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

 


Forgot your password?
Close
typodupeerror
×
Intel

Intel Demos Lightning Fast 13.8 GBps PCIe 5.0 SSD with Alder Lake (tomshardware.com) 40

Posted by msmash from the moving-forward dept.
Intel has demonstrated how its Core i9-12900K Alder Lake processor can work with Samsung's recently announced PM1743 PCIe 5.0 x4 SSD. The result is as astonishing as it is predictable: the platform demonstrated approximately 13.8 GBps throughput in the IOMeter benchmark. From a report: Intel planned to show the demo at CES, however, the company is no longer going in person. So, Ryan Shrout, Intel's chief performance strategist, decided to share the demo publicly via Twitter. The system used for the demonstration included a Core i9-12900K processor, an Asus Z690 motherboard and an EVGA GeForce RTX 3080 graphics board. Intel hooked up Samsung's PM1743 SSD using a special PCIe 5.0 interposer card and the drive certainly did not disappoint. From a practical standpoint, 13.8 GBps may be overkill for regular desktop users, but for those who need to load huge games, work with large 8K video files or ultra-high-resolution images will appreciate the added performance. However, there is a small catch with this demo. Apparently, Samsung will be among the first to ship its PM1743 PCIe 5.0 drives, which is why Intel decided to use this SSD for the demonstration. But Samsung's PM1743-series is aimed at enterprises, so it will be available in a 2.5-inch/15mm with dual-port support and new-generation E3.S (76 Ã-- 112.75 Ã-- 7.5mm) form-factors, so it is not aimed at desktops (and Intel admits that).
This discussion has been archived. No new comments can be posted.

Intel Demos Lightning Fast 13.8 GBps PCIe 5.0 SSD with Alder Lake More

Intel Demos Lightning Fast 13.8 GBps PCIe 5.0 SSD with Alder Lake

Comments Filter:

  • Yeah? At what temperature? (Score:4, Funny)

    by fustakrakich ( 1673220 ) on Friday December 31, 2021 @02:23PM (#62131617) Journal

    Gonna need some cryogenics for this one

  • apple is in trouble with locked & over priced ssd's.
    While stuff like this will work on more system and give end users more control

    • As many will be quick to point out, Apple users don't buy Apple products based on specs. The only time an Apple fan will know or care about a spec is when it trounces a non-Apple product and/or Apple's marketing team gives the spec a fancy adjective, like "Retina".

    • apple is in trouble with locked & over priced ssd's.
      While stuff like this will work on more system and give end users more control

      Apple's competition is the whole reason why Intel released Alder Lake. This is how Apple not only gives us faster Macs, but faster PCs.

      • Re: (Score:2)

        by Luckyo ( 1726890 )

        Apple is utterly irrelevant in this one. Alder Lake is because AMD has been crushing Intel over last couple of years in CPU market, selling basically everything they can make even before the current problems with semiconductors.

        Apple has nothing to do with this. It's not even in the high performance general purpose CPU market.

    • What is stopping Apple from using this in their next line of Macbooks and iPads? They are a PCI-SIG member and can promise any NAND manufacturer millions of units orders and this is a great selling point for a new generation. I imagine PCIe 5.0 is already in the requirement list for their "M2" or whatever it will be called next gen CPU.

      • Re: (Score:2)

        by Entrope ( 68843 )

        How many Macs have PCI slots now? Why would Apple care about PCI for Macs that don't have expansion slots? They could use a custom flash controller, maybe even on the SoC, and users wouldn't see a difference except maybe in performance.

        • Re: (Score:1)

          by Anonymous Coward

          Many people are unaware that Thunderbolt [wikipedia.org] is PCIe in an external cable form factor -- first Mini DisplayPort, now USB-C. Nearly all Macs since 2011 have Thunderbolt ports.

  • Would love to know how they managed to control the jitter of 13.8GHz PLLs.

    • Would love to know how they managed to control the jitter of 13.8GHz PLLs.

      They switched to decaffeinated coffee and herbal tea...

    • Re: (Score:2)

      by gweihir ( 88907 )

      it is 4x PCI-E 5.0, i.e. more like 4GHz.
      As to PCI-E 5.0, the spec are here https://pcisig.com/specificati... [pcisig.com]
      Seems to require membership for download though, so not actually an open spec.

      • PCIe Gen 5 is 32.0 Gbps per lane with 128/130 but encoding. The 4x refers to the number of lanes that the card supports. So 32.0 Gbps * 4 lanes / 8 bits * 128 / 130 gets a raw unidirectional bandwidth of 15.754 GBps. So getting realized transfers of 13.8 GBps is really efficient.
    • Its not that bad. We built some 12.8 Gbit links (JESD). The biggest problem was the PC board layout and especially connectors. We've been able to get ~1ps single shot timing jitter with discrete logic chips for many years now, and better recently. I'm sure it was serious engineering to integrate that onto a chpset for PCIE, but its not beyond state of the art in general. (I think single shot trigger noise down to ~100 femtoseconds is possible now). (we regularly lock lasers to
      The PLLs need to be good,

    • Re: (Score:2)

      by Agripa ( 139780 )

      Would love to know how they managed to control the jitter of 13.8GHz PLLs.

      Decoding each serial data stream relies on recovering the clock rather than a separate source synchronous clock, which is the reason serial data buses can be faster than parallel data buses. The disadvantages include higher complexity and higher latency while decoding. It only became practical with higher integration.

  • Intels marketing people are doing a nice job (Score:4, Interesting)

    by ITRambo ( 1467509 ) on Friday December 31, 2021 @02:57PM (#62131703)
    There is nothing practical about this, other than to gain mindshare on Intel's Alder Lake processors. That's smart marketing. Intel has done a great job in that respect. Within a couple of years when they have additional access to newer fabs, including their own new ones, I expect them to pull ahead of AMD in power use/performance efficiency. They have the money to seriously do a much better job and a CEO that knows that he's doing, unlike the idiots they had as CEO's for the past six years or so.
    • That is such wishful thinking it is ridiculous.
    • The desktop utility is non-obvious at this point; people who buy really expensive storage devices or very fast NICs or other demanding peripherals usually tend to be the ones who also want ECC support; but I suspect that Intelâ(TM)s server people want this to happen as fast as possible: faster PCIe helps do a lot to justify buying optane based SSDs and expensive NICs(one of which is an Intel exclusive, the other they will happily supply). PCIe 5 is also the basis for CXL, which Intel was on board wit

    • There is nothing practical about this

      I once heard there's no point in getting cable because a 56k modem was more than enough to use the internet.

  • Samsung really doesn't like it when consumers buy its enterprise SSDs, but they do offer pretty good value if you want power loss protection any way. Bought a pm9a3 for that reason.

    The current outrageous GPU prices don't seem to stop people buying high end PCs. Enthusiast motherboard manufacturers should just start supporting these enterprise SSD form factors.

    • Itâ(TM)s not ideal to have to use an adapter (and some can be surprisingly expensive, even for PCIe 3; Iâ(TM)d expect that the premium for 5 is still pretty stiff); but at least they didnâ(TM)t pull any irksome segmentation games like we had to deal with with SAS vs. SATA. Even downright antique boards, from the days when PCIe was new enough that you had to update your WinXP install disk to a certain version so that PCIe GPUs wouldnâ(TM)t cause a crash will talk to an NVMe device with an

    • Samsung really doesn't like it when consumers buy its enterprise SSDs
      And why would that be the case?

    • Re: (Score:2)

      by Luckyo ( 1726890 )

      >The current outrageous GPU prices don't seem to stop people buying high end PCs.

      What are you basing this on? Because there's a very large body of anecdotes of people just straight up leaving the enthusiast PC building for other hobbies specifically because you can't get certain parts for any reasonable cost. Everything from being unable to sell certain things that miners don't actually need while being able to sell parts that they do need, to the enthusiast PC gaming video channels reporting unexpectedl

  • We need a new speed standard. The speed of light in vacuum is 300 million meters per second.
    Lightning is way slower at only about 30 meter per microsecond.

    • Converting lightning's speed to a data transfer rate is tricky enough to begin with. It's unclear whether one should count the speed of a lightning strike's leader at 61 km/s [archive.org], the return stroke at 100,000 km/s [zenodo.org], or average both. These are of course slower than light's vacuum speed at 300,000 km/s, but without any idea of lightning's bandwidth one can't make a reasonable estimate of the data transfer rate.

      A more reasonable comparison might be the proprietary Lightning connector, which can be said to have a da

  • So Intel had to do a "special interposer" in order to connect CPU PCIe lanes to the storage device. I suppose a lot of addle dolts will fall for this shit and expect similar performance when connected to the "shared" PCIe lanes -- performance which cannot be obtained except by switching to high performance CPU's from AMD. Or perhaps Intel will finally get around to having their PCIe actually attached to the CPU rather than being merely a shared serial port.

    • Intel would rather you buy an AMD CPU and an Intel SSD than nothing from.Intel at all, which without a better SSD is what currently makes sense.

    • So Intel had to do a "special interposer" in order to connect CPU PCIe lanes to the storage device.

      Of course they did. The drive doesn't come in M.2 format, it comes in PCIe.
      This means you need an interposer.

      I suppose a lot of addle dolts will fall for this shit and expect similar performance when connected to the "shared" PCIe lanes

      M.2 is usually connected to the CPU lanes.
      So, usually, are 1-2 ports on the motherboard.
      Anyone who's been in the SSD game for the last, oh I don't know, 5 years now, understands the difference between CPU and Northbridge lanes.

      performance which cannot be obtained except by switching to high performance CPU's from AMD.

      Wait, what?
      What the fuck are you talking about here?
      Even DMI-fed lanes are oodles faster than the PCIe 3 lanes on all extant AMD APUs.
      AMD APUs work the same way as curren

  • When you're seeing plaid, you know that you're there.

    https://www.youtube.com/watch?... [youtube.com]

    All kidding aside, I have three SATA SSDs, a PCIe 3.0 NVMe drive, and a PCIe 4.0 NVMe drive. The extra speed is nice when it's being utilized, but I rarely appreciate the advantage of the greater transfer speeds of my two faster drives, as compared to the slower solid state drives.

    They are all mostly used for storage, my fastest is for Windows, and when launching videos or games they perform about the same.

    While I have zer

  • Huh? (Score:3)

    by nbritton ( 823086 ) on Saturday January 01, 2022 @06:24PM (#62134709)

    Huh, did they typo the abstract, I could build an array of spinning rust that could sustain 25GBps random I/O at 500k IOPs using antiquated SAS2 6Gbps PCIe 3.0 technology. Not impressed.

Slashdot Top Deals

Top Ten Things Overheard At The ANSI C Draft Committee Meetings: (10) Sorry, but that's too useful.

Close