I am honestly not quite sure about the usefulness of this metric in actual laptop use.
Unless you are doing very io heavy things, how fast it enters an idle state and how low the idle states go is much more interesting there and quite a bit harder to test.
It is about the accumulation of these low power states that makes the difference.
A colleague had to replace his laptop NVME SSD with a Samsung 990 Pro (gen 4 - even though his laptop does not have pcie gen 4) and he practically lost 1h30 of battery time.
And that is what I want to check out. We “moan” here about the power utilization of many components - but didn’t have a big view on the disks here.
I have P41 running in my FW13 12th gen Intel, and it has been everything the P31 foreshadowed. My P31 was running in my previous daily driver a Lenovo Thinkpad T480. The P31 had replaced a Samsung 970 EVO Plus and that laptop received roughly an extra 45-75 minutes of runtime on battery depending on wor load.
The P41 in my FW13 while I have no direct comparison to another drive undoubtedly contributes to my great battery life. I always get better than 9.5hrs usually averaging 10.5hrs on battery. If I were buying today I would actually probably go with the P44 Solidigm which appears to be their new consumer/prosumer line (SKHynix+Intel’s old SSD division whihc was bought by SKHynix) which will apparently target the non OEM market.
I hear you on this one. Even though the P44 seems faster, the P41 seem to have still lower power usage. Check the power usage section below
Still not sure which version I’ll take though :} P41 or P44 … The 41 is currently way cheaper :}
When I’ll have the mail from Framework, I’ll check out which one to take. Will probably a “feeling” matter
While it does have a lower power consumption while active, it does have to stay active a lot longer to shovel your data from the slc cache into the actual qlc storage so it very well may consume a lot more wh per gb written. Read on the other hand should not make a huge difference.
Still though, as long as you don’t write a lot, the ssd is not going to make a huge difference as long as the low power states work and are used aggressively and for the price the p41s are pretty appealing if I hadn’t gotten an amazing deal on a 980 pro.
I don’t understand where you got QLC storage out of all that. Like most nvme of recent vintage it uses TLC. We are also refering to the Solidigm P44 and the SKHynix P41 not the Solidigm P41 Plus.
How would one test this with some degree of repeatability without having specialized hardware? I got a P41 Platinum in my FW13 and had a SN850X in there for some time beforehand (now in my desktop). No discernable difference to me.
With the degree of difference the power configuration of the OS (and battery state and power supply used and display brightness and …) makes I doubt that the SSDs power use can be easily pinned down when not controlling for a lot of variables during testing.
With normal laptop use there should not be all that much disk access so unless a disk as no or broken apst you’d see very little difference between drives.
Yeah you’d need to do a lot of variable controlling there. Testing sequential and random reads/writes is relatively easy (even that needs to control for caching behaviors and stuff if you want actual data though).
On my linux laptops I always fine-tune the power management of the entire system when under linux.
Under Windows, there is not really much I can do. Windows just does its thing.
On my Latitude 7400 (i7 8665U, 32Gb Ram, Samsung 970 Pro 1TB NVme SSD, 7950mAh battery), I get around 6 hours under Windows and around 9h30 under Linux. It is just that I can control what is started under Linux (services, adapt the scheduler, disk accesses when under battery), and don’t want to bother with the “company” software running on it while under windows.
I’ll second everything he just said, except he’s had better luck with Windows than I have. My laptops also get 9.5hrs on linux but usually under 2 on Windows.
Under windows, I get the 6 hours when I just browse and there are no updates etc. coming in.
Note also, that inside my LAN, I’m using a predecessor of piHole, advertisements etc. don’t get through. So that limits a lot what Windows does in the background.
UPDATE: And I have a dedicated rpz rule disabling all known Microsoft telemetry links:
The majority of my drain comes from the hybrid GPU turning on for no reason, even after spending hours working on it. It’s part of why I ditched windows. When the GPU worked as expected I would get 4-6 hours.
Good catch. Will check that when I get hands on mine :}
I eventually will put SteamOS on one partition for the games.
But for Star-Citizen, even though they should bring out a Linux port, I may need windows.
You can run Star Citizen on Linux, Lutris makes the install easy enough. There’s also a helper script, but I don’t think it’s necessary anymore.
You can also skip Lutris and install directly through steamos via proton, you just need to add a launch parameter that sets the game name to star-citizen so that proton knows to turn on the work around they implemented in proton 8ish
Throwing my 2C into the thread: I’m in the same scenario as Magic, except I’ve got my P41 in an enclosure so I can use it to hold games for my current laptop while I wait for the FW16. The enclosure is only USB 3.2 10Gbps, but the drive saturates that and stays there pretty much constantly, so I have high hopes for it as a boot drive. I’d say it’ll pair well with a nice new laptop like ours!