Or they simply got it wrong: there was an AMCC/APM X-Gene demo at the Red Hat Summit.Semiaccurate very briefly had a story about AMD demoing Fedora running on ARM servers, but it was quickly taken down. I think they may have accidentally broken an embargo...
Looks like there is a broad based ecosystem in place, nice. Hmm and Red Hat is a member of the HSA foundation I believe, interesting. Well it looks like AMD won't be 'going it alone' after all, nice!HSA is almost certainly going to have a big impact.
![]()
Wow, NV pushing OpenCL - now that's a shocker! The specifically didn't implement OpenCL 1.2 on Kepler (which was annoying). I wonder what they'll do with Maxwell? Would love to see them implement v2.0. Off topic I know...
Are you sure about that? Do you have some reference about that? Because random Google search brings threads like this where a developer explicitly say that OpenCL support in NVIDIA drivers is poor as far as optimization goes. The lack of OpenCL sub-forum on devtalk is to be noted too.Nvidia is less vocal about its support for OpenCL, but it supports OpenCL quite good. Better drivers, better support, better everything. Their consumer cards also support OpenCL since long time ago.
I don't think they will miss OpenCL 2.0 in future hardware.
Are you sure about that? Do you have some reference about that? Because random Google search brings threads like this where a developer explicitly say that OpenCL support in NVIDIA drivers is poor as far as optimization goes. The lack of OpenCL sub-forum on devtalk is to be noted too.
	Care to share the link? I think you stumbled on the X2100 series, based on Jaguar cores.Well, I was I little hopeful about this, till I found AMD's spec page on Berlin and it TDP is listed as 11-22 watts!!! I'm sorry, I just don't get it. What is the size of the market for ~15W SoCs with this level of performance? On the plus side, it won't be bandwidth limited anymore :|
that like chicken and egg problemDon't you find a little strange that software companies, not OEMs are testing this type of servers? doesn't Calxeda ring a bell?
Well, I was I little hopeful about this, till I found AMD's spec page on Berlin and it TDP is listed as 11-22 watts!!! I'm sorry, I just don't get it. What is the size of the market for ~15W SoCs with this level of performance? On the plus side, it won't be bandwidth limited anymore :|
that like chicken and egg problem
but then amd already had hardware platfom from seamicro
what missing is the software to showcase it
Wow yeah just noticed that, 11 watts for nearly 1 TFLOPS is incredible.
AMD mentioned that the Berlin APU with a 35W TDP and 512 SPs would deliver 700 GFLOPS of compute performance
Care to share the link? I think you stumbled on the X2100 series, based on Jaguar cores.
AFAIK Berlin will use 4 Steamroller cores.
Berlin is SR based, not Excavator.It's worse than you think. Berlin is 65W, is the excavator APU.
That has been the future since the 60s. Whether it has become the past, or will do so soon, depends on what kind of use cases you're looking at.A few months ago I attended a keynote by the guy behind HSA who's now at nVidia (Norm Rubin), and while he couldn't talk specifically about how great HSA is because of his current affiliation, he did an excellent job of selling the notion that the future lies in ISA-independent languages, like Java and OpenCL. I'm personally aboard this particular hype train because of my own research, and how universal languages fit into it.
A few months ago I attended a keynote by the guy behind HSA who's now at nVidia (Norm Rubin), and while he couldn't talk specifically about how great HSA is because of his current affiliation, he did an excellent job of selling the notion that the future lies in ISA-independent languages, like Java and OpenCL. I'm personally aboard this particular hype train because of my own research, and how universal languages fit into it.
If you hate AMD, that's fine. I personally think it's silly to hate a company, but I won't stop you. AMD might even totally fail in this particular implementation and attempt at HSA, but I think the history books will show that something like HSA + high level languages are the way to go. The cool thing isn't how you can run code on a CPU or GPU (IMO GPUs are almost as lame as CPUs), but how this will maybe make it easy and practical to run our general purpose code on something way more powerful and efficient than either.
Sorry if there are mistakes, I wrote this out on my phone in a busy airport.
Java hasn't been the future since 1990. High-level languages have advanced immensely past Java. Hell, I could start listing off a bunch of really good languages/frameworks built directly on top of the Java VM that are superior to Java in programmer friendliness in any way you can think of. Clojure, Groovy/Grails, Scala... And that's just languages built on the JVM, much less newer languages like Python. As far as I'm concerned if they want speed they can improve the C (or port to C++) API. If they want programmer friendliness, use a popular and nice language like Python. Java is like x86 in that its primary value is the massive library of things built upon it
