PCIe Express 7.0 specs version 0.5 now available to PCI-SIG members, on track to launch in 2025

erek

[H]F Junkie
Joined
Dec 19, 2005
Messages
10,917
1712153703366.png


Source: https://videocardz.com/newz/pcie-ex...to-pci-sig-members-on-track-to-launch-in-2025
 
Presumably this is exclusively useful for enterprise since most consumers aren't even leveraging the current max speeds for storage or GPUs. The Direct Storage thing is still very niche so games for example are bottlenecked in other ways before storage speed.
 
games for example are bottlenecked in other ways
Games, like almost every other app that runs on computers, are bottlenecked by 1) sloppy, inefficient coding, and 2) Legacy crap code from days gone by (16/32bit pieces) and 3) primitive backwards compatibility with old stuff like pcie gens 1 & 2...

Until/unless these things are fixed, they can bring out pcie v332.973, and it won't matter :)
 
Great! I'll try and pick an pcie 7 ssd up around cyber Monday 2033.
 
Last edited:
And 6.0 isn't even released yet lol.
Specs have to precede rollout by quite a bit. It takes time to do all the design and engineering. So often you will see something like whatever is being rolled out now, the spec for the next version is either final or near final, and the spec for the version after that is under development. Remember that once a spec is complete it isn't like companies just start building it. Once a spec is complete companies can start working on solving the problems of implementing it, which can take awhile, then actually design products that use it, then manufacture those products. A spec that gets ratified today can easy be 2 years or more from having hardware on the market that implements it.

With interconnects, we want to try and stay ahead of the game. You want a situation where the PCIe sockets on your motherboard are faster than you need, because you don't want to get to a situation where we have neat new hardware that is hamstrung by interconnect speed.
 
Specs have to precede rollout by quite a bit. It takes time to do all the design and engineering. So often you will see something like whatever is being rolled out now, the spec for the next version is either final or near final, and the spec for the version after that is under development. Remember that once a spec is complete it isn't like companies just start building it. Once a spec is complete companies can start working on solving the problems of implementing it, which can take awhile, then actually design products that use it, then manufacture those products. A spec that gets ratified today can easy be 2 years or more from having hardware on the market that implements it.

With interconnects, we want to try and stay ahead of the game. You want a situation where the PCIe sockets on your motherboard are faster than you need, because you don't want to get to a situation where we have neat new hardware that is hamstrung by interconnect speed.

For a average gamer, I doubt it matters much at all. I mean it might constrain the theoretical limit of your NVME drive but that's about it.
 
Last edited:
Maybe we will not care, but maybe we will get GPU using 4-8 lane, hard drive using 1 lane, making possible for simple CPU to become quite the workstation affair (but there would a lot of incentive for that never happening and the industry continuing to push 4 lane harddrive, x16 gpus and keeping entry level board to one GPU-3/4 drive with not much else to keep an high end)..

1x Gen 6 would already be quite a lot of a lot of stuff, a single lane would maybe be ok for a 40gbs network card, most harddrive case, and so on.
 
Maybe we will not care, but maybe we will get GPU using 4-8 lane, hard drive using 1 lane, making possible for simple CPU to become quite the workstation affair (but there would a lot of incentive for that never happening and the industry continuing to push 4 lane harddrive, x16 gpus and keeping entry level board to one GPU-3/4 drive with not much else to keep an high end)..

1x Gen 6 would already be quite a lot of a lot of stuff, a single lane would maybe be ok for a 40gbs network card, most harddrive case, and so on.
Yeah, that.

Seems a bit pointless until we actually have devices catch up with even PCIe gen4 and then the mobos are going to have to do a better job at divvying out PCIe lanes.

I have a avermedia capture card. It runs PCIe gen 2 x4. Their best run card looks to run at Gen3 x4.

Zarathustra is looking for a fancy nic that does 40G that is gen4 x4. They don't seem to exist.

That being said, there are some 10G nics that run Gen4 x2....but the way lanes are divvied out on the motherboard, I'm not sure what good x2 is versus x4.
 
Once a spec is complete companies can start working on solving the problems of implementing it, which can take awhile, then actually design products that use it, then manufacture those products.
Well. what happens if the designers discover a flaw in the specs, or a spec detail that is very hard/impossible to implement?
 
Well. what happens if the designers discover a flaw in the specs, or a spec detail that is very hard/impossible to implement?
Why do you think there's so much time when it is getting revised and set up? They spend time working on the spec in no small part to make a spec that is going to be doable with the technology that is available.
 
they sure are pumping these out. there's people that dont have pcie5 yet

There's a pretty big pipeline on these things, and I think the consortium decided to push towards a faster release cycle after pci-e 4.0, because 7 years was too long.
 
There's a pretty big pipeline on these things, and I think the consortium decided to push towards a faster release cycle after pci-e 4.0, because 7 years was too long.
Well for so long there was barely anything pushing the limits of PCIE3, and the very few things that were used custom solutions to work around it. So PCIe4 wasn't really needed until solid-state memory really became a mainstream item. And now it's AI and other accelerators that are pushing the envelope.
 
Well for so long there was barely anything pushing the limits of PCIE3, and the very few things that were used custom solutions to work around it. So PCIe4 wasn't really needed until solid-state memory really became a mainstream item. And now it's AI and other accelerators that are pushing the envelope.
That's a reason to keep on top of it with interconnects though: you never know when the next thing will come that'll need it.

SATA was like that. When it first came out, it was slightly faster than PATA, which was already faster than magnetic drives needed, and it quickly doubled in speed. Great... until SSDs hit the scene. They instantly maxed SATA 2, and then instantly maxed SATA 3. We suddenly had a new tech and out interconnects weren't up to snuff for it.

I'd much rather that our interconnects were faster than we need than for something new to come out and have great potential but be hamstrung by insufficient interconnect speed.
 
I'd much rather that our interconnects were faster than we need than for something new to come out and have great potential but be hamstrung by insufficient interconnect speed.
What he said.
 
Back
Top