• 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.

AMD RDNA3 unveiling event

6900 XTX > 7900 XTX £999. (Excluding any gouge)

3080 FE £650 > 4080 Fe £1200.

Was the 4080 more than double price initially or did I imagine that? Maybe it was AIB models only not sure.
 
Last edited:
6900 XTX > 7900 XTX £999. (Excluding any gouge)

3080 FE £650 > 4080 Fe £1200.

Was the 4080 more than double price initially or did I imagine that? Maybe it was AIB models only not sure.
6800XT £610 > 7900XT £900, not as bad as Nvidia but still a 50% price increase for just a 35% performance uplift.

Will AMD or nvidia be able release anything this gen that's actually faster than a 6800XT or 3080 for the same price or less?
 
I maintain the 7900XTX was meant to be more a 4090 competitor


Yes R&D is a part of it, if you spend $500 Million on developing a GPU architecture you need to get that money back or you wont have it to develop the next architecture, or you're dipping in to your savings until that dries up.

In a way Nvidia are much better situated for that because they sell a lot more GPU's than AMD, so they could sell them for less over BOM cost to get that money back, i very much doubt the money to develop RDNA came from Vega, that came from Ryzen sales, that cannot continue, Radeon must stand on its own feet.


Leaving aside the tops from the last mining bubble, for a 2 year period lasting q1 2019 to q1 2021, there are around 80mil. units shipped. At that time, it could have been more, but that bubble flooded back then the market with plenty of second hand GPUs.

Let's say AMD sticks to its premium branding strategy and only stay around 10% market share, that's 8mil units, making that 500mil R&D budget about $62.5/card on average. Of course, gaming is only a part of AMD and nVIDIA's sales, so that $500 mil will result in even less cost / card.

BOM costs should go down over a 2 year life cycle and no one says you can't still produce older gen cards from the lower/mid end if the costs are low enough...

Shipping and logistics I don't know how much they are, but I doubt they're more than 30-50 dollars / card.

So, let's say $80/card R&D to account for lower cost attributed to lower end cards (although the overall costs would most likely be less), about $200 to produce and ship and you'll end up with $350+/- for a high end card to produce and recoup the initial investment. You can still add $200 on top and sell it for $550-$600 as per usual and still make a profit at the high end.

Of course, when you divide that $500mil cost for a bigger market share, let's say 20-30% (which is doable even for AMD if they would actually bother with it), then would be 24mil units sold (for 30% market share), and the R&D price goes down to $20,83. For those holding 70% market share, $500mil R&D is around $8,93 / card. Again, considering you're not doing just gaming cards, that R&D is only large initially.

If you take into consideration the shipments from last 2 years, it goes to around 105 million units shipped, ergo R&D costs / unit are relatively low...

So, all in all... greed from both companies, even more so from nVIDIA.

Intel, by their own admission spent $3 Billion on developing ARC, they are never getting that back.

That doesn't mean much at all. Could be that either you need to invest more initially since you don't have the know-how, or, most likely, they're not the best at spending money efficiently and since AMD managed to best them with a significant less budget in the CPU department I wouldn't put much trust in Intel's ability to spend wisely all the time.
 
The 6900 XT was a 3090 competitor and they sold it for $1000.

If you listen to what AMD has been saying on Twitter they are blaming Nvidia for the weird naming. They say it's called 7900xtx because that's their best GPU and it's $999 because that's where their best was price last time.

So why does it not compete with Nvidia's 90 series like it did last time - blame Nvidia, AMD says they didn't do anything differently, it's Nvidia who decided change their stack - implying that the 4090 is basically a Titan being sold as a 90 series product
 
If you listen to what AMD has been saying on Twitter they are blaming Nvidia for the weird naming. They say it's called 7900xtx because that's their best GPU and it's $999 because that's where their best was price last time.

So why does it not compete with Nvidia's 90 series like it did last time - blame Nvidia, AMD says they didn't do anything differently, it's Nvidia who decided change their stack - implying that the 4090 is basically a Titan being sold as a 90 series product
So AMD is basically saying the 4090 is a bargain compared to their top product.

Well at least they’re being honest.
 


Leaving aside the tops from the last mining bubble, for a 2 year period lasting q1 2019 to q1 2021, there are around 80mil. units shipped. At that time, it could have been more, but that bubble flooded back then the market with plenty of second hand GPUs.

Let's say AMD sticks to its premium branding strategy and only stay around 10% market share, that's 8mil units, making that 500mil R&D budget about $62.5/card on average. Of course, gaming is only a part of AMD and nVIDIA's sales, so that $500 mil will result in even less cost / card.

BOM costs should go down over a 2 year life cycle and no one says you can't still produce older gen cards from the lower/mid end if the costs are low enough...

Shipping and logistics I don't know how much they are, but I doubt they're more than 30-50 dollars / card.

So, let's say $80/card R&D to account for lower cost attributed to lower end cards (although the overall costs would most likely be less), about $200 to produce and ship and you'll end up with $350+/- for a high end card to produce and recoup the initial investment. You can still add $200 on top and sell it for $550-$600 as per usual and still make a profit at the high end.

Of course, when you divide that $500mil cost for a bigger market share, let's say 20-30% (which is doable even for AMD if they would actually bother with it), then would be 24mil units sold (for 30% market share), and the R&D price goes down to $20,83. For those holding 70% market share, $500mil R&D is around $8,93 / card. Again, considering you're not doing just gaming cards, that R&D is only large initially.

If you take into consideration the shipments from last 2 years, it goes to around 105 million units shipped, ergo R&D costs / unit are relatively low...

So, all in all... greed from both companies, even more so from nVIDIA.



That doesn't mean much at all. Could be that either you need to invest more initially since you don't have the know-how, or, most likely, they're not the best at spending money efficiently and since AMD managed to best them with a significant less budget in the CPU department I wouldn't put much trust in Intel's ability to spend wisely all the time.

In 2019 1GB Micron 14GB/s Memory IC's where $11.50 each if buying 2000 at a time

The cost of those will have gone up, lets assume the memory IC's on the 7900XTX cost the same, because they are buying in maybe 50,000 bulk, they are 12X 2GB, 20GB/s, at $11.50 that's $138, just for the memory IC's, the rest of the PCB contains about 100 other components costing ranging from tens of cents to a few $ each, assembly costs, the cost of the chip its self, the cooler, the shroud, fans, backplate...

I think you're looking at $350 before you have a product, then shipping, the retailer wants their cut.

I think they could sell them for $600, get their completely assumed $500 Million R&D investment back (Bear in mind that AMD R&D spend for 2022 alone is $6 Billion, twice what Intel spent on ARC over 4 years) and make $600 Million profit over the two years, they made that profit in a single quarter of Ryzen sales.
 
Last edited:
In 2019 1GB Micron 14GB/s Memory IC's where $11.50 each if buying 2000 at a time

The cost of those will have gone up, lets assume the memory IC's on the 7900XTX cost the same, because they are buying in maybe 50,000 bulk, they are 12X 2GB, 20GB/s, at $11.50 that's $138, just for the memory IC's, the rest of the PCB contains about 100 other components costing ranging from tens of cents to a few $ each, assembly costs, the cost of the chip its self, the cooler, the shroud, fans, backplate...

I think you're looking at $350 before you have a product, then shipping, the retailer wants their cut.

I think they could sell them for $600, get their completely assumed $500 Million R&D investment back (Bear in mind that AMD R&D spend for 2022 alone is $6 Billion, twice what Intel spent on ARC over 4 years) and make $600 Million profit over the two years, they made that profit in a single quarter of Ryzen sales.
Why do you assume the cost of Memory will have gone up? For a start they use 2gb IC's which are not double the cost of 1gb and generally tech gets cheaper as time passes.

A friend bought 16gb 3600/16 Bdie kit in 2019 for £200 and now you can get a 3600/14 for around half that, while the price of DDR5 has been dropping fast so why would VRAM be going up?
 
In 2019 1GB Micron 14GB/s Memory IC's where $11.50 each if buying 2000 at a time

The cost of those will have gone up, lets assume the memory IC's on the 7900XTX cost the same, because they are buying in maybe 50,000 bulk, they are 12X 2GB, 20GB/s, at $11.50 that's $138, just for the memory IC's, the rest of the PCB contains about 100 other components costing ranging from tens of cents to a few $ each, assembly costs, the cost of the chip its self, the cooler, the shroud, fans, backplate...

I think you're looking at $350 before you have a product, then shipping, the retailer wants their cut.

I think they could sell them for $600, get their completely assumed $500 Million R&D investment back (Bear in mind that AMD R&D spend for 2022 alone is $6 Billion, twice what Intel spent on ARC over 4 years) and make $600 Million profit over the two years, they made that profit in a single quarter of Ryzen sales.
We don't know the price of the card itself (including the memory which isn't necessary that expensive), that's true, but like you've said, they're not buying those by the one at the time and no one was forcing them to make 24GB cards for nothing...
Of course they've made their Ryzen profit, after all, they pushed the prices up on those as well. ;)
 
Last edited:
  • Like
Reactions: TNA
We don't know the price of the card itself (including the memory which isn't necessary that expensive), that's true, but like you've said, they're not buying those by the one at the time and no one was forcing them to make 24GB cards for nothing...
Of course they've made their Ryzen profit, after all, they pushed the prices up on those as well. ;)
I'd have been happy with a 12gb XTX if it would have meant prices at £650-700, if anything 24gb is overkill and an unnecessary addition to costs.

It's the same with the large overengineered coolers on the 4080 although I believe this is done more so to hide what's under the hood similar to the kids who stick an RS badge and a body kit on their 1.0L fiesta.
 
Last edited:
Back
Top Bottom