I even said there are people more knowledgeable than I. I just know from what actual Engineers at AMD, and Third party vendors have assessed about NAVI as consumer product line. There is a Navi 20 in the mix which will more than likely by their high end, but leaks from AdoredTv and a couple others that talk about TSMC process, and the kind of yields they are seeing have caused confusion since from what reputable leakers have said that it was re-spun hence the no show at CES 2019.
From what you have said you don't know anymore than I , I just know from people that I trust in their knowledge, and their connections like actual vendors that Navi will be price/perf conscious,.
I literally was replying to someone who said Navi will be more powerful than Radeon 7. ANd currently with what has leaked and what we know from AMD's actual Pipeline they layed out high end NAVI is later.
Radeon 7 is a stop gap card, hence why it's using MI50 chips in it and HBM2 memory and not GDDR6. It did not cost them more R&D t make it.
We know that RX 590 was literally a last minute product like the Radeon 7 that they are trying to get the best perf/watt they can out of the first lineup of NAVI. So them putting out those cards as just stop gap cards while they get NAVI ready for initial launch this year that will 100% replace VEGA/Polaris lineup as time goes on.
What they will release first is what we don't know. But from what AMD engineers have said them selves the focus is on perf/watt which means efficiency in their focus. Which is what Polaris was for it's time.
You are right there isn't enough info out there, as in knowing how many CU's we will have, currently their high end has 64-60 with Radeon 7 having 60 but vega have 64.
There's nothing incoherent about my argument if you actually took the time to see the things that have literally been talked about by AMD themselves, and knowing what their roadmap is.
I won't derail any further, i just reacted to a comment someone made saying navi would be more powerful than radeon 7. And that would be true if navi 20 had a lot of what your are talking about. But it also would have to have over 1tb of memory bandwidth which so for no card has except the RADEON 7. So in raw performance maybe the navi 20 is way more effecient has more CU's, better memory controller with GDDR6. But lacks the high band width of 1tb that radeon 7 has because of the 16gb of memory.
I mean it cost them close to $600 alone to build a radeon 7 I don't think they want another product which is that expensive. ANd seeing GDDR6 prices are still not as cheap as what GDDR5/GDDR5X were I'm inclined to believe they may be going for more price conscious cards, with maybe the RADEON 7and Navi 20 being their high end.
TechRadar using couple people that are hit or miss for sources is in the same boat I a, thinking that Navi 20 probably does exist but is not on the roadmap till next year.
So for when in May they announce the new chips we will have a lineup that will at some point compete with high end, but nothing in the 2080ti range until 2020.
Your original reply saying "Navi won't be faster than Radeon 7" was posted against my post in which I was discussing the problems with the Vega architecture and why I think Navi won't see the same ills befall it.
Go back and check. Hence why I called that post incoherent because it didn't follow logically from what I posted and led us off on this somewhat heated tangent.
Perhaps you responded to thw wrong person?
Reagrdless, I can see you're well read up on all the Navi rumours, I would simply only caution to take all that stuff with a pinch of salt.
Internet guys and even tech outlets can be wrong as sources can be bogus and plans can change.
Point is, it's not wise to plant your flag in the sand and make definitive statements about an unannounced GPU architecture when you even admit we KNOW nothing about it.
All we know is rumours and speculation and that is subject to change, embellishment and bullshit.
The anti-SSD folks should consider the trend on SSD prices and where prices will be 2 or 3 years from now.
Cost isn't the only factor though. Reliability while very good these days, may still be a concern for a console.
Can someone give me a simple layman explanation for the follow two queries?
1. Why has it been a trend for AMD GPUs to always have lower performance/watt compared to nVidia GPUs (since forever)?
2. Why do AMD GPUs always have higher raw "TF" value which do not translate into performance compared to their nVidia GPU counterparts?
With no insider knowledge whatsoever I would guess the biggest reasons are a combination of the following:
a) Market share and Sales Volumes -
Market share means NVidia gets paid and takes more per GPU than their competitor. More revenue means deeper coffers for larger R&D spend which is everything in semi-conductor development - you can't get around it.
Their higher sales volumes also mean bigger production volumes and this can be a huge advantage when speed binning parts for the various market sectors you intend to market products to. E.g. if you imagine a normal distribution of chips coming off the line based on how high they'll clock stably, with higher overall volumes you can shift your product boundaries for speed bins to the right.
Higher overall production volumes allows you to have more of your highest end parts meaning you can skew your enthusiast level cards to the absolute cream de la creme perf/watt dies coming off the line.
b) Microarchitectural Advantages
The advantages here pretty much play into the above, because a richer company with bigger coffers can afford more for researching and developing (and patenting) the best technology.
If next gen consoles do not come with NVMe SSDs, next gen consoles are already dead.
:-O...
*slowy backs away from this post*
