News for hackers. Stuff that matters.
Re:Inside Pascal: Nvidia 's Newest Computing Platform by pavlov on Thursday April 21, 2016 @02:04PM (#11544127)
Hasn't half precision (16-bit float) been in NVidia GPUs forever? I could swear it was available back in the very first shader-capable Geforce FX days already.
Re:Inside Pascal: Nvidia 's Newest Computing Platform by bsprings on Thursday April 21, 2016 @07:51PM (#11546155)
(Post author here.) Yes, FP16 has been supported in NVIDIA GPUs as a texture format "forever" -- since before it was incorporated into the IEEE 754 standard. Indeed what is new in GP100 is hardware ALU support (and note denormals are full speed, which is even more important for lower precision formats).
Re:Inside Pascal: Nvidia 's Newest Computing Platform by wyldfire on Thursday April 21, 2016 @02:07PM (#11544160)
IIRC what's new here is native support in the ALUs, etc. I think the older support was probably in software.
Re:Inside Pascal: Nvidia 's Newest Computing Platform by rsp1984 on Thursday April 21, 2016 @06:32PM (#11545809)
FP has always been 16- or 20-bit precision in the Tegra (mobile) chips up to Tegra 4.
Re:Inside Pascal: Nvidia 's Newest Computing Platform by pavlov on Thursday April 21, 2016 @03:17PM (#11544663)
Makes sense, thanks.