AMD Announces New Records Accelerating Ecosystems With Xilinx

Published by

Click here to post a comment for AMD Announces New Records Accelerating Ecosystems With Xilinx on our message forum
https://forums.guru3d.com/data/avatars/m/172/172560.jpg
impressive. useless.
https://forums.guru3d.com/data/avatars/m/243/243702.jpg
gx-x:

impressive. useless.
Ever heard of NSA?
https://forums.guru3d.com/data/avatars/m/172/172560.jpg
what use would they get out of this? Besides, this is per interface throughput, you can have a "million" connections that are 10x slower and get more done in the same time. Especially when you are a trillion dollar spy agency. edit: since you mentioned NSA, they would certainly benefit more from accuracy then from throughput don't you think? Some good AI is what they need, not senseless image overload 🙂
https://forums.guru3d.com/data/avatars/m/243/243702.jpg
gx-x:

what use would they get out of this? Besides, this is per interface throughput, you can have a "million" connections that are 10x slower and get more done in the same time. Especially when you are a trillion dollar spy agency. edit: since you mentioned NSA, they would certainly benefit more from accuracy then from throughput don't you think? Some good AI is what they need, not senseless image overload 🙂
What do you think is information provided by all those cameras on streets? Something Good AI has to process for face recognition or just dumb mass processing? Idea is to know who, where, when and with whom. After you track that, you can use some advanced AI for implications.
https://forums.guru3d.com/data/avatars/m/236/236838.jpg
I think it will be very useful to improve urban secure monitoring if you mix this recording and storage capability with Face Detection AI. Maybe something like "Person of Interest", with real-world technological limitations.
https://forums.guru3d.com/data/avatars/m/172/172560.jpg
@Fox2232 Exactly. They already have that. They don't need this. What they have is miles ahead of this.
https://forums.guru3d.com/data/avatars/m/243/243702.jpg
gx-x:

@Fox2232 Exactly. They already have that. They don't need this. What they have is miles ahead of this.
So, basically you are saying that this king of market is fulfilled and done? What they have is some unknown computational capacity within some space and thermal envelope. Only reason for exchanging is to get same computational capacity into smaller space and smaller thermal envelope, so you can add more as your needs grow. Apparently, capacity planning decides on ROI (based on time expected for given HW to serve).
https://forums.guru3d.com/data/avatars/m/172/172560.jpg
I haven't said that. What I am saying is that this is impressive amount of data processing using only a single link. I am implying that it is useless because it looks like it is oriented more towards small business / end users, that will never need this kind of stuff (no market). And as you well know, companies have server farms (market for this, one day, in the far far future) with more than one connection and thus don't need this link. Maybe in the future they will use this. But someone like NSA, CIA , FBI etc (not market) is NOT going to go to AMD and buy this tech. So again, useless.
https://forums.guru3d.com/data/avatars/m/243/243702.jpg
gx-x:

I haven't said that. What I am saying is that this is impressive amount of data processing using only a single link. I am implying that it is useless because it looks like it is oriented more towards small business / end users, that will never need this kind of stuff (no market). And as you well know, companies have server farms (market for this, one day, in the far far future) with more than one connection and thus don't need this link. Maybe in the future they will use this. But someone like NSA, CIA , FBI etc (not market) is NOT going to go to AMD and buy this tech. So again, useless.
Sometimes I feel bit slow... "The XPO200 3U PCIe Expansion System is designed to deliver outstanding flexibility and scale-out GPU computing, as well as highly cost effective VDI solutions, via AMD Radeon Instinct MI25 GPUs and EPYC CPUs."
data/avatar/default/avatar30.webp
Impressive.
https://forums.guru3d.com/data/avatars/m/271/271700.jpg
They should put this in AMD's next GPU and ask game developers to support it. AMD could charge $2000.00 and call it a high end GPU. Trump sent me a MSG on my phone today. I don't like him playing on my phone.