$NVIDIA (NVDA.US)$ People don't understand that the testing ...
$NVIDIA (NVDA.US)$People don't understand that the testing speed of a processsor doesnt mean jack, if your codes can't compile against the processor. Most programs that's been invented by the human race still run on Intel. If you want to do the same thing on AMD, it not just click a button lmao. You will have to have the compilers for it. And then you will have to have so much shit that works together such as InfiniBand for parallel processing. Some of these stuff doesn't always work easily with non-Intel products. Same with GPU. You can't just click a button and pray your codes compile lmao.
Disclaimer: Community is offered by Moomoo Technologies Inc. and is for educational purposes only.
Read more
SharesGrow com
:
Well said, NVDA invented CUDA for intensive AI learning model, and, the moat is, only NVDA supports CUDA code compilation. That is why NVDA can ask for whatever price they demand from MSFT, META, etc, who have no choices but to buy and to take leads in this AI competition.
151654992
OPSharesGrow com
:
In addition to that, some of the mathematical models run slower on GPU, simply because of the inter-node/processsor communication lag. GPU seems to be very good with AI, and AI is only good for graphic and text processing, a very small portion of work we have to do.
151654992
OPETONA
:
For now, it is a small thing in turns of portion in engineering, but it has the largest user base. The rest can't catch up with AI in this regard.
ETONA
151654992
OP
:
yes mobile phone, note book, all things using chips will need AI but still quiet far waiting for the right price to hold super long for my kids later haha
🎙️Discussion: 1. How will tariff policies affect the movement of key assets such as U.S. stocks, gold, and Bitcoin? 2. Given this context, Show More
Moo Live
Jan 23 16:54
MicroStrategy Q4 2024 earnings conference call
Reassessing Chinese Assets
Following the introduction of China's groundbreaking DeepSeek technology, Wall Street giants have revised their investment outlooks for the Chinese market.
103970650 : then layoff programmer. ask AI
151654992 OP 103970650 : The tasks AI can do is actually a very small portion of all the tasks we have at hand. AI can't manufacture a new compiler.
SharesGrow com : Well said, NVDA invented CUDA for intensive AI learning model, and, the moat is, only NVDA supports CUDA code compilation. That is why NVDA can ask for whatever price they demand from MSFT, META, etc, who have no choices but to buy and to take leads in this AI competition.
151654992 OP SharesGrow com : In addition to that, some of the mathematical models run slower on GPU, simply because of the inter-node/processsor communication lag. GPU seems to be very good with AI, and AI is only good for graphic and text processing, a very small portion of work we have to do.
ETONA 151654992 OP : that is why AI is just small thing now
far from what it should be
5 years ahead may see the change
151654992 OP ETONA : For now, it is a small thing in turns of portion in engineering, but it has the largest user base. The rest can't catch up with AI in this regard.
ETONA 151654992 OP : yes
mobile phone, note book, all things using chips will need AI
but still quiet far
waiting for the right price to hold super long for my kids later
haha