Who's planning to buy 5700(?) GPU

tangoseal

[H]F Junkie
Joined
Dec 18, 2010
Messages
9,743
I'm excited to see what the new rdna plus pcie 4 can deliver.

The bond between the two is that AMD specified that Navi RDNA has performance gains when paired with pcie4.0. It's not a guess as to whether or not it can saturate bandwidth or if it is needed. It was specifically mentioned that the architecture is vastly more powerful over pcie4.

I'm probably going to get one. I'm not comparing to rx cards pound for pound.
 
Last edited:
I doubt the mid-range Navi stuff is going to need PCI-E 4.0 bandwidth. I don't think PCI-E 3.0 x16 was a bottleneck for midrange cards.
 
i'm interested in seeing what RDNA does but going to wait for navi 20.
 
  • Like
Reactions: Ronin
like this
i want to know the specifics as to what rDNA is actually is. Rumor reports were throwing NAVI as the last of GCN, while Lisa said GCN is still the compute KING GPU while NAVI is this thing that no one knows about but is geared towards gamers. Marketing lingo or actual stuff. I`m in depending on reviews and $/performance. But if its the same old same old as Radeon has been putting out, I`ll go last gen 580/90 new or team green.

Tight budget, and the rig its going into would be a slap on and never look back type of system.
 
i want to know the specifics as to what rDNA is actually is. Rumor reports were throwing NAVI as the last of GCN, while Lisa said GCN is still the compute KING GPU while NAVI is this thing that no one knows about but is geared towards gamers. Marketing lingo or actual stuff. I`m in depending on reviews and $/performance. But if its the same old same old as Radeon has been putting out, I`ll go last gen 580/90 new or team green.

Tight budget, and the rig its going into would be a slap on and never look back type of system.
There was some discussion of this on the Phoronix forums.

Navi is GCN ISA but not GCN architecture is what they explained.

https://www.phoronix.com/forums/for...-in-linux-5-3-mesa-19-2?p=1103266#post1103266

RDNA is GCN ISA but not what you think of as GCN architecture.
 
Thanks, however for the layman like myself, it still does not clarify what that means in terms of performance. The simple question is "rdna" the answer to the price AMD will be putting on these cards? If they ask 2070 levels of $, will they deliver
Impossible to answer yet :) The portion of your question I was answering is "the specifics as to what rDNA actually is". It's the name of their new architecture.

GCN and RDNA are architectures, just like Turing or Pascal etc. Navi is a specific chip built on the RDNA architecture.
 
I'm excited to see what the new rdna plus pcie 4 can deliver.

I'm probably going to get one. I'm not comparing to rx cards pound for pound.

It depends a little on pricing I was checking Vega 64 (sapphire nitro+) pricing and they end up just under 400 Euro. According to someone you can under volt then move up the power slider and you get excellent performance.
 
I really hope AMD finally takes the lead in efficiency. I mean, it would suck big time if Turing at 12nm ends up being more efficient than Navi at 7nm.
 
Navi are on a short list, as i am on a almost brand new X399 motherboard pcie 4 brings me no joy, and if pci 4 mean jump to DDR 5 also, then really i just jumped on DDR4 and 64 GB of it.
 
I see no reason to buy one. I buy top-end cards for my gaming box and cheap cards or hand me downs for my other boxes.
 
I doubt the mid-range Navi stuff is going to need PCI-E 4.0 bandwidth. I don't think PCI-E 3.0 x16 was a bottleneck for midrange cards.

Look at my original post at top. I added why 4.0 is necessary.
 
No me but, that is because I already have good stuff and do not game like I used too.
 
I am awaiting more details. I don't need 2080ti performance (would of course be nice), so if AMD can drop a bombshell and give 2070 performance for a great price I may be a buyer.

I have stayed away from AMD gpu's for a long time, but I am curious to play with one again.
 
Look at my original post at top. I added why 4.0 is necessary.
Yeah but how much of it is true ;)
You can stretch it and do it like AMD and show the bandwidth demo and claim it needs it to perform better, but in reality how much would that card suck if it did not perform near as well on PCIE 3.0 that would seem the most stupid thing for AMD to do cripple your own products especially since most people still run Intel motherboards.

Because when people see this and the first thing that would come to mind is: screw this I'm buying Nvidia ...
 
The bond between the two is that AMD specified that Navi RDNA has performance gains when paired with pcie4.0.

Where did they specify that? You aren't talking about that horribly misleading PCI Bandwidth test they ran on stage?
 
Where did they specify that? You aren't talking about that horribly misleading PCI Bandwidth test they ran on stage?

I'm talking about literally them stating that the real benefit was coupling Navi with pcie 4.0.

I'm not making any claims, just reporting what was said.
 
Even 2080Ti does not need full speed PCI-E 3.0 so what use could lower performing card have from PCI-E 4.0?
It is there because it is there and other than that and maybe ~1% of performance gains there will be nothing gained
 
Even 2080Ti does not need full speed PCI-E 3.0 so what use could lower performing card have from PCI-E 4.0?
It is there because it is there and other than that and maybe ~1% of performance gains there will be nothing gained

I don't know. Maybe there is some super high bandwidth required to make up for the really small GPU die? Well know when embargoes are over. But it's probably just marketing voodoo for now. Too early to tell.
 
If the purported performance and price is true, there's really no reason for anyone to get a 5700 if they have a 1080/Vega 64 or above. And for anyone with a 1080 ti or above, this would be a downgrade.

You can throw out the window those theories on PCI-E 4.0 giving any real benefit. As mentioned, PCI-E 4.0 will get you maybe ~1% improvement on a 2080 ti: PCI-E 3.0 is by no means bandwidth limited, even with top-end cards. That's not to say AMD won't talk about supposed benefits of Navi and 4.0 until the cows come home to sell cards, chips and mobos...
 
Thats true, if you have a newer high end card, then what AMD dish up with soon will probably not be for you.
I was on a GTX 570 that then died so i had to get a replacement 1060, which kinda sucked in regard to timing, but at least i can compute and sit here on my ass waiting for things i can burn my piggy-bank content on.

I am now thinking maybe AMD release something more hard hitting at a steeper price later on, as it have been said again and again that the up coming navi models are just mid range, so the question are ? is that mid range like now mid range or will mid range need to be redefined again after navi.
As you know mid or high / low end, are constantly changing more or less.
 
I'm talking about literally them stating that the real benefit was coupling Navi with pcie 4.0.

I'm not making any claims, just reporting what was said.

Then provide the exact quote where they say this.
 
At this point I doubt I am buying a 5700. There look to be a new crop of used 2070s about to appear in addition to the newer faster SKUs. Right now I've got nothing though, so waiting for actual numbers doesn't hurt me.

But if Navi isn't going to be >2070 performance or <2070 money, why wouldn't a non-fanboy just buy one of those instead?
 
I don't know. Maybe there is some super high bandwidth required to make up for the really small GPU die? Well know when embargoes are over. But it's probably just marketing voodoo for now. Too early to tell.
smaller die size does not affect anything, especially external communication

PCI-E 4.0 is something new AMD platform will support so obviously good opportunity to advertise Navi as the first GPU to support it also despite it in itself doing nothing except maybe for situations which do not support full 16 lanes... which given death of multi-GPU is not as importand

still... its new shiny tech : )
 
I am glad i am just messing around mid range, 1000 - 1200 USD for high end GFX cards thats 2 X or more of the money i am willing to part with, and such a product i am sure would not bring me 2 X the joy.
At least buying computer hardware are not as stupid a investment as buying a new car here in Denmark. :)

More dedicated people will probably find it much more easy to part with that kind of money, i never did not even when computers was pretty much the only thing in my life aside for drugs.
 
Pci 3 does not hold back a 2080ti in anyway. Pci 4 is nothing more then a marketing gimmick for AMD.
 
Would PCI-4 be beneficial in a 4 GPU type workstation scenario?
 
if it offers more than 16GB vram then maybe, else i'd be set on vega.
 
Pci 3 does not hold back a 2080ti in anyway. Pci 4 is nothing more then a marketing gimmick for AMD.

on consumer GPU's it's definitely a gimmick but also helps sell x570 boards so why not as long as it doesn't significantly increase the cost of producing the cards.
 
on consumer GPU's it's definitely a gimmick but also helps sell x570 boards so why not as long as it doesn't significantly increase the cost of producing the cards.
From rumors the x570 boards are going to be pretty pricey.
 
From rumors the x570 boards are going to be pretty pricey.

do i have to answer your post like you answered snowdogs? cough cough... lol jk :p (sorry had to do it)



yes they're expensive due to the trace lengths and higher quality pcb's needed for pcie 4.0, that shouldn't be an issue on gpu's.
 
Last edited:
well with the price of my X399 motherboard ( Aorus extreme ) i would assume it was made with the very highest standards in existence at the time, and really i have heard nothing new in regard to new innovations in the field of PCB making.
I would assume AMD would leave it to their board partners weather or not to offer this if they have made boards that have the hardware that make it possible at all.
 
well with the price of my X399 motherboard ( Aorus extreme ) i would assume it was made with the very highest standards in existence at the time, and really i have heard nothing new in regard to new innovations in the field of PCB making.
I would assume AMD would leave it to their board partners weather or not to offer this if they have made boards that have the hardware that make it possible at all.


Even a mid-range x570 board is going to come in North of $200+. PCI-E 4.0 requires etremely short copper runs in order to keep the signal intact...it can go 3.5-4" or so, but after that you need either redriver (costly) and PCB with much thicker copper traces, which means many more layers, which then drives the cost of the board up...

Part of the issue is that AMD didn't just do a first gen launch and have PCI-E hanging off the CPU for say one x16 slot and a x4 NVME slot...They went ahead and made the entire CPU-SOUTHBRIDGE use 4.0, which is a long run that requires agsin a higher quality layer board.

AMD did this as a slap in the face to Intel, since Intel is stuck on PCI-E 3.0, especially for their CPU-SB interconnect. This gives them a huge marketing point, as they are again then the "first" int the field.
 
  • Like
Reactions: Dan_D
like this
Back
Top