NVIDIA supreme boss JHH announced at the GPU Technology Conference 2014 a new GTX Titan as well as Maxwell successor.
1 – GeForce GTX Titan Z
The GTX Titan Z is a dual-GPU card that packs two full GK110 GPUs (Kepler GPU arch.) totalizing non less than 5760 CUDA cores (2880 cores per GPU). This monster card comes with a gigantic 12GB of video memory (6GB per GPU and ready for 5K as they said!). Of course such a behemoth has a price: $3000…
2 – Pascal GPU
Pascal (from French mathematician Blaise Pascal) is Maxwell successor. In this news, we learnt that Volta was the post-Maxwell architecture, but it seems that Pascal is the new official name. One of the main feature of the Pascal architecture is the 3D memory or stacked DRAM that should provide terabyte bandwidth.
Update (2014.03.26): according to techreport, Volta is the successor of Pascal:
Turns out Volta remains on the roadmap, but it comes after Pascal and will evidently include more extensive changes to Nvidia’s core GPU architecture.
9 thoughts on “GTC 2014: NVIDIA Pascal GPU Architecture (Post-Maxwell) and the $3000 GeForce GTX Titan Z”
“Turns out Volta remains on the roadmap, but it comes after Pascal and will evidently include more extensive changes to Nvidia’s core GPU architecture.
Nvidia has inserted Pascal into its plans in order to take advantage of stacked memory and other innovations sooner. (I’m not sure we can say that Volta has been delayed, since the firm never pinned down that GPU’s projected release date.) That makes Pascal intriguing even though its SM will be based on a modified version of the one from Maxwell. Memory bandwidth has long been one of the primary constraints for GPU performance, and bringing DRAM onto the same substrate opens up the possibility of substantial performance gains.”
JeGX, can you please correct your post? Volta still coming in the future.
Jen-Hsun announcing Pascal.
I am sure that the next generation architecture will be called Pasteur or Babbage. In our CUDA applications, we use a lot of memory accesses with little math on them. I hope that 3D memory helps with apps that have the same all-memory-no-math problem.
Pretty lame show overall!
awfully overpriced ….
not so Beefy GFX …
a well overclocked gtx 690 (mine oced up to 1297 mhz) can reach up to 7 TFlop/s MAD FMA/SP 32-bit with only 3072 cuda cores …
8 Tflops/s for the titan Z(ero)
3000 $ ca sera sans moi ! je garde ma 690 🙂
2x Performance but 3x Price
Head -> Wall
can i use such a device efficiently for custom made softeare even without proper support in the drivers? isnt any 3dgeek or fp64geek going to drop a little review and a little example how to extract all the available horsepower from it on this website? hack the titz!
if I was 50cent I’d order dat swag decorated with diamonds and gold 😛
Rapacity hidden under the name of Progress !
In rest, fairy tales/many words…
Comments are closed.