Soldato
-
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.
When the Gpu's prices will go down ?
- Thread starter sys6em
- Start date
Soldato
Will hold him to that one. 
Caporegime
- Joined
- 18 Oct 2002
- Posts
- 30,849
Aka 'We still love milking those gaming cows, we won't forget about you!!' 
Yea. But why not both? They still milk us at the same time. My issue is, milk the people who buy 4090's and provide the rest of us the 4080 at £749 which is what really should be.
Which is also the reason there was no FE 4070ti, they'd already had the heatsinks etched as 4080 12 gig then realised people seen through their bs.
So far its the only gpu in the 40x range not to have an fe variant.
Last edited:
Everyone knew that was no 4080. Hence why it changed. The current 4080 just about deserves it's name as it is. Plus you know what I meant
A Founders Edition was never on the cards even when it was still the 4080 12GB. Right from the announcement it was stated to be a partner model only.Which is also the reason there was no FE 4070ti, they'd already had the heatsinks etched as 4080 12 gig then realised people seen through their bs.
https://www.kitguru.net/components/...2gb-to-be-aib-only-no-founders-edition-model/
Same goes for the 4060 Ti 16GB, which won't have one either. It's been said that the 4060 Ti 16GB was a last-minute decision by Nvidia that wasn't originally planned, and that's why it doesn't have a FE. Which makes me think that the 4080 12GB was a fairly late choice they made too, and they'd originally planned to only use AD104 for 70-series cards. Then someone there had the brilliant idea of trying to sell it as a 4080 so they could charge more. That meant it was too late to do a FE model (which we know were manufactured long before launch, as we saw leaked pictures), so it became a partner-only card. Would also explain why they ditched the idea so quickly after the backlash, as it's something they weren't wedded to in the first place and was just an exercise in seeing what they could get away with.
Soldato
I read between the lines that the cast off and surplus yields that aren't quite good enough for the AI grade bins will then appear as a gamer grade stock.
What's odd about the Ti releases is they tended to be following the regular launch models. Now they have been used to cover up the mistakes but at least they are conscious of the vram amounts.
What's odd about the Ti releases is they tended to be following the regular launch models. Now they have been used to cover up the mistakes but at least they are conscious of the vram amounts.
That is probably also the reason why the tensor sensor have made it into gamer cards despite being of questionable use there. While not all gaming dies are useful for selling on as pro cards, there is some overlap. And they get to heavily market DLSS.
All those PCMR comments about the iGPU on Intel CPUs throughout the years ("I'd rather have more cores on my i7/i9") has strangely been absent from GPU discussions, although I am sure PCMR would prefer more GPU shaders instead.
Having graphic cards that reflect how much they actually cost to make would be a start, instead of adding something like 30% to what it cost them we're seeing that they're trying to charge what they think the markets will bear.
e: 4060 Ti 8/16GB being the perfect example of that. It costs them maybe $30-40 to add an extra 8GB but instead of trying to make a reasonable profit on that by charging customers an extra $50 they go full on greed and try to get away with twice that amount.
Last edited:
Associate
Nvidia basically talking about the beginning of the end of a relationship. It's inevitable.
Hopefully if Nvidia sell it's "gaming division" side of things the new "company" has success similar to what happened when Lenovo bought IBM's PC division.
Of course, just to shaft PC gamers it would probably just be Intel "jumping the queue" and we are back to square one
Hopefully if Nvidia sell it's "gaming division" side of things the new "company" has success similar to what happened when Lenovo bought IBM's PC division.
Of course, just to shaft PC gamers it would probably just be Intel "jumping the queue" and we are back to square one
I don’t see why they would split it, the businesses are too intertwined. The AI side is all being driven by GPU architecture, it wouldn’t make sense to split them. All of their major products are GPU related.
Why would they sell a portion of their business that still generates high revenue and is seemingly linked so specifically to other areas of their business.
Gaming focused GPU’s will become even more AI related if anything.
Why would they sell a portion of their business that still generates high revenue and is seemingly linked so specifically to other areas of their business.
Gaming focused GPU’s will become even more AI related if anything.
Last edited:
I agree. Consumer GPUs will continue to make it possible to run open source AI models on consumer hardware.
However, look at the data companies as an example. Twitter and Reddit now charge millions for data that used to be freely available. Evidently trillions in revenue is up for grabs. The big AI players have essentially struck oil.
Future AI optimised hardware becoming incredibly expensive benefits the rulers of society so of course they will make that happen and NVIDIA will have many years of multiplying their recently acquired $1T market cap.
AI hardware decision making needs to be the result of capitalism because it's the way it's meant to be played.
Soldato
- Joined
- 14 Aug 2009
- Posts
- 3,102
I really doubt those extra shaders would offer the same performance improvement that DLSS does. Then there's RT and you can see how bad AMD does without proper dedicated hardware.
Soldato
I think probably what will happen in the future is that their pro cards will continue to get more tensor/AI performance but they'll let it stagnate in the gaming segment. It'll exist for "good enough for DLSS" levels of performance but not good enough for pro applications.
This way they'll be able to have better product segmentation without relying on e.g. VRAM so much. Although they will continue to worry about cannibalising the creative professional market somewhat, but only if that doesn't start leaning hard on ML inference (which it probably will!).
We're already starting to see this at the very high end with H100 being much more capable than AD102 in this respect to an increasingly exaggerated degree and we should continue to see that in the future.
This way they'll be able to have better product segmentation without relying on e.g. VRAM so much. Although they will continue to worry about cannibalising the creative professional market somewhat, but only if that doesn't start leaning hard on ML inference (which it probably will!).
We're already starting to see this at the very high end with H100 being much more capable than AD102 in this respect to an increasingly exaggerated degree and we should continue to see that in the future.
Last edited:
Soldato
No but if you recall AMD had very compute heavy resources with Vega and when it came to gaming the excuse then was they were not focusing on it (or maybe the angle was the resources were being wasted). The compute element was actually quite strong then they shifted design again.
Last edited:
Soldato
Can someone explain to me what use is AI to the average gamer?