Monday, February 24th 2020

PCI-Express Gen 6 Reaches Development Milestone, On Track for 2021 Rollout

The PCI-Express gen 6.0 specification reached an important development milestone, with the publication of its version 0.5 first-draft. This provides important pointers to PCI-SIG members on what features and design changes gen 6.0 hopes to bring, and what its all important number is - bandwidth. PCIe gen 6.0 quadruples per-lane bandwidth over gen 4.0 to 64 GT/s (double that of gen 5.0), resulting in bi-directional bandwidth of 256 GB/s in an x16 configuration.

The spec also introduces a new physical layer change, with PAM4 (pulse amplitude modulation) signaling replacing NRZ (non-return to zero), a key ingredient in the generational bandwidth doubling effort. Despite this, PCIe gen 6.0 retains backwards-compatibility with all older generations of PCIe, which could mean the PCIe slot on motherboards may not look any different. PCIe gen 6.0 also introduces FEC (forward error-correction), and has similar per-channel reach as PCIe gen 5.0. Our older article on Intel's proprietary CXL outlines a key feature of PCIe gen 5.0 besides its bandwidth doubling over gen 4.0 - scalability. Although targeting completion in 2021, it could take several more years for the technology to transcend enterprise computing segments and reach the client. PCI-SIG anticipates the need for gen 6.0 kind of bandwidth in the industry by 2025.
Source: AnandTech
Add your own comment

33 Comments on PCI-Express Gen 6 Reaches Development Milestone, On Track for 2021 Rollout

#1
ARF
That is interesting but quite meaningless because there are quite many instances when users ignore the older PCIe 4.0 standard......
Now the talk about 6.0, what for? :laugh:
Posted on Reply
#3
delshay
What happened to more power from the slot, has it increased?
Posted on Reply
#4
TheLostSwede
delshay
What happened to more power from the slot, has it increased?
Not sure it's possible with current slot design. Hence why Apple designed a proprietary connector for the Mac Pro that can deliver more power.
ARF
That is interesting but quite meaningless because there are quite many instances when users ignore the older PCIe 4.0 standard......
Now the talk about 6.0, what for? :laugh:
Progress of technology as always? It's not as if this is going to appear in consumer devices any time soon. I have a feeling we'll be using PCIe 4.0 for the next 4-5 years at the very least. This is more for data centers and high-end servers for the time being.
Besides, new, improved standards are being developed all the time, no? So how is this different?
Posted on Reply
#5
bug
Seeing what PCIe4 did to motherboards and considering PAM4 is more sensitive, I can't say I'm eager to see PCIe6 in the consumer space :p
Posted on Reply
#6
Aquinus
Resident Wat-man
Meanwhile we can wait half a decade for implementation. :laugh:
Posted on Reply
#7
bug
Aquinus
Meanwhile we can wait half a decade for implementation. :laugh:
In the consumer space, yes.
But isn't it refreshing that at least one component in your PC is so fast, you needn't worry about it for a few more years?
Posted on Reply
#8
Aquinus
Resident Wat-man
bug
In the consumer space, yes.
But isn't it refreshing that at least one component in your PC is so fast, you needn't worry about it for a few more years?
Refreshing? I've been rocking 3.0 for 8 years and only now looking at a machine with 4.0 where I'll still be using 3.0 devices for probably the first couple years. At this rate, I'll get around to using PCIe 6 in well over a decade. :p
Posted on Reply
#9
ARF
Aquinus
Refreshing? I've been rocking 3.0 for 8 years and only now looking at a machine with 4.0 where I'll still be using 3.0 devices for probably the first couple years. At this rate, I'll get around to using PCIe 6 in well over a decade. :p
Probably in 50 years. Especially, when today there is no significant if any practical difference between using a SATA SSD and NVMe PCIe SSD.

I am not sure where the problem is, though. Is the software too ancient or other components like the CPUs bottleneck the processing time and performance?
Posted on Reply
#10
bug
ARF
Probably in 50 years. Especially, when today there is no significant if any practical difference between using a SATA SSD and NVMe PCIe SSD.

I am not sure where the problem is, though. Is the software too ancient or other components like the CPUs bottleneck the processing time and performance?
This is off-topic, but the SSDs are held back by random performance. Sequential numbers are much bigger for NVMe, but random access is need needed way more frequently. Random numbers for NVMe are about 20% better which is something you will feel (but only just), if you can get a NVMe drive without a hefty premium, get it. Whether random access is held back by flash itself or the controllers available today, I wouldn't know.
Posted on Reply
#11
TheDeeGee
I guess motherboards in the future will have 6x 40MM Fans.
Posted on Reply
#12
TheinsanegamerN
ARF
Probably in 50 years. Especially, when today there is no significant if any practical difference between using a SATA SSD and NVMe PCIe SSD.

I am not sure where the problem is, though. Is the software too ancient or other components like the CPUs bottleneck the processing time and performance?
Ioading data is only one step of software. Initializing that software takes time, loading buffer to video RAM takes time, ece. All these things have gotten faster, in terms of games they are still written around consoles with jaguar CPU cores and 5400 RPM hard drives. Next gen consoles will ease up on this restriction. In terms of other software, there is only so much you can benefit without totally re-writing software to take advantage of new hardware, and given that sata SSD performance is already "fast enough" for 99% of people, it is likely said companies see no reason to spend the money. Programs like photoshop that deal with massive files already naturally see improvements from NVMe storage.

Unless programs start ballooning in size, there is only so fast a time you can load a 500MB piece of software.
bug
This is off-topic, but the SSDs are held back by random performance. Sequential numbers are much bigger for NVMe, but random access is need needed way more frequently. Random numbers for NVMe are about 20% better which is something you will feel (but only just), if you can get a NVMe drive without a hefty premium, get it. Whether random access is held back by flash itself or the controllers available today, I wouldn't know.
Where are you getting 20% from? A casual look at the crucial MX 500 of any other sata SSD would show random IOPS for read and write both top at most 110K/s. Most are below 100k. Meanwhile, the first generation of NVMe drives from samsung pushed reads to 300k/s, and second gens pushed the write above 200k/s. Currently, sabarent's rocket is showing 750K read and 700K write. The only random performance where NVMe does not significantly outperform SATA is 4KB random read/write, but there is only so much you can spead up reading/writing such small drives. Even then, the samsung 950 pro was 33% faster then a high end SATA SSD in 4k, and the 960 and 970 evos are faster then that.
Posted on Reply
#13
ARF
TheDeeGee
I guess motherboards in the future will have 6x 40MM Fans.
This and the need for repeaters every 1 cm, which will make them prohibitively expensive, too.
TheinsanegamerN
Ioading data is only one step of software. Initializing that software takes time, loading buffer to video RAM takes time, ece. All these things have gotten faster, in terms of games they are still written around consoles with jaguar CPU cores and 5400 RPM hard drives. Next gen consoles will ease up on this restriction. In terms of other software, there is only so much you can benefit without totally re-writing software to take advantage of new hardware, and given that sata SSD performance is already "fast enough" for 99% of people, it is likely said companies see no reason to spend the money. Programs like photoshop that deal with massive files already naturally see improvements from NVMe storage.

Unless programs start ballooning in size, there is only so fast a time you can load a 500MB piece of software.
Well, I do have a SATA SSD and to be honest the games loading is pretty bad, still. Hardly any difference in loading times over an HDD 5400 rpm. F1 2018 with M.2 SATA WD Blue 3D.

I still wait 5 minutes for the game to load.
Posted on Reply
#15
fynxer
Yea yea, people talking down new tech is nothing new. In Sweden we have an old saying from the eighteen hundreds and call them "backstugesittare" and it means low IQ peasant, reluctant to change, living in a one room mini house or small earth cellar on the corner of other peoples property because they didn't know how to do any better in life.

I for one welcome new tech and relize we need new tech to be ready and availabe to all users so they can invent new ways to use it.

It's like Intel for years tried to convice main stream consumers that they had no need for more than four cpu cores out of greed so they could continue to peddle four core cpu's to main stream for insane money.

Now AMD has over run Intel and are changing the consumer landscape forever with up to 16 cores for main stream and 64 core for workstation for starters, near future they will go 32 core main stream and 128 core workstation.

I am looking forward to SSD with 32GB/s bandwidth and all the other ways we can invent ways to use this amazing new bandwidth to enhance our every day life.
Posted on Reply
#16
rtwjunkie
PC Gaming Enthusiast
There’s nothing like getting a new technology standard that will be “ancient” by the time it’s actually used in the consumer arena. Servers maybe before then.
Posted on Reply
#17
bug
@fynxer I don't think this is talking down as much as acknowledging it's something aimed at enterprise with years before it will be needed in the consumer space.
As an engineering achievement, hats off when they set it in stone.
Posted on Reply
#18
Hardware Geek
ARF
This and the need for repeaters every 1 cm, which will make them prohibitively expensive, too.



Well, I do have a SATA SSD and to be honest the games loading is pretty bad, still. Hardly any difference in loading times over an HDD 5400 rpm. F1 2018 with M.2 SATA WD Blue 3D.

I still wait 5 minutes for the game to load.
What processor and how much RAM do you have? It sounds like there is a major bottleneck somewhere in your system.
Posted on Reply
#19
dj-electric
Hi, i'm a typical moronic poster who thinks PCIE standards are intended for consumer motherboards only and nothing else in the electronics and semiconductor industry.
Where is my PCIE5 already? Where are my compatible gaming CPUs?

Come on now, PCI SIG, that's totally lame.
Posted on Reply
#20
ARF
Hardware Geek
What processor and how much RAM do you have? It sounds like there is a major bottleneck somewhere in your system.
No, I guess I need to re-install it because it was originally installed on an HDD and then moved to the new drive with Macrium Reflect Free.
I guess it's fragmented all over the partition.
Posted on Reply
#21
Chloe Price
WTF? 4.0 is still coming for us consumers, we all don't have a X570 board and a R 3000 series CPU :D
Posted on Reply
#22
ARF
Chloe Price
WTF? 4.0 is still coming for us consumers, we all don't have a X570 board and a R 3000 series CPU :D
True story. Intel and Nvidia don't even support PCIe 4.0.

That's why I said 50 years.

5 years, at the very least, to adopt PCIe 4.0.
Then 15 years to happily use it.
And then 15 years more to use PCIe 5.0.

After all these years, maybe PCIe 6.0.
Posted on Reply
#23
Chloe Price
ARF
True story. Intel and Nvidia don't even support PCIe 4.0.

That's why I said 50 years.

5 years, at the very least, to adopt PCIe 4.0.
Then 15 years to happily use it.
And then 15 years more to use PCIe 5.0.

After all these years, maybe PCIe 6.0.
Is this the first time when there's hella many PCIe gens coming when we haven't even fully adapted the newest one?
Posted on Reply
#24
TheinsanegamerN
ARF
No, I guess I need to re-install it because it was originally installed on an HDD and then moved to the new drive with Macrium Reflect Free.
I guess it's fragmented all over the partition.
Fragmenting HELPS SSDs. The more scattered the files are around different NAND clousters, the faster the SSD can read them, to a point, as a SSD can read from multiple clusters at once, but not multiple files from the same cluster at once.

Sounds like your CPU/RAM are a much bigger bottleneck. What CPU/RAM do you have?
fynxer
Yea yea, people talking down new tech is nothing new. In Sweden we have an old saying from the eighteen hundreds and call them "backstugesittare" and it means low IQ peasant, reluctant to change, living in a one room mini house or small earth cellar on the corner of other peoples property because they didn't know how to do any better in life.

I for one welcome new tech and relize we need new tech to be ready and availabe to all users so they can invent new ways to use it.

It's like Intel for years tried to convice main stream consumers that they had no need for more than four cpu cores out of greed so they could continue to peddle four core cpu's to main stream for insane money.

Now AMD has over run Intel and are changing the consumer landscape forever with up to 16 cores for main stream and 64 core for workstation for starters, near future they will go 32 core main stream and 128 core workstation.

I am looking forward to SSD with 32GB/s bandwidth and all the other ways we can invent ways to use this amazing new bandwidth to enhance our every day life.
For one thing, just because technology is "new" doesnt mean it is ready for consumer use, nor is it cheap enough for consumer use. New =! better. PCIe 5/6 have incredibly tight timings that, until a cheap method of implementation is established, would dramatically increase the costs of consumer motherboards with no real world benefit.

Lets bring up that quad core argument, shall we? Intel "convinced" consumers fo that, did they? No, I'm pretty sure the market decided that for itself. AMD offered 6/8 core mainstream CPUs for 9 years now, and only in the last 2 years or so did this start making any kind of difference in games. It's only been fairly recently that games have taken advantage of 6 cores in workloads, the vast majority STILL dont use more then 6, which is why many gaming focused builds use i5 9600k/r5 3600 CPUs. Only a single gen, the i 7000 series, really ignored the advancement of 6 core usage, up to that point there was no benefit to having more then 4 cores, as both intel's HDET platform and AMD's construction core showed. Ryzen happened to hit at the right time for multi core advancement.

You can clamor for 32GB/s of bandwidth all you want, but given that NVMe drives are already stupid fast, the question becomes why do you need more? Is a 3 second boot time just too slow for you? Keep in mind that most people have 125MB/s network speeds maybe 10% of that in internet speeds, and maybe 500MB/s over USB 3, all that NVMe speed is worthless outside of its own box. Internal storage stopped being the bottleneck with SATA III drives, and NVMe 4.0 drives are already ludicrous at 6GB/s, not to mention the 7-8 GB/s they will be hitting in a gen or 2.

We have a saying in english for someone who constantly jumps on new,expensive tech before it is useful or needed: "a fool and his money are soon parted"
Posted on Reply
#25
bug
TheinsanegamerN
Fragmenting HELPS SSDs. The more scattered the files are around different NAND clousters, the faster the SSD can read them, to a point, as a SSD can read from multiple clusters at once, but not multiple files from the same cluster at once.
You can't fragment a SSD, the OS doesn't see the physical locations. Those are taken care in the background for you ;)
Posted on Reply
Add your own comment