Slashdot is powered by your submissions, so send in your scoop

 


Forgot your password?
Close
typodupeerror
×
Data Storage Intel

Intel's New Optane SSD P5800X Is the Fastest SSD Drive Ever Made (hothardware.com) 24

Posted by BeauHD from the off-the-charts dept.
MojoKid writes: Intel recently shifted its storage strategy somewhat and is now catering its flagship Optane SSD P5800X, which was formerly targeted solely at data centers, to workstation users. The Optane SSD P5800X is based on a proprietary PCIe Gen 4x4 native controller and it features Intel's second-generation Intel Optane memory. In terms of performance, in some of the first benchmark numbers to hit the web, the drive is an absolute beast in the workloads that matter most for the vast majority of workstation users and enthusiasts. Random reads and writes are exceptionally good and access times at low queue depths are best-of-class. The Optane SSD P5800X's sequential transfers, while strong, aren't quite on the same level as some of today's fastest NAND-based PCIe 4 solid state drives, but they do exceed 7GB/s, which is still extremely fast. Overall, it's essentially the fastest SSD ever made. Endurance is off the charts too. However, all of that SSD horsepower comes at a price though, at a little over $2.50 per Gig and over $2,000 for an 800GB drive. With capacities of 400GB, 800GB and 1.6TB, the new Intel Optane SSD P5800X is shipping and available now.
This discussion has been archived. No new comments can be posted.

Intel's New Optane SSD P5800X Is the Fastest SSD Drive Ever Made More

Intel's New Optane SSD P5800X Is the Fastest SSD Drive Ever Made

Comments Filter:

  • Addicted to speed. (Score:3)

    by Ostracus ( 1354233 ) on Wednesday June 30, 2021 @08:51PM (#61539050) Journal

    Just throw that into the fastest [youtu.be] computer and have some fun.

  • Not even fastest according to the article? (Score:5, Insightful)

    by niftydude ( 1745144 ) on Wednesday June 30, 2021 @09:23PM (#61539130)
    Am I going crazy, or is this sort of gaslighting the new standard for slashvertisements?

    According to the benchmarks I'm looking at in TFA the drive is slower than the drives it was compared to in most situations.

    The only time it had a slight edge was when there was a low number of concurrent IOs, But that advantage dropped off quickly once you had more than four IO operations running.

    Which with the multi-threaded operating systems we've had for the last couple of decades is something that occurs literally all the time.

    This article is a new low.

    • I have the 905p Optane, 480 GB unit in my daily driver. Sure, there is a price premium. I got mine on a special from NewEgg for about $420 I think. Consumer line was canceled a year ago maybe? Micron bailed, too.

      The Optane does have some faster specs on some things, but with a bunch of RAM cache it is hard to actually see it shine, ever..

      High endurance and low latency makes it ideal for DB temp, system swap, but again, RAM should be maxed out first.

      Flash killed SCSI 15k boot drives, and this is better,

      • Re:Not even fastest according to the article? (Score:4, Interesting)

        by Rockoon ( 1252108 ) on Thursday July 01, 2021 @02:38AM (#61539612)
        I think exactly this about all the "high end" SSD's really.

        The biggest improvement SSD's give is the effective 0 latency compared to a spinning platter (I realize its not 0 latency but compared to 16ms drive seek times...), which translates into a thousand-fold improvement in random i/o...

        Going "high end" doesnt really improve the situation all that much beyond that. You may think level loading in that game will fly if you give it significantly more bandwidth than a regular SATA3 SSD, but it wont in practice, because the games loading routines, which nearly always include significant parsing (xml and other insane idiocy forced into game design by fucking retards), already struggle to keep up with the 600MB/s being thrown at it by a SATA SSD.

        An extra $300 in ram will get you more meaningful performance improvement than a "high end" 256GB SSD, and thats just about how much you can expect to save by going the less speedy 256GB SSD route.

        • Re:Not even fastest according to the article? (Score:4, Interesting)

          by AmiMoJo ( 196126 ) on Thursday July 01, 2021 @04:24AM (#61539742) Homepage Journal

          Windows 11 will support streaming from SSD direct to GPU memory, similar to the PS5. That should see a decent boost in game performance but also mean that game installs will get significantly larger due to the need for uncompressed, pre-processed data.

          We are getting to the point where the line between RAM and SSD is becoming less meaningful. Modern NVMe SSDs are already faster in raw throughput terms than RAM from the Core 2 era, say 15 years ago. While random access speeds are not quite there yet for some applications that is very significant.

    • Re: (Score:2)

      by AmiMoJo ( 196126 )

      Intel has been doing a lot of these paid exhibition articles lately. Their tech is kinda meh, CPUs have fallen behind AMD, GPUs are behind AMD and Nvidia, storage products have a poor reputation...

      We are seeing ever more desperate gimmicks from Intel, good numbers in one specific benchmark to mask overall crappiness and poor pricing.

    • They're not even that much faster than the ones designed by God [computer.org] nearly forty years ago in the X-MP.
    • What puzzles me is how the 1TB Samsung 980 Pro they tested shows lower performance than I get with either of my 500GB 980 Pros (one on CPU, one on chipset). And somehow, they show more bandwidth for the 512GB PCI gen3 970 pro than for the 1TB gen 4 980? That just isn't believable. NVMe drives get faster as they get bigger, and PCIe4 has what, twice the bandwidth of 3? No, something is wrong. Either I'm reading it completely wrong, or they screwed something up.

      • It looks like the 980pro switched out MLC nands for TLC, which makes it slower in real world demanding tasks. It will still burst faster than a 970pro due to a small amount of SLC storage tacked on however. See anand's coverage: https://www.anandtech.com/show... [anandtech.com]

        The trick often used is fewer higher density chips at the same total drive capacity, which can make a newer SSD in the same series weaker overall.

  • Primarily for the reputed endurance of the Optane Memory over the standard NAND memory. Being able to buy a drive with TBW measured in Petabytes would be very reassuring if you got into or were into something that did lots of writes to the SSD.
    • I fail to understand how its reasonable to imagine an SSD being written to so much.

      When you do the math, you find out that even if you were trying to kill an SSD by using up its write endurance, and you also did so theoretically optimally (such that the block erase time is the limiting factor), it would still an unreasonable amount of time unless your SSD is really small (the days of 32GB SSD's for instance)

      Its hard to imagine a situation where you accidentally achieved sustained optimal block erase rat

      • A quick google search suggest that you could buy a 500GB Optane drive (the 905P) for some $1500.
        Meanwhile, 4x64GB RDIMM (i.e. server) RAM cost about the same (again, quick google search).

        So, it all depends on platform. Is your platform able to run with 4x64GB? Or maybe 8x32? A "High Performance Desktop" processor might have 6 channels of RAM.

        All in all, if you've filled the memory on your computer, your only choice is another computer or the Optane. In this case, the $1500 price for some 500GB is cheaper th

  • If the price was closer to NAND, I'd always opt for Optane SSDs. I think everyone would. Instead, they price it ridiculously high, it flops completely in the consumer space, and Micron exits the space entirely. Is it really that much more expensive to manufacture? Or are they artificially holding it back so that it performs just a little better than NAND so they can milk the market for a long time? Judging by Micron losing $400 million and selling it's facility in Utah makes me think it's just expensiv

    • It is expensive to manufacture. They cannot vertically stack it to the same degree it is possible with VNAND.

  • It's almost like Intel didn't want money! (Score:3)

    by ZorinLynx ( 31751 ) on Wednesday June 30, 2021 @11:52PM (#61539376) Homepage

    Releasing a product like this and only selling it to datacenter customers is nuts. There's so many enthusiasts out there happy to pay $thousands for the fastest thing on the market. I'm glad they changed their mind.

  • At what cost of writes?

  • I've been following the creation and development of solid state drives since the very beginning reading all the in-depth articles on AnandTech.com and other websites.

    I still have a generation one Intel drive along with a generation 2 that is functioning but their generation 3 has failed on me.

    I noticed very quickly that after the initial creation of SLC memory then we had DLC and then MLC and now QLC got the regular SSD manufacturers like Samsung were completely happy with just doing evolutionary increases

  • Problem for Intel is that RAM is about 5 bucks per gigabyte. This is about 10x slower than RAM but only 2x cheaper. So if this is targeted at "workstation" users, then in that class a lot of these will have motherborads capable of terabyte or more RAM.

Slashdot Top Deals

To the systems programmer, users and applications serve only to provide a test load.

Close