NVIDIA RTX 3080 PCIe 3.0 vs. PCIe 4.0 x16 Benchmarks (& Intel vs. AMD for 3080 Discussion)

preview_player
Показать описание
The NVIDIA RTX 3080 GPU uses PCIe Gen4, and so we have another opportunity to benchmark PCIe 3.0 vs. 4.0, using an AMD X570 platform for benchmarks as to whether PCIe gen matters.

CORRECTION: Slide at 05:10 for Quake II RTX should read "R9 3900XT and X570 Crosshair VIII Hero," not 10700K. Sorry about that copy/paste error.

PCIe generation testing has always been a hot topic for GPU launches, but this time, it's about PCIe 3 vs. 4. We've most commonly talked about PCIe 3.0 x16 vs. PCIe 3.0 x8, and as Ryzen has matured, that has presented opportunities for Gen4 testing -- except only the 5700XT could leverage it, and so it didn't particularly matter. That's not to say the 5700 XT wasn't good, but just that it was unable to leverage the additional bandwidth in any meaningful way. Our previous testing with Titan RTX cards also showed limited gains, and so we're revisiting this storyline today with our NVIDIA RTX 3080 GPU (the Gigabyte Eagle also makes a brief feature). This tests PCIe generational bandwidth for gaming and for workstation workloads.

RELATED PRODUCTS (Test Bench Parts)

GPU - This is what we're testing!

TIMESTAMPS

00:00 - RTX 3080 Generational Testing
00:53 - This Is Not a CPU Benchmark
02:47 - Marketing Value of Bigger Number Better
04:07 - 3DMark PCIe Bandwidth Test Validation
04:36 - Quake II RTX
05:37 - Red Dead Redemption 2 4K DX12
06:20 - Red Dead Redemption 2 1440p DX12
07:05 - Rainbow Six Siege 1440p Ultra
07:26 - Rainbow Six Siege 1080p
08:07 - Shadow of the Tomb Raider 4K
08:35 - Tomb Raider 1440p
08:50 - Microsoft Flight Sim 2020 PCIe Gen 3 vs 4 Benchmarks
09:20 - Hitman 2 PCIe Gen Benchmarks
09:40 - GTA V PCIe 3 vs 4 Benchmarks
10:12 - Blender GPU Cycles Benchmarks
10:25 - Conclusions


** Please like, comment, and subscribe for more! **

Links to Amazon and Newegg are typically monetized on our channel (affiliate links) and may return a commission of sales to us from the retailer. This is unrelated to the product manufacturer. Any advertisements or sponsorships are disclosed within the video ("this video is brought to you by") and above the fold in the description. We do not ever produce paid content or "sponsored content" (meaning that the content is our idea and is not funded externally aside from whatever ad placement is in the beginning) and we do not ever charge manufacturers for coverage.

Follow us in these locations for more gaming and hardware updates:


Editorial, Testing: Steve Burke
Additional Testing: Patrick Lathan
Video: Keegan Gallick, Andrew Coleman
Рекомендации по теме
Комментарии
Автор

Slide at 05:10 for Quake II RTX should read "R9 3900XT and X570 Crosshair VIII Hero, " not 10700K. Sorry about that copy/paste error. FYI, we will visit this with RTX IO once we have something to work with for it!

GamersNexus
Автор

The variation between gen 3 and gen 4 is like my annual pay increment this year: it exists, it is real, but it is also functionally irrelevant and no human being would notice the difference.

Veni_Vidi_Vortice
Автор

Lol that benchmark at 4:11 gets absolutely murdered by the youtube compression

vaguedirector_
Автор

"...doesn't mean it would mean anything to a human player"
*shows cat player*

I love it.

thebakedgoods
Автор

Great video, I’m almost done getting all my pc parts and I was debating the vertical GPU connection because it is 3.0 only. Seeing how negligible the difference are helped make my choice. Thank you for this vid.

ejvargas
Автор

8:33 "the difference is real... but its also irrelevant" very good summary of gen 3 vs 4 lmaoo

alanschwartz
Автор

Steve: "No human will notice these 0.04 milliseconds gap."
Bounce Alerts: We'll notice that 3080 stock in 0.01 millisecond.

soapa
Автор

"I don't think anyone knows what a Linus is, but I digress". This is the real content, right here.

TheKazragore
Автор

The camaraderie between the top tech channels is genuinely nice to see.

Doofunator
Автор

This guy has 1.07M followers, but he is still underrated IMO. Thanks a lot for this great benchmarks. Can't imagine how much work he put on each of his videos.

mhnoni
Автор

Much thanks for this Steve. There's currently an issue with PCIE Gen4 cards and the USB ports on certain Gigabyte X570/B550 boards. Setting PCIE to Gen3 in the BIOS resolves all these issues for now. Good to know that we're not taking too much, if any performance hit when using this workaround for now.

cellar-door
Автор

04:07 - Compression quality has left the room

lauril
Автор

4:16 The YouTube compression is strong with this one.

shadowkiller
Автор

😌 this is what I've been waiting for

JarrodsTech
Автор

Would love to see this revisited (even if briefly) with ryzen 5000, just to verify the 3900x didn't skew the data in any way

MaxNakfoor
Автор

the definition of a Linus: He who drops expensive hardware.

ryant
Автор

I’m a professional cuda programmer. My job was to literally test the gpu performance in a certain system with dual cpu socket. I’ve made custom software that test memory bandwidth between the ram and gpu, wrap size optimisation, kernel optimisation ect... if you ever need help for workstation testing I can help you with that. For instense, we can obtain more acurate results by runing a certain program that will acces differents memory range of the graphic card and Mostly show the difference between cpu memory and pined memory.

rndargis
Автор

4:09 RIP YouTube's compression algorithm.

justicesportsman
Автор

"The difference is real but also irrelevant" - My favorite phrase from the video.

kryztovalyn
Автор

has anyone else noticed a marked increase in mentions and interaction between Steve, Linus, and Jay this year? I'm loving it personally.

jtlewis