What's really happening is that Nvidia's top end GDDR5 cards have a 384-bit memory bus, so they can pack in at most 24 memory chips. AMD's top end GDDR5 card has a 512-bit memory bus, so it can have 32 memory chips. Whenever memory vendors are able to offer new chips of double the capacity of the old, both AMD and Nvidia can double the memory capacity of their cards--if they care to. Whoever has the widest bus can have the most capacity.
The number of (professional) applications that need decreases as the memory goes up. To date applications requiring more than 12Gb are few - Maya and the odd 3D CAD applications can use the extra. For most applications needing 12Gb or less however the Quadro card is faster, in some cases much faster. What has kept the FirePro "in the hunt" is its much lower price.
Doubling the Quadro's memory will - presumably - address those very few applications that "require / can use" more than 12Gb. And it won't matter if AMD double the memory to 64Gb the need "probably" isn't there.
Significantly the announcement indicated that there would be no increase in the price - no reduction was suggested either so AMD will still be able to compete on a price/performance basis.
The FirePros do compete against the Quadros. You just have to do your research more than with a standard video card since the results vary wildly application to application. It means the nVidia is better at doing somethings and the AMD better at others. There is no wins every time card in professional graphics. For example, a couple years ago the FirePro was about 20% faster than the Quadro in Maya. If you are a system purchaser for devices that run Maya, why would you spend 3x more for a product that is slower? At the same time, the Quadro was 30% faster in 3D Studio Max.
if you read the article, you will notice that those are NO gaming gfx cards
"Instead, the cards are specifically engineered for non-gaming purposes, many of which run much better on the processing structure found in a GPU than a traditional CPU. That could mean graphical work like CAD and rendering, but also extends to weather prediction, facial recognition, and Bitcoin mining. They’re also typically equipped with certified drivers that ensure complete compatibility with high-end rending and 3D modeling software suites."
"I'll never grow up, never grow up, never grow up! Not me!"
the next gaming cards have already been announced why report on a non gaming card? http://wccftech.com/nvidia-pascal-gpu-gtc-2015/ pascal cards will be mean and also have 24g mem, projected release 2nd quarter 2016
Comments
AMD had 32 GB on a card last July.
What's really happening is that Nvidia's top end GDDR5 cards have a 384-bit memory bus, so they can pack in at most 24 memory chips. AMD's top end GDDR5 card has a 512-bit memory bus, so it can have 32 memory chips. Whenever memory vendors are able to offer new chips of double the capacity of the old, both AMD and Nvidia can double the memory capacity of their cards--if they care to. Whoever has the widest bus can have the most capacity.
거북이는 목을 내밀 때 안 움직입니다
거북이는 목을 내밀 때 안 움직입니다
It will be 10 years before a game uses that much video memory.
Doubling the Quadro's memory will - presumably - address those very few applications that "require / can use" more than 12Gb. And it won't matter if AMD double the memory to 64Gb the need "probably" isn't there.
Significantly the announcement indicated that there would be no increase in the price - no reduction was suggested either so AMD will still be able to compete on a price/performance basis.
So NVidia will announce/show 28nm GDDR5 maxwell rebrand. For 5k$. mmmmmmmmmkay
Pascal? HBM? MIA?
For example, a couple years ago the FirePro was about 20% faster than the Quadro in Maya. If you are a system purchaser for devices that run Maya, why would you spend 3x more for a product that is slower? At the same time, the Quadro was 30% faster in 3D Studio Max.
Please do not respond to me, even if I ask you a question, its rhetorical.
Please do not respond to me
"Instead, the cards are specifically engineered for non-gaming purposes, many of which run much better on the processing structure found in a GPU than a traditional CPU. That could mean graphical work like CAD and rendering, but also extends to weather prediction, facial recognition, and Bitcoin mining. They’re also typically equipped with certified drivers that ensure complete compatibility with high-end rending and 3D modeling software suites."
"I'll never grow up, never grow up, never grow up! Not me!"