beaten already.

pc handheld = zero nvidia

console = zero nvidia


nvidia dont made good cpu is the problem

at least nintendo love them.
recent rumor said nvidia want nintendo to clock the switch2 chip high so it could give 4tflop so it didnt appear weak to the competition, but nintendo said no.

Hari2 say igpu can blow off GPU but nowhere to be seen. Same like apple silicon killer

bodo

igpu and gpu is the same product

how i3 can beat same gen i9???

iGPU always a tuned down ver.

nvidia caters to real pc master race. high res high refresh rate gaming is what we need.

Later us sanctions Malaysia then intel, Nvidia, AMD semua pun tak dapat , later gt1030 become RM 10k per piece in Malaysia 😂😂

Nvidia bring their gpu out side gaming world

medical
industrial
labs
automobil

..... ai

Nintendo switch is using Nvidia tegra sei sohai

The nxt switch will also most likely be using Nvidia

This post has been edited by Baconateer: May 4 2024, 01:24 PM

Bruh
Igpu matching 1050 pun already pipedream once upon a time.
Matching 1060 is already big at it is.

Matching 4070 is a big claim but seeing how amd managed to achieve something once thought impossible, i would give them benefit of the doubt.

We already know it's not about processing power nor is it it about power consumption/efficiency issue, it's always been a bandwidth problem, seeing how m1 and consoles do it.

Maybe quad channel ddr5 or even ddr6 if amd managed to rush it could solve bandwidth issue.

calculation of memory bandwidth seems wrong. 256bit of 8533 memory is 273GB/s not 500.

on paper it should be around the performance of 7600xt, since total 150/175w also caters to the zen cores too unlike the discrete 7600xt can use as much power it wants. Similar memory bandwidth too.

EDIT: my mistake i only saw rtx4070 laptop and not desktop, no way it is faster than rtx4070 desktop, laptop can be close

This post has been edited by terradrive: May 4 2024, 02:57 PM

misleading title, you wrote rtx4070 but didn't specify rtx4070 laptop

effective bandwidth, thats including cache too.
there are probably ipc increase and clock uplift on rdna3.5. yup they are targeting mobile variant of rtx4070 which pretty much closer to rtx4060ti level.

isn't effecttive bandwidth already written there, the 8533MT/s.

thats ddr speed.
the reason why you could see 7800x3d yield great number of fps even with ddr5 6000. big cache do helps alot

This post has been edited by zerorating: May 4 2024, 03:12 PM

every gpu has it's own cache. the thing that matters is the memory bandwidth the most, unless you are using like 192-256MB of big size caches. iinm amd already has big caches and it still didn't give huge fps advantages over nvidia on comparable segment

do remember that amd dont have access to gdddr6x, thats why they need to compensate to big cache. nvidia high end chip also have big chungus cache, so their advantage become larger plus superior compression tenchique too.
32mb l3 is a decent amount for those 200mm soc, they didnt tell if l2 cache got increase or not.

anyway its a compromised product but if amd charge their client low, then thats a decent product, you can cut product cost because no need to purchase gddr module, more vrm, chokes and capacitors and simplified cooling system.

This post has been edited by zerorating: May 4 2024, 03:29 PM

Good explanation!