LinkBack URL
About LinkBacksAfternoon.
Any idea when your going to be playing with these new puppies?
http://www.nvidia.com/object/fermi_architecture.html
Doh, Joe I spelt directhex wrong lol - can an admin do a sneaky edit
Afternoon.
Any idea when your going to be playing with these new puppies?
http://www.nvidia.com/object/fermi_architecture.html
Doh, Joe I spelt directhex wrong lol - can an admin do a sneaky edit
These could come in handy at present at work. Send a free sample and I am sure we will tell you if they are worth stocking or not
Ditto
I have to admit, this looks **** hot. Proper MIMD, innit.
Now, let's all get used to writing code without branches
PHP Code:$s = new signature();
$s->sarcasm()->intellect()->font('Courier New')->display();
I want a sample as well
I dunno what I would use them for at home. Maybe a little bit of folding @ home to help the hexus team jump up the tables, thats if folding can be accelerated by them.
Several for my underground base wouldn't go amiss, either.
World Domination is a demanding business.
The ECC stuff is very interesting as well.Originally Posted by Steve
I have to admit, this looks **** hot. Proper MIMD, innit.
Now, let's all get used to writing code without branches
Well it is and it isn't. Modern fabrication processes are technically capable of handling a LOT more transistors than we put down onto them (an order of magnitude I think). The problem is with us - we can't design and (most importantly) validate that much stuff and get a product out of the door in a competitive amount of time.
However, something like this is a lot of repeated cells so a lot easier to verify if you approach it in a sensible way. You still put a hell of a lot of effort verifying some of the higher level stuff such as the L2 sharing and rest of the busses, but I bet throwing in ECC wasn't that costly. Well, evidently it wasn't, otherwise I can't see why they'd do it!
PHP Code:$s = new signature();
$s->sarcasm()->intellect()->font('Courier New')->display();
It'd be nice to see how all this is going to translate into the next generation of graphics cards
____
(='.'=)
(")_(")
will they play ARMA2 above 30fps? is al I need to know
Will ANYTHING ever manage that?
Don't be silly, that's just not possible!will they play ARMA2 above 30fps? is al I need to know
Will ANYTHING ever manage that?
____
(='.'=)
(")_(")
.will they play ARMA2 above 30fps? is al I need to know
Will ANYTHING ever manage that?
Well ArmA 1 was only just about playable when I had an E6600/GTX260, and that was about 6 months back... Given ArmA 1 is what... 3 years old or more?
Looks odd. I have no doubt that we'll see something along these lines eventually - we do work with a professor from computational finance who's very cozy with NVIDIA, and got us our last four Teslas
Hmm.
ECC is something a lot of people bang on about wanting, so thats a tick i guess.
But for what it is we are doing, we still find its often the overhead of shifting the Market Data that is the killer, if you've got a correlation matrix style problem that requires 5GB+ of correlation data, the performance of the CUDA system we last tried it on (295 iirc) was just not as good as a high end more expensive CPU, but effecting the total system cost by a few % when you take into account an SSD RAID.
I'm sure they have a market for it, its just I can't see it in most Trade Finding or classic Risk engines in the financial world.
throw new ArgumentException (String, String, Exception)
monte carlo.
Is it going to make Crysis playable?
[
There are currently 1 users browsing this thread. (0 members and 1 guests)
Posting Permissions
Reply With Quote
Steam: (
