Test Setup
We are using an AMD Ryzen 7 9800X3D for testing. We are using the motherboard’s current default BIOS settings on the latest release BIOS. We have enabled PCI-Express Resizable BAR in the motherboard BIOS. We have enabled EXPO I in the BIOS for the memory. We are using a fresh install of Windows 11 Pro with the latest Windows Updates, including the 24H2 update. We have the latest drivers installed including the latest chipset drivers from AMD. We are using the “Balanced” power profile in Windows Settings.
NVIDIA provided a press driver for the NVIDIA GeForce RTX 5080, and we are using this driver (572.12) also on the NVIDIA GeForce RTX 4090 testing data and GeForce RTX 4080 Super testing data. All data was re-tested in this review with this new driver. We are using the latest Radeon RX 7900 XTX driver available.
Discussion (19 replies)
Join Discussion →Long story short get an RTX4080 while you can.
No surprise here, given the specs, I didn't expect the RTX 5080 to do any better.
The RTX5070Ti hopefully will be a much better option as it's supposed to be right behind the RTX4080.
I'm very disappointed, I can't remember the last time a next gen x80 card wouldn't beat or at least meet the next up tier.
I remember when I got my 1070 that it was supposed to be on par with the 980ti
Looks like 5080 is a dud. I hope people read reviews before buying them on ebay for hundreds more. Maybe holding on to the 4090 is a good idea. But, the 4090 just increased in resale value!
So why is the 5090 not anywhere twice as fast as the 5080? what's holding it back?
Honestly, I'd love to see some deep dive on if the 5090 is being constrained by CPU. I've seen some side mentions on a couple sources saying it seems that way but no deep dive.
Reminds me of the 1000 series to 2000 series jump, only without the DLSS carrot there to give you a reason to bite.
On a side note, I've seen a few posts claiming that the new transformer DLSS mode is much more improved quality than the old one. So much as DLSS balanced mode being indistinguishable from quality mode and even performance mode being much better than before.
That should be pretty easy to test with time available.
System 1: Ryzen 9700X Down clocked, or even just set to 65w
System 2: Ryzen 9800X3D pushed as high as you can go with clock
run both systems at 4k and see if there is any significant difference
I mean it's a BIT more than that. You might want to use Intel's monitoring tools to see how things are working and better project where there is a bottleneck between the platforms for differences.
That's an interesting idea, give me some ideas (examples) of what to test to accomplish this. I also think it is true that the 5090 is being held back by the CPU, I said as much in my review.
Thanks for that suggestion, as I said above, I personally think that is a great idea to test. Right now, something like that wont be soon, but I've got it written down in my potential reviews doc.
It would need to be basic. at 4k whatever refresh is the GPU at 100% while the CPU is running at max ghz high utilization across cores? That's how I would logically think a held back device would look.
But it would need to be a slim test.. like perhaps raster only... I'm not sure on that.
Jay has a video on how to use the intel tool to track and see if the CPU or GPU is throttling. *shrugs*. It seems kind of esoteric to me. All the info I've seen calling out CPU throttling has been at higher res where NORMALLY that would be at lower rests... so it's interesting.
[embedded media]
Presentmon is the tool that you can use... just a idea. Jays video has some info on how to use it but there are better tutorials online.
Gotta wonder if a cutdown GB202 die is on the table for a $1499 5080 Ti w/ 24 GB VRAM.
It appears the 5080 is a good overclocker. If that is consistent across a bunch of samples then it's a great candidate for water cooling to open up some possible additional headroom.
Good point. A lot of folks are complaining about how close this is to a 4080 SUPER but things could get really interesting when OC'd and liquid cooled since you don't have to worry about hitting a 600W power limit for the cable.