By using this site, you agree to our Privacy Policy and our Terms of Use. Close
HollyGamer on 13 March 2019
Edit Thread
Pemalite said:

Not as major as you would think. It's evolutionary not revolutionary improvements.
Even then, on the odd occasion there was actually regression.

For example... Gamer Nexus did a comparison of Fiji and Vega's IPC. Vega didn't look all that pretty...
https://www.gamersnexus.net/guides/2977-vega-fe-vs-fury-x-at-same-clocks-ipc

Architecturally Maxwell and Pascal are brothers, Pascal did introduce a few improvements, namely an update to Delta Colour Compression which meant a good reworking of the ROPS.
However, Pascals main focus was on reducing leakage and noise so as to drive up clockrates instead, nVidia was pretty successful on that front.

Also 24nm wasn't a thing.
It was 28nm and 16nm respectively.
https://hexus.net/tech/reviews/graphics/74849-nvidia-geforce-gtx-980-28nm-maxwell/
https://hexus.net/tech/reviews/graphics/103213-nvidia-geforce-gtx-1080-ti-16nm-pascal/

Increasing the clockrates has a better overall effect than just blowing out the CU count... Because like I alluded to prior, Graphics Core Next is limited to 64 ROPS (Although independent of the DRAM), 256 TMU's and 4x geometry units... So every-time you increase clockrates, you increase the performance of those components.

There is a reason why Vega 7 took the approach it did, it cut down on the CU count, but dramatically increased clockrates over Vega 64, it's a balancing act.
But for Graphics Core Next to scale beyond 64 CU's, there needs to be a significant reworking of the architecture to fix it's inherent bottlenecks that holds back gaming performance.

Navi will perform better than Polaris, but less than Vega. That's the reality of the situation.

Navi is Polaris's replacement, not a high-end GPU.

It's the same rubbish that happened prior to Polaris's release, everyone was hyping it, but at the end of the day... It was a GPU that was destined for the mainstream markets, just like Navi, so don't expect high-end performance.

And I know this for a FACT because of AMD's statements... And when it comes to hardware, I tend to be right more often than not.

Doesn't matter? It's not meant to test performance, it's meant to test load power consumption.
You only need the one gaming test, which is all that Anandtech provided, unless you are suggesting that you have a better knowledge-base than even Anandtech on this topic?

https://www.anandtech.com/show/10446/the-amd-radeon-rx-480-preview/6

I mean, you could cling to the power-virus "furmark" if you wish, but AMD's GPU's throttle in that.

If we take a look at Polaris Version 2. Aka. The Radeon RX 580. Power consumption is higher yet again, higher than a GTX 1070 which out-performs it.
https://www.anandtech.com/show/11278/amd-radeon-rx-580-rx-570-review/16


What?

A youtube video based on a youtube video rumor.
Grains of salt shall be had on this day.

 The leaks I have seen pegs Navi at Vega 56 levels. RTX 2070 beats Vega 64... So...
https://www.theinquirer.net/inquirer/news/3072523/amd-navi-benchmarks-leak

 A revolutionary or not it's still a big jump in improvement even if it's just because of small die size.

Also why the hell you comparing Vega with Fiji, you should compare it to Polaris instead Vega? 

Pascal and Maxwel on architectural level is the same the same with all GCN across the AMD GPU after Fiji , the are improving on the driver level , you just mentioned by yourself.

And why the hell are u using power load consumption as benchmark? and on top of that comparing it to GTX 1070 . Fiji to Polaris as an improvement can only be compared to GTX 970 or 980 and it need  games benchmark not power consumption benchmark. The point of the benchmark is to prove that GTX 970 and GTX 980 as high end GPU can be performed with mainstream GPU like Polaris. 

Navi will perform better or not we still don't know yet, it's only rumor, this just pure speculation based on the latest rumor and on top of that trusted source across the Internet, anyone can agree or not .

Also We don't know if AMD can increase their ROp's or their shader processor , they probably cannot break 64 CU limit but they can increase shader processor.  And yes it's rumor but a rumor or not i have a source , nobody can debunked either right or wrong, because that what we have.

You just assuming AMD is already in the dead end , but we don't know yet, many expert like Adore TV out there who also expert in this, has a lot insider and got a lot of info and newsleak ,think that AMD making a big improvement of GCN for Navi.

Your logic is "AMD is performing bad now, so in the future they will perform bad as well " that's how you think on this matter.  

So yes It's still in a realm of possibility. Also we also got a new benchmark that suspected as Navi GPU and on top of that it's an  igpu or APU https://wccftech.com/amd-navi-gpu-radeon-rx-benchmarks-graphics-performance-rumors/. You need to remember it was mentioned as  low end Navi GPU (navi 12 or 15 probably) , and PS5 are rumored to be using Navi 9 ( a cut down version) .

Also your standar of mainstream GPU for near future is very pessimistic , all High end GPU from 2016 and 2017 will just be a mainstream GPU on 2020 , it's moore law. Even if you say it's more then that AMD is not that stupid. 

Last edited by HollyGamer - on 13 March 2019