Return to Level1Techs.com

Nvidia RTX 20XX Thread


#102

image

Don’t be upset.

It is time that we realize that we the market have time and time again demonstrated to Nvidia that they as a business can charge whatever they want for their products.

We keep buying and even pre-ordering them en mass.

Let’s all wait for the the later part of September :slight_smile:


#103

and that is sad


#104

Now this is a painful read:


#105

Written by Avram Piltch aka Editor in Chief of Toms Hardware aka https://twitter.com/geekinchief
Now I’m not usually one to call shill on someone, But in the past he’s been a well known tech shill with no particular brand affiliation, I think I’ve seen him using the #BeForTheGame nvidia hashtag in a few places.

If you need to promote your tech, Avram is your guy.


#106

#107

Which really makes me hate having to buy NVidia in the future if AMD can’t get their graphics shit together.


#108

buy this because of features you can’t use for the next few months until they get patched into games you may or may not even have!

Oof


#109

Did not happen for the features Vega has, won´t happen for RTX features.
Hell, it is not even happening for the DX12 or Vulkan features!

Article is… Oof indeed…


#110

Well this oughta be good… Tech Jesus has caught wind of it.


#111

This may need its own thread, but for now, I park it here:


#113

Im smiling :slight_smile:
Maybe Avram Piltch makes editorials drunk like I post some times :slight_smile:


#114

The glaring problem is the RTX chip is massive so expensive. Most high end gamers are 1060 or RX580 budget gamers or cheaper cards. Look at the top steam stats for gamers. The top gamers are using budget cards. That includes the consoles.

One of the biggest chips ever built for a GPU will not be tech on a budget card next year or even 2020. Nvidia whole gamble on this chip with redundant tech that is not needed to game is people will pay to play high end.

RTX needs int32 processing and tensor cores for AI. That takes a lot of room on a chip if its not used.

Budget cards will be all FP32 (or with AMD already FP16 at x2 speed).

IF AMD gets into this it will be a multi chiplet thing with navi + (AMD RTX) or just navi +navi chiplet’s ETC. It will not be some massive chip that costs $1200+


#115

Play this in sync with the Super Mario 2 Overworld theme and you have yourself a thread simulator.


#116

Suppose they are terrified of getting killed by nVidia like HardOCP did for recommending you not preorder.


#117

Ever since that video, Avram’s twitter feed has been a train wreck with him trying to backtrack in a rather clumsy way.


#118

Take it as unconfirmed for now, but allegedly nvidia is spiteful as usual.


#119

Following these social endeavors is exhausting

My only question is that, without tracing, is this the performance order?


#120

Nah.
I would refrain from making such comparisons at this time :slight_smile:

But you can count on the Titan V.


#121

Wait, I just realized that Titan V “Volta” is 12nm as is this 20 series “Turing”

But 10 series “Pascal” are 16nm, and 1050 Ti downwards are 14nm, so there is that tiny thing


#122

You also aren’t accounting for Turings included Tensor cores or improved async compute.