BRIGHT SIDE OF NEWS About | Advertise | Contact BSN USER Login
| Register 		SUBSCRIBE Newsletter | RSS Feeds
Saturday, May 25, 2013
Email this to a friend.
Your friend's e-mail:
Your Name:
Your e-mail:
Message subject:
Comments
by: Anonymous on 1/9/2010
The US should go to war with Taiwan and occupy all those small companies and give the to Intel and Nvidia.

Just a joke.
by: Anonymous on 1/5/2010
Good luck with hearing from "representatives of said companies"! You sound about 10, blowing your own trumpet etc etc! Yes, you are absolutely wonderful.
by: Anonymous on 1/3/2010
Well, nVidia did lock people with AMD and nVidia cards out of hardware-accelerated PhysX.
The acceleration only works when all GPUs in the system are nVidia.

But that was after AMD declined to support Cuda/PhysX, even though nVidia was open to it.
AMD wanted to go their own way, first through their proprietary Stream, and now through OpenCL.
So far, nothing much has materialized.

I hope Fermi will be a success, and give GPGPU a push into the mainstream. That might finally get AMD to realize that they really DO need a good software stack for OpenCL and physics acceleration (among other things).
Huh? by: Anonymous on 1/3/2010
Hold on there - I thought nvidia offered both cuda and physx to amd? But amd refused saying "they would make it run slower on amd hw", or something silly.

So physx and cuda are closed because of amd, not nvidia - nvidia just wants to sell its hw, and that's why they locked out amd hw so that people wouldn't use high-end amd hw for graphics and consumer-level nvidia hw for physx, which is kind of weak, but they did not lock amd out of physx or cuda - on the contrary, amd didn't want in.

As for opencl - it's basically a copy of cuda.

And for fermi - I hope it does not fall on its face - why would anyone? If it does, then hw progress will slow down, amd prices go up and we are left with worse hw with a higher price tag on our hands.

And of course, the benchmarks should be compiled specifically with the best compiler for each hw. Benchmarking is such a difficult subject and there are a lot of bad benchmarks out there.
RE: CUDA / OpenCL by: Theo Valich on 1/2/2010
Ask Apple why are they supporting OpenCL.

OpenCL was first demonstrated to Apple by nVidia using nVidia hardware, after which the contract was sign between the two companies.

It is no secret who is the largest investor in Khronos group and who the chairman is.

But yeah, the way how nvidia locked everybody else out with PhysX in general sucks.

Ed.
by: Anonymous on 1/2/2010
"push their OpenCL API to go against CUDA"

Eh, 'their' OpenCL API? OpenCL was Apple's idea, and is governed by Khronos.
nVidia is the one who supports OpenCL in public drivers, AMD doesn't yet.
by: Anonymous on 1/2/2010
I don't think the problem is with Intel and its compiler, but rather with benchmark developers who choose the Intel compiler rather than a more neutral one. Unless ofcourse the purpose of the benchmark never was to evaluate the performance between CPUs of different vendors in the first place, then the problem is with whoever misinterpreted and misused the benchmark.
I mean, it's perfectly valid to create a benchmark to measure how different Intel systems scale (eg varying memory configurations, number of cores/CPUs, clockspeed etc). In that case I don't see a problem with using the Intel compiler for a benchmark.
Licenses galore by: Anonymous on 1/2/2010
What I hope to see is more open x86 licensing, I'd like to see more x86 companies.

I would also like to see nVidia open up CUDA and Physx as well as change their TWIMTBP so it doesn't lock out graphic settings on non-nVidia GPUs.

But yeah, I am dreaming. The reality is that many companies in IT are just patent trolls, money hordes, and douchebags who can't seem to open their platforms for more competition to benefit the consumer.

I hope Fermi falls on its face, Intel get's their asses handed by the FTC, and AMD to actually get their CPUs on the high-end again, push their OpenCL API to go against CUDA, and get some open physics platform out, and increase their relationship with game developers.
by: Anti-DRMintosh on 1/2/2010
Scumvidia and VermIntel, 2 ****sucking monopolist piece of $hit a$$holes.
Compilers by: Anonymous on 1/2/2010
Compilers by definition optimize code for the HW the code is running on. Of course an Intel compiler will make SW run better on Intel HW--thats the whole point! I don't get what the big deal is.
by: Anonymous on 1/2/2010

So benchmarks have been favourable towards Intel because people use Intel´s software to run them... I´m i the only one that sees what´s wrong here? Shouldn´t people use a software that isn´t provided by one of the companies being evaluated? Did someone expect Intel or any other company in the same situation to act differently???
© 2009 - 2011 Bright Side Of News*, All rights reserved.
© 2009 - 2013 Bright Side Of News*, All rights reserved.