http://wccftech.com/nvidia-pascal-gp100-gpu-compute-performance/
Here's a summary of current info:
What we know so far about Nvidia’s flagship Pascal GP100 GPU :
Pascal graphics architecture.
2x performance per watt estimated improvement over Maxwell.
To launch in 2016, purportedly the second half of the year.
DirectX 12 feature level 12_1 or higher.
Successor to the GM200 GPU found in the GTX Titan X and GTX 980 Ti.
Built on the 16nm FinFET manufacturing process from TSMC.
Allegedly has a total of 17 billion transistors, more than twice that of GM200.
Will feature four 4-Hi HBM2 stacks, for a total of 16GB of VRAM and 8-Hi stacks for up to 32GB for the professional compute SKUs.
Features a 4096-bit memory bus interface, same as AMD’s Fiji GPU power the Fury series.
Features NVLink (only compatible with next generation IBM PowerPC server processors)
Supports half precision FP16 compute at twice the rate of full precision FP32.
*The best estimate I have is October 2016, unfortunately availability, pricing, and what models will launch probably won't be known before September IMO. I'm not expecting to get the card I want until sometime in Q1 2017, though I refuse to preorder.
(note that AMD at 14nm doesn't necessarily mean it's better than NVidia at 16nm. It's quite complicated.)
Really not much info there, though the POWER USAGE is interesting. That means smaller, quieter cards (A Zen/Polaris APU should also make sense in laptops and cheaper desktops assuming no DDR4 bottleneck. Something with over 3X the performance of the PS4 maybe.. obviously not optimized the same but a big deal regardless). Also, unless buying the top-end models the value may actually make more sense with current GPU's. Having said that, we won't initially have games that support the new architecture fully so if similar to a GTX980Ti on average on launch we'll see games come later that do a lot better on Pacal.
There's minor things of course that we know will be a fact like:
- H.265 support
Basically there's not much to worry about apart from the release date an the pricing (relative value to current GPU's). I'm planning to buy a card with specs similar to:
a) 8GB VRAM
b) 1.25X or better vs GTX980Ti (current game average)
It's hard to say if the top-end cards will wait, but I suspect we'll see something similar to the past where the best card will take up to a YEAR (closer to 2018) for a few reasons:
1. Maturation of new process node required to increase yields
2. Binning and building up enough supply for a launch takes several months after reliability of node is good enough
I won't wait that long. One final point:
Things we won't know for quite a while:
1. How well dual-GPU will be supported in most games (ideally the new method of splitting work load for CURRENT frame is used)
2. How well Virtual Reality can be optimized for (requires 90FPS, at effective HUD resolution so it's pretty demanding). also:
a) will we get cards with chips on them SPECIFICALLY to optimize for VR (if so they may be only for dual-GPU cards)
b) what VR options will exist? (for example, a new card just came out for GTX980Ti that has HDMI on the other end to run short cable to front 5.25" addon tray so you can plug HUD into front of PC).
Long story short is addititional info is likely several MONTHS away.
Facebook
Twitter
Instagram