• 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 Navi 23 ‘NVIDIA Killer’ GPU Rumored to Support Hardware Ray Tracing, Coming Next Year

Status
Not open for further replies.
That would make sense of course. But looking at potential projected power draws of the top navi, which could be around 320-360W, a three(3) fan 2.5 slot is bare minimum by the look of it.

Water cooled edition could be incoming to cool that puppy,

If the firmware stuff that was leaked is legit there were two profiles 230 watt and 250 watt and I assume they are the graphics core and not total board but I have no idea on that. Suggests to me they only have two products on the top tier core or very close derivatives if there are more.
 
The PC version was running on its Ultra settings and XSX demo (which was thrown together in 2 weeks) ran the same settings with additional effects and outperformed the PC. I suggest you go watch the event if it's still available before spouting off.

Dont want to go too much off topic here LePhurroni, seeing as this is a navi thread not xbox thread.
But i will say this, a modified XSX Microsoft demo designed to promote their new gaming console aint going to cut any mustard here, whatever setting they used.

Get back to me when it runs something other than ported game so we can have real data to compare it to. Till then its nothing more than advert for their new console with streamed line code slapped on.
 
I just have my doubts that if the card Lisa Su showed was actually a 6800xt that it would be just 72CUs.
A 3 fan, 2.5 slot card for just 72CUs? An 80CU card would seem more appropriate if that rumor was true.

But we will see in a few days.
If the recently rumoured 320W+ are true, that seems perfectly reasonable. I'd expect AIBs to be releasing 3 slot or close to cards.
 
1) Speculation and estimation are not lies so stop being dramatic.

2) That is what the AMD / NV driver teams do in the PC space. In the console space devs optimise for the console hardware.

On a different note I am pretty convinced that N21 cannot be 500+mm^2 @ 3080/3090 performance. The money is just not there for that large a GPU. A wafer of N21 dies @ 500mm^2 would create 83 good dies. If all bin to 6900XT requirements then that is around $58k in revenue (split between AMD, retailers, distribution etc). If they used that wafer for Zen3 dies instead they would make 782 good CCDs which even if used in 5600X's would generate around $230k in revenue (again split). I think that kind of disparity is too much for AMD and their shareholders to swallow.

Yes misstating what DF said is a lie (or a mistake).

Funny how in this thread when people do the same but to the detriment of AMD they don't get that benefit of the doubt.

Note that it turns out to be a 2080 AMD fanboys are on the backfoot again trying to caveat DF's statement (not necessarily you though).

Whatever pal, whatever. Not sure "we put the Xbox demo together in 2 weeks" counts as "streamlined code", but whatever floats your boat.

How quickly do you think game ready drivers are released on PC? PCs don't get streamlined code for each GPU. If anything it's more of a valid comparison. Consoles otherwise outperform the hardware they have.
 
Last edited:
Yes, the 255W TGP was debunked in favour of an actual 300W TBP.

It's not really debunked. TGP is the chip power you would see in GPU-Z and software monitoring.

It was mentioned several times when that number came out earlier in the thread. It was logical that this number would leak. However, as is par for this thread, it was ignored for a more favourable interpretation.
 
On a different note I am pretty convinced that N21 cannot be 500+mm^2 @ 3080/3090 performance. The money is just not there for that large a GPU. A wafer of N21 dies @ 500mm^2 would create 83 good dies. If all bin to 6900XT requirements then that is around $58k in revenue (split between AMD, retailers, distribution etc). If they used that wafer for Zen3 dies instead they would make 782 good CCDs which even if used in 5600X's would generate around $230k in revenue (again split). I think that kind of disparity is too much for AMD and their shareholders to swallow.

Usually the way around that is to put multiple dies on a wafer with a small number of larger and larger number of smaller dies - but the AMD 7nm wafers I've seen so far have been all one product (not necessarily the same model) which I found a little surprising.

I think people are going to be disappointed this round if they expect AMD to be disruptive on pricing TBH unless they've pulled out quite a feat of engineering they are likely going to slot into nVidia pricing given the costs and the disparity vs Zen3 production.

EDIT: On the subject of MCM though not entirely relevant to this thread:

https://twitter.com/kopite7kimi/status/1318781796360675330
 
https://hexus.net/tech/news/graphics/146311-amd-partner-preps-radeon-rx-6000-card-2577mhz-boost/

The 'very special board partner card', which is the source of Patrick Schur's BIOS, sets the GPU boost clock to 2,577MHz.

abeedb3a-6f7d-4775-a5bb-6d59d73a3e1f.png


https://www.igorslab.de/en/the-can-...-works-oc-may-maybe-a-glance-in-bios-details/

The power limit for the socket.Power is already at 255 watts. If you add the maximum allowed 15% (without MPT intervention), you end up at 293 watts. That would be plus memory up to 315 watts plus the rest (partial voltages, converter and board losses, fans etc.).

Screen-3.jpg
 
It's not really debunked. TGP is the chip power you would see in GPU-Z and software monitoring.

It was mentioned several times when that number came out earlier in the thread. It was logical that this number would leak. However, as is par for this thread, it was ignored for a more favourable interpretation.
I mentioned it earlier in the thread if it was TGP there would be additional wattage to add (TBP).

As you say, unfortunately it started a mini hype train here and elsewhere on the net.
 
It's not really debunked. TGP is the chip power you would see in GPU-Z and software monitoring.

It was mentioned several times when that number came out earlier in the thread. It was logical that this number would leak. However, as is par for this thread, it was ignored for a more favourable interpretation.

I mentioned it earlier in the thread if it was TGP there would be additional wattage to add (TBP).

As you say, unfortunately it started a mini hype train here and elsewhere on the net.

It is not just chip/socket power..
its basically chip+gddr6 power
which patrick clarified later
so i would be adding few 10's of watts above it possibly 30w
however, igor puts TBP at 320-355W for navi 21 xt which requires ton of salt.. i will believe it when i see it

On a different note I am pretty convinced that N21 cannot be 500+mm^2 @ 3080/3090 performance. The money is just not there for that large a GPU. A wafer of N21 dies @ 500mm^2 would create 83 good dies. If all bin to 6900XT requirements then that is around $58k in revenue (split between AMD, retailers, distribution etc). If they used that wafer for Zen3 dies instead they would make 782 good CCDs which even if used in 5600X's would generate around $230k in revenue (again split). I think that kind of disparity is too much for AMD and their shareholders to swallow.

Decisions like those have to factor capital deployment and return metrics, other than the fact that you cant just be dumping the most profitable product in excess of market demand.. part of the reason why Intel would be making the HPG line of GPUs on TSMC 6nm instead of allocating capacity inhouse.

Infact the capital deployed in GPU biz is lot lower than CPUs cuz you can push the inventory much faster to AIBs than directly to distributors
 
Last edited:
Anyone noticed relatively inactive accounts always seem to activate around hardware launches...?

You also get an influx of new, low post count, users around that time as well. Some of them are almost certainly dupes of existing longer term members from a general observation of posting styles :s
 
Status
Not open for further replies.
Back
Top Bottom