Sapphire To Release Vapor-X edition for the Radeon RX 7900 XTX & XT

Published by

Click here to post a comment for Sapphire To Release Vapor-X edition for the Radeon RX 7900 XTX & XT on our message forum
https://forums.guru3d.com/data/avatars/m/271/271560.jpg
cucaulay malkin:

nvidia's mcm is different, can't remember the name for it but they're chiplets designed so that each one can have a different function. edit: it's called composable on-package architecture (COPA), basically allowing them to make many different variations.
yes. Nvidia's difficulty is in finding both a substrate and an interposer so the lag between different chiplets becomes infinitesimal. think of it as having the signal have to travel the long way round instead of point-to-point direct access
https://forums.guru3d.com/data/avatars/m/271/271560.jpg
cucaulay malkin:

what ? a 60% functional ad102 is still 15% more powerful than 4080 - amd's target for 7900xtx. that is really impressive.
this is a business. while high end GPUs have the greatest margin @ retail, the manufacturer not-so-much which is why Nvidia makes FE now and why AIB's (like EVGA) have their margin squeezed by competition with their supplier. when AMD introduced Polaris it was for the mid-market only. the Titan was the initial halo and the 1080 was the enthusiast card. they had no intention of competing at either the level of the Titan or the 1080 and your math is off. GPUs do not work the way you suppose with a linear scaling of function a 60% AD102 may be completely ruined
https://forums.guru3d.com/data/avatars/m/282/282473.jpg
tunejunky:

this is a business. while high end GPUs have the greatest margin @ retail, the manufacturer not-so-much which is why Nvidia makes FE now and why AIB's (like EVGA) have their margin squeezed by competition with their supplier. when AMD introduced Polaris it was for the mid-market only. the Titan was the initial halo and the 1080 was the enthusiast card. they had no intention of competing at either the level of the Titan or the 1080 and your math is off. GPUs do not work the way you suppose with a linear scaling of function a 60% AD102 may be completely ruined
still, there are a lot of options for a gpu that with just 88% functional die flattens everything on the market. I'd much rather wait for a nicely priced cut of ad102 somewhere in a year's time than get a 4080.
https://forums.guru3d.com/data/avatars/m/271/271560.jpg
cucaulay malkin:

still, there is a lot of options for a gpu that with just 88% functional die flattens everything on the market.
yes, if it works the particular problems w/ AD102 re: yield are circuit density (a record for Nvidia) and physical size of the die. you have to remember when you're squaring a circle the radius will not be reaching the same areas creating limited functionality in differing areas, some fatally, plus the wasted space of gaps cost just as much as a real chip
https://forums.guru3d.com/data/avatars/m/282/282473.jpg
tunejunky:

yes, if it works
it's called 4090
https://forums.guru3d.com/data/avatars/m/271/271560.jpg
please don't get me wrong, the only things i hate about the 4090 are in order: price and size. i love the technology, i know it's the shiniest shiny. i just know it's not profitable. halo cards sell the mid-range cards, which is where the money (and wafers) are at
https://forums.guru3d.com/data/avatars/m/282/282473.jpg
tunejunky:

please don't get me wrong, the only things i hate about the 4090 are in order: price and size. i love the technology, i know it's the shiniest shiny. i just know it's not profitable. halo cards sell the mid-range cards, which is where the money (and wafers) are at
well it's more profitable than not having a card that competes with it at all. amd are pitting their 533mm2 die against nvidia's 400mm2 ga103
https://forums.guru3d.com/data/avatars/m/271/271560.jpg
cucaulay malkin:

well it's more profitable than not having a card that competes with it at all. amd are pitting their 533mm2 die against nvidia's 400mm2 ga103
no just no again your math is not just off, it's spurious. you cannot compare the die sizes between a MCM and a monolithic design it makes no sense. MCM's can have whatever the heck die size they want and it doesn't affect yield. monolithic designs are exactly opposite
https://forums.guru3d.com/data/avatars/m/282/282473.jpg
tunejunky:

no just no again your math is not just off, it's spurious. you cannot compare the die sizes between a MCM and a monolithic design it makes no sense. MCM's can have whatever the heck die size they want and it doesn't affect yield. monolithic designs are exactly opposite
to me intel being quite competitive on price, not being able to put up a good fight against ad102 and slacking on rt yet again would be a final nail in the coffin scenario for amd's dgpu sales.
https://forums.guru3d.com/data/avatars/m/271/271560.jpg
cucaulay malkin:

to me intel being quite competitive on price, not being able to put up a good fight against ad102 and slacking on rt yet again would be a final nail in the coffin scenario for amd's dgpu sales.
AD 102 is totally irrelevant to the marketplace except for marketing purposes (its designed role) there won't be any cut-down GPU products from it (until much much later and maybe not even then). AD 103 has higher yield and has had more wafers fabbed (over 2X) than AD 102 AMD and Intel are going to do what they do and both will have profitable product lines especially after Nvidia's disastrous launch of Lovelace.
https://forums.guru3d.com/data/avatars/m/282/282473.jpg
tunejunky:

when AMD introduced Polaris it was for the mid-market only. the Titan was the initial halo and the 1080 was the enthusiast card. they had no intention of competing at
and n31 was intended to not compete with ad102 too,right ?
tunejunky:

Nvidia's disastrous launch of Lovelace.
tunejunky:

halo cards sell the mid-range cards, which is where the money (and wafers) are at
can't keep track of the things you're saying https://www.guru3d.com/articles_pages/geforce_rtx_4090_founder_edition_review,33.html
https://forums.guru3d.com/data/avatars/m/271/271560.jpg
cucaulay malkin:

and n31 was intended to not compete with ad102 too,right ? can't keep track of the things you're saying https://www.guru3d.com/articles_pages/geforce_rtx_4090_founder_edition_review,33.html
it's very simple when you stop trying to force a different conclusion. AMD does not compete with Nvidia across Nvidia's entire product line and never has. Nvidia has several divisions with zero AMD competition and AMD has divisions with zero Nvidia competition (and a couple that might as well be zero) as David Byrne sang "same as it ever was". the RTX 4090 was specifically designed to stay out of AMD's wheelhouse by virtue of cost of production. if you think the retail is high on the 4090 you should see the manufacturing costs. and of course the FE costs more to engineer and produce but Nvidia is getting the cash directly and making more (profit) than the sales to AIBs - and it still isn't profitable. what IS profitable is using the 4090 in advertising (its real commercial purpose) to drive sales of the entire stack (though i use the word disastrous to reflect the unloved status of the rest of the stack)
https://forums.guru3d.com/data/avatars/m/282/282473.jpg
tunejunky:

it's very simple when you stop trying to force a different conclusion. AMD does not compete with Nvidia across Nvidia's entire product line and never has. Nvidia has several divisions with zero AMD competition and AMD has divisions with zero Nvidia competition (and a couple that might as well be zero) as David Byrne sang "same as it ever was". the RTX 4090 was specifically designed to stay out of AMD's wheelhouse by virtue of cost of production. if you think the retail is high on the 4090 you should see the manufacturing costs. and of course the FE costs more to engineer and produce but Nvidia is getting the cash directly and making more (profit) than the sales to AIBs - and it still isn't profitable. what IS profitable is using the 4090 in advertising (its real commercial purpose) to drive sales of the entire stack (though i use the word disastrous to reflect the unloved status of the rest of the stack)
Okay makes sense. I guess I got caught up in the usual hype train for AMD, as many people said n31 on mcm will demolish ad102,with 2.5x 6900 performance,and be a lot cheaper. I guess that's all it always is, hype. I still have high hopes for 7800/7700,7900 is waaaay to expensive starting at 900eur.
https://forums.guru3d.com/data/avatars/m/201/201426.jpg
I see malkin in here with a hard on for hating on AMD. Not gonna buy a vapor X, cool but piss off already. Im insanely excited for a vapor X. Does it matter if it beats the 4090, hell no. Do i care if its slower in RT then a 4080, hell no. Im not a simp for Nvidia and their overpriced stuff these days. People said AMD was dead in the water with the 5700xt, yet it was succesfull and said AMD coulndt do well against Nvidia with RTX 3 series and their first RT gen cars kicked ass.