Intel does it on their own products which use combination of Intel and TSMC tiles. And even Lunar Lake on which both tiles are TSMC
NVIDIA discussion [2024]
- Thread starter pharma
- Start date
I think the demand for nVidia is so high right now that they'll use any and all available fabs to produce chips. Pretty sure for at least the next year or so they'll be able to sell as many as they can make.
https://videocardz.com/newz/nvidia-...stigation-over-alleged-market-dominance-abuse
NVIDIA hit with U.S. DOJ antitrust investigation over alleged market dominance abuse
a price to pay for beiing successfulYou're right but it's still Intel silicon. I don't know of IFS packaging any non Intel silicon but presumably they could and should if they truly are independent.
But still, it doesn't exactly make sense for Nvidia.
DavidGraham
Veteran
Former Google CEO Eric Schmidt on the CUDA moat that everyone completely missed, it took place on April 9th, 2024.
Full lecture.
Full lecture.
Last edited:
DavidGraham
Veteran
Former Google CEO Eric Schmidt said in a recent talk at Stanford that tech giants are planning for increasingly large investments into Nvidia-based AI data centers.
“I’m talking to the big companies, and the big companies are telling me they need $20 billion, $50 billion, $100 billion” in AI infrastructure, Schmidt said.
Schmidt said it will be difficult for competitors to catch up with Nvidia because many of the most important open source tools that AI developers use are based on the company’s CUDA programming language. He said AMD’s software that translates Nvidia’s CUDA code for its own chips “doesn’t work yet.”
“I’m talking to the big companies, and the big companies are telling me they need $20 billion, $50 billion, $100 billion” in AI infrastructure, Schmidt said.
Schmidt said it will be difficult for competitors to catch up with Nvidia because many of the most important open source tools that AI developers use are based on the company’s CUDA programming language. He said AMD’s software that translates Nvidia’s CUDA code for its own chips “doesn’t work yet.”
DavidGraham
Veteran
I wouldn't call it an "insignificant delay" since they were aware of cooling issues with the first GH200 system they shipped in late Jan(?).
+6months of cooling issues not being resolved is significant considering the ~12-18month release cadence.
It's possible, but I think most market analysts probably don't think the recent downturn has much to do with these rumors, as it's market-wide, not just AI related or even semiconductor only. It's probably more likely because the surprise raise of interest rate by BOJ and thus causing the unraveling of a lot of yen carry trades. Now yen has settled down a bit so people are now rebuilding their carry trade portfolio (after all, even at 0.25% it's still much lower other places), with probably less leverage than before (for smart people anyway, stupid people will always do risky trades).
https://wccftech.com/mediatek-nvidi...-dedicated-modules-bring-down-monitor-prices/
MediaTek & NVIDIA Partnership Brings G-Sync Technology To A Lot More Gamers, No Dedicated Modules To Bring Down Monitor Prices
That means some new tech must be coming to drive/keep prices up. 50 series exclusive of course.
No, it means that Nvidia doesn't want to make Gsync h/w anymore as they don't see enough profits on that.
They could have stopped far sooner than this if that was the case. They have something else in the pipeline.
They kinda did. The Gsync h/w hasn't been updated in years, it's still using HDMI 2.0 inputs. Moving from that to just licensing the IP is a logical step when the h/w itself isn't interesting to produce anymore. It's the same with their mobile SoCs - they don't have anything in the pipeline, they just license that out to a manufacturer who is willing to make use of the tech in their h/w.
maybe it´s what the colab was all about ....
Yeah, though the Mediatek scaler chip can be used in a variety of devices (auto, mobile, pc, display, tablet, industrial, console). Licensing the IP like this allows a footprint into more otherwise restricted devices. Don't know what the future of console will be though it should provide the option to integrate different graphical capabilities via the MediaTek scaler chip.
From the link above:
Last edited:
MTK wanted to make a SoC for tablets/laptops with Nvidia's GPU in it, that's still in the oven I guess.
DavidGraham
Veteran
The NVIDIA/MediaTek collaboration has resulted in two things so far:
-An automotive chip with RTX graphics (Ada based) as it supports ray tracing + DLSS frame generation.
-A display scaler with integrated G-Sync capabilities.
Rumors point to a third thing, a laptop SoC with ARM + RTX graphics.
-An automotive chip with RTX graphics (Ada based) as it supports ray tracing + DLSS frame generation.
-A display scaler with integrated G-Sync capabilities.
Rumors point to a third thing, a laptop SoC with ARM + RTX graphics.
Similar threads
