NVIDIA ‘Ampere’ 8nm Graphics Cards

[Perfect_Chaos]

Maybe it's a reference to giving us dial-up performance in a world where we need broadband?

They better not

 

[SirPsychoZe0]

I have been thinking about how Nvidia will spin the 3000 series at the lunch event / announcement. I remember how Jensen talked about needing a new way to measure performance because of RTX at the 2018 launch event for the 2000 series.

I reckon they will promote the 3000 flagship on two points:

1) If the rumours are true, then they will start by focusing on RTX running at x4 times that which the 2080 Ti was capable of. This will be showcased with a C2077 demo probably. The tagline will therefore be "4 times the RTX performance from last gen".
then
2) They will use a combination of DLSS 3.0 and raw performance improvements to create a slightly 'fudged' metric for comparison with the 2080 Ti. So all comparisons will start with TAA. If rumours are true and the 3000 flagship can do 40% more raw performance, and DLSS 3.0 can create substantial increases (again, only compared with TAA), then I imagine they might get to a fudged position that says "2 times the (non-RTX) performance from last gen".

So before we get to the price reveal, Jensen will hammer home the point that the 3000 flagship overs 4 times the RTX performance and 2 times the traditional performance. With that in mind, he will then be able to justify a price increase. I expect it will be 25% more than the 2080 Ti - and why not if you are getting (apparently) 100% performance increase and 300/400% RTX performance increase? Would seem reasonable if you didn't know any better.

I also think they will move to name the flagship 'RTX 3090' on the basis that a XX90 > XX80 Ti, thus helping to justify the price increase.

 

[BigBANGtheory]

Yeah I was thinking of holding off until AMD cards are released as there will likely be an uplift in the first couple of months with no competition.

I think that is wise it also gives time for the true results to surface vs Jensen 4x faster than Turing inflated marketing waffle

The only good reason to get in early being (impatient doesn't count btw) is if you get to realise the benefits early so a 3080Ti on TSMC 7nm at a sensible price (not cheap, but sensible) or a good value 3080 but not an over priced one that Big Navi 'could' out perform and under cut.

 

[Random Guy Esq.]

15702340003

Is that a real telephone number anywhere?

It's the price!

 

[varkanoid]

15702340003

Is that a real telephone number anywhere? Isn't 1 the code for dialing internationally from the US? And 57 is the country code for Columbia.

So we get 1-57-02-34-00-03. Is that a real Columbian telephone number?

I googled that number and it actually comes up a couple of times on old posts in a couple of forums and the subject they are talking about is dial up. Which is slightly weird.

 

[Satchfanuk]

Another teaser from Nvidia:

https://twitter.com/x/status/1293090205054902273

Do we have 20 more days of these obscure twitter posts to suffer before they actually tell us anything meaningful

 

[Perfect_Chaos]

It's the price!

It's the number ET phones to go home, because the price will be out of this world

 

[Seearbe]

15702340003

Is that a real telephone number anywhere? Isn't 1 the code for dialing internationally from the US? And 57 is the country code for Columbia.

So we get 1-57-02-34-00-03. Is that a real Columbian telephone number?

1 is the international number for USA (and a couple of other countries)
570 is the area code for Pennsylvania
234-0003 is the number being dialled

I think it's just the number modems used to dial. There's absolutely nothing in the number.

 

[Satchfanuk]

It's the number ET phones to go home, because the price will be out of this world

Comment of the day, take a bow

 

[JediFragger]

It's the number ET phones to go home, because the price will be out of this world

LOL!!!

 

[Kelt]

It's the number ET phones to go home, because the price will be out of this world

Heh heh

 

[Deleted User 456458]

The price will depend on what AMD can deliver and when they can deliver it.

 

[Kaapstad]

I am unable to get excited about anything to do with the Ampere launch at the moment.

I won't be buying any of the 3XXX cards and will just wait for the Titan to arrive.

 

[Seearbe]

I'm excited about the cards as I need to upgrade from my 670. But I'm far too tight-fisted and will probably shrink away from actually buying one, especially if we're looking at silly prices!

My Radeon 9800 Pro was the last time I had a top-of-the-line card.

 

[Illuminist]

I have been thinking about how Nvidia will spin the 3000 series at the lunch event / announcement. I remember how Jensen talked about needing a new way to measure performance because of RTX at the 2018 launch event for the 2000 series.

I reckon they will promote the 3000 flagship on two points:

1) If the rumours are true, then they will start by focusing on RTX running at x4 times that which the 2080 Ti was capable of. This will be showcased with a C2077 demo probably. The tagline will therefore be "4 times the RTX performance from last gen".
then
2) They will use a combination of DLSS 3.0 and raw performance improvements to create a slightly 'fudged' metric for comparison with the 2080 Ti. So all comparisons will start with TAA. If rumours are true and the 3000 flagship can do 40% more raw performance, and DLSS 3.0 can create substantial increases (again, only compared with TAA), then I imagine they might get to a fudged position that says "2 times the (non-RTX) performance from last gen".

So before we get to the price reveal, Jensen will hammer home the point that the 3000 flagship overs 4 times the RTX performance and 2 times the traditional performance. With that in mind, he will then be able to justify a price increase. I expect it will be 25% more than the 2080 Ti - and why not if you are getting (apparently) 100% performance increase and 300/400% RTX performance increase? Would seem reasonable if you didn't know any better.

I also think they will move to name the flagship 'RTX 3090' on the basis that a XX90 > XX80 Ti, thus helping to justify the price increase.

I think you are taking a lot of sense right there

 

[Yaayuh!]

I have been thinking about how Nvidia will spin the 3000 series at the lunch event / announcement. I remember how Jensen talked about needing a new way to measure performance because of RTX at the 2018 launch event for the 2000 series.

I reckon they will promote the 3000 flagship on two points:

1) If the rumours are true, then they will start by focusing on RTX running at x4 times that which the 2080 Ti was capable of. This will be showcased with a C2077 demo probably. The tagline will therefore be "4 times the RTX performance from last gen".
then
2) They will use a combination of DLSS 3.0 and raw performance improvements to create a slightly 'fudged' metric for comparison with the 2080 Ti. So all comparisons will start with TAA. If rumours are true and the 3000 flagship can do 40% more raw performance, and DLSS 3.0 can create substantial increases (again, only compared with TAA), then I imagine they might get to a fudged position that says "2 times the (non-RTX) performance from last gen".

So before we get to the price reveal, Jensen will hammer home the point that the 3000 flagship overs 4 times the RTX performance and 2 times the traditional performance. With that in mind, he will then be able to justify a price increase. I expect it will be 25% more than the 2080 Ti - and why not if you are getting (apparently) 100% performance increase and 300/400% RTX performance increase? Would seem reasonable if you didn't know any better.

I also think they will move to name the flagship 'RTX 3090' on the basis that a XX90 > XX80 Ti, thus helping to justify the price increase.

Sounds realistic to me, unfortunately.

We'll find out models and pricing but as usual any realistic performance information we'll have to wait for the reviews.

 

[danlightbulb]

Presumably someone has actually dialled the number? Maybe there is an answerphone message on the other end.

 

[SirPsychoZe0]

I think you are taking a lot of sense right there

Thanks. You can imagine the websites reporting on this who will simply grab the headline "x2 performance, x4 RTX performance" and then report that this is a 'monster' graphics card that is so god-like that they had to make a new tier (XX90) above the XX80 Ti to place it.

If you were offered all of that for 25% more cost compared to the 1080 Ti / 2080 Super, you would be justified as accepting that as a good deal. However for those of us 'in the know', who understand the trickery with DLSS, the lack of adoption of RTX, and the fact that 2080 Ti itself was never a good proposition from which to use as a stepping stone to justify to another new proposition/price point, we can hopefully see through this.

It is funny how I used to dream as a student of being able to afford the top-tier GPUs once I got a good job. Now that I have a good job, the price of GPUs has ran away from me again!

 

[kazuya1337]

I think you are taking a lot of sense right there

I don't.

£2000 for a 3080Ti when AMD most likely have a very competitive card and their flagship I doubt will be over £1200

 

[Devrij]

It is funny how I used to dream as a student of being able to afford the top-tier GPUs once I got a good job. Now that I have a good job, the price of GPUs has ran away from me again!

I feel you there. I remember buying a 5850 2nd hand for 150 quid when it was still current gen and thinking it would be great to be able to afford one of those big dual gpu cards like the 59xx one day. Now the equivalent card would be a 2070S and to get a 2nd hand one would cost nearly 400 quid, and that flagship gpu is over £1k. it's enough to make me want to go back to 1080p!