• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

** The AMD Navi Thread **

Soldato
Joined
25 Nov 2011
Posts
20,639
Location
The KOP
If that's the case then it must be doing something that only Navi can do so not just Sharpening.


Well say i had this feature for my Vega 64 on my 4k screen. I am running it to the wire. If i could drop to 1440p and make the image look somewhere close with this sharpening then it's in effect doing the same thing for me as DLSS is for those using it to claw performance they don't have. Basically i would be gaining the performance i needed without having to drop image quality to much. It's 2 very different ways of achieving a similar goal. Any how until this gets released and reviewed we won't really know. DLSS looked and sounded great until it got in games and was then slated for a bit. People are still not overly impressed with DLSS atm and it may turn out the same here.

This is what I keep trying to say but am sure you know on here it gets ignored. We really won't know much till July 7th just how good/bad this is.
 
Soldato
Joined
27 Mar 2010
Posts
3,069
How's Navi 10 a step in the right direction? It's worse price/performance than a Vega 64/56 and they've already got a stronger (Albeit EOL) GPU with the VII.
If it £300 then sure it'd be brilliant. It however is closer to £450.

Sure there's people who'd be happy with a Navi 10. But people who are happy to buy a Navi 10 price/performance wise would have bought a 2070 assuming no bias.
If my Vega 64 turned into a Navi 10 over night I'd be happy too.

Remember though Amd have addressed 3 main issues,(clock speed/current draw, cu performance and front end scheduling). From the looks of it their front end now has 60% less shader capability, (2560 vs 4096) but performance is still scaling around V56/64, there's probably another 10% to gain from clocking too. If the next jump is to a 6x or 8x wide geometry then we should see a die size around 330-375mm comparable to an Rtx2080/+15%.

Amd since Fiji/Tonga always needed massive dies and a high cluster of cu's to compete with Nvidia. This is the thing nearly everyone is forgetting, Navi is small and refined stepping stone towards making money as quick as possible, whilst simultaneously writing off the Loss making Vegas, whilst they gear up for a rework on phase 2.

But the problem is people aren't prepared to pay Navi prices.
 
Caporegime
Joined
18 Sep 2009
Posts
30,097
Location
Dormanstown.
Remember though Amd have addressed 3 main issues,(clock speed/current draw, cu performance and front end scheduling). From the looks of it their front end now has 60% less shader capability, (2560 vs 4096) but performance is still scaling around V56/64, there's probably another 10% to gain from clocking too. If the next jump is to a 6x or 8x wide geometry then we should see a die size around 330-375mm comparable to an Rtx2080/+15%.

Amd since Fiji/Tonga always needed massive dies and a high cluster of cu's to compete with Nvidia. This is the thing nearly everyone is forgetting, Navi is small and refined stepping stone towards making money as quick as possible, whilst simultaneously writing off the Loss making Vegas, whilst they gear up for a rework on phase 2.

But the problem is people aren't prepared to pay Navi prices.

If this had been released instead of the Vega 64, then fair enough.
Or had it been a price/performance part like the 48XX series then brilliant.
But this is just AMD releasing something that's literally a generation behind what it needs to be.
 
Soldato
Joined
27 Mar 2010
Posts
3,069
If this had been released instead of the Vega 64, then fair enough.
Or had it been a price/performance part like the 48XX series then brilliant.
But this is just AMD releasing something that's literally a generation behind what it needs to be.

Not wrong in anything you say, I agree but the 7nm node wasn't ready for the Navi to work. So they had to march on with Vega 56/64.

It's identical to the Polaris situation where Amd were making loss on grenada (390/x) and the 20nm node was cancelled, Polaris came out late but was a cheaper to manufacture more efficient hawaii/grenada replacement, whilst Fiji was a complete and utter failure.

Like yourself i'm stuck on a 2 year old v64, radeon VII isn't worth it and I'm not paying 600-1200+ fir a nvidia. Navi is expensive but at least for now Amd are in a healthier position to manufacture reasonable gpu's. It's just they should have had all tiers ready by now considering the delays and at a (low/mid and high) rather than just the mid range.
 
Soldato
Joined
27 Mar 2010
Posts
3,069
Not sure they really need a low end part - rx570 at £120 or whatever it is now is decent enough as a low end part. Only benefit a new part would offer would be to reduce power consumption

I guess you are right, it depends on the manufacturing cost vs market tier. I'd rather see 7nm Navi widely adopted in mobile (mxm) than the 14/12nm polaris which never got there though. But on the other side Gf12nm is still very efficient and capable at lower clocks/voltage, which would be perfect for mobile/entry level desktop.

This is where Nvidia still have the edge they can mark up the same chip for laptop and desktop.
 
Soldato
Joined
26 Sep 2010
Posts
7,146
Location
Stoke-on-Trent
570's surely going to be made EOL with it not being a 7nm part.
Perhaps depends on how many more ties to GloFo they have. I'm still not 100% convinced RX 590's shrink to 12nm was a one-off. I know it's been reported as largely "trivial" to produce 16nm GloFo designs on 12nm GloFo process because it's the same equipment, but it always felt weird to just do a single product with it. Perhaps instead of cutting Navi down even more AMD could respin the 570 and 580 on 12nm and take the power benefit (rather than ramping it up again for the 590). Maybe even drop in GDDR6 as it's supposedly pin-compatible with GDDR5.

That'd give some nice little cards in the £150-£200 bracket. Rebranded as RX 5500 and 5600 of course :p
 
Soldato
Joined
27 Mar 2010
Posts
3,069
Perhaps depends on how many more ties to GloFo they have. I'm still not 100% convinced RX 590's shrink to 12nm was a one-off. I know it's been reported as largely "trivial" to produce 16nm GloFo designs on 12nm GloFo process because it's the same equipment, but it always felt weird to just do a single product with it. Perhaps instead of cutting Navi down even more AMD could respin the 570 and 580 on 12nm and take the power benefit (rather than ramping it up again for the 590). Maybe even drop in GDDR6 as it's supposedly pin-compatible with GDDR5.

That'd give some nice little cards in the £150-£200 bracket. Rebranded as RX 5500 and 5600 of course :p

Yep I agree with this, especially if wafer costs are in the favour of Gf12nm.
 
Soldato
Joined
23 May 2006
Posts
6,723
Remember though Amd have addressed 3 main issues,(clock speed/current draw, cu performance and front end scheduling). From the looks of it their front end now has 60% less shader capability, (2560 vs 4096) but performance is still scaling around V56/64, there's probably another 10% to gain from clocking too. If the next jump is to a 6x or 8x wide geometry then we should see a die size around 330-375mm comparable to an Rtx2080/+15%.

Amd since Fiji/Tonga always needed massive dies and a high cluster of cu's to compete with Nvidia. This is the thing nearly everyone is forgetting, Navi is small and refined stepping stone towards making money as quick as possible, whilst simultaneously writing off the Loss making Vegas, whilst they gear up for a rework on phase 2.

But the problem is people aren't prepared to pay Navi prices.
And the thing is why should we be prepared to? I know I am repeating myself but I just can't grok that the same people who are hating on Nvidia for being greedy are whilst not championing AMD prices but are giving them a pass. Stepping stone or not these navis are just as poor value as anything from Nvidia and by the time they are out will likely be worse because current rtx cards will be cut in price and the super cards will out perform them as well.
Hopefully this Navi will be the beginning of a great GPU line but right here right now they are offering just as bad value as NV and NV have yet to show their hand with 7nm parts which will surely be what the full fat Navi will be up against next year
 
Soldato
Joined
27 Mar 2010
Posts
3,069
And the thing is why should we be prepared to? I know I am repeating myself but I just can't grok that the same people who are hating on Nvidia for being greedy are whilst not championing AMD prices but are giving them a pass. Stepping stone or not these navis are just as poor value as anything from Nvidia and by the time they are out will likely be worse because current rtx cards will be cut in price and the super cards will out perform them as well.
Hopefully this Navi will be the beginning of a great GPU line but right here right now they are offering just as bad value as NV and NV have yet to show their hand with 7nm parts which will surely be what the full fat Navi will be up against next year

Agree price/perf is the killer, but it will improve slightly just give it a little time. Unfortunately Nvidia and Amd formulated the price fixing with the rx590 and gtx1660-1660ti. With Nvidia out the gate first the rtx2060/70 then price matched Vega 56/64 (pre eol vega pricing).

The days of £150, £250 and £4/500 max tiers are long gone. There's always a gimmick to lock you in, last time it was g-sync, this time it's Rtx. It's clear now that we are entering the same situation that hifi did back in the 90's. Mini-midi systems replaced Hifi seperates, whilst hifi became more niche and pricey. The lower end of hifi stagnated .

(Consoles met parity with pc when the internet was good enough for multi player gaming).
(Gaming Mobile phone/tablets is a growing market)
(Game Streaming services, with the latency cure will be the future).
 
Last edited:
Back
Top Bottom