Where the hell is NAVI???????????

EastCoastHandle · 2 Mar 2019 at 17:42

Poneros said:
uh oh

I don't get it? This is the text book definition of FUD. Is there any insight to any of this?

Gerard · 2 Mar 2019 at 19:19

EastCoastHandle said:
I don't get it? This is the text book definition of FUD. Is there any insight to any of this?

The usual "tech press" BS imo.

Nasher · 2 Mar 2019 at 19:37

EastCoastHandle said:
I don't get it? This is the text book definition of FUD. Is there any insight to any of this?

Post contains "Jensen said..." so probably contains a large amount of BS.

EastCoastHandle · 2 Mar 2019 at 20:44

Nasher said:
Post contains "Jensen said..." so probably contains a large amount of BS.

I will believe that Jensen knows RTG unreleased SKU just like AMD knows Nvidia unreleased SKU. But what I don't understand is why this has been allowed for so long now?
However, it's very clear I wouldn't believe any bad news coming from Jensen no different then I would if it came from RTG about Nvidia unreleased sku.

bru · 2 Mar 2019 at 20:54

Nasher said:
Post contains "Jensen said..." so probably contains a large amount of BS.

Ok now I'm just confused, Poneros posted what Jim from adoredtv was talking about on twitter in regard to Navi, it has nothing to do with Jensen from NVidia. Jensen is only mentioned in relation to what he said about RadeonV11 being rushed.

humbug · 3 Mar 2019 at 12:10

nashathedog said:
So it was okay when AMD were diverting all the resources they could into the cpu side of the business to create Zen but now that succeeded they can't do it the other way round? For the last 6 or 7 years the gpu side of the business has suffered while R&D funds were diverted for Zen, As Lisa Su said some time ago they've been able to increase RTG's budget, so they're in the process of getting RTG back in the game, it just takes time.

Well no... AMD GPU's post the Bulldozer era, Hawaii (R9 290X) Polaris (RX 480 RX 580 RX 590) Fiji (Fury X)

That's 3 architectures during the development of Zen, successful architectures so i don't buy the idea that Zen was at the expense of GPU's.

Aside from that you're missing the point i'm making, they are making cards for retail to sell at a loss, not just Vega 56 and 64 but one of AMD's own people said Vega VII cost $700 to make, GN calculated the 8GB of HBM2 on Vega 56/64 costs them $160, add the GPU ontop of that and then the cooler, PCB partners make who want profit ontop of that and then the retailer also want a profit, by the time they get to Gibbo they have about $250 worth of components profits in them so they have to start at AMD's end selling to board partners at below cost, Vega Vii has twice as many HBM2 modules on it.

This is money thrown away on retail cards that should go on developing Navi, getting it ready

CuriousTomCat · 3 Mar 2019 at 12:10

humbug said:
Yeah, its a great card for little money, but i can't see AMD making any money on those now, i suspect they are losing money on everyone they sell

I agree with respect to the Vega price drop and I can't work out why they did that. I suppose they'll benefit by having brand awareness in the lower mid tier secter so when Navi launches, people will take notice. Essentially it's advertising for the AMD graphics brand.

humbug said:
its really not good having the profitable part of the business prop up the failing part.

Before Ryzen launched, the profitable part of the business was proping up the failing CPU part. R&D for Ryzen would have came from a couple of GPU archtectures plus the PS4 and Xbox One.

humbug said:
I would rather AMD stopped making retail cards than run their business like that, we can't have the retail GPU part of the business draining resources away from everything else.

A lot of resourses went into designing Navi for the PS5 and Xbox Two so it makes sense to drop that GPU onto a retail card seeing as they have it anyway. Then they could release a refresh after 2 years for very little money.

humbug · 3 Mar 2019 at 12:14

Right ^^^ that's why they are doing it, its screaming "Hey we are still here"

humbug · 3 Mar 2019 at 12:32

Poneros said:
uh oh

I hope not....

LoadsaMoney · 3 Mar 2019 at 12:45

humbug said:
Right ^^^ that's why they are doing it, its screaming "Hey we are still here"

They aren't though

bru · 3 Mar 2019 at 12:45

humbug said:
I hope not....

Your not alone, I think most of us on here want AMD to do well with Navi, but if Jim is hearing bad tiding from different sources, then something is probably up.
We can only hope it is not something disastrous.

As for why AMD did the cut price VEGA's for a short stint recently it was purely to rain on NVidia's parade of the 1660ti, which in a lot of the benches and reviews I have seen leaves the 590 and lower bracket standing in the dust. Of course it is too expensive but very quick for the segment it is slotting into.

nashathedog · 3 Mar 2019 at 12:50

humbug said:
Well no... AMD GPU's post the Bulldozer era, Hawaii (R9 290X) Polaris (RX 480 RX 580 RX 590) Fiji (Fury X)

That's 3 architectures during the development of Zen, successful architectures so i don't buy the idea that Zen was at the expense of GPU's.

Hi, We've been all but directly told this via interviews of AMD staff over the last couple of years ending with Lisa Su stating that the RTG's R&D budget would be going up now Zen's in the wild & doing okay. Why do you think we've been getting high-end gpu's using chips that were made for the Professional market?

humbug said:
Aside from that you're missing the point i'm making, they are making cards for retail to sell at a loss, not just Vega 56 and 64 but one of AMD's own people said Vega VII cost $700 to make, GN calculated the 8GB of HBM2 on Vega 56/64 costs them $160, add the GPU ontop of that and then the cooler, PCB partners make who want profit onto of that and then the retailer also want a profit, by the time they get to Gibbo the have about $250 worth of components profits so they have to start at AMD's end selling to board partners at below cost, Vega Vii has twice as many HBM2 modules on it.
This is money thrown away on retail cards that should go on developing Navi.

That's a theory based on a lot of presumptions. As you've admitted in other posts

humbug said:
i can't see AMD making any money on those now, i suspect they are losing money on everyone they sell and as you yourself said its really not good having the profitable part of the business prop up the failing part.

I would rather AMD stopped making retail cards than run their business like that, we can't have the retail GPU part of the business draining resources away from everything else.

For the last few years AMD have been unable to make a competitive high end gpu that was developed solely for gaming, They couldn't afford too as they needed to focus on replacing the FX cpu's etc,

Let's use Vega as an example: First we got the 14nm pro cards & then we got exactly the same chips with half the memory for the RX Vega gaming range. Now AMD have started selling the cutdown Pro chip as the VII gaming card, You say at a loss but there's other factors that need to be considered, What if sales of the pro card that the VII gets its chips from have declined? If so they're left with only a few options, 1, Build up stock that may never get sold, 2, Slow down production or 3, add a new card to the roster, I think the first and second choices are bad ones, With the first you could end up with stock you can't sell, with the second you've entered a contract so slowing down production won't change what it's costing you to build them & the third is what I think's the smart option, Add another way to sell them & that's as the VII.

We don't know if any of this is accurate which is why claiming it be fact is detrimental to what people know about the subject, On the other hand some of it is fact, It's just a case of not blurring the line between them, which is something that happens a lot on the forum.

nashathedog · 3 Mar 2019 at 13:04

ANDREW GREEN said:
I reckon Navi is nearer than we might expect and that's why this generations prices are dropping so much

I think it'll be close. Last time AMD haved it's NM process we had the 480 outperforming the 380 by 40%. So if top level Navi does the same with the 580 replacement it'll be on par with Vega 56 and the 2060 performance wise.

This is true but what you're not considering is where it sat in AMD's new 14nm line up not how it compared against it's namesake, The 480 slotted in where the 390 & 390x sat in the previous line up and at the time it was generally slower than those cards, Then you also need to consider how it took around a year to improve on it in the 14nm lineup. As for top level Navi I think you're right, from what I've seen I think it'll only offer 14nm Vega performance, It could be that there's a completely different high-end Navi chip coming to market as well, each with multiple skus but I've seen nothing to suggest this other than unsubstantiated rumours.

humbug · 3 Mar 2019 at 13:04

LoadsaMoney said:
They aren't though

Meh...

bru said:
Your not alone, I think most of us on here want AMD to do well with Navi, but if Jim is hearing bad tiding from different sources, then something is probably up.
We can only hope it is not something disastrous.

As for why AMD did the cut price VEGA's for a short stint recently it was purely to rain on NVidia's parade of the 1660ti, which in a lot of the benches and reviews I have seen leaves the 590 and lower bracket standing in the dust. Of course it is too expensive but very quick for the segment it is slotting into.

I know.

@nashathedog i hear what you are saying and i agree, i don't suppose AMD have much choice given that as Bru points out Polaris just isn't competitive now, i would really rather AMD didn't sell cards at a loss but if they didn't they wouldn't have a card on the market for gamers.

nashathedog · 3 Mar 2019 at 13:08

humbug said:
@nashathedog i hear what you are saying and i agree, i don't suppose AMD have much choice given that as Bru points out Polaris just isn't competitive now, i would really rather AMD didn't sell cards at a loss but if they didn't they wouldn't have a card on the market for gamers.

Sadly true, It's just a case of waiting for Navi to replace it and get them back in the game.

ANDREW GREEN · 3 Mar 2019 at 14:52

Jim from AdoredTV strikes me as someone trying to make a career as a fake In The Know. He admits being inaccurate on his leaked Turing Specs, Polaris and Navi being at CES and his defense for all this is below:

https://www.youtube.com/watch?v=c8EONokJTdU

The '1500 word email' and 'a call from a number withheld' are hilarious as is the talk of 'they're risking losing their jobs but they say they're doing it because they want me to succeed, they all do and there's loads of them'. Yes this definitely happened.

Okay Navi has been a nightmare. In what sense? Component costs? Drivers? Stability? Performance? Or the fact that Jim from AdoredTV was wrong about it being at CES? Xbox Two is supposed to be

I actually think it might be the opposite. If they're shifting old stock at a loss now, it kind of means Navi is almost ready. If prices for Vega and 580 jump back up, see it as a sign Navi is seriously delayed, otherwise it's coming 3rd quarter of 2019 or earlier. We'll see but Microsoft and Sony are wanting their new consoles in 2020 so it has to be ready.

humbug said:
Well no... AMD GPU's post the Bulldozer era, Hawaii (R9 290X) Polaris (RX 480 RX 580 RX 590) Fiji (Fury X)

That's 3 architectures during the development of Zen, successful architectures so i don't buy the idea that Zen was at the expense of GPU's.

Aside from that you're missing the point i'm making, they are making cards for retail to sell at a loss, not just Vega 56 and 64 but one of AMD's own people said Vega VII cost $700 to make, GN calculated the 8GB of HBM2 on Vega 56/64 costs them $160, add the GPU ontop of that and then the cooler, PCB partners make who want profit ontop of that and then the retailer also want a profit, by the time they get to Gibbo they have about $250 worth of components profits in them so they have to start at AMD's end selling to board partners at below cost, Vega Vii has twice as many HBM2 modules on it.

This is money thrown away on retail cards that should go on developing Navi, getting it ready

This would be true if they were buying in new parts to sell at a loss. This isn't what's happening imo. Every large manufacturer buys components in bulk to manufacture. AMD has invested bucket loads in HDM2 memory and other components that they will have loads of excess stock of.

The Radeon 7 cards are not good enough for the Pro Cards that they had written off and binned. Now they are selling because they can sort of match the 2080 in gaming and destroy it in mining and processing. So while it maybe sold at a loss of say 90% of cost it's much better than at a loss of 100%. It has to be 16gb of HBM2 because of it's design and AMD don't care because they have shedloads of HBM2 they need to shift.

Vega and Polaris are now end of life and are being replaced. In 6 months something better and more will replace it and be highly profitable. So AMD has to shift all this excess stock in the next 6 months. They need to monitorise obsolute components now or otherwise write them off

So selling at a slight loss makes perfect sense for AMD financially.

humbug · 3 Mar 2019 at 15:14

@ANDREW GREEN

To quote Jim from the earlier linked Twitter feed.

Try to understand the difference between me "thinking out loud" and making a statement of fact. Way, way too many of you guys have ~@£$!^ comprehension issues. I throw a lot of stuff "out there" just because it helps to get closer to the eventual truth.

I'm just talking in general. Every time I talk about something even slightly "out there" or a tiny bit controversial it's immediately assumed and declared as a 100% statement of fact.

Of course he is trying to carve out a career in Hardware Journalism, i wish him all the best in that and i happen to think that as a Journalist he's better than any of the established ones because he has a better understanding than almost all of them and puts the research in for complex analysis, most of which are his opinions based on said analysis.

He is what Hardware Journalism used to be and largely no longer is.

bru · 3 Mar 2019 at 17:13

We all have had this assumption that the Radeon V11's are failed MI50 instinct cards, but Ill just point out that the MI50 instinct cards are listed with lower clocks on the core, the same clocks on the memory and consequently slightly lower peak throughputs.

Radeon V11

MI50 Instinct

So Ill just ask what part of the Radeon V11 is a failed chip?

Personally I have no idea, as far as I can tell, if it works then it could be used for either.

Anyway this is off topic for this thread, so apologies.

Davedree · 3 Mar 2019 at 17:18

bru said:
We all have had this assumption that the Radeon V11's are failed MI50 instinct cards, but Ill just point out that the MI50 instinct cards are listed with lower clocks on the core, the same clocks on the memory and consequently slightly lower peak throughputs.

Radeon V11

So Ill just ask what part of the Radeon V11 is a failed chip?

Personally I have no idea, as far as I can tell, if it works then it could be used for either.

Anyway this is off topic for this thread, so apologies.

No even before launch I always acknowledged the difference between Mi50 and V7,totally different (cards).
Mi50 is headless, has no I/o for display and has an extra slot for the Infinity link, V7 is just die harvesting MI50 packages onto a redesigned pcb.

bru · 3 Mar 2019 at 17:23

Davedree said:
No even before launch I always acknowledged the difference between Mi50 and V7,totally different (cards).
Mi50 is headless, has no I/o for display and has an extra slot for the Infinity link, V7 is just die harvesting MI50 packages onto a redesigned pcb.

Sorry, totally agree, yes the cards are different but the chip/memory packages are not.