Wednesday, November 28th 2018

Intel Detailing Their Arctic Sound Discrete GPU This December; Aiming for 2020

According to DigiTimes, Intel's top graphics executive Raja Koduri and other senior Intel partners will be hosting a discrete GPU-focused conference this December. The conference aims to instill confidence in shareholders and customers alike in that Intel is pursuing its high-performance discrete entry into the graphics card market at a fast pace. The GPU architecture, codenamed Arctic Sound, is expected to debut by 2020, aiming for the gaming, AI, and machine learning sectors - much like any GPU solution these days. It remains to be seen which details - if any - can be gleaned from this conference, but we'll keep you up to date when those surface.
Source: DigiTimes
Add your own comment

36 Comments on Intel Detailing Their Arctic Sound Discrete GPU This December; Aiming for 2020

#27
Ubersonic
SlizzoWhat are you talking about in reference to PhysX?
In addition to handling all the graphics and the PhysX acceleration Nvidia GPUs have the ability to act purely as a PhysX co-processor (as the original Ageia PhysX cards did), this means you can (for example) use a GTX1080 as your main GPU and also use an additional GTX1050 purely to handle the PhysX (resulting in better performance than the GTX1080 alone would due to offloading the PhysX hardware acceleration).

Now, if you try and use an Nvidia card as a pure PhysX accelerator in a system with a main AMD GPU it will not work (anymore, it used too) as Nvidia have coded the driver specifically to disable it in this scenario. However once Intel start shipping high end GPUs it will work alongside those, which means at that point Nvidia will have no choice but to remove the AMD block or they would find themselves in an extremely precarious legal position.
Posted on Reply
#28
R0H1T
That's interesting, I thought the hacks still worked. Though why do you say that Nvidia will allow their cards to work as PhysX accelerator with Intel dGPU(?) cards?
Posted on Reply
#29
lemonadesoda
Intel Manager desperate to keep his job makes false promisesin that Intel is pursuing its high-performance discrete entry into the graphics card market at a fast pace
er, sounds like 2008 again and Larrabee.
Posted on Reply
#30
Fouquin
Vayra86You'll have to guide me a bit more... I'm not seeing the relation to video cards. :cry:
It's as much related as having the names of trees be the codename for video cards. Or islands.
Posted on Reply
#31
Blueberries
FordGT90Concept"A narrow stretch of water forming an inlet or connecting two wider areas of water such as two seas or a sea and a lake."
a.k.a. a "strait" and, as far as I can tell, there's no "strait" called "Arctic Sound." Intel is making stuff up.
As far as I know there's no Sky Lake or Coffee Lake either.
AssimilatorJust like there isn't a body of water called Sky Lake or Kaby Lake or Cannon Lake... nor are there bridges named Sandy or Ivy. Codenames don't have to mean anything, Intel has seemingly gone for water-based ones since 2015.
There IS a Kaby Lake in Ontario, Canada and a Cannon Lake in Rice County, Minnesota

I believe Intel chose geographical naming schemes to emphasize their global outreach. Sandy Bridge was definitely a play at Silicon.
Posted on Reply
#33
Prima.Vera
Who's the retard doing all those naming schemes for GPUs and CPUs ???
Posted on Reply
#34
eidairaman1
The Exiled Airman
BwazeAnd cryptomining, don't forget cryptomining!
It went bust
Posted on Reply
#35
Ubersonic
R0H1TThat's interesting, I thought the hacks still worked. Though why do you say that Nvidia will allow their cards to work as PhysX accelerator with Intel dGPU(?) cards?
Because they have sort of walled themselves in there. Over the past 5-10 years they have sold millions and millions of laptops which, in order to perform well on battery and reduce thermals, use the Intel iGPU as their primary display output and their Geforce card as a standalone 3D accelerator which boosts performance as and when needed. So if Intel were to update their drivers to disable PhysX when the primary GPU is AMD or Intel, then it would cripple millions an millions of existing Nvidia products, that's class action territory and considering many of these laptops cost in excess of four figures its BIG lawsuit territory.

The end result is it would be far less damaging to Nvidia just to not block it, which means they would also have to unblock AMD as they would then have grounds to sue Nvidia for specifically targeting them (as opposed to targeting every discrete GPU manufacturer like they do now).

We are in interesting times as Intel turning this into a three way dance will be good for consumers even if they don't want an Intel GPU.
Posted on Reply
Add your own comment
Nov 16th, 2026 08:46 CST change timezone

New Forum Posts

Popular Reviews

TPU on YouTube

Controversial News Posts