Nvidia products for AI

[FP91G]

At COMPUTEX 2023, NVIDIA announced NVIDIA DGX GH200, which marks another breakthrough in GPU-accelerated computing to power the most demanding giant AI workloads. In addition to describing critical aspects of the NVIDIA DGX GH200 architecture, this post discusses how NVIDIA Base Command enables rapid deployment, accelerates the onboarding of users, and simplifies system management.
https://developer.nvidia.com/blog/announcing-nvidia-dgx-gh200-first-100-terabyte-gpu-memory-system/

If you still dream of using your graphics cards in the field of AI, then Nvidia is already releasing more powerful and modern solutions.

[OgNasty]

Nvidia is no doubt the current leader in AI technology and if you’re looking at using a lot of computing power for your own AI cluster then you have no option but Nvidia products. I don’t think this early lead will last though. AMD and Intel will no doubt be pouring a ton of money into R&D to make their own AI focused products. Exciting times watching GPU use cases evolve.

[FP91G]

Nvidia is no doubt the current leader in AI technology and if you’re looking at using a lot of computing power for your own AI cluster then you have no option but Nvidia products. I don’t think this early lead will last though. AMD and Intel will no doubt be pouring a ton of money into R&D to make their own AI focused products. Exciting times watching GPU use cases evolve.

Intel currently has no competitive solutions for modern Nvidia products. Their processors are inferior in performance to new solutions from Nvidia. AMD has not yet presented competitive solutions either, and they still have not figured out the overheating problems of the new X3D processors.

[babymethyl2013]

Isn't most Ai written for CUDA? Therefore NVidia has a monopoly on it in the Ai scene. AMD would need to create their entire version of something CUDA like and then try to steal market share when NVidia is known for the currently hyped ChatGPT.

AMD seems really behind. They would almost have to create something better than CUDA.

[FP91G]

With no big customers named, AMD's AI chip challenge to Nvidia remains uphill fight

Advanced Micro Devices Inc (AMD.O) on Tuesday gave new details about an artificial intelligence chip that will challenge market leader Nvidia Corp (NVDA.O), but the company left out what Wall Street wanted to know - who plans to buy it.

Santa Clara, California-based AMD said the forthcoming chip, which will start trickling out in the third quarter followed by mass production beginning in the fourth quarter, will have 192 gigabytes of memory.

That could help tech companies get a handle on the spiraling cost of delivering services similar to ChatGPT, AMD Chief Executive Lisa Su told Reuters in an interview. She spoke following a keynote presentation in San Francisco during which Su showed an AI system on the MI300X chip writing a poem about the city.

Nvidia, whose shares have surged 170% so far this year, dominates the AI computing market with a market share of 80% to 95%, according to analysts.
https://www.reuters.com/technology/amd-likely-offer-details-ai-chip-challenge-nvidia-2023-06-13/

[joker_josue]

With no big customers named, AMD's AI chip challenge to Nvidia remains uphill fight

If they are not even able to rival in the gamers area, how will they be able to rival in the AI area?
They really had to do something extraordinarily far superior - which I doubt.

In the early 2000s, they started an excellent path to reach the gamer market. But it seems that in the last 6-7 years, they simply forgot to invest in this market, losing almost all of the quota they had achieved years before.

Therefore, unfortunately for consumers, I see AMD with some difficulty in recovering much of the market they had. And it will be more difficult to conquer new ones.


In the case of Intel, they are willing to invest a lot in this area, and maybe that's why they end up being at a better advantage than AMD to compete with Nvidia. Even because it has a very strong part of the processors, which is also necessary for this area of AI.

[FP91G]

With no big customers named, AMD's AI chip challenge to Nvidia remains uphill fight

If they are not even able to rival in the gamers area, how will they be able to rival in the AI area?
They really had to do something extraordinarily far superior - which I doubt.

In the early 2000s, they started an excellent path to reach the gamer market. But it seems that in the last 6-7 years, they simply forgot to invest in this market, losing almost all of the quota they had achieved years before.

Therefore, unfortunately for consumers, I see AMD with some difficulty in recovering much of the market they had. And it will be more difficult to conquer new ones.


In the case of Intel, they are willing to invest a lot in this area, and maybe that's why they end up being at a better advantage than AMD to compete with Nvidia. Even because it has a very strong part of the processors, which is also necessary for this area of AI.

If AMD is forced out of the gaming graphics market, Nvidia will become a monopolist and charge very high prices in that market. Intel can't compete with Nvidia either. AMD share 12% and Intel share 4% in the gaming graphics card market in Q1 2023.

https://www.ixbt.com/news/2023/06/10/intel-nvidia-amd.html

[joker_josue]

If AMD is forced out of the gaming graphics market, Nvidia will become a monopolist and charge very high prices in that market. Intel can't compete with Nvidia either. AMD share 12% and Intel share 4% in the gaming graphics card market in Q1 2023.

Then! I know.

That's why I say that AMD should invest more in the gamer market, than venture into a market that will have more difficulty growing - AI.
I say this, because if they haven't been able to grow in a gamer market, which theoretically will be easier. I think they will have a much more difficult time with AI, because it will be a much more demanding market.

In the case of Intel, they still have a long way to go, but as they are making a big investment in just over a year in the area, and being one of the largest processor companies, they manage to be better positioned to grow in AI than the AMD. It starts right away with the customers, being much easier for Intel to reach the big players, since it is already there with the processing and servers.

[FP91G]

If AMD is forced out of the gaming graphics market, Nvidia will become a monopolist and charge very high prices in that market. Intel can't compete with Nvidia either. AMD share 12% and Intel share 4% in the gaming graphics card market in Q1 2023.

Then! I know.

That's why I say that AMD should invest more in the gamer market, than venture into a market that will have more difficulty growing - AI.
I say this, because if they haven't been able to grow in a gamer market, which theoretically will be easier. I think they will have a much more difficult time with AI, because it will be a much more demanding market.

In the case of Intel, they still have a long way to go, but as they are making a big investment in just over a year in the area, and being one of the largest processor companies, they manage to be better positioned to grow in AI than the AMD. It starts right away with the customers, being much easier for Intel to reach the big players, since it is already there with the processing and servers.

Where did AMD spend all the profits from the sale of video cards for mining?
The X3D processors were unsuccessful. Intel is better and without problems with overheating.

[joker_josue]

Where did AMD spend all the profits from the sale of video cards for mining?
The X3D processors were unsuccessful. Intel is better and without problems with overheating.

Good question!
How is it that a company that was growing, did not know how to take advantage of the profits obtained to improve even more and be a real option in the graphics market. This cannot be understood.

That's why I say that at this moment I see an Intel much more capable of trying to compete with Nvidia than AMD.

[sxemini]

If AMD is forced out of the gaming graphics market, Nvidia will become a monopolist and charge very high prices in that market. Intel can't compete with Nvidia either. AMD share 12% and Intel share 4% in the gaming graphics card market in Q1 2023.

Then! I know.

That's why I say that AMD should invest more in the gamer market, than venture into a market that will have more difficulty growing - AI.
I say this, because if they haven't been able to grow in a gamer market, which theoretically will be easier. I think they will have a much more difficult time with AI, because it will be a much more demanding market.

In the case of Intel, they still have a long way to go, but as they are making a big investment in just over a year in the area, and being one of the largest processor companies, they manage to be better positioned to grow in AI than the AMD. It starts right away with the customers, being much easier for Intel to reach the big players, since it is already there with the processing and servers.

Where did AMD spend all the profits from the sale of video cards for mining?
The X3D processors were unsuccessful. Intel is better and without problems with overheating.

The overheating problem is solved, and it was not only AMD´s fault - The Mainboard Manufacturers do something wrong and the biggest issues has ASUS. I don´t see Intel in front in CPU market - both AMD and Intel do well jobs.

[joker_josue]

The overheating problem is solved, and it was not only AMD´s fault - The Mainboard Manufacturers do something wrong and the biggest issues has ASUS. I don´t see Intel in front in CPU market - both AMD and Intel do well jobs.

In terms of processors, AMD does a good job, but even so, it has lost ground to Intel.
Now, not even at the price level they can be competitive.

Check this review and see how Intel as a whole turns out to be better in almost every scenario:
https://www.tomshardware.com/features/amd-vs-intel-cpus

[JayDDee]

In terms of processors, AMD does a good job, but even so, it has lost ground to Intel.
Now, not even at the price level they can be competitive.

Check this review and see how Intel as a whole turns out to be better in almost every scenario:
https://www.tomshardware.com/features/amd-vs-intel-cpus

That's only for gaming where the CPU plays a minor part, and it includes price, so a powerful CPU doesn't necesarilly rate well.

Every decade or so AMD comes out with a winner CPU architecture like Ryzen that surpasses Intel for a while. The last time was AMD64 (x86_64) the first 64 bit desktop,
and previous to that was Athlon the first desktop CPU to hit 1GHz.

That pattern doesn't seem to be the case with GPUs.

[joker_josue]

That's only for gaming where the CPU plays a minor part, and it includes price, so a powerful CPU doesn't necesarilly rate well.

Yes, for gaming, AMD's processors at some point outperform Intel's - by a little while.
Now, for work, Intel's processors tend to be the best.

If you look at the table that this analysis presents, it makes it clear that Intel's chips are better for productivity (work).

[joker_josue]

Well... although theoretically Intel processors perform better. It looks like AMD's sales have been going really well:

CPU Retail Sales Week 24 2023 (mf)
ℹ️ Sales:
AMD: 2415 units sold, 74.77%, ASP: 274
Intel: 815, 25.23%, ASP: 330

Twitter: @TechEpiphany

[FP91G]

Chinese publication Jitwei revealed that ByteDance has already ordered around $1 billion worth of Nvidia GPUs in 2023 so far, which amounts to around 100,000 units split between Nvidia's A100 (ordered before the US government told Nvidia to stop selling its top-performing HPC cards to China, back in August 2022) and H800 cards - that last series number corresponding to a Hopper-based custom accelerator Nvidia built to comply with export restrictions — a nerfed cameo of the H100 accelerator. Nerfed or not, Nvidia's Average Sale Price (ASP) seems to sit at around $10,000 per accelerator.
https://www.tomshardware.com/news/chinas-bytedance-has-gobbled-up-dollar1-billion-of-nvidia-gpus-for-ai-this-year

ByteDance (TikTok) will make money on AI by selling video cards 2 times more expensive, because in China the prices for such models have already increased