[NSW, ACT] PNY CS2241 2TB PCIe Gen 4 NVMe M.2 2280 SSD $125 C&C / In-Store Only @ Umart & MSY

Related Stores

Umart

MSY Technology

• 

  Mostly Fanless on 10/08/2023 - 00:14

  Wonder if I'm missing something, says 145 for me.

  • 

    FDA2021 on 10/08/2023 - 00:26

    +2

    I think it's only for Click and Collect, $125 shows up when I choose a pickup location

    • 

      Mostly Fanless on 10/08/2023 - 00:30

      +1

      Ohhh, right, thanks :)

• 

  Clear on 10/08/2023 - 00:17

  Available in TAS and VIC for $125 too. At least on the MSY side.

• 

  johnkash on 10/08/2023 - 00:37

  This deal is not available in SA and WA

• 

  aaronsd on 10/08/2023 - 09:36

  Getting so excited seeing these NVME drive prices go down in general, it'll be great to see the 4TB drives come down in value as well. Fingers crossed for a cheap sub $200 4TB NVME drive for games/software that I don't mind failing on me.

  • 

    Associatedpearcekk @ MSY Technology on 10/08/2023 - 11:13

    Lexar: Hold my beer

    • 

      Look Up on 10/08/2023 - 11:31

      Are you associated with both BPC and MSY/Umart?

      • 

        Associatedpearcekk @ MSY Technology on 10/08/2023 - 12:50

        +3

        I used to work for both BPC and MSY/Umart

• 

  combatant on 10/08/2023 - 19:00

  At this price I'd be getting the Lexar deal as it's TLC, better endurance and read/write speeds. Wouldn't even consider this given the prices are within a couple of $.

  https://www.ozbargain.com.au/node/792769

  • 

    Look Up on 10/08/2023 - 21:48

    +1

    At this price I'd be getting the Lexar deal as it's TLC, better endurance and read/write speeds. Wouldn't even consider this given the prices are within a couple of $.

    Drives with the Maxio controller like the NM790 are optimised for PS5 marketing, not real world performance

    Direct comparison of the TLC NM790 vs QLC CS2241
Lexar poor latency = 2x slower on reads, 3x slower on writes
Lexar poor non-sequential write speed = how garbage is the controller when a "6500MB/s" TLC drive has write performance 25 ~ 40% slower than a "4200MB/s" QLC drive

    Garbage 4K random write performance
The Gen 3 WD SN570 smashes the Gen 4 Maxio drives (Netac NV7000-T, Acer Predator GM7) but is only rated at "3500 MB/s" read/write

    Read/write latency is slower than a Crucial MX500 SATA drive
4K random read
4K random write

    Drives currently have major issues being used in Linux and by extension, all mainstream NAS platforms
For a mainstream controller vendor to prioritise testing on PS5 versus Linux says it all

    The NM790 is exactly like the Kingston NV2: good for gaming storage, ghetto PS5 setups (add cheap heat sink) and storing any data that is safe to lose

    • 

      cr4pm4n on 12/08/2023 - 00:34

      Wish I would've seen this on the original thread. I probably would've held out longer instead of buying.

      However, one thing I don't understand is how the random read/write test look so good in reviews of the NM790? Is it the queue depth that they test for which causes this difference (Compared to those you've linked)? How do the different queue depth tests reflect in real-world performance?

      Also at the end, what do you mean by 'data that is safe to lose'? How can the drives have such highly rated TBW then?

      Not trying to come off as rude at all. It's just kinda hard to know what to look out for when it comes to these newer NVMe drives.

      • 

        netsurfer on 12/08/2023 - 01:57

        +1

        NM790's random read performance is good at low queue depth (and even at higher queue depth, beats CS2241). Random write is NM790's weakness. Technically, that's shrewd because most real world application testing don't depend heavily on that, except for consistency tests (which need to run for long period of time so most reviewers don't run those).

        Data that's safe to lose is a way of saying the NAND used on that SSD and/or the components used on the SSD is cheap. However, QLC NANDs have less writing cycles available and harder to store data reliably compared to TLC. It's brave to rave about QLC being more reliable. QLC SSDs use really aggressive SLC cache (to hide QLC write being slow) but that has an issue, the data has to be written as QLC eventually so double writes can be quite common.

        NM790, on paper, appears to perform better than CS2241. ChatGPT managed to find a lot of useful information on the weak side of NM790. Problem is, NM790 does well in majority of the benchmark tests, especially ones most people care about.

      • 

        Nom on 12/08/2023 - 20:38

        Also at the end, what do you mean by 'data that is safe to lose'?

        ANY drive can die at any time - even the most reliable drives are not immune from failures.

        For this reason, you always need backups no matter what drive you choose - so don't spend time worrying how reliable your drive is. It just doesn't matter - the worst case scenario is that you need to restore your backup 👍

        Not trying to come off as rude at all. It's just kinda hard to know what to look out for when it comes to these newer NVMe drives.

        For normal desktop use, you don't really need to look for anything - all modern drives are plenty fast for normal use. Drive specifics only become important for some very specific workloads - for things like windows boot time then there's barely any difference between the fastest and slowest drives.

    • 

      netsurfer on 12/08/2023 - 01:22

      Let's really look at the results:

      • 4K random read, NM790 wins. That's a problem for CS2241 because for general public, 4K random read matters more.
      • 4K latency issue. While that's an issue, majority of the time, most apps people use don't heavily depend on that. A simple way to explain this is that DDR5 RAM modules generally have higher latency than DDR4 so does that mean DDR5 memory modules are trash?
      • The ugly side of QLC, where the write speed is about 100MB/s. This is not something that can be ignored, not for a 2TB SSD.
      • NV2 being trash, while I agree, the problem is TechPower reviewed Phison E21 version and for some reason, I cannot get the same good result on my NV2 Phison E21 version. That's an issue because if you looked at TechPower's results, you get better than expected NV2 results. Good luck getting a NV2 with Phison E21 in 2023.

      If we trash NV2, then unfortunately, we should trash CS2241 also. Phison E21 with QLC and then use SLC cache results to cheat.

      Silicon Power's equivalent of CS2241 is a piece of junk according to you, but PNY CS2241 somehow (with the same components) is super duper. That makes no sense. I've been on the receiving end of a recent PNY component swap (TLC to QLC and it is also Micron QLC) so PNY is following SP too.

      PS5 "compatible". Unfortunately, whether we like it or not, NM790 meets Sony's recommendation. CS2241 does not.

    • 

      netsurfer on 12/08/2023 - 02:44

      From Lexar rep's trash talk, he hinted the QLC version is coming too AND Lexar will play the same game CS2241 is doing. Basically same sequential read/write as the TLC version with random read/write QD1 figures.

    • 

      netsurfer on 12/08/2023 - 09:52

      Also, do bear in mind, if CS2241 QLC's max sequential write when SLC cache is all used up is 100MB/s then its max 4K random write cannot possibly exceeds that.

      That means QLC, without SLC cache, will suffer a significant drop in 4K random write as well. If you really care about high queue depth 4K random write, then clearly QLC is a big no no. With reviewers consistently run most benchmark tests with empty SSD (Crystal DiskMark, AS SSD), we, consumers, indirectly have the best case figures in mind.