-
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.
RX 7900XT, 15,360 cores, MCM, Tapeout Q4
- Thread starter humbug
- Start date
Soldato
@KompuKare Fair assessment 
Let's see if this is more than just a rumor. Maybe they decided to do a big one to compete with Nvidia's excessive cards and saw there's a market for them.
Let's see if this is more than just a rumor. Maybe they decided to do a big one to compete with Nvidia's excessive cards and saw there's a market for them.
Iirc Intel's full fat DG2 GPU with four 512 Xe tiles comes in at about 1000mm2 now and they've even had test GPUs at 1200mm2 last year
1000mm2 sound like it's too big, and it would be for a single monolithic die but MCM allows a lot more silicon to fit onto the same PCB
And yes more silicon = more $$$.
anyone who thinks the days of $200 entry level cards and $600 flagships are coming back is crazy
The use case may be unclear but if AMD does make a card 2-2.5x faster than a 6900XT I doubt a 630W power draw will scare anyone that can afford it.
I was actually thinking of someone else but yes, you're sitting there with more than 630W power draw from your sli 3090s so huge power draw isn't going to be a problem if the performance is there. For whatever its good at.
Associate
- Joined
- 4 Nov 2015
- Posts
- 250
Sucks fake, I have to build another one?!i see the psu requirements have leaked for next gen flagship cards
Soldato
At $2500, it will need to be able to print money via crypto mining or something.
If it doesn't print money, it will sit on shelves at $2500. -Even if it can do a million FPS @8k.
Now that mining is becoming less profitable, cards are just sitting on websites at over inflated prices.
Soldato
Lol nope
Lovelace shipping Q2 2022, 60 to 100% faster than RTX3090.
RDNA3 100% faster than 6900XT.
Internally AMD expects to beat Lovelace.
Internally, Nvidia has a backup plan in case they lose the performance crown - and that is to drop their prices and ship huge numbers of cards to hold overwhelming market share.
AMD intends to own the position of top dog performance, with execs floating the idea of pricing RDNA3 cards higher up than RDNA2 - an example is give of $2000 RRP for the 7900xt
RDNA3 100% faster than 6900XT.
Internally AMD expects to beat Lovelace.
Internally, Nvidia has a backup plan in case they lose the performance crown - and that is to drop their prices and ship huge numbers of cards to hold overwhelming market share.
AMD intends to own the position of top dog performance, with execs floating the idea of pricing RDNA3 cards higher up than RDNA2 - an example is give of $2000 RRP for the 7900xt
Last edited:
Soldato
Nope what?
Almost right. Well according to MLID anyway RDNA3 could be anywhere between 0% and 40% faster than Nvidia in rasterization, but where it lands depends on what Nvidia decides to do in terms of how much it cuts down its gaming gpu core design from the x100 GPU and what TDP they go with. There is a path that Nvidia has to get it close to 0% but it involves making a huge GPU with massive TDP higher than the RTX3090's 350w
As for price, MLID says the 7900xt at this stage has a placeholder price of $2000usd and the final price will naturally depend on where it's performance advantage lands compared to Nvidia.
You really love the clickbait channels, MLID is a snake oil salesman and talks a lot of rubbish, his whole channel is based on guessing and unrealistic rubbish to get people to click.
Can you imagine what will happen when AMD says the MSRP is $2000-$2500 for a 7900xt compared to the 6900xt at $999 ... ? AMD will have to fire sell them in the end as no one will buy them..
People laughed hard when the 6900xt came out at $999 (people have just forgotten because of the silly prices we had after because of scalpers and miners, even the 3090 was laughed at but it had the ability to be a good work card, where the 6900xt was useless for serious work) and especially after the reviews when people realised it is basically an old gen card with good performance and not much else and the 6800xt was the one to buy. All they did was make the 3090 look like a good buy as it was just a better card with next gen features and actually usable for work too.
If AMD say $1500 msrp for a 7900xt they can forget being in the market especially with all the fake msrp stuff going on and imagine if we have another mining craze around then what will the prices be from MSRP.. 2x 3x.. You honestly think anyone is crazy enough to pay $3000-$4000 for it then ? It would have to be able to mine $100 a day for people to buy it.
Last edited:
Caporegime
- Joined
- 18 Oct 2002
- Posts
- 29,851
Rubbish, they've gone that high before with the 295x2. They did however drop them quite a bit in the end then they sold quite healthily.
It doesn't seem like good business sense to release a 7900XT that's twice as powerful as a 6900XT. All they need to do is make a card thats around 50% faster and can play 4K games at over 80fps with full RT and it will sell. If the 7900XT is really double the performance of a 6900XT then I suspect whoever buys it will not need to upgrade for quite a while.
Intel just called, they want their CEO back.
I think theres a point being made that is being mis-understood.
Many of us here used to buy 2,3,4 cards and cross-fire/SLI them, I have a feeling that once it became clear to AMD/Nvidia that this was a dead-end i.e. a) gaming support would evaporate over time and b) data-centre gpu grids are must have, the interposer replacement for cross-fire/SLI became a priority and here we are, 12 months away from them becoming available.
Making GPU's scale like CPU's with the scale-up of price ranges makes sense if you see GPU horsepower in the same way as CPU.
It also allows AMD/Nvidia to get buyers like myself re-engaged with buying multiple GPU's.
I think its a mouth-watering prospect and yes, I would imagine an 4x GPU card would be wallet-busting but so is a 64 core CPU. The downside is that for games, the game devs. still target 6 core CPU's because that is what is mainstream and therefore, in a way, a 4x4 GPU will probably only boost your fps rather than the detail level. For most this will be more than a big enough carrot to stump up.
Something that has 4 x the GPU power of a 3080/6800 class card (minus the vram bottleneck and lower clock speeds) would be pretty mental but the euphoria wears off most buyers, like CPU's, will ask 'what do i need' versus 'what do i want?' and will probably settle for a 2x GPU card. A bit like when you get 1gbit broadband and realise 300mbit is probably enough.
The real winner here is probably VR, getting to an 8k upsample @ 120hz is probably grail territory but mgpu's will open the door to make it a possibility.
Associate
There is of coure another way to approach pricing:
We can plug that into a wafer yield calculator (taking the 0.09 defect per cm which is something TSMC mentioned for 7nm - although that was a few years ago now):
https://caly-technologies.com/die-yield-calculator/
Now 600mm² is 24.5mm x 24.5mm, which gets us this:
For the die (ignoring testing, packaging, VRAM, board etc.), the only other consideration is how much wafers cost.
TSMC 7nm was meant to be about $10,000 per wafer but prices are rumourd to actually have gone up a bit.
6nm might be very similar rather than costing more as EVU might be super expensive but throughput and amount of masks is reduced.
5nm will probably be more, maybe $15,000 per wafer.
Anyway, $10k get us to $190 per die, $15k to around $280.
(This ignores that with redudancy not all dies with defects will be wasted, although against that not every good die is able to hit clocks/power targets.)
Those numbers aren't that bad actually: in theory that means a dual chiplet card might be doable for around $1,200 or so even with the $15k figure.
However, AMD at least have more profitable things they can do with wafers and that would be Zen3 or Zen4 CCDs or even APUs.
The current Zen CCD is 83.7mm² which means the wafer costs between $15 and $23 ($10k / $15k). Zen 4 might be a bit bigger, and of course the IO dies come on top of that.
In other words, CPUs have far higher margins.
We see that with Zen3 vs Navi21. A 5600X @ $300 versus a RX 6800 @ $579 using the $10k figure gives margins of 95% and 72% (with $15k per wafer it is 92% and 58%).
Hard to see how things will play out.
AMD seem to madly focused on margins ATM even at the cost of marketshare and mindshare. The design and mask costs are fixed, so once wafer are available it makes no sense to price too high, but what do we know?
- How much would it cost to make
- And, if wafers are still in short supply (which they will be...) how much profit could they make producing CPUs instead (this is mainly revelant for AMD)?
We can plug that into a wafer yield calculator (taking the 0.09 defect per cm which is something TSMC mentioned for 7nm - although that was a few years ago now):
https://caly-technologies.com/die-yield-calculator/
Now 600mm² is 24.5mm x 24.5mm, which gets us this:
For the die (ignoring testing, packaging, VRAM, board etc.), the only other consideration is how much wafers cost.
TSMC 7nm was meant to be about $10,000 per wafer but prices are rumourd to actually have gone up a bit.
6nm might be very similar rather than costing more as EVU might be super expensive but throughput and amount of masks is reduced.
5nm will probably be more, maybe $15,000 per wafer.
Anyway, $10k get us to $190 per die, $15k to around $280.
(This ignores that with redudancy not all dies with defects will be wasted, although against that not every good die is able to hit clocks/power targets.)
Those numbers aren't that bad actually: in theory that means a dual chiplet card might be doable for around $1,200 or so even with the $15k figure.
However, AMD at least have more profitable things they can do with wafers and that would be Zen3 or Zen4 CCDs or even APUs.
The current Zen CCD is 83.7mm² which means the wafer costs between $15 and $23 ($10k / $15k). Zen 4 might be a bit bigger, and of course the IO dies come on top of that.
In other words, CPUs have far higher margins.
We see that with Zen3 vs Navi21. A 5600X @ $300 versus a RX 6800 @ $579 using the $10k figure gives margins of 95% and 72% (with $15k per wafer it is 92% and 58%).
Hard to see how things will play out.
AMD seem to madly focused on margins ATM even at the cost of marketshare and mindshare. The design and mask costs are fixed, so once wafer are available it makes no sense to price too high, but what do we know?
Last edited: