NVidia Ada Speculation, Rumours and Discussion

No it's not. AMD has explicitly stated that the years on their roadmaps include the whole year in both ends.
In this case those roadmaps are from beginning of 2019 to end of 2022.

 

Agreed. If both IHV's are on the same node Nvidia definitely has the performance/efficiency advantage.

 

They will age horrifically.

 

That roadmap shows from the beginning of 2019 to the beginning of 2022. Look at their previous roadmaps.

 

All the AMD stuffs on this thread should be cleaned up.

 

Some of the leakers feel like they're just proving they're lying right now.

Nvidia's going to "wait" until AMD's next generation comes out, like, specifically? That's not how it works, their roadmap is laid out years ahead of time and if there's anything they can change last minute at all it's just price and maybe, if they've months upon months of warning, the clocks/power/cooling. They're definitely not "hiding" some sort of improved architecture in the back just in case competition is better than they think.

I'll still maintain that there's no reliable looking leaks concerning Nvidia until any real looking details show up, and so far there's none.

 

Nah, kimi's good.

 

Doesn't look like a roadmap to me.

Also, NVidia will have had at least two years to decide how to respond to RDNA 2 and whatever path they predict AMD is now on.

Don't pretend that NVidia will continue to ignore AMD.

 

I can't believe for an instant that these companies don't shift their schedules and products based on competition.

 

Of course they do. But there's only so much, you can do. Those options materially differ whether or not you're running fab-full or fab-less. At least (for Nvidia), they are not at a process disadvantage, if they did not chose to be, as was Intel for 2 years now. They have wafer allocation at TSMC and they have allocation at Samsung, should either one suffer from some unforeseeable event. They can not change, however in the bat of an eye nor would they give up wafers scheduled for A100 in order to combat GeForce shortages. Same goes for AMD. ATM the most profitable die for them should be their 7 nm CCD qualifying for use in Epyc processors. They won't touch that just in order so appease some gamers like me to get more stock of RX6k into the stores.

What companies (other than Intel up until 2019 apparently) can and will do, is have contingency plans for their chips. They'd be crazy not to.

 

They really can't, not significant shifts anyway. Delaying launch because competition isn't ready is beyond stupid and with the foundries being as full as they are and production taking as long as it takes you can't really move any earlier either

Note: with launch I mean bringing the cards for sale, not announcing they will be available at date x

 

You mean "Nvidia got to decide some couple years ago how they're going to react to AMD product that's going to be released in a year or so and what comes after it to compete with whatever they decided to build."
Just like with every other generation.

 

Exactly it’s simply not possible to have a major shift in strategy after competition has hit the market. Any response that late in the game will be down to turning the usual knobs when carving out price & performance for each SKU. If Nvidia did have a change in strategy to counter AMDs chiplet move that decision has likely already been made but even then there’s only so much they can do.

 

According to the latest rumor roundup on videocardz, the wait-and-see approach is to determine whether they only use cheaper Lovelace chips (AD102) or if they resort to a more expensive Hopper solution (GH202) to compete at the high end.
It seems Hooper might be a chiplet solution too.

Nvidia isn't deciding if how they develop their solutions based on RDNA3's performance.

 

AMD has to redesign their GPUs to provide better support for Raytracing and DL. Yet people here believe that this huge shift in graphics doesnt exist for AMD. nVidia has changed the game like Apple with the iPhone.
Ampere delivers 50% more compute, twice the raytracing and 4x performance for DL.

nVidia is far ahead and they will continue to push towards raytracing.

 

It doesn’t make sense that Hopper and Lovelace will be ready at the same time with the same graphics feature set. If that’s the case there’s no reason for Lovelace to exist.

 

bs.

H100 yes.
MI200 but moar stuff.

 

Yeah, NVidia saw that AMD has zig-zagged sometime around spring 2020 (if not earlier) and a few months later was quite happy to stop being "budget king".

I guess the people who always buy NVidia are just mad that AMD only made the prices of NVidia cards cheaper for 2 or 3 months. And most of those poor people are so out of luck because they missed those beautiful prices.

If NVidia floods the market with cards that have the same performance as 3070 and 3080 in Q1 2022, by launching Lovelace "104" first, does that count as a major shift in strategy?

Yes, we saw that with Ampere.

NVidia would be pretty stupid to have assumed that AMD wasn't going to use chiplets, as for the past five years it's been a hot topic.

Welcome to 30 odd posts ago.

RDNA 3 rumours indicate some cards are chiplet based and some aren't. Hopper and Lovelace might form a similar pairing.

But I give no credence to the "Hopper is NVidia's plan B for desktop (top tier prosumer or consumer)" theory/rumour. NVidia will be selling every last Hopper to data centre people who are quite happy to pay 10x more than consumers.

 

If the constellation is such, that only with a heavily overclocked Hopper they can claim perf kingship in the desktop and/or gaming space, they are likely to divert a couple of hundred cards, make a lot of marketing fuss around it and then say "sorry, demand is far greater than we estimated. We will be working with our partners to improve supply." just because of ego and to keep the halo. They won't do this, if they can achieve the same with Lovelace or if they'll be only second best anyway.

 

No such thing is possible.

H100 has no display outs or RT gear.
Impossible.