Will we see PCI-E 4.0 this year?

[rwnrwnn7]

• PCI-SIG Finalizes and Releases PCIe 4.0, Version 1 Specification: 2x PCIe Bandwidth and More

https://www.anandtech.com/show/11967/pcisig-finalizes-and-releasees-pcie-40-spec

We do not allow link dropping
without personal commentary.

AT Mod Usandthem

 

[thecoolnessrune]

Yes, the first swath of IBM POWER9 based Servers like the L922 that released last month all have a smattering of Generation 4 PCIe slots.

https://www-01.ibm.com/common/ssi/c...897/ENUS9008-22L&appname=totalstorage#addinfo

 

[2is]

From a home/small business/gamer perspective, my interest in PCIe 4 isn't even on the radar. Even the main-stream products from Intel with limited PCIe 3.0 lanes have plenty of bandwidth for NVMe and multi-gpu configurations, which I may add, have increasingly lacking support every year. If they can crank up the power delivery to reduce the number of cards that need external power, I suppose that would be one very small benefit, beyond that it's just a shiny new spec that won't have any real world implications for most of us.

 

[thecoolnessrune]

From a home/small business/gamer perspective, my interest in PCIe 4 isn't even on the radar. Even the main-stream products from Intel with limited PCIe 3.0 lanes have plenty of bandwidth for NVMe and multi-gpu configurations, which I may add, have increasingly lacking support every year. If they can crank up the power delivery to reduce the number of cards that need external power, I suppose that would be one very small benefit, beyond that it's just a shiny new spec that won't have any real world implications for most of us.

This is almost every new technology. The high end starts, and it slowly filters down to the the lower end. It took the better part of a year for PCIe 3.0 to start showing up on high end consumer motherboards.

As always, you build the platform, and things taking advantage of it will follow. If anything for high-end motherboards, Using PCI-e 4.0 as your root gives you more bandwidth downstream. Currently High I/O Enthusiast boards use PLX switches that even on "high end" Intel platforms require a sacrifice of lanes from 16x to 8x. Using 4.0 Ports as the Root to 3.0 Downstream would allow a 6 or 7 slot mobo to have a nearly entire 16x slot stack offered vs 8x or outright cutting ports.

Not to mention, as 4.0 becomes more common place, 3.0 gear becomes cheaper. It most certainly has real world implications for most of us, but like any new technology the farther down the pole you are, the longer you'll have to wait to see the impact.

 

[whm1974]

From a home/small business/gamer perspective, my interest in PCIe 4 isn't even on the radar. Even the main-stream products from Intel with limited PCIe 3.0 lanes have plenty of bandwidth for NVMe and multi-gpu configurations, which I may add, have increasingly lacking support every year. If they can crank up the power delivery to reduce the number of cards that need external power, I suppose that would be one very small benefit, beyond that it's just a shiny new spec that won't have any real world implications for most of us.

I'm wondering if AMD and Intel will skip v4.0 and to go straight to v5 instead?

 

[LTC8K6]

4.0 was announced in November 2011.
5.0 was announced in June 2017.

 

[2is]

This is almost every new technology. The high end starts, and it slowly filters down to the the lower end. It took the better part of a year for PCIe 3.0 to start showing up on high end consumer motherboards.

As always, you build the platform, and things taking advantage of it will follow. If anything for high-end motherboards, Using PCI-e 4.0 as your root gives you more bandwidth downstream. Currently High I/O Enthusiast boards use PLX switches that even on "high end" Intel platforms require a sacrifice of lanes from 16x to 8x. Using 4.0 Ports as the Root to 3.0 Downstream would allow a 6 or 7 slot mobo to have a nearly entire 16x slot stack offered vs 8x or outright cutting ports.

Not to mention, as 4.0 becomes more common place, 3.0 gear becomes cheaper. It most certainly has real world implications for most of us, but like any new technology the farther down the pole you are, the longer you'll have to wait to see the impact.

Not really. NVMe benefits were immediate. New GPU's that get released have immediate benefits. USB 3.0 benefits were immediate for things like external hard drives which over time trickled down to other products. Taking time to trickle its way down to consumer products is one thing, having a tangible benefit anytime in the near future is another.

 

[thecoolnessrune]

Not really. NVMe benefits were immediate. New GPU's that get released have immediate benefits. USB 3.0 benefits were immediate for things like external hard drives which over time trickled down to other products. Taking time to trickle its way down to consumer products is one thing, having a tangible benefit anytime in the near future is another.

Yes really. NVMe has been kicking around in special Oracle systems and other High end Enterprise since 2014. Tensor Cores exist in Enterprise where we have have no idea what sort of consumer workload could even benefit from them.

You're asking for the chicken before the egg, and that's simply not how growing technologies work. You're speaking as if these things dropped and suddenly we were using them, but in these cases the tech is floating around in Enterprise, then by the time it gets down to us we find uses for it. The same can be said of PCIe 4, if for no other reason then as a price reduction for PCIe 3. To say there's no benefit anytime in the near future is redonkulous.

 

[2is]

Yes really. NVMe has been kicking around in special Oracle systems and other High end Enterprise since 2014. Tensor Cores exist in Enterprise where we have have no idea what sort of consumer workload could even benefit from them.

You're asking for the chicken before the egg, and that's simply not how growing technologies work. You're speaking as if these things dropped and suddenly we were using them, but in these cases the tech is floating around in Enterprise, then by the time it gets down to us we find uses for it. The same can be said of PCIe 4, if for no other reason then as a price reduction for PCIe 3. To say there's no benefit anytime in the near future is redonkulous.

I think you need to spend more time understanding what I'm actually saying instead of arguing a point I'm not even discussing. I'm not arguing about the timeline of when technologies trickle down to consumers. I'm arguing their benefits when they do trickle down to consumers.

NVMe had immediate benefits as soon as it hit consumer products
New GPU's have immediate benefits as soon as they are available to consumers
USB 3.0 had immediate benefits to consumers as soon as it becaome available.
The next iteration of PCIExpress will not, and because it will not, it's not something on my radar and I don't think it's a feature most people here are salivating for either.

So no, not really.

 

[thecoolnessrune]

I think you need to spend more time understanding what I'm actually saying instead of arguing a point I'm not even discussing. I'm not arguing about the timeline of when technologies trickle down to consumers. I'm arguing their benefits when they do trickle down to consumers.

NVMe had immediate benefits as soon as it hit consumer products
New GPU's have immediate benefits as soon as they are available to consumers
USB 3.0 had immediate benefits to consumers as soon as it becaome available.
The next iteration of PCIExpress will not, and because it will not, it's not something on my radar and I don't think it's a feature most people here are salivating for either.

So no, not really.

Whens PCIe 4.0 hitting consumer systems?

 

[2is]

Whens PCIe 4.0 hitting consumer systems?

No idea, but I get where you're going. Would you like for the both of us to bookmark this thread until it does? Seems you're trolling towards "you can't prove me wrong yet so I'm right" I have no problem telling you "i told you so" when it releases.

 

[thecoolnessrune]

No idea, but I get where you're going. Would you like for the both of us to bookmark this thread until it does? Seems you're trolling towards "you can't prove me wrong yet so I'm right" I have no problem telling you "i told you so" when it releases.

It's strange for you to accuse me of trolling when it's yourself who's making such a strange statement with no qualifiers.. So what qualifies as a consumer workload? USB 3.0 Gen 2 for instance requires a 4x lane to maintain full speeds to the CPU, since 1x isn't enough. If we get a 1x or say 4x quad port USB card on PCIE 4, are we good?

 

[2is]

It's strange for you to accuse me of trolling when it's yourself who's making such a strange statement with no qualifiers.. So what qualifies as a consumer workload? USB 3.0 Gen 2 for instance requires a 4x lane to maintain full speeds to the CPU, since 1x isn't enough. If we get a 1x or say 4x quad port USB card on PCIE 4, are we good?

Lets just keep it simple. Wait until it's available, there will be myriads of PCI 3 vs 4 performance reviews and we'll see who the clueless one is. Then instead of you having to take my word that you're wrong, i'll just post links that prove it.

 

[thecoolnessrune]

Lets just keep it simple. Wait until it's available, there will be myriads of PCI 3 vs 4 performance reviews and we'll see who the clueless one is. Then instead of you having to take my word that you're wrong, i'll just post links that prove it.

Why do we need to wait for the benchmarks to be published to see if there's any difference? Do you not agree that a 1x 3.0 interface would starve a 10gb USB port, or a 4 port controller? Do you disagree with the workload? If you disagree, can you think of some workloads that for average home users benefit from say, 2.0, but not from 3.0 or 4.0?

 

[2is]

Why do we need to wait for the benchmarks to be published to see if there's any difference? Do you not agree that a 1x 3.0 interface would starve a 10gb USB port, or a 4 port controller? Do you disagree with the workload? If you disagree, can you think of some workloads that for average home users benefit from say, 2.0, but not from 3.0 or 4.0?

Because instead of talking about theoretical workloads and arguing what is and isn't a consumer work load, we'll have actual numbers across a vast array of tests. Why are you so worried? I've already favorited this thread. Have your quote notifications turned on please, i wouldn't want you to miss out on the 0-3% average difference. lol

 

[thecoolnessrune]

Because instead of talking about theoretical workloads and arguing what is and isn't a consumer work load, we'll have actual numbers across a vast array of tests. Why are you so worried? I've already favorited this thread. Have your quote notifications turned on please, i wouldn't want you to miss out on the 0-3% average difference. lol

There's little point in saying things like "Why are you so worried?". The questions are merely to specify what we're going to be discussing here. You seem to be more interested in finding a "gotcha" than common ground. For instance, you make these two statements:

"Because instead of talking about theoretical workloads and arguing what is and isn't a consumer work load, we'll have actual numbers across a vast array of tests."
"i wouldn't want you to miss out on the 0-3% average difference"

If that's the banner mark, I assume you've got some figures already to show how 4.0 is markedly different from 1.0 to 2.0 or 2.0 to 3.0 right? What do you think the average difference is in a broad array of consumer workload between PCIe 1.0 and 2.0? More pointedly, do you think the average worldwide consumer in an average worldwide range of workloads would notice a difference between PCIe 1.0 and 4.0?

 

[2is]

What we'll be discussing is pretty obvious. The benefits of PCI Express 4 vs 3. Don't worry, I'll quote you with some reviews just as soon as they're available. I think we both know essentially what the end result will be, which is why you're in a bit of panic mode.

 

[thecoolnessrune]

What we'll be discussing is pretty obvious. The benefits of PCI Express 4 vs 3. Don't worry, I'll quote you with some reviews just as soon as they're available. I think we both know essentially what the end result will be, which is why you're in a bit of panic mode.

See, you're doing this again because you don't want to be specific with what you're asking for. You want to leave you that portion up to your interpretation because you're more concerned about being perceived as incorrect. In other words, what you're looking for is room to spin. It's a forum, so you're welcome to do what you like, but I hope you understand that if you don't set any specificity, I can't know if I agree with you or not. Depending on the spin you apply in the future we might end up with you saying "I'm right", and I'd say "Yeah, I never disagreed". But we don't know that because you just want to spin.

"The benefits of PCI Express 4 vs 3"

Under what circumstances for the above statement do you say "I'm right"?

 

[2is]

There are no specifics currently, it's not even available yet all we can be is general. There is no spinning here. I will post actual reviews of actual numbers using actual PCI Express 3 vs Actual PCI Express 4 performance figures. You are the one who completely misunderstood what I was initially trying to say, and instead of acknowledging that and moving on, you wanted to flip the argument to try to be "right" about something. So we'll play that game too, and you'll be wrong again. I do apologize I can't "prove" you wrong now, but don't worry. It'll come.

 

[thecoolnessrune]

There are no specifics currently, it's not even available yet all we can be is general. There is no spinning here. I will post actual reviews of actual numbers using actual PCI Express 3 vs Actual PCI Express 4 performance figures. You are the one who completely misunderstood what I was initially trying to say, and instead of acknowledging that and moving on, you wanted to flip the argument to try to be "right" about something. So we'll play that game too, and you'll be wrong again. I do apologize I can't "prove" you wrong now, but don't worry. It'll come.

You still haven't specified anything to "prove me wrong" by, so you're still spinning. Here's something more ballpark. A statement from you:

"home/small business/gamer perspective, my interest in PCIe 4 isn't even on the radar"

Another statement from you:

"NVMe had immediate benefits as soon as it hit consumer products"

So when you talk about PCIe 4, are you talking about home/small business/gamer perspective, or are you talking about any consumer product? When you talk about NVMe having immediate benefits in consumer products, are you saying from a home/small business/gamer perspective, or that you could see a difference in literally any consumer workload? If you said "I can't see PCIe 4 doing any good any of my workloads" I'd be inclined to agree, if by the fact that from most home/small business/gamer perspectives, there is little to be seen from PCIe 2 (at this time). But in a later statement, you talk about NVMe, as if it had a benefit in the same workload group that PCIe 4 would not. Yet there is little to no difference in home/small business/gamer workloads with an NVMe drive vs. a SATA SSD. Certainly being within the "0-3%" you expected the average to be on average workload between PCIe 3 and PCIe4. https://youtu.be/ecCA0gx_eZk

If you averaged the standard light consumer average workload, do you think there would be a meaningful difference between NVMe, which you note has differences, vs. a SATA SSD?

 

[2is]

You still haven't specified anything to "prove me wrong" by, so you're still spinning. Here's something more ballpark. A statement from you:

I know I haven't, and I even apologized that I'm not able to do it right away. but like I said... It'll happen, be sure not to unsubscribe from this thread and you'll see the notification with "proof" when the proof becomes available.

I do apologize I can't "prove" you wrong now, but don't worry. It'll come.

That was the very last sentence of my previous post.

 

[thecoolnessrune]

I know I haven't, and I even apologized that I'm not able to do it right away. but like I said... It'll happen, be sure not to unsubscribe from this thread and you'll see the notification with "proof" when the proof becomes available.

That was the very last sentence of my previous post.

Again, what are you proving? Not having proof is not the same as specifying your belief that needs proof. If you don't know what you're proving, then what is it you're trying to prove? 🙂 That PCIe 4 will have less of an impact on home/small business/gamer workloads than the small percentage gain NVMe has?

 

[2is]

That the benefits of PCIe 4 over 3 will be marginal to non-existent for the vast majority of people here.

 

[thecoolnessrune]

That the benefits of PCIe 4 over 3 will be marginal to non-existent for the vast majority of people here.

For what duration? Upon release? 6 months? 12 months? 5 years? Never?

 

[whm1974]

For what duration? Upon release? 6 months? 12 months? 5 years? Never?

I give five years.