• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

1st Fermi review

Pendu said:
Apple is apart of the OpenCL consortium, Apple have been buying up all the 48XX stock and sticking them in their PCs. Doesn't take a genius to realise that OpenCL is going to become a big deal. I bet Intel will also adopt OpenCL on their Paralell compute device "Larabee"
Click to expand...

Larrabbe will support ANY api throuh a wrapper, that's the whole design of it (and it will add support for any future revisions too without hardware upgrade). But that's primarily for graphics API's, why on earth would you want to run OpenCL on an x86 architecture that natively supports C, C++ etc..? You would lose perfromance.

I forgot to mention that Fermi actually supports native C++ btw, so they are already making moves towards a Larrabee like solution.

edit: also PC's out number Macs by 10:1 or so, they are not irrelevant but they certainly are not a driving force in API adoption.
 
Last edited:
Intel will probably can it completely as a GPU device, their approach to GPU processing has not got a hope in hell of competing, but keep the ard as a parallel compute device.
 
p0rks0da said:
Larrabbe will support ANY api throuh a wrapper, that's the whole design of it (and it will add support for any future revisions too without hardware upgrade). But that's primarily for graphics API's, why on earth would you want to run OpenCL on an x86 architecture that natively supports C, C++ etc..? You would lose perfromance.

I forgot to mention that Fermi actually supports native C++ btw, so they are already making moves towards a Larrabee like solution.
Click to expand...

Therefore CUDA is still pointless, perhaps in time OpenCL will become also, if C++ is also supported in ATi's next gen cards. AMD has OpenCL drivers setup with their CPU Architecture, you can decide what to send to the CPU or GPU to be processed.

Without having to deal with data moving from C++ to OpenCL.
 
Last edited:
Orangey said:
They ARE using the latest Nvidia drivers.
They're NOT using the latest ATI drivers.

But there's a million things wrong with this review/propaganda piece other than that.
Click to expand...

You missed my point. Drivers never come out at the same time. You'll always get fans of both camps defending their card being behind with 'it'l be faster with the next driver release' argument.
 
Pendu said:
Therefore CUDA is still pointless, perhaps in time OpenCL will become also, if C++ is also supported in ATi's next gen cards.
Click to expand...

Adobe is using CUDA in it's CS5 suite to net major gains in certain programs. It's far from pointless. If you want that performance now you have to use the CUDA solution. We can chalk up many things as pointless if we're comparing them with a superior future non-existant product.
 
Adrianr said:
Adobe is using CUDA in it's CS5 suite to net major gains in certain programs. It's far from pointless. If you want that performance now you have to use the CUDA solution. We can chalk up many things as pointless if we're comparing them with a superior future non-existant product.
Click to expand...

Pointless from the point of view that all this investment into CUDA and nVidia go and put C++ into Fermi. C++ existed when CUDA was being thougnt up.
 
Pendu said:
Therefore CUDA is still pointless, perhaps in time OpenCL will become also, if C++ is also supported in ATi's next gen cards. AMD has OpenCL drivers setup with their CPU Architecture, you can decide what to send to the CPu or GPU to be processed.
Click to expand...

Well it's not pointless because it's being widely used by people today, that's the point. Everything has the possibility of being the next big thing, but we just have to wait and see what actaully happens.

Still hoping Intel make a Larrabee announcement this year though, supposedly they will do.
 
p0rks0da said:
Well it's not pointless because it's being widely used by people today, that's the point. Everything has the possibility of being the next big thing, but we just have to wait and see what actaully happens.

Still hoping Intel make a Larrabee announcement this year though, supposedly they will do.
Click to expand...

It will take a constant leap in manufacturing node sizes for Intel to make Larrabee competative with nVidia or ATi. Using x86 cores is just not efficient.
 
Pendu said:
It will take a constant leap in manufacturing node sizes for Intel to make Larrabee competative with nVidia or ATi. Using x86 cores is just not efficient.
Click to expand...

No offence, but i think i'll wait and see. Intel have some moderately clever people working for them :)
 
You must log in or register to reply here.
Back
Top Bottom