** The AMD Navi Thread **

[Panos]

These are so called AMD fans, blind to the obvious nature of bussiness and economy, let alone AMD's own product history.

AMD released the Fury at the same RRP bit higher street prices than the 980ti despite being slower.
AMD released the 480 at the same price as the 1060 despite being slower.
AMD released Vega 64 as same price as the 1080 but slower.


Wasn't hard for me to project that AMD would release Navi and not offer any price-performance advantage, despite being nearly a year late and offering less features.



IMO, AMD absolutely should be selling GPUs at the market value, which is what nvidia is exploring. What AMD should also do is release competitive products at the same time frame. If you are a year late then good luck seeing any reasonable sales.

Hahaha. Seems you missed something there. The reference blower V64 beats the 1080 even on Nvidia sponsored games, and on couple of game losing by handful fps is still the blower card. Nitro, Red Devil or even an undervolted reference gain the difference without oveclocking at all.

Pick any game from the list, from the latest up to date review.
https://www.guru3d.com/articles_pages/gigabyte_geforce_gtx_1660_gaming_oc_review,12.html

 

[Martini1991]

Hahaha. Seems you missed something there. The reference blower V64 beats the 1080 even on Nvidia sponsored games, and on couple of game losing by handful fps is still the blower card. Nitro, Red Devil or even an undervolted reference gain the difference without oveclocking at all.

Pick any game from the list, from the latest up to date review.
https://www.guru3d.com/articles_pages/gigabyte_geforce_gtx_1660_gaming_oc_review,12.html

You judge the card on launch day, not 2 years in the flipping future. (Where game depending, they trade blows, as they did at launch).

Vega 64 at launch was parity performance at parity pricing with caveats the 1080 didn't have.

The cards available today don't change that.

 

[Panos]

You judge the card on launch day, not 2 years in the flipping future. (Where game depending, they trade blows, as they did at launch).

Vega 64 at launch was parity performance at parity pricing with caveats the 1080 didn't have.

The cards available today don't change that.

I can judge the card merely 10 months later also....... Or just 2 months after release when the first AIB showed it self.

And something interesting. All Vega 64s sold on September 2017 (when the product went on sale) and afterwards, had different bios with higher clocks, compared to those given for reviews at the end of August who had almost 100mhz less default clocks.

My second hand reference Vega 64, is beating out of the box the one guru3d and hexus have by good margin. Made a post last year in the Vega discussion

 

[Ross Thomson]

I can judge the card merely 10 months later also....... Or just 2 months after release when the first AIB showed it self.

Thats true but given we are discussing the launch of a new card, AMD's history on launch day is pertinent to the discussion not 10 months or so down the line.

 

[D.P.]

But they're not effectively anything in terms of this current generation. Nvidia made a choice to use the 12nm node and that suits them fine. Right now why does it matter what process each is on?

It's not like saying AMD are less efficient and using 7nm suddenly makes the physical card use more power or be less efficient than it is.

I don;t think you are understanding the original posters point. AMD have reduced the power requirements by moving to the 7nm node. Their architecture is still a long way behind Nvidia's. So as a consumer of this generation you may not care but come next generation the differences will be very important in defining viable products.

As a consumer, you should also care that because AMD had to go down the 7nm route, their products are being released 1 year late.

 

[D.P.]

Was the 2080 a year late? It offered the same performance as the 1080ti for the same price. Was the 2070 a year late? It offered the same performance as the 1080 for the same price

Turing released first. You can;t be late when you are first to market. Navi can only be compared to Turing.

The poor price-performance of Turing is irrelevant because Nvidia was only competing with themselves and the Pascal sales. Nvidia priced Turing specifically to continue Pascal sales and clear an supply chain issue. Turing was never priced to increase market share rapidly.

AMD on the other hand have very different market positions and product competitiveness. They absolutely would have wanted NAVI to be released a year ago so they could stop Vega production with the expensive HBM memory. AMD is loosing market share, so offering competitive products earlier with better profit margins was paramount.

 

[D.P.]

Hahaha. Seems you missed something there. The reference blower V64 beats the 1080 even on Nvidia sponsored games, and on couple of game losing by handful fps is still the blower card. Nitro, Red Devil or even an undervolted reference gain the difference without oveclocking at all.

Pick any game from the list, from the latest up to date review.
https://www.guru3d.com/articles_pages/gigabyte_geforce_gtx_1660_gaming_oc_review,12.html

And what happened at launch day and in the follwing 6 months when most of the reviews, sales and decisions are made? the 1080 was faster.

 

[Panos]

And what happened at launch day and in the follwing 6 months when most of the reviews, sales and decisions are made? the 1080 was faster.

LOL. Launch day Vega 64 reference were different than the ones sent to reviewers a month before. That's what I am saying, the ones sent to reviewers had 100mhz lower clocks and older version of BIOS.

 

[4K8KW10]

Will the Image Sharpening and FidelityFX filters be available for older cards, too?

 

[Sargatanas2511]

I don;t think you are understanding the original posters point. AMD have reduced the power requirements by moving to the 7nm node. Their architecture is still a long way behind Nvidia's. So as a consumer of this generation you may not care but come next generation the differences will be very important in defining viable products.

As a consumer, you should also care that because AMD had to go down the 7nm route, their products are being released 1 year late.

Sure the architecture is still behind and AMD are playing catchup but they used what they can to mitigate that issue. It doesn't make the product less viable just because they used what they could in the short term to bridge a gap.

Now with Nvidia going down the 7nm route AMD can work on improving the architecture and IPC for their next card while already having 7nm experience. I still don't see why it's 'bad' that they used a short term bridge to try to catch up when it actually worked for the most part.

It would be like complaining that a car company used an older engine in a newer car, but managed to get it to similar capability and efficiency. Who cares how they do it as long as it's done.

 

[Panos]

Will the Image Sharpening and FidelityFX filters be available for older cards, too?

<shrug> Some of the features announced are going to be supported by Vega and Polaris, however we do not know about the rest.

 

[D.P.]

Sure the architecture is still behind and AMD are playing catchup but they used what they can to mitigate that issue. It doesn't make the product less viable just because they used what they could in the short term to bridge a gap.

Now with Nvidia going down the 7nm route AMD can work on improving the architecture and IPC for their next card while already having 7nm experience. I still don't see why it's 'bad' that they used a short term bridge to try to catch up when it actually worked for the most part.

It would be like complaining that a car company used an older engine in a newer car, but managed to get it to similar capability and efficiency. Who cares how they do it as long as it's done.

It is bad for AMD because they had to wait 1 year for 7nm to be mature enough for large dies at high yields.

I would also be far more optimistic of AMD's GPU future if they radically changed the architecture and improved efficiency on the same 14nm node as Vega. AMd desperately need their own Maxwell jump, 70% plus efficiency gain.

 

[Martini1991]

I find it interesting that you base most things on your own V64 @Panos
It's a flipping watercooled and tweaked card that bares little resemblance to the performance you'd actually receive from running a reference V64 at stock (Because a stock V64 throttles to hell and back).

Sure, I imagine an AIB card right now (Especially at £300) is absolutely unbeatable in what it offers. But that doesn't change the launch of the V64.

 

[Martini1991]

As for Navi, it's the old adage that there's no such thing as a bad graphics card, rather bad pricing. And that's exactly what the 5700XT is.

If the 5700XT launch at £300-£350 I'd have swapped out my Vega 64 for it, but unfortunately it's closer to £450.

 

[shankly1985]

As for Navi, it's the old adage that there's no such thing as a bad graphics card, rather bad pricing. And that's exactly what the 5700XT is.

If the 5700XT launch at £300-£350 I'd have swapped out my Vega 64 for it, but unfortunately it's closer to £450.

I totally agree with you.

 

[opethdisciple]

With these tech companies, why are they sometimes not able to do what is blindingly obvious to us consumers?

 

[Martini1991]

If this Nvidia super and price drop stuff actually happens, then Navi can't launch at $450. It'll be DOA.

If it drops down Navi to $350 and it's available between £300-£350 because of that, then I'll get one.
I'd be tempted to get two and a 3900X at that case tbh.

 

[Panos]

I find it interesting that you base most things on your own V64 @Panos
It's a flipping watercooled and tweaked card that bares little resemblance to the performance you'd actually receive from running a reference V64 at stock (Because a stock V64 throttles to hell and back).

Sure, I imagine an AIB card right now (Especially at £300) is absolutely unbeatable in what it offers. But that doesn't change the launch of the V64.

No I am talking about reference cooler performance, not watercooled. Made an extensive post about it back in October.
The performance of the reviewers resembled the Balanced setting. Not Turbo let alone custom. And did also tests with untouched heating, liquid metal and Kryonaut with the same reference cooler. Took me days to go through everything testing various things for "science"

 

[LePhuronn]

These are so called AMD fans, blind to the obvious nature of business and economy, let alone AMD's own product history.

AMD released the Fury at the same RRP bit higher street prices than the 980ti despite being slower.
AMD released the 480 at the same price as the 1060 despite being slower.
AMD released Vega 64 as same price as the 1080 but slower.

And never once did they gain significant market share by doing so. You would think then that, at some point, somebody would say "hang on, perhaps we should change our approach somewhat". But then a great man once said the definition of insanity is repeating the same action yet expecting a different result.

Wasn't hard for me to project that AMD would release Navi and not offer any price-performance advantage, despite being nearly a year late and offering less features.

It's funny what bankruptcy will do to R&D budget and the capability to design and produce graphics cards. For every "blind AMD fan" there's similarly a "blind AMD detractor"; you accuse the former of being oblivious to "the nature of business and economy" and yet you lambast them for being late to the party when they almost went under. Pray tell, how exactly do you propose a company struggling financially keep parity with a juggernaut competitor with budget and operating capital significantly larger? And you claim "fans" are blind to the nature of business and economy?

IMO, AMD absolutely should be selling GPUs at the market value, which is what nvidia is exploring.

Except that Nvidia has skewed market value. None of Nvidia's offerings are "worth" the price point, and Navi is not "worth" the price point. So yes, AMD should be selling GPUs at market value, but the actual market value which will generate sales, not Nvidia's bloated vision.

What AMD should also do is release competitive products at the same time frame. If you are a year late then good luck seeing any reasonable sales.

Um...bankruptcy? Tell you what, you go pay off AMD's $1B debt so they can ramp up their R&D and production and expedite the next few generations of graphics cards. Then if/when AMD still can't keep up with Nvidia you're welcome to cry about how they get it so wrong.

AMD's financial history as of late as precluded them from pushing their GPU tech harder. It is that simple. Now that is not an excuse, it is merely a fact. If their cards are poor because of the near-bankruptcy then that's AMD's own fault. Or you can accept they've done the best with what they have available and judge it on its own merit. But you cannot have it both ways and cry about how woeful they are yet ignore the circumstances which contributed to it.

 

[Panos]

As for Navi, it's the old adage that there's no such thing as a bad graphics card, rather bad pricing. And that's exactly what the 5700XT is.

If the 5700XT launch at £300-£350 I'd have swapped out my Vega 64 for it, but unfortunately it's closer to £450.

I agree with you, though I will pass judgment when see the actual product.
There are a lot of things under NDA for both Ryzen 3000, X570 and Navi still. Especially in relation to Infinity Fabric.

After all the games I play either are Vulcan (X4), WOT (140 fps on max out settings) or ESO. Where RTX2080Ti and Vega 64 have the same performance on 5Ghz Coffee Lake CPUs. So Vega 64 will be used until I deem a worthy update