NVIDIA GeForce RTX 3080 PCI-Express Scaling

[W1zzard]

NVIDIA Ampere finally brings PCI-Express 4.0 support to the high-end graphics market. The new interface promises twice the bandwidth of PCI-Express 3.0. We've setup an AMD Ryzen 3900XT system to test how various PCIe generations and lane widths affect gaming performance.

Show full review

 

[deu]

Practically no bottleneck even on PCI-E 2.0! :0 PCI-E is holding up quiet well; GPU manifactures get your s*** together and bottleneck them faster!

 

[BorisDG]

Was expected PCI-E 3.0 to be enough.

 

[Jism]

And even if the PCI-E 3.0/2.0 was "limited" in some way, you could always throw a marginal OC towards it, i.e 110Mhz which increases the bandwidth even more.

 

[TechLurker]

Considering how well the GPU still tends to scale even on 2.0 and even 1.0 in some cases, I'd like to see, just mostly for fun, how much FPS is possible in an ancient rig running 1.1 PCIe and the top CPU of the time period at 1080p minimum (an extreme case of blowing the budget on GPU-only upgrades for old rigs). Given that earlier Ryzens ran on PCIe 2.0, it's not too surprising to see GPUs still able to provide respectable numbers on PCIe 2.0 in still a fairly modern setup.

That said, I could still see FPS chasers waving around these graphs and insisting that they absolutely must upgrade to Ryzen 3000 or the upcoming 4000 NOW for that extra .5% FPS boost. Which would perfectly benefit AMD's CPU division and those mobo companies slightly burned by Intel's delay on 4.0 capable CPUs (from a video where GN mentioned the topic).

 

[jeremyshaw]

Any chance of frametime/lows analysis? It was done in the main RTX 3080 article, and it wouldn't have to be too broad - just covering max PCIe 3.0, 4.0 configs would be more than enough to conclusively settle the matter in a pre-DirectStorage timeframe.

 

[Charcharo]

I think we need 1% Lows on these since those tend to be slightly more impacted by PCIE bandwidth.

 

[Jism]

Considering how well the GPU still tends to scale even on 2.0 and even 1.0 in some cases, I'd like to see, just mostly for fun, how much FPS is possible in an ancient rig running 1.1 PCIe and the top CPU of the time period at 1080p minimum (an extreme case of blowing the budget on GPU-only upgrades for old rigs). Given that earlier Ryzens ran on PCIe 2.0, it's not too surprising to see GPUs still able to provide respectable numbers on PCIe 2.0 in still a fairly modern setup.

That said, I could still see FPS chasers waving around these graphs and insisting that they absolutely must upgrade to Ryzen 3000 or the upcoming 4000 NOW for that extra .5% FPS boost. Which would perfectly benefit AMD's CPU division and those mobo companies slightly burned by Intel's delay on 4.0 capable CPUs (from a video where GN mentioned the topic).

You should understand that PCI-E 4.0 was released and 5.0 is coming primarily due to the enterprise demands / standards. Not for gamers. AMD is the only one that implemented PCI-E 4.0 into their consumer products because they are not going to drive 2 different CPU lines with different chipsets. They use IP for both standards where quality per product is different.

PCI-E 3.0 is'nt even fully taxed, and if it was you could still overclock it increasing it with another x16 times your OC'ed speed. But on the other side of the spectrum, if you had a system now with 2x NVME SSD's tapping into the 4.0 bus including a graphics cards, now those situations would be different as PCI-E 4.0 would actually come in handy.

But it's all within ~ 1% of margin; nothing to worry about either. I dont miss much when i play at either 186 fps or 194 fps.

 

[BorisDG]

But on the other side of the spectrum, if you had a system now with 2x NVME SSD's tapping into the 4.0 bus including a graphics cards, now those situations would be different as PCI-E 4.0 would actually come in handy.

That's why HEDT Intel platform come in handy with it's many PCIe lanes.

 

[dirtyferret]

I can squeeze out an extra 1.1% FPS via PCIE 4.0?! THIS CHANGES EVERYTHING!

 

[Aretak]

Really need to see results for the games that other outlets have found to suffer when dealing with limited PCIe bandwidth, like Horizon: Zero Dawn. Some titles seem to lose a lot more performance than others.

 

[BorisDG]

Really need to see results for the games that other outlets have found to suffer when dealing with limited PCIe bandwidth, like Horizon: Zero Dawn. Some titles seem to lose a lot more performance than others.

Horizon is terrible port. Stutters as hell. The game is PS4 era so I really doubt it matters PCIe gen... See Death Stranding with same engine (probably even modernizied, because it's newer) performs excellent.

 

[HenrySomeone]

I love how the "we only cater to AMD fanboys these days" Hardware Unboxed used a Ryzen test platform despite it being almost 10% slower even at 1440p, lmao!

 

[phill]

Thank you @W1zzard

 

[Cheeseball]

It looks like the main advantages of PCI-E 4.0 are only for storage use. Makes sense if you do a lot of sequential data transfer.

 

[Vader]

I love how the "we only cater to AMD fanboys these days" Hardware Unboxed used a Ryzen test platform despite it being almost 10% slower even at 1440p, lmao!

Apparently you missed the community post where they asked people which platform to test RTX 3000 with. The amd config was what people wanted to see

 

[HenrySomeone]

Nope, seen it and that's just it - "what people wanted to see" (those people undoubtedly being team red fans to a worrying degree - that channel is now little better than the laughably biased Moore's law is Dead and the likes) instead of what is objectively better. I love how Steve Gamers Nexus made a subtle but unmistakable reference to it by saying that they made sure to use the "fastest cpu for every appropriate test"

 

[Papahyooie]

I love how the "we only cater to AMD fanboys these days" Hardware Unboxed used a Ryzen test platform despite it being almost 10% slower even at 1440p, lmao!

Literally on the first page of the review:
"Since 3rd Gen Ryzen is the only desktop processor series with PCIe Gen 4 as of this writing "

How exactly are you going to test PCI-E 4.0 scaling on the competition, bud?

 

[dirtyferret]

I love how Steve Gamers Nexus made a subtle but unmistakable reference to it by saying that they made sure to use the "fastest cpu for every appropriate test"

Steve calling out another site and proclaiming himself better then that site? Ridiculous! Next you will be telling us he went an entire 20 seconds without brushing his hair out of his face in one of his videos!

 

[Bubster]

NVIDIA knows that that pcie 4 will matter more for RTX IO later on pcie 3 @ 32gb/s is still fast. i load games from a raid 0 SSD sata drive under 10 seconds at a fifth of the time it took with legacy 7200 rpm HDD. i can imagine it will be much faster with pcie 3 nvme at least double the speed that i have...the only deterrence is the cost of 2tb or more NVME M2 ssd.

 

[HenrySomeone]

Literally on the first page of the review:
"Since 3rd Gen Ryzen is the only desktop processor series with PCIe Gen 4 as of this writing "

How exactly are you going to test PCI-E 4.0 scaling on the competition, bud?

No bud, not for the reasons of pcie scaling - Hardware Unboxed tested the 3080 exclusively on Ryzen!

 

[Papahyooie]

No bud, not for the reasons of pcie scaling - Hardware Unboxed tested the 3080 exclusively on Ryzen!

You're commenting on a review of PCI-E 4 scaling.... Who cares what hardware unboxed did?

 

[juular]

Any chance of frametime/lows analysis? It was done in the main RTX 3080 article, and it wouldn't have to be too broad - just covering max PCIe 3.0, 4.0 configs would be more than enough to conclusively settle the matter in a pre-DirectStorage timeframe.

This ! Average framerate data isn't nearly as relevant as frametime analysis when all that changes is the interface bandwidth, can't believe authors of this article didn't thought of that. I imagine we see such a big difference with PCIe 1.0 x8 just because it stutters as hell trying to move all that data through.

 

[HenrySomeone]

You're commenting on a review of PCI-E 4 scaling.... Who cares what hardware unboxed did?

They did this exactly on the pretext of "needing a pcie 4.0 platform" and it's no use commenting there as the channel is absolutely flooded with rabid fanboys.

 

[Papahyooie]

They did this exactly on the pretext of "needing a pcie 4.0 platform" and it's no use commenting there as the channel is absolutely flooded with rabid fanboys.

That's completely irrelevant to the thread here.