.anandtech.com/show/12349/the-intel-ssd-760p-512gb-review/8

Review shows 760p faring worse.

TH battery test:
http://www.tomshardware.com/reviews/intel-760p-ssd-review,5435.html

But actual battery life test shows 760p faring better than 600p.

 

[Billy Tallis]

Another case of watt numbers being misleading.

Anandtech 760p review: https://www.anandtech.com/show/12349/the-intel-ssd-760p-512gb-review/8

Review shows 760p faring worse.

The 760p only came off looking worse in my idle power tests because it has some bugs with its implementation of the NVMe APST feature. It's relatively common to find NVMe SSDs whose power management barely works under Windows, and does crazy things if the configuration is even slightly different from how the Windows NVMe driver does things. The Linux NVMe driver has a disappointingly long list of drives that require special treatment such as not using the deepest sleep states or not using APST at all. I've even put together a spreadsheet that lets me fairly accurately identify the NVMe SSD controller based solely on the power management behavior of the drive.

I could set up the test to measure the idle power as experienced under Windows in its default configuration, but then the idle power measurements wouldn't match up with the wake-up latency test. Instead, I hope to be able to soon replace the current idle power test with a more in-depth one that actually tests all the different NVMe power states, but that level of testing will require a lot more manual intervention than the current test.

 

[whm1974]

The reviews show SATA-based SSDs are generally better in battery life than NVMe ones. The top HDDs can beat lots of the NVMe SSDs in battery life..

I'm am not surprised by this as NVMe SSDs are built for speed to begin with, and should really only be use with desktops and portable workstation class laptops.

 

[IntelUser2000]

I could set up the test to measure the idle power as experienced under Windows in its default configuration, but then the idle power measurements wouldn't match up with the wake-up latency test.

Thanks for the reply. It's nice to see reviewers actually read responses.

It's silly. Even TH's tests are not what it should be. They should have used a U-class system. But they at least try. And because nearly every new SSDs that get tested by them go through battery life testing, they only need little extra work to get many SSDs facing off against each other.

Power management for drives are only really relevant for Notebooks, because they might correlate to battery life. The 1-2W difference really does not matter for Desktops, because the absolute number is too small to care anyway. That means absolute power ratings are no more useful than Sandra benchmarks for CPUs, because they don't show the numbers the people really need to see.

I'm am not surprised by this as NVMe SSDs are built for speed to begin with, and should really only be use with desktops and portable workstation class laptops.

It can work, and laptop manufacturers do use them. They use Intel 600p and Samsung 960 Pro, among others on ultrabook-class systems where storage differences might contribute more than an hour of difference. Perhaps because they feel users think 8-10 hours is enough and want more performance.

 

[whm1974]

It can work, and laptop manufacturers do use them. They use Intel 600p and Samsung 960 Pro, among others on ultrabook-class systems where storage differences might contribute more than an hour of difference. Perhaps because they feel users think 8-10 hours is enough and want more performance.

I doubt I could noticed the performance difference between SATA and NVMe SSD with normal notebook use, and I rather have an hour or two of battery life anyway as that is more useful to me. Now that said, for portable workstations I can really see where the higher speeds of NVMe could be needed instead as those types of laptops are meant to be plugged in while being used.

 

[Yuriman]

Consider that load power for an SSD when compared to a HDD is misleading. A mechanical disk will use the most power when seeking. Maybe an NVMe SSD will use more power, but throughput may be 1,000-10,000x higher. With a non-continuous workload (e.g. read / transfer one file), the SSD will finish far sooner and quit consuming power.

 

[whm1974]

Consider that load power for an SSD when compared to a HDD is misleading. A mechanical disk will use the most power when seeking. Maybe an NVMe SSD will use more power, but throughput may be 1,000-10,000x higher. With a non-continuous workload (e.g. read / transfer one file), the SSD will finish far sooner and quit consuming power.

You do have a very valid point there. But I'm thinking that HDDs shouldn't be used in laptops anyway, for both reliability and performance reasons.

 

[Charlie22911]

What about running the nvme drives in SATA mode? Surely that would have a positive effect on power consumption?

 

[UsandThem]

What about running the nvme drives in SATA mode? Surely that would have a positive effect on power consumption?

NVMe and SATA are completely different beasts (a NVMe drive can't run as SATA, and a SATA drive can't run as NVMe).

 

[Charlie22911]

NVMe and SATA are completely different beasts (a NVMe drive can't run as SATA, and a SATA drive can't run as NVMe).

Oh I see. I’d assumed that they could do both since I can set either mode in bios, now I see that it depends on the drive too... learn something new everyday, thanks!

 

[UsandThem]

Oh I see. I’d assumed that they could do both since I can set either mode in bios, now I see that it depends on the drive too... learn something new everyday, thanks!

No problem. A lot of people get them confused since both generally work in a M.2 slot. A SATA drive will go through the SATA controller (and often disable some on the motherboard's SATA connections). A SATA drive is hust that regardless if it's plugged into a M.2 slot or into a SATA port.

A NVMe drive is just another way to say PCI-E, and that uses PCIe lanes much like a video card does. There's more variables than how I just summed it up (like through the CPU or through a controller), but it's the easiest way to remember the difference.

 

[whm1974]

So if I replace the 5400rpm HDD in my T430 with an SSD will I see an increase in battery life, and by how much?

 

[IntelUser2000]

Consider that load power for an SSD when compared to a HDD is misleading.

That's why we need battery life tests, not power tests.

Maybe an NVMe SSD will use more power, but throughput may be 1,000-10,000x higher. With a non-continuous workload (e.g. read / transfer one file), the SSD will finish far sooner and quit consuming power.

The number is an exaggeration. That's assuming worst case scenario for HDDs and best case for SSDs.

In actual modern day systems, OS does a pretty good job of using RAM to place most of the data on repetitive tasks which will be common with average folks using their computers. It allows the HDD to be powered down.

Decades of programming and mindset using HDDs also mean coding is optimized for it. It's hard to find a scenario that's 10x faster. We do not have millisecond boot times, and microsecond program launch times with SSDs. Instead, we may get speed boost of 2-4x.

A properly designed SSD will do better than HDD, but as my original post has shown, there are SSDs that use more battery than HDDs.

So if I replace the 5400rpm HDD in my T430 with an SSD will I see an increase in battery life, and by how much?

Maybe 30 minutes to an hour, if you were getting 10+ hours already. That's only applicable on a bursty usage low load scenario like web browsing and video watching. At higher load it would be negligible. It would also depend on the SSDs you get.

 

[UsandThem]

So if I replace the 5400rpm HDD in my T430 with an SSD will I see an increase in battery life, and by how much?

I replaced the HDD in my Dell Inspiron 7559 with a SSD, and I get about an extra 20-30 minutes of battery life. It differs with every laptop depending on the CPU/GPU etc. My Dell has an Nvidia 960m and a i7-6700HQ so not the most battery sipping laptop out there.

 

[whm1974]

I replaced the HDD in my Dell Inspiron 7559 with a SSD, and I get about an extra 20-30 minutes of battery life. It differs with every laptop depending on the CPU/GPU etc. My Dell has an Nvidia 960m and a i7-6700HQ so not the most battery sipping laptop out there.

Thanks, I'm planning getting a SDD since using a slow 5400rpm HDD is extremely annoying. I was wondering if I would see an increase in battery life.

 

[UsandThem]

Thanks, I'm planning getting a SDD since using a slow 5400rpm HDD is extremely annoying. I was wondering if I would see an increase in battery life.

You will, but the biggest benefit is the much better system responsiveness. We have 4 laptops, and I replaced all the hard drives in them. It's almost painful for me to use a system with a regular hard drive, especially on a laptop since the processors are already slower than their desktop counterparts.

 

[IntelUser2000]

You will, but the biggest benefit is the much better system responsiveness.

Yea, hard drives cap the performance of rest of the system. At some point, your CPU/RAM doesn't scale because the drive is holding it back.

 

[cbn]

Looking at the hard drive results:

I wonder how newer hard drives would compare?

 

[Arushi]

SSDs have better battery life than HDDs