• 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.

Isn't it possible that AMD will save it's most powerful GPU designs for 5nm TSMC fab process?

Soldato
Joined
17 Nov 2005
Posts
3,583
I hope they have something GPU wise. Same a I hope intel come back CPU wise.

It's all good for the consumer.

Yes To much of a lead stagnates progress, if Intel had giving people more of a increase in performance over the last 10 years Amd would have not caught up to them.
 
Soldato
Joined
13 Jan 2010
Posts
6,354
Location
Manchester
Yes To much of a lead stagnates progress, if Intel had giving people more of a increase in performance over the last 10 years Amd would have not caught up to them.

I don't think it's about not giving. I think that's all they had!

So with no competition, the price went up and they got lazy with the (can't think of the word, very drunk) research and upgrades.

I'm on old tech x99 8 core.... But it's still epic, what can actually beat a old rig like this? if you take into account the age and the time Vs money spent over the years? It still holds its own.

Btw I got the 5960x used for £200. Sold my 5630kfor £130 ISH.so I didn't pay the RRP
 
Soldato
OP
Joined
30 Jun 2019
Posts
7,875
There's also rumours of a Navi 31 being developed, for potential release in 2021. If true, I wonder if this design and others could be 5nm GPUs?

MCM designs might only be used for compute CDNA based GPUs, at least initially in 2021.
 
Last edited:
Soldato
Joined
24 Jul 2004
Posts
22,594
Location
Devon, UK
I thought Apple got there first with a 5nm TSMC order and bought up all the capacity? In which case I expect we'll see it late next year when they can spare some for AMD.
 
Soldato
OP
Joined
30 Jun 2019
Posts
7,875
I could be wrong, but TSMC seems to be investing heavily into 5nm production. I think there will be enough capacity to go around, by the time the N5p fab process has passed risk production.
 
Soldato
OP
Joined
30 Jun 2019
Posts
7,875
The hype train is now running at 'full steam ahead' and with about the same level of sophistication.

There is absolutely no chance AMD will 'delay' some GPUs, to wait for TSMC 5nm availability according to the main Navi hype mongering thread :D
 
Last edited:
Soldato
OP
Joined
30 Jun 2019
Posts
7,875
Last edited:
Soldato
OP
Joined
30 Jun 2019
Posts
7,875
Last edited:
Soldato
OP
Joined
30 Jun 2019
Posts
7,875
Anyone going to weigh in on this? What do you make of the TSMC 5nm / '5nm+' plans outlined here:

https://hexus.net/tech/news/industry/142480-list-tsmc-5nm-customers-orders-published/

Perhaps unsurprisingly, it looks like both Nvidia and AMD have booked 5nm well in advance, for 2021-2022.

Personally, I don't think too much attention should be paid to the RDNA 3 / Hopper architecture names, the names themselves (and any associated rumours) probably don't matter much in my view.

Speculation - AMD and Nvidia are waiting for volume production of N5+ rather than standard 5nm, as this is better suited to higher spec parts, such as GPUs and CPUs.

Maybe even 3/4nm wide usage by 2023, if all goes to plan?
 
Last edited:
Man of Honour
Joined
13 Oct 2006
Posts
91,147
(not at you G6) The freakin slide has a picture of a RX6900 in it and some people still think the performance numbers are for a different card

If you look at https://hexus.net/tech/news/graphics/146008-amd-outs-radeon-rx-6000-performance-rtx-3080/ for instance those numbers are pretty much bang on what you'd expect over a 5700XT with arch refinements, node improvements and the inclusion of things like RT for big Navi at the kind of commercially viable core sizes for a gaming GPU. Highly unlikely to be a lower tier than the big Navi core though possible it isn't the fully specced version of the core but I think people are getting their hopes up.

(Personally I'd expect to see like 2-4% improvement in the numbers with launch drivers)

Combined with the way nVidia has used the 3090 (an unusual move) I'm pretty sure that card is aimed at keeping the crown and there is no way they are completely out of touch with what AMD are doing.
 
Soldato
Joined
8 Jun 2018
Posts
2,827
(not at you G6) The freakin slide has a picture of a RX6900 in it and some people still think the performance numbers are for a different card

I think we have a failure to communicate here.
Lets look at the definition of "Series" which is found on that chart you mentioned: "RX 6000 Series
In part series means:
a number of things or events of the same class coming one after another in spatial or temporal succession

Since we can conclude that if the term RX 6000 Series doesn't designate it as a stand alone product. However, lets ask ourselves a question about it. Has AMD used the term XXXX Series before? Yes, yes they have with the RX 5700 Series as linked below:
https://www.amd.com/en/partner/amd-radeon-rx-5700-graphics-card

Eureka!! It appears that when AMD uses that term, RX XXXX Series, they are implying more then one product. As there is no such thing as a Radeon card called RX 5700 Series by itself. It shows that within the 5700 Series you have both a 5700 and 5700 XT, for example.

I do hope that clears up why people still think that the performance numbers might be for a different card. It's because the term RX 6000 Series doesn't designate a specific card. And since AMD is being nebulous about what card is used it leaves room for speculation until the release date.
:)
 
Man of Honour
Joined
13 Oct 2006
Posts
91,147
I think people are kidding themselves if they think this wasn't big Navi being shown off (as some still believe) - I'll concede it is possible it isn't the biggest of big Navi (though I have a lot of reasons to doubt that) - but they'd have to have pulled off a miracle to get this kind of performance out of the kind of silicon that would be commercially viable as a Navi 22 type part - the physics just don't work.

For instance a lot of their 50% perf/watt improvements will come from things like more optimised input to setup engines allowing lower power draw at the same performance, the ability to better gate power usage on unused parts of the GPU during processing, etc. which might allow for a small reduction in area and a small improvement in frequency but it is going to be single digit percentage type stuff versus the 10s of percent for reduced wattage.
 
Back
Top Bottom