• 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.

Zen 3D V-Cache Ryzen CPUs May be Available in December 2021

Joxeon said:
AMDs success is really down to TSMC. Their Gpus while good only match Nvidia on a far worse Samsung node and they also have a node advantage against Intel.

If China invades Taiwan then what happens to AMD?
Click to expand...
The fact little Johnny won't be getting an Xbox for Christmas if China invades Taiwan is the last the thing we should all worry about in such a situation.

I don't where you getting this assumption that Samsung's process is 'far worse' then TSMC, Ampere draws many parallels to GNC in terms of design rationale which is focused around the data centre. Cramming all those cores into the chip creates issues with utilisation and occupancy which has meant Nvidia having to ramp up the core clocks resulting in the GPU having a massive power draw and using GDDR6X isn't helping matters either in terms of efficiency. None of those things have anything to do with Samsung's process, if they tried to build a GA102 at TSMC if would cost Nvidia a lot of more money to make meaning even more expensive GPU's (Nvidia won't accept a lower margin product).
 
Freddie1980 said:
The fact little Johnny won't be getting an Xbox for Christmas if China invades Taiwan is the last the thing we should all worry about in such a situation.

I don't where you getting this assumption that Samsung's process is 'far worse' then TSMC, Ampere draws many parallels to GNC in terms of design rationale which is focused around the data centre. Cramming all those cores into the chip creates issues with utilisation and occupancy which has meant Nvidia having to ramp up the core clocks resulting in the GPU having a massive power draw and using GDDR6X isn't helping matters either in terms of efficiency. None of those things have anything to do with Samsung's process, if they tried to build a GA102 at TSMC if would cost Nvidia a lot of more money to make meaning even more expensive GPU's (Nvidia won't accept a lower margin product).
Click to expand...
TSMC 7nm+ Can squeeze in almost 50% more transistors per mm2 over Samsung 8nm so it's not a supprise it's by far the more power efficient node and the reason Nvidia avoided it and went with TSMC when making the data center cards.

As you say Nvidia went with Samsung to wring out the most profits and they have certainly done that, power efficiency was of little concern.
 
Last edited:
Freddie1980 said:
The fact little Johnny won't be getting an Xbox for Christmas if China invades Taiwan is the last the thing we should all worry about in such a situation.

I don't where you getting this assumption that Samsung's process is 'far worse' then TSMC, Ampere draws many parallels to GNC in terms of design rationale which is focused around the data centre. Cramming all those cores into the chip creates issues with utilisation and occupancy which has meant Nvidia having to ramp up the core clocks resulting in the GPU having a massive power draw and using GDDR6X isn't helping matters either in terms of efficiency. None of those things have anything to do with Samsung's process, if they tried to build a GA102 at TSMC if would cost Nvidia a lot of more money to make meaning even more expensive GPU's (Nvidia won't accept a lower margin product).
Click to expand...

I don’t think Nvidia will have much choice from this point on. The competition is very strong and they don’t the same supply constraints.
 
Joxeon said:
TSMC 7nm+ Can squeeze in almost 50% more transistors per mm2 over Samsung 8nm so it's not a supprise it's by far the more power efficient node and the reason Nvidia avoided it and went with TSMC when making the data center cards.

As you say Nvidia went with Samsung to wring out the most profits and they have certainly done that, power efficiency was of little concern.
Click to expand...

I am just happy to see that they are going back to TSMC.
 
pete910 said:
Because 10 is higher that 7 :p
Click to expand...

i hate these marketing terms can you guys please switch to transistor density instead "7nm" and "10nm" doesn't mean anything.

If you look at transistor density first and then compare that to the marketing name you'll see that tsmc "7nm" has the same density as Intel's "10nm" and Samsung's "8nm" has the same density as TSMCs "10nm".

The reason Intel is behind is because their architecture doesn't scale well with more cores and their process node doesn't have a good failure rate.

The reason AMD is so close to Nvidia in GPus is because TSMC's "7nm" has a nearly 50% higher transistor density than Samsung's "8nm" - and if you looked at density first and compared like for you like you'd more easily see why that is
 
Grim5 said:
i hate these marketing terms can you guys please switch to transistor density instead "7nm" and "10nm" doesn't mean anything
Click to expand...

We lost that war to marketing guys. Just look what marketing has done Cyber-Mav. He seems to genuinely believe what he’s saying.
 
You must log in or register to reply here.
Back
Top Bottom