RJARRRPCGPI feel like this might just be "a newfangled version of an HD 5450"!

... it is, what's your point?

trsttteWhat you select in the bios doesn't matter, the card will only run at the specs it supports, gen 3.0 in your case

Been here since 2007, So you aint telling me nothing new, point being is AMD didnt need to make the cards 4X etc.

AssimilatorIt would indeed be impressive if they could force the card to use lanes that it physically does not have.

lol, I just meant I wish AMD would instead of putting PCIe 4.0 x4, put PCIe 3.0 x8 or PCIe 2.0 x16 on this, but I do understand this graphics card is using a GPU that was meant for laptop applications where the CPU PCIe lanes are capped at x4 (for ultrabook CPUs, this is the case). To be honest though, y'all can clown on AMD for making this a PCIe 4.0 x4 GPU, I'm over here running my Quadro M4000 at PCIe 3.0 x2 lmao

Berfs1I'm over here running my Quadro M4000 at PCIe 3.0 x2 lmao

That's the real shit, right there :laugh:

In all seriousness, though, I take it for your usage bandwidth doesn't matter much?

just found this about pricing

videocardz.com/newz/amds-199-radeon-rx-6500xt-officially-gets-e300-324-msrp-from-asus

DeathtoGnomesjust found this about pricing

videocardz.com/newz/amds-199-radeon-rx-6500xt-officially-gets-e300-324-msrp-from-asus

Well, fuck.

On the other hand, I've heard of ASUS as a more expensive brand? So maybe there's hope other manufacturers' products will be less expensive. Though I don't really expect anything anymore.

Gotta quote myself on that one:

if someone is looking at bottom of the barrel stuff, they should not even be thinking about this card for a while to begin with.
Because, let's face it, it WILL be price inflated, even if it won't make a good miner. Demand WILL be crazy, and sellers WILL take advantage of it regardless

Was not even a bold prediction.

£300+ for this POS! lolololol

Mind you, it does have two fans that go weeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee

Assimilator... it is, what's your point?

I don't know what his point was, but I gotta say: assuming that's true, I look forward towards finding these guys for cheap and free down the road. Huge upgrade over the 5450s, etc. that common eWaste bins.
Also, if that's true, then that implies very good 6nm yields.

By being only 4.0 x4 allow the PCI lanes to be used else were, instead of 8\16 being taken up and not even being needed ?.

TomorrowThe point is that people who use it on PCIe 3.0 or perhaps even 2.0 board will also be limited to x4 link but wil much less bandwidth than 4.0 x4 would provide. Obviously 4.0 x4 is just fine for this card but it may not be for 3.0 or 2.0 users.

Based on TPU's GPU database and assuming 6500XT has roughly the performance of GTX 980 it could lose up to 14% with 2.0 and up to 6% with 3.0: www.techpowerup.com/review/nvidia-gtx-980-pci-express-scaling/21.html

"TPU's GPU database and assuming 6500XT has roughly the performance of GTX 980 it" WHAT>>>>???????????? GTX 980... ID BE BETTER JUST STICKINGWITH MY RADEON R9 Nano 4GB WTF IS THIS DOG $H!T AMD.... F**K YOU AMD........... MY 6 year OLD R9 Nano 4GB or R9 FuRy X WOULD DESTROY THIS PIECE OF $H!T!!!!! RX 6500XT WTF THIS IS F**CKING STUPID AMD AMD = MAD

GAMING = DEAD

Hardware unboxed did simulated test using 5500XT with similar scenarios:


And holy crap, it's awful.
In fact in many games even AMD's intended PCI-E 4.0 x4 configuration cripples performance significantly, nevermind 3.0 x4.

Between this, gimped decoding options, and undoubted overpricing, only one conclusion: it's garbo. AMD is taking a piss.

The 5500XT has 2MB L2 while 6500XT has 1MB L2 and a 16MB L3 which along with 4gbps higher effective memory speed so at x4 it should get filled and buffered by the L cache on the GPU better with the 6500XT. There is also driver level FSR coming to the hardware and perhaps we'll see driver level FSR on par with Godfall's quality.

There is probably a reason AMD didn't cripple the 5500XT to PCIE 4.0 x4 and video certainly highlights that, but the underlying hardware is a fair bit different. There is also a reason AMD didn't make the 6500XT 8GB VRAM as well which is you look at the hardware between both there are other reasons behind that it would be quite anemic if they had done so at least w/o even faster GDDR memory and even then hardware itself would struggle more in other import area's in regard to scaling performance towards higher image quality settings and resolutions.

6500XT is just a card fill a part of the product stack that's never very exciting and at the same time short of the pandemic and crypto wouldn't have been launched as a discrete GPU offering in the first place or with a even cheaper MSRP if it had.

windwhirlThat's the real shit, right there :laugh:

In all seriousness, though, I take it for your usage bandwidth doesn't matter much?

Sorry for the late reply, I mainly use it as a display card, but I also use it for video editing. PCIe 3.0 x2... IIRC that's 2GBps (half of an x4 NVMe for example), eh I haven't had any noticeable issues, but then again it could also be that I have never actually gotten to see the true performance of it when ran in x16 mode lol

Does anyone have two RX 6500 XT to test in a multi gpu enviroment ?
Thanks

eledx710Does anyone have two RX 6500 XT to test in a multi gpu enviroment ?
Thanks

Haven't seen any reviews looking at that. There were a couple looking at the 6800 XT in multi-GPU configuration, but that's about it.

But, if you have the money for two 6500 XT, get a 6600 XT. The 6600 XT has double the cores, double the VRAM, and uses 8 PCIE lanes. Less complications that way.

Or look at Nvidia's offerings. That's an option too.

Just thinking about MCM performance of navi 24, sort of.

eledx710Just thinking about MCM performance of navi 24, sort of.

Provided the MCM approach actually happens (I think it will, but who knows), a set of two RX 6500 XTs would be a bad comparison to draw conclusions from.

If AMD goes ahead with an MCM approach for a future GPU, it would likely implement some sort of Infinity Fabric, which would avoid some potential issues of going over PCIE (IF can provide more bandwidth, though it would depend on the implementation, plus reduced latency)