Lexar hits 7 GB/s with new M.2 PCIe 4.0 SDD
Click here to post a comment for Lexar hits 7 GB/s with new M.2 PCIe 4.0 SDD on our message forum
pegasus1
And for free.
Something new being faster is a selling point, and also not everybody goes from the last gen to the next gen, some jump several generations, i went from a 2700k (5ghz) to a 3800x and noticed a huge improvement. If id have upgraded several times instead i doubt the perceived improvement would have been as great.
Faster also sells, and that's the whole point.
nosirrahx
FYI, not a college grad, self taught IT career.
I mentioned it above but I will mention it again. Your PC spends almost all of its IO time working with small random files so speed in the 4KQ1T1 metric is what you feel.
From a marketing perspective though this is tough as 300meg per second is insanely fast for 4KQ1T1, yet the number is smaller than the advertised 3000 to 6000 megs per second sequential speed.
The average Joe/Jane sees a huge number and buys the drive never knowing that this number applies to .1% of their use case.
In the car world, this is like HP, top speed and lap times. Massive HP and insane top speed gets attention even though a small lap time actually better reflects how the car drives and handles.
Need does not dictate as much as you think.
nizzen
My 3x Optane 900p In raid-0
Everything else feels so slow 😛
https://www.diskusjon.no/uploads/monthly_01_2018/post-42975-0-17312900-1516234836.png
Happy christmas 😀
nosirrahx
I do something similar with 4 905Ps in my workstation. You have me beat on 4K.
https://i.imgur.com/MCa857u.jpg
wavetrex
That's not the same CDM version... the results cannot be compared, the author completely changed the algorithm.
Here's another with 6.0.2:
https://i.imgur.com/QXRJr4M.png
nizzen
You are aparently benching RAM. Using cache program?
No problem comoaring 2 different versions. Close to no difference on the same computer 🙂
wavetrex
Guilty 🙂
Which is also why I'm in no hurry to upgrade the "old" SATA SSD, because other than very large linear transfers (which very rarely happen), just 4 GB of RAM caching pretty much makes everything instantaneous.
http://www.romexsoftware.com/en-us/primo-cache/index.html
nizzen
"Everything"
You mean data in cache...
It's a reason we use Optane and not software cache program. Reading data for the first time in one session is slow with cache, but fast with Optane 😉
Cacheprograms make benchmarks fast, and that's pretty much it. Maybe faster loading of often used programs and windows startup.
wavetrex
... but happens in background after booting, with the feature called "prefetch"
It can be slow, but it contains the most useful things which I've used during last session.
CPU caches, GPU caches, OS caches (RAM), Flash caches in SSHD, RAM cache in higher-end SSDs... they all exist and work just fine for a reason:
A large chunk of the data that is being exchanged is reused over and over again in normal usage patterns.
Caches work and engineers have figured this out a long, long time ago.
Obviously the effects of RAM caching of SSD/HDD can't be directly measured, as shown by those ridiculously high numbers, which don't reflect reality.... but the effect is noticeable enough to worth using it.
I'm not a professional video editor, I don't need gigantic amounts of data being moved all the time from the storage... for me even the little RAM cache has enormous effects in speed in all general use.
Anyway, wrong topic to discuss all this...
nosirrahx
Does not work on boot and if you do very heavy work you need that RAM. Handing a pile off to persistent cache ends up paging more, turns into an IO mess.
I experimented with RAM cache and it resulted in very inconsistent performance, boot times were obviously just at the speed of the drive.
Optane gives you consistently fast everything, can't even tell what is and is not cached into RAM currently.
