Actually they don't need to explain the name since Enrico Fermi actually developed the 1st nuclear reactor :S thats ... [en.wikipedia.org]
there's the little poodle for ya
alright, to all you dorks that keep calling matt out and seem to be new to the interwebs -- look at the damn article title before *gasp* saying matt doesn't know who fermi is. it's not just a coincidence. friggin' 'tards.
GT300Runs:
DX11, DirectCompute, OpenCL, OpenGL, CUDA, C code, C++ code and FORTRAN code. It can basically do what your CPU does now, but a whole lot freaking better. It won't be hard at all for someone to port their windows applications to run on Nvidia's GPU's now.
@Kakkoii: "It can basically do what your CPU does now, but a whole lot freaking better. It won't be hard at all for someone to port their windows applications to run on Nvidia's GPU's now"
I hate to poop your party, but GPUs are not just super awesome CPUs.
@hfutrell: And I love to poop on your party, because I never said they were. I said this new one is. Nvidia has configured their cores to now finally do C and C++ operations without need for rewriting your whole code. Read the damn article.
@Kakkoii: "It can basically do what your CPU does now, but a whole lot freaking better" Read your own comment instead.
I can name you quite a few things your CPU can do better than Fermi will: branching, recursion, or any task that can't be parallelized into thousands of threads. The fact that Fermi can run C++ code does not imply it will do it better than the CPU for a given task. #nvidiafermi
i can finally rub everything about the new nvidia gpu specs to my friend's face!he kept blubing about how ati is better cause he has an ati card and that they have gddr5 and it's double,triple,quadzillion pumped!hell yeah!
I've worked with an "open standard" committee before and like to contribute here.
I think there is A LOT of confusion due to the complexity of the issues.
First, there is some confusion between "open standard" and "proprietary" - the two concepts are not mutually exclusive, but are often used interchangeably carelessly. In many cases, that is ok, but in other cases, it just adds to misunderstanding.
Proprietary simply refers to who owns it - and most standards are owned by somebody - a company, an alliance, a non-profit group. For example, "Microsoft proprietary standard" simply means Microsoft owns it. It could still be an open standard.
An open standard is one that is accessible by a other entities, usually for little or no fee, but usually still requires some license agreement to acknowledge ownership, define "compliance" with the standard, and how licensees can claim support of the standard.
The structure of fees is usually hidden from consumers - the manufacturers pay - whether a flat fee or a per unit fee - so just because it's open doesn't mean that its free.
For many standards "developed" by non-profit organizations (Bluetooth, for example) - there are actually companies who participate in supporting the non-profit, both by contributing funds as well as engineering resources. (Standards don't just write themselves) Sometimes, companies license or transfer their technology (patents) to the non-profits as part of their contribution and to help speed along development. (After all, why keep reinventing the wheel?) But sometimes, committees staffed with people from competing companies have conflicts and need to work through them - sometimes resulting in compromises that no one is happy with.
Obviously, the companies that are interested in helping develop a standard have something to gain by it - by selling chips or connectors, or by being able to plan for incorporation of new standards into their products.
For companies like Apple, there are many factors to consider in whether to go proprietary or to adopt an open standard. Is the open standard a quicker way to market? (maybe, maybe not - some committees take a very long time to get consensus to ratify a new standard) Open standards may mean greater choice of suppliers - there are many USB controller chip suppliers, for example. Open standard components may cost less due to more competition among suppliers. An open standard may mean more potential partners. On the other hand, proprietary standard may mean more control, quicker to market and better fit for the company's goals, as well as competitive advantage.
There are strategic reasons to consider on top of all that. Microsoft didn't sell more Windows OS by developing IE - but it didn't want to be left behind and let potential competitors develop substantial revenue sources. Microsoft's consistent strategy has been to develop their own standards and products to compete with any market leader - witness IE, MSN, WMV, Zune, Bing. Apple, on the other hand, adopts open standards when it makes sense, and develops proprietary when available open standards don't exist or aren't a good fit with it's (Steve's) vision for new products.
Basically, my point is that, as much as we all want to distill complex situations down to the simplest terms - good/bad - the reality is almost never that simple. Life would be boring if it were, no?
Actually, Apple charged a license fee to use the term "FireWire". That's why there are 3 different names for it. FireWire cost 25 cents a device (it's free now). That's why PCs called it IEEE1394 (the standard behind it). Sony did their own thing and introduced the unpowered version, which they called i.Link.
And USB 3.0 can't beat Firewire in a few areas, notably instant networking (plug two computers together via one FireWire cable - instant network), or remote debugging (kernel debuggers often use FireWire or serial).
It won't die out - it's a better a/v spec than USB - even 3.0 won't beat it in that regard.
10/01/09
10/01/09
09/30/09
09/30/09
there's the little poodle for ya
09/30/09
09/30/09
09/30/09
09/30/09
GT300Runs:
DX11, DirectCompute, OpenCL, OpenGL, CUDA, C code, C++ code and FORTRAN code. It can basically do what your CPU does now, but a whole lot freaking better. It won't be hard at all for someone to port their windows applications to run on Nvidia's GPU's now.
09/30/09
I hate to poop your party, but GPUs are not just super awesome CPUs.
10/01/09
10/22/09
I can name you quite a few things your CPU can do better than Fermi will: branching, recursion, or any task that can't be parallelized into thousands of threads. The fact that Fermi can run C++ code does not imply it will do it better than the CPU for a given task. #nvidiafermi
09/30/09
09/30/09
"How many teraflops does it have?"
"....."
09/30/09
But if I had to guess, it's at least 4 Teraflops+
09/30/09
09/30/09
09/30/09
09/30/09
09/30/09
09/30/09
09/30/09
09/30/09
Now give me benchmarks.
09/30/09
12!
09/30/09
09/30/09
09/30/09
09/30/09
Vegeta, what does 3dMark06 say about it's power level?
IT"S OVER X90000!!!!!
09/30/09
"IT'S... NOT... POSSIBLE!!"
HEART CLICKYUP!
09/30/09
09/30/09
09/30/09
09/30/09
09/30/09
09/30/09
09/03/09
I think there is A LOT of confusion due to the complexity of the issues.
First, there is some confusion between "open standard" and "proprietary" - the two concepts are not mutually exclusive, but are often used interchangeably carelessly. In many cases, that is ok, but in other cases, it just adds to misunderstanding.
Proprietary simply refers to who owns it - and most standards are owned by somebody - a company, an alliance, a non-profit group. For example, "Microsoft proprietary standard" simply means Microsoft owns it. It could still be an open standard.
An open standard is one that is accessible by a other entities, usually for little or no fee, but usually still requires some license agreement to acknowledge ownership, define "compliance" with the standard, and how licensees can claim support of the standard.
The structure of fees is usually hidden from consumers - the manufacturers pay - whether a flat fee or a per unit fee - so just because it's open doesn't mean that its free.
For many standards "developed" by non-profit organizations (Bluetooth, for example) - there are actually companies who participate in supporting the non-profit, both by contributing funds as well as engineering resources. (Standards don't just write themselves) Sometimes, companies license or transfer their technology (patents) to the non-profits as part of their contribution and to help speed along development. (After all, why keep reinventing the wheel?) But sometimes, committees staffed with people from competing companies have conflicts and need to work through them - sometimes resulting in compromises that no one is happy with.
Obviously, the companies that are interested in helping develop a standard have something to gain by it - by selling chips or connectors, or by being able to plan for incorporation of new standards into their products.
For companies like Apple, there are many factors to consider in whether to go proprietary or to adopt an open standard. Is the open standard a quicker way to market? (maybe, maybe not - some committees take a very long time to get consensus to ratify a new standard) Open standards may mean greater choice of suppliers - there are many USB controller chip suppliers, for example. Open standard components may cost less due to more competition among suppliers. An open standard may mean more potential partners. On the other hand, proprietary standard may mean more control, quicker to market and better fit for the company's goals, as well as competitive advantage.
There are strategic reasons to consider on top of all that. Microsoft didn't sell more Windows OS by developing IE - but it didn't want to be left behind and let potential competitors develop substantial revenue sources. Microsoft's consistent strategy has been to develop their own standards and products to compete with any market leader - witness IE, MSN, WMV, Zune, Bing. Apple, on the other hand, adopts open standards when it makes sense, and develops proprietary when available open standards don't exist or aren't a good fit with it's (Steve's) vision for new products.
Basically, my point is that, as much as we all want to distill complex situations down to the simplest terms - good/bad - the reality is almost never that simple. Life would be boring if it were, no?
09/03/09
And USB 3.0 can't beat Firewire in a few areas, notably instant networking (plug two computers together via one FireWire cable - instant network), or remote debugging (kernel debuggers often use FireWire or serial).
It won't die out - it's a better a/v spec than USB - even 3.0 won't beat it in that regard.