Possibly a better Vega: My take on what AMD should do

[Vya Domus]

You missed the point where more Shader Engine increases geometry processor. As you can see yes current design can fully utilize all the CU's but that only good for compute thats why they so good in mining. They dont solve the geometry performance of the design and this hurts in gaming performance. Its an unbalanced design

But my point is that this isn't their main limitation , there are other more fundamental aspects that make GCN somewhat undesirable for gaming.

They probably won't bother making such a small change to Vega and releasea new chip.

 

[R-T-B]

What level of standards? WCCFtech?

https://wccftech.com/amd-navi-gpu-spotted-in-linux-drivers/

https://www.techpowerup.com/239794/amd-navi-found-secretly-hiding-in-linux-drivers

I think even WCCFTech would've refused what you posted. Not that I would've personally relayed WCCF techs works... but TPU still has to maintain some level above blatant ranting.

It's been a bit since I was an editor, so yeah I can say things here slip from time to time too. But at least when I was around, we tried, and I'm pretty sure they still try.

 

[Apocalypsee]

ALL architectures are a compromise, you haven't solved the yeild issues of aprox 30% more chip per chip.
And you couldn't, no one can at this point that's the singular reason all three big chip pc companies are ALL talking modular mcm chips from here on in, because it increases die area efficiency to do separate bits separately and cheaper if possible.

And as is obvious big chips cost big money on small nodes.

That is why I 'created' the small Vega first, when the yield is better and/or a node shrink they could proceed with big Vega next.

But my point is that this isn't their main limitation , there are other more fundamental aspects that make GCN somewhat undesirable for gaming.

They probably won't bother making such a small change to Vega and releasea new chip.

Its not a small change, they been on 4 Shader Engine since Hawaii back in 2013 because of GCN limitation, on NCU they removed this but why do it if they dont bother increasing it?

Again this is just my take on this, what I've seen from reviews and throretical tests to elevate GCN weakness.

 

[xkm1948]

I think even WCCFTech would've refused what you posted. Not that I would've personally relayed WCCF techs works... but TPU still has to maintain some level above blatant ranting.

It's been a bit since I was an editor, so yeah I can say things here slip from time to time too. But at least when I was around, we tried, and I'm pretty sure they still try.

So all the Vega owners' concern for a specific Vega feature that should be working is all made up stuff? Like one troll registered several hunderd reddit account to post these things and upvote themselves as well?

That is why I 'created' the small Vega first, when the yield is better and/or a node shrink they could proceed with big Vega next.


Its not a small change, they been on 4 Shader Engine since Hawaii back in 2013 because of GCN limitation, on NCU they removed this but why do it if they dont bother increasing it?

Again this is just my take on this, what I've seen from reviews and throretical tests to elevate GCN weakness.

I believe instead of a small Vega, a larger Polaris would be better.

 

[TheoneandonlyMrK]

That is why I 'created' the small Vega first, when the yield is better and/or a node shrink they could proceed with big Vega next.


Its not a small change, they been on 4 Shader Engine since Hawaii back in 2013 because of GCN limitation, on NCU they removed this but why do it if they dont bother increasing it?

Again this is just my take on this, what I've seen from reviews and throretical tests to elevate GCN weakness.

Because four dies with four Aces and four shader arrays appiece will need to accomodate more shader arrays , making one chip with more array would net no performance gain and increase die size but yes they now need to be able to address more then four arrays at once.

 

[R-T-B]

So all the Vega owners' concern for a specific Vega feature that should be working is all made up stuff? Like one troll registered several hunderd reddit account to post these things and upvote themselves as well?

Social Media trends are not newsworthy unless the numbers are somehow able to be validated.

 

[eidairaman1]

Social Media trends are not newsworthy unless the numbers are somehow able to be validated.

Aka FAKENEWS LOL

 

[Apocalypsee]

Because four dies with four Aces and four shader arrays appiece will need to accomodate more shader arrays , making one chip with more array would net no performance gain and increase die size but yes they now need to be able to address more then four arrays at once.

Hawaii comes with 8 ACE, only recently they reduce them to 4 and add a couple of hardware scheduler so its a non issue if they wanted to increase them to feed additional Shader Engines.

 

[TheoneandonlyMrK]

Hawaii comes with 8 ACE, only recently they reduce them to 4 and add a couple of hardware scheduler so its a non issue if they wanted to increase them to feed additional Shader Engines.

I am not saying they can't or wouldn't just if im right amd gain with scale, like threadripper and its pro brethren they gain performance and vast amounts of scaling potential , to be honest I think my four vega analogy extreme it's more likely amd will design a 1-2000 ish shader design with infinity fabric a hbm controller/v3 and a ddr6 controller , scaled to 7nm it would make perfect sense to package it in various forms from bottom to mid to high including four on an interposer with four stacks of hbm3/?
At this point what you suggest doesn't make sense to me they have so many devs working on gcn as it is but they have and are opening up features ,the pro driver now support primary shaders (hello geometry) its a start ,i expected devs to see these NEW features first or how would any game start using them.
Really explains the founders edition to me it's going to get these features first, roll on Rpm mining joy perhaps.