Stories
Slash Boxes
Comments

SoylentNews is people

posted by janrinok on Thursday June 23, @03:42AM   Printer-friendly
from the PCI-has-the-zoomies dept.

PCI-SIG Announces PCI Express 7.0 Specification to Reach 128 GT/s

PCI-SIG today announced that the PCI Express (PCIe) 7.0 specification will double the data rate to 128 GT/s [gigatransfers per second] and is targeted for release to members in 2025.

PCI-SIG technical workgroups will be developing the PCIe 7.0 specification with the following feature goals:

  • Delivering 128 GT/s raw bit rate and up to 512 GB/s bi-directionally via x16 configuration
  • Utilizing PAM4 (Pulse Amplitude Modulation with 4 levels) signaling
  • Focusing on the channel parameters and reach
  • Continuing to deliver the low-latency and high-reliability targets
  • Improving power efficiency
  • Maintaining backwards compatibility with all previous generations of PCIe technology

Also at Phoronix and Tom's Hardware.


Original Submission

This discussion has been archived. No new comments can be posted.
Display Options Threshold/Breakthrough Mark All as Read Mark All as Unread
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
(1)
  • (Score: 0, Troll) by Anonymous Coward on Thursday June 23, @04:45AM (3 children)

    by Anonymous Coward on Thursday June 23, @04:45AM (#1255526)

    Ok... are there any consumer applications for this?

    • (Score: 1, Informative) by Anonymous Coward on Thursday June 23, @04:55AM

      by Anonymous Coward on Thursday June 23, @04:55AM (#1255528)

      Possibly not. GPUs are getting to where they don't need the bandwidth that's already available, and it seems like SSDs are nearing that point as well.

      It might be that these high speed interconnects end up only being useful in data centers. But, if the bandwidth is available, maybe someone will find a use for it that no one has thought of yet.

    • (Score: 2) by takyon on Thursday June 23, @03:45PM (1 child)

      by takyon (881) <takyonNO@SPAMsoylentnews.org> on Thursday June 23, @03:45PM (#1255603) Journal

      https://www.pcworld.com/article/704893/the-first-pcie-gen-5-0-ssds-are-almost-ready-for-consumers.html [pcworld.com]
      https://arstechnica.com/gadgets/2022/02/pcie-5-0-ssds-promising-up-to-14gb-s-of-bandwidth-will-be-ready-in-2024/ [arstechnica.com]

      PCIe 5.0 support will spread fast in the next year on the latest Intel and AMD (AM5) motherboards. SSDs will be the first component to make use of it, and won't be too far from maxing out PCIe 5.0 x4 (15.754 GB/s).

      It could take a long time for PCIe 7.0 to come to consumers, but for example PCIe 7.0 x1 should allow for up to 121 Gbps for an Ethernet NIC with 1x 100 Gbps and some other ports.

      SSDs are going to start to have cooling problems, and motherboards with PCIe 6.0 and above could need expensive design changes. PCIe 5.0 seems to be well handled on AMD's side by using dual chipsets so signals don't have to travel as far.

      --
      [SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
      • (Score: 0) by Anonymous Coward on Thursday June 23, @08:16PM

        by Anonymous Coward on Thursday June 23, @08:16PM (#1255672)

        > x1 should allow for up to 121 Gbps

        I wonder if the processors can keep up (yeah, DMA is a thing, but there is still a lot that has to go through the processor).

        I guess the low power ARM stuff can forever stick with pcie gen 3 and gen 2.

  • (Score: 0) by Anonymous Coward on Thursday June 23, @07:59AM (5 children)

    by Anonymous Coward on Thursday June 23, @07:59AM (#1255540)

    What's the very-non-SI "T" quantity? We're not Magnetic Flux Density here obviously, and googling only gives "GT" as a trademark.

    • (Score: 1, Interesting) by Anonymous Coward on Thursday June 23, @09:09AM (3 children)

      by Anonymous Coward on Thursday June 23, @09:09AM (#1255545)

      It's gigaTransfers/s referring to the raw bit rate as in without any encoding schemes. It's been used since the first PCIe version.

      • (Score: 0) by Anonymous Coward on Thursday June 23, @09:45AM (2 children)

        by Anonymous Coward on Thursday June 23, @09:45AM (#1255547)

        can you please confirm: "128GT/s and compression could lead to >"?

        • (Score: 0) by Anonymous Coward on Thursday June 23, @09:47AM (1 child)

          by Anonymous Coward on Thursday June 23, @09:47AM (#1255549)
          idiotic html filters. I chose "plain old text"!

          anyway. "128GT/s and compression ould lead to <<up to 512GB/s>>".
          • (Score: 3, Informative) by Anonymous Coward on Thursday June 23, @10:37AM

            by Anonymous Coward on Thursday June 23, @10:37AM (#1255552)

            No compression necessary. 128Gt/s * 2bits/t (PAM4) * 16 lanes = 4096Gb/s / 8bits/byte = 512GB/s

    • (Score: 0) by Anonymous Coward on Thursday June 23, @02:14PM

      by Anonymous Coward on Thursday June 23, @02:14PM (#1255582)

      When saw GT I thought to myself "GigaTerabytes??", which although stupid would be more meaningful than Gigatransfers because it's measureable and comparable since the unit of transfer is know.

      GT/s is meaningless by itself. Gigatransfers per second of what? 1 bit? 1 byte? 1024 btyes?

  • (Score: 0) by Anonymous Coward on Thursday June 23, @01:44PM (2 children)

    by Anonymous Coward on Thursday June 23, @01:44PM (#1255577)

    This is a new term for us, albiet may have been in use before.
    However, even the Wiki reference link sites no confirmed sources for the metric.

    What happened to Hz, you know, the cycles per second SI unit?

    Is that not a lot easier than the introduction of yet another unit of measure?
    What am I missing here?

  • (Score: 2) by acid andy on Thursday June 23, @11:42PM

    by acid andy (1683) on Thursday June 23, @11:42PM (#1255701) Homepage Journal

    Maintaining backwards compatibility with all previous generations of PCIe technology

    That's the best bit.

    --
    Master of the science of the art of the science of art.
(1)