AMD NAVI Radeon 5700 Product Announcement thread

I'm thinking you got your prices, or namings messed up. 2070 was never $599, i could point out more, but again, i'm assuming you miss-typed something, somewhere there, over and over.

Unless you're talking about the FE cards, which, makes no sense to talk about FE cards when you don't have to purchase FE cards. That'd be like regarding any GPU released as Galax's HOF cards that generally cost much, much more then the base cards MSRP.

 

Hi @Hilbert Hagedoorn, more typos

5th page, under anti-lag img "it can bring ou an up-to 32ms decrease in latency"

 

Yeps, i think that FE version.
grab those reference from other site.

 

They stated multiple times that for each single game they shown results for API which delivered best fps for each compared GPU. That's fair play.

No, they did not show nVidia's Vulkan results because they were worse than DX11.
So, it is other way around in reality.
AMD leveraged modern API while nVidia had better results with legacy API.

 

I am quite excited by new GPU hardware. Interesting new techs that will hopefully boost gameplay experience in mid gpu space + looking forward to seeing true benchmarks and dx12 performance (well spotted about dx11 api btw), and hopefully big navi around the corner. The increasing homogeneity between consoles and pc through amd as well gives me hope in future that games will move across platforms more easily while retaining higher end graphics options for those systems that can support it.

 

Wait! If Navi can OC like Radeon VII then it may hold ~30% OC headrom from the "Game Clock"?

 

If 5700XT actually delivers similar or better performance stock then 2070. AMD actually has caught up in terms of performance imo. And perf/watt. And well it seems like no need to have way more shader cores either. Quite nice actually.

Especially if that TDP is typical board power since 2070 reaches roughly similar usage.

 

Disappointing. Can't believe those power figures at 7nm and the prices are about $100 too high to be interesting. Sad that I am hoping for Intel to put out a competitive product at this point.
When will a decent upgrade for my 1070Ti be available and under $400?

 

Sadly no they did not caught up per/watt wise nvidia is still using the 14nm node i guess as far the performance crown is not contested they do not care to move to a more expensive node.

 

I do not think so. Automatic boost range is so high due to ability to achieve much higher utilization of everything inside CU for semi-new games => high power draw => TDP limit => lower clock.
And then this scheduling can do pretty impressive things for older games which do not have ability (requirements ratio) to use CUs fully => low power draw => not reaching TDP => higher clock.

I would say that instead of doing some manual OC one can benefit most from increased TDP/fine tuned voltage.

Question is:
Will reference/AIB cards have dual BIOS? Will it be unlocked?
If not, we'll not be able to get most of those GPUs from get go.

 

no it isn't, proper approach is to bench both and show both.

api vs api is just fudging numbers, since most games are still dx11 where AMD is weak at still (maybe that will change now they have a multithread primitive capable card?)

 

They sad that the "Game Clock " is the actual clock while playing, so OCing can improve this, it maybe limited by thermal/TDP so this is why it's much lower the max boost.

 

Considering the fact that AMD is still behind on performance, they will push hardware to its power and thermal limits.

 

5700 XT has 10.3B transistors. 2060 has 10.8B transistors.
I do not think they are behind. Performance per clock per transistors is on AMD's side. It only remains to be seen if it is by 10% or 20%.

 

AMD missed with pricing here. No current Vega64 or even Vega56 owner is going to switch.
And for current RX580 owners those Navi prices are out of their budget zone.

Ryzen CPU specs and pricing looks good tough, and will definetly buy one. Probably 12C one.

 

If a GPU does better in a certain API vs another, THAT will be what the consumer will (logically) be using, and therefore the metric that matters most to buyers. Anything other than that only serves to mislead as to a GPUs fullest potential.

 

That is NOT how things work, it gives a invalid idea that all games can use vulkan natively and gain fps from it.

You will never see nvidia provide comparison references using different API's because its a blatant misrepresentation.

 

Article Page 2 says they have HDMI 2.1, but the Page 5 says it's only HDMI 2.0b.

If you care about the price competitiveness, you shouldn't hope for anything from Intel. All decent and good Intel products have always been and will always be expensive. Only the crappy ones are cheaper. Assuming they can immediately make a good GPU on their first try (which is a lot hoped, but they do have unlimited budget works, so who knows), it will be expensive if the performance is at all comparable to AMD, let alone Nvidia. One thing Intel would rather die than do is charity.

 

You do see how this can be misused, dont you? A DX11/12 game that does well with Nvidia in 11 but poorly with AMD, yet better in 12 with the latter. It would be poor form for any site to show just one API and not the other in such a case. That is the crux of the argument, nothing more.

 

You can't misuse it if you share the performance results of both api's in a multi-api game.

But it is a misuse to share vulkan vs d3d11 since vulkan is far higher grained for multithreading vs dx11 which is mostly hacked in.