Nvidia shows signs in [2020]

Status
Not open for further replies.
Is he leading the buzzword division?

Based on the twitter messages it's research team looking into future. Maybe something that could become part of geforce now in distant future i.e. games in cloud using heavier hw instead of running regular pc games in cloud?
 
December 7, 2020
By applying a breakthrough neural network training technique to the popular NVIDIA StyleGAN2 model, NVIDIA researchers reimagined artwork based on fewer than 1,500 images from the Metropolitan Museum of Art. Using NVIDIA DGX systems to accelerate training, they generated new AI art inspired by the historical portraits.
...
The technique — called adaptive discriminator augmentation, or ADA — reduces the number of training images by 10-20x while still getting great results. The same method could someday have a significant impact in healthcare, for example by creating cancer histology images to help train other AI models.
NVIDIA Research Achieves AI Training Breakthrough | NVIDIA Blog
 
I’m guessing Nvidia will be sticking with Samsung for the foreseeable future. Things are just too tight at TSMC.

Taiwan Semiconductor Manufacturing Company (TSMC), one of the largest semiconductor manufacturers in the world, is reportedly ending its volume discounts. Now, every customer will pay full price for the wafer, without any exceptions. For now, it is unclear what drove that decision at TSMC's headquarters, but the only thing that we could think is that the demand is too high to keep up with the discounts and the margins are possibly lower.

https://www.techpowerup.com/276029/tsmc-ends-its-volume-discounts-for-the-biggest-customers-could-drive-product-prices-up
 
It has been reported in the Korean news media (via WCCFTech) that Jen-Hsun and Co. has just signed a second contract with Samsung to produce GPUs "worth hundreds of billions of won."

For reference, one billion won is about ninety million dollars, so we're not talking about a drop in the silicon ocean either.
...
And this is where Nvidia's pick of foundry partner could pay dividends.

Given the strain on TSMC's 7nm production capacity, with all of AMD's CPUs and GPUs, as well as a host of other customers' products built in its Taiwan fabs, going back to ask for more could be nigh-on impossible. But if Nvidia can simply throw more money at Samsung for more Ampere GPUs it's going to be in a far better position than AMD is with its Big Navi supply.
Nvidia just signed a multi-million dollar deal to create a whole load more RTX 30-series GPUs | PC Gamer
 
Graphcore Challenges Nvidia With In-House Benchmarks | EE Times
December 20, 2020
By publishing an array of in-house benchmark figures, British AI chip startup Graphcore has mounted a challenge against the market leader for AI acceleration in the data center, Nvidia. Graphcore is claiming significant performance advantages for its second-generation IPU versus state-of-the-art Nvidia GPUs. However, Graphcore has put systems of different sizes head-to-head, saying it has instead compared the Nvidia product that’s closest in price.
...
“The Graphcore numbers are misleading,” said Kevin Krewell, principal analyst at Tirias Research. “Many companies self-publish benchmark and performance data, but those should always be viewed sceptically. The use of performance per dollar is not a good measure for AI systems purchases because there are many other factors in the cost of ownership. Often, performance per rack space is a critical factor.”
...
The majority of Graphcore’s benchmarks compare the IPU-M2000, a system with four IPU-MK2 chips, against a single Nvidia A100 GPU. The company also compares its IPU-Pod64, a system with 64 chips, against one or two Nvidia DGX-A100 systems (8x or 16x A100 chips). The scale of the systems compared in Graphcore’s announcement seems inconsistent, but as with all performance benchmarks, the devil is in the details.
...
Graphcore has said previously that the IPU-M2000 has a recommended retail price of $32,450, though this does not include a CPU server also needed to run the system (Graphcore says this enables freedom of server choice). By comparison, the 8-GPU DGX-A100 starts at $199,000. An Nvidia A100-accelerated server with 4x A100 GPUs (Supermicro A+ Server 2124GQ-NART) including CPU starts in the region of $57,000.
...
Graphcore also announced last week that it will join MLCommons and plans to submit benchmark scores to MLPerf in 2021. The next round of MLPerf inference benchmarks scores will be published in the later part of Q1 2021, with the next round of training scores following in Q2.

“I am glad to see Graphcore join MLCommons and promise to publish further benchmarks in 2021,” said Tirias Research’s Kevin Krewell. “Those benchmark scores will have far more scrutiny by the community. I believe [Graphcore’s self-published] benchmarks will age poorly.”
 
Status
Not open for further replies.
Back
Top