AMD are crowing about something which is late to the game and completely unproven. Right now all we have is a puff PR piece from AMD with a bunch of claims and nothing to back it up. In comparison the NV solution is a tested and proven solution available in production and hardened enough to by used by some very big clients. No amount of hand waving by somebody who literally cannot accept that AMD might be behind in anything (see also the increasingly desperate and reaching claims made each time AMD announce continued financial losses) will change that. AMDs solution is bringing nothing new in terms of function or form, they are playing semantics.
I guess you are going to go off on a bunch of increasingly tedious and incorrect rants yet again. Just like you did the other day when you raged about me daring to suggest NV had something similar to HSA functionality cooking away, well just take a read of this, seems highly respected technology sites agree with me and not the angry AMD troll.
http://arstechnica.co.uk/gadgets/20...-and-radeon-veteran-phil-rogers-joins-nvidia/ well fancy that. The Ars Technical hardware editor is saying exactly what I was. Where as the guy who wrote for some unnamed failed hardware blog spent ages writing increasingly mad walls of text raged otherwise. Do you want to play this game again? Really? As each time you end up wrong and either disappear from the thread in shame or attemp to bury the facts beneith a mountain of words that actually surprisingly little.
Whelp. That was far more than I wanted to say, but I'm really tired and annoyed that individuals like that are constantly allowed to derail threads.
So you post in a thread, are entirely wrong again, I asked you specifically by the way which parts of AMDs information was inaccurate which you claimed and I'm shocked to find you haven't responded.
If I leave a thread it's because I've proven beyond a doubt you were wrong. You posted nothing factual or useful in the HSA thread and I posted information from AMD and Nvidia that agreed with what I was saying. In fact 99% of people in the thread said you were wrong, when everyone knows you're wrong and your argument as to why you're right is basically just saying "i'm right because I say I'm right", people get bored and leave, that doesn't make you right.
You derail EVERY thread, you derailed the HSA thread by coming in and pointing out how Nvidia were doing the same things, you derailed this thread, you came in and straight away said AMD were lying. Called on it, with proof from Nvidia, AMD and other sources, you don't refute anything, you can't prove anything, you post no information to back you up. Nothing at all in that article 'agrees' with you or your view point.
If anything it says
which boasts its NVLink CPU and GPU interconnect, allowing for heterogeneous computing setups that are similar to what AMD's been pushing out for years.
which is directly saying Nvidia needs those things to match things AMD has been doing for years, and HSA is well beyond anything AMD has done previously. If you want to be literal arstechnica are saying Nvlink helps Nvidia achieve with the NEXT generation what AMD have been pushing out for multiple generations already, so it's saying Nvidia are miles behind and Nvidia brought him in to attempt to help to catch up. More to the point, it's inaccurate. Nothing about NVlink allows on die scale communication, level of latency or ability to send small bits of code for gpu acceleration. Nothing about NVLink enables anything like AMD has been pushing for years. Nvlink is there to improve access to gpus to system memory and decrease the bottleneck in multi gpu systems, it's inherently designed for systems where 99.999% of the load is on the gpu, it's also exclusive to SOME IBM chips and SOME IBM motherboards and has no chance at integration in the desktop market. Sure Nvidia might get some stupid companies to put some NVlink chip on the motherboard so you get 16x pci-e to the chip then the chip connects to 2-4 gpu slots with increased bandwidth between the slots, which isn't entirely pointless but is already done with a pci-e bridge chip. Minimal gains and absolutely no need for higher power links providing a level of bandwidth that isn't required.
You use irrelevant posts to attempt to prove your side of the argument when your links directly dispute what you are claiming.
YOu go into every AMD thread and say "this is tech Nvidia have, this is AMD way behind or AMD are lying here", every single one.
Once again, you accused AMD of lying with their information and you claimed Nvidia had 'this' for years, this thread is about
HARDWARE virtualisation, not software. You are the one making bold claims about Nvidia's capabilities.
