SK hynix Gold P31 2TB PCIe NVMe Gen3 M.2 2280 SSD $328.76 + Delivery ($0 with Prime) @ Amazon US via AU

180

Affiliate

Go to Deal

Cashback

LNil Einne on 26/01/2022 - 13:17 amazon.com.au (343 clicks)
Last edited 26/01/2022 - 13:21 by 1 other user

This post contains affiliate links. OzBargain might earn commissions when you click through and make purchases. Please see this page for more information.

I'm in NZ but been looking at Amazon AU for a good SSD deal. Noticed the SK hynix Gold P31 2TB is at $328.76 which is a decent price, not far off the current US price once GST etc is taken into account. This isn't available me since it's Amazon Global (US) via Amazon AU but thought it might of interest here.

The P31 is a Gen 3 M.2 2280, but it's very popular especially for laptops due to excellent efficiency and performance, often considered the best current Gen 3 consumer drive. Uses Hynix own controller and 128-Layer TLC flash. Gen 3 means it's not compatible with the PS5. Has 5 year warranty and 1200 TBW but don't know what dealing with warranty with Amazon US will be like. Most popular alternative is the Samsung 970 EVO Plus which has similar performance but less efficient, currently $299.00 sold by Amazon AU but been cheaper last month.

The Gen 4 Phison E16 TLC drives which can sometimes be found at a similar price. These are a bit controversial, while they have better sequentials on Gen 4 systems than these Gen 3 drives and so show better performance in some synthetics, the chipset isn't considered that good since it's barely an upgrade from the E12 which wasn't considered that great. And their Toshiba 96 layer flash is also not as good. So some suggest good Gen 3 drives like the P31 especially but also the 970 Evo Plus are generally a better choice for most users performance wise. See my comments and links here https://www.ozbargain.com.au/comment/11667512/redir And personally I wouldn't choose a Gen 4 QLC drive over this unless it's quite a bit cheaper or for a PS5. The newer E18 and other good Gen 4 drives are still quite a bit dearer from what I've seen.

The 1TB and smaller P31s have been around for a while but the 2TB only came in September last year after a lots of delays. I think it might have been cheaper on Amazon.com before, but these are were coupon or real limited time deals which may not make it to Amazon AU via Global. It seems to take a while for Amazon to adjust the AU price as it's been at the US price for a while but it only recently changed on AU. (I've been watching it not just relying on CamelCamelCamel). You also cannot buy from Amazon.com to Australia, NZ or pretty much anywhere in the world. It's actually sold by SK hynix and although fulfilled by Amazon.com, they don't let them ship outside the US. SSD prices seem to be headed downward at the moment so might be cheaper in the future, but who knows?

Mods: As said mention I'm an overseas user. Not associated with Amazon or Sk Hynix in any way, have an account on Cheapies you can check out https://www.cheapies.nz/user/535 just to show I'm not a shill. Decided to try posting this directly since I believe my post is supposed to be auto-rejected if the store isn't whitelisted but apologies if I misunderstood.

Price History at CamelCamelCamel.

Related Stores

Amazon AU

Marketplace

Amazon Global Store

• 

  fastnet on 26/01/2022 - 13:28

  +3

  Why not get a Samsung 970 Evo Plus 2TB instead? They’re around $287 on eBay with current coupon codes.

  Even with the controller changes being discussed online they still offer top notch performance with negligible differences for 99% users.

  • 

    bono on 26/01/2022 - 14:31

    +5

    The Hynix P31 gold is meant to have by far the lowest power usage for equivalent performance drives, and I have the 2TB 970 Evo plus and it runs quite hot. So the Hynix may be cooler with slightly longer battery life for the laptop. Of course, DYOR!

    • 

      fastnet on 26/01/2022 - 17:10

      Yeah I agree. I have a 970 Evo Plus 2TB in a gaming laptop as the second SSD and it runs hot. I've installed a thermal pad and it still gets to 63 °C when transferring a lot of files. Idle temps are reasonable at around 41°C.

  • 

    Nil Einne on 27/01/2022 - 01:58

    Having looked in to it, I'd actually prefer prefer the newer controller and especially newer flash on 2TB, but AFAICT, most 2TBs are still older controller.
    (If you look online there's a lot of panic because of a packaging change. But when people check the part number or firmware, seems clear it's the old controller.) Anyway forgot to make this clear, but even with Amazon's price of $299 for the 970 Evo Plus (whichever controller), I'd choose it over the P31 because I plan to use in a desktop and I don't think the performance advantages are worth ~$30.

    For a laptop, it might be worth it, since wasn't my use case, never really looked into it well. Just found that even at a higher price the P31 was often suggested in the US by people who seemed to know what they were talking about. I'd imagine it depends on your laptop and usage patterns. E.g. a low power thin and light where you're mostly doing internet and office work it's more likely to be worth it then if you have a beast with fancy GPU and spend most of your time gaming even on battery, but this is a guess rather than based on any data.

    (Since Amazon is pretty much the only Australian store which will ship to me in NZ for a reasonable price, haven't been paying that much attention to non Amazon prices.)

• 

  garetz on 26/01/2022 - 13:51

  +1

  Overpriced for gen3

• 

  justtoreply on 26/01/2022 - 14:29

  +3

  Well, that's an informed post!

• 

  thlunasa on 26/01/2022 - 14:29

  +1

  I get that the power efficiency is fantastic, but how many minutes extra will I actually get on battery?
  I could not find any evidence like this online.
  Some forum posts say 2%, others say 10%.
  Worth it for sure if you get 10% more battery, but where is the evidence?

  • 

    shellshocked on 26/01/2022 - 14:37

    +5

    Have you looked at the Anandtech review of the P31 Gold?

    https://www.anandtech.com/show/16012/the-sk-hynix-gold-p31-s…

    I would mainly look at putting this in a laptop due to performance per watt.

    In relation to laptops and battery life, there was a two year review conducted by Tweak Town in 2019 which showed measurable differences in battery life, up to 1hr, of a laptop with different types of SSD. Unfortunately it hasn't been updated to include newer SSD's.

    • 

      thlunasa on 26/01/2022 - 16:12

      +2

      Hey, thanks for the information.
      So from the 2019 tt article we can see that the best nvme has a 20% difference from the worst nvme which is really surprising.

      Let's do some bad strawman math like this
      Intel 660p 62p 320min
      Adata sx8200 pro 82.6p 331min

      20 points = about 3% difference

      Hynix p31 172.2p = 90p delta = 13% difference over the sx8200 (!) = 376 minutes

      45 minutes (!) difference to a 5.5 hour battery is huge, that's 20 minutes difference to a 3-hour gaming laptop (!)

      Is that worth the $60-80 premium over the 970 Evo plus?… Maybe!

      It's paying 60-80 for a 2500mah battery bank, which would be horrendously expensive except they don't make battery banks in 300w so it's technically priceless

      • 

        Nil Einne on 27/01/2022 - 02:42

        I didn't make this clear enough in my post, I wouldn't get this even if it were available to me since the 970 Evo Plus is cheaper and I'm putting in a desktop. The minor performance advantages aren't worth the higher price IMO and that's without considering warranty risks. And thanks for the stats, since I'm not planning on putting it in a laptop, didn't look in to this aspect, just knew it was a popular choice/recommendation even at a higher price because of the efficiency.

    • 

      Danneo on 26/01/2022 - 16:32

      I currently have an unopened 970 Evo plus that I’m looking to put into my Lenovo Yoga 14” Evo, should I return the 970 for the henix?

      • 

        thlunasa on 27/01/2022 - 19:01

        Is a 13% (derived entirely from strawman math) battery boost worth the price difference compared to what you paid for the 970?

• 

  NotAnAudiophile on 26/01/2022 - 15:48

  +2

  Still waiting for cheap 2tb Gen 4 drives…

  • 

    Nil Einne on 27/01/2022 - 02:36

    +1

    FWIW, the Silicon Power US70 2TB (SP02KGBP44US7005) (Phison E16 + Toshiba 96L) is at a similar price on Amazon.com. Not available to AU unfortunately even from Amazon.com, and Amazon AU are charging a bit more. But it's available to me in NZ from Amazon.com for effectively same price as this would be. I was originally quite tempeted to get the US70, it has better sequentials (on a Gen 4 system) and sort of newer so thought it'll be better. Heck even better or the same on a gen 3 system. But having looked in to it (part of my research in the link above), I think I'd still chose the Evo Plus 970 since it's cheaper. I think I'd even chose the Evo Plus 970 or P31 at the same price. This is a more contentious point though and doesn't apply if you have a special use case e.g. PS5. If you mean the "true" Gen 4 drives e.g. Phison E18 etc then yeah I agree.

• 

  mikezillakind on 27/01/2022 - 00:09

  how Big dos a File have to Be before the write Speed Falls off A cliff ?

  • 

    Nil Einne on 27/01/2022 - 04:19

    +2

    It's complicated. The P31 uses a hybrid SLC cache (dynamic + static) so when empty something between 166 - 184GB https://www.techpowerup.com/review/hynix-gold-p31-2-tb/6.htm… https://www.tomshardware.com/reviews/sk-hynix-gold-p31-m2-nv… (not sure why Tom's Hardware & TechPowerUp got different values, even if TechPowerUp are using GiB it doesn't seem to completely account for the difference). Although it's for the 1TB, the Anandtech review might as be worth a look if you're interested. For the 2TB you can roughly double capacities https://www.anandtech.com/show/16012/the-sk-hynix-gold-p31-s…

    When not empty, it's more complicated. I assume the size of the dynamic SLC cache will depend on the empty space like all dynamic cache. (I couldn't find someone specifically commenting on this and I could be wrong since I didn't see any signs of a folding state in whole drive writes.) But also the dynamic SLC cache can take a while to recover. You can see in Tom's tests it didn't recover even after 30 minutes. The small 12GB static SLC cache recovers fast so you should have this again very quickly after stopping writing.

    While 12GB may seem tiny, performance even with direct to TLC is 1.4GB/s+ or a bit below half the SLC speeds. I.E. well above a SATA SSD let alone any HD. We aren't talking about a crappy tiny old QLC drive here, those can be slower than HDs. Also if you're using Windows, note Explorer and some other tools limit transfer speeds to 2GB/s per transfer anyway.

    BTW, massive dynamic SLC caches like the all drive ones used in the E16 devices may seem better at first glance. since when empty you can write 666GB (for a 2TB drive) at maximum speed. But t's complicated. If you do manage to fill the entire SLC cache, you now have to fold the this back to TLC, you can't just write direct to TLC, you've already filled all blocks (let's discount wear levelling etc). So probably worse performance when you exhaust the SLC cache and also means more write amplification. Further I think drive treat it more urgent to fold your SLC to TLC so they tend to be more aggressive which may have a small effect on reads (they will generally stop folding for reads), write amplification and energy consumption.

    If you're confused what folding is https://www.reddit.com/r/NewMaxx/wiki/basics might be worth a read. Ultimately just rememeber you're using a portion of the TLC as SLC so this may need to be "converted" back to TLC. If it's static, it's always treated as SLC so you just write it out to TLC at some stage or keep it as SLC, but not if its dynamic. There are a lot of considerations which go into folding the dynamic cache SLC back into TLC hence I think why the P31 is so slow at recovering while others do it faster. Ultimately it's a series of trade-offs depending on the expected workload and usage pattern.