PC Perspective Podcast #409 – 07/21/2016
Join us this week as we discuss the GTX 1060 review, controversy surrounding the async compute of 3DMark Time Spy and more!!
You can subscribe to us through iTunes and you can still access it directly through the RSS page HERE.
The URL for the podcast is: https://pcper.com/podcast – Share with your friends!
- iTunes – Subscribe to the podcast directly through the iTunes Store (audio only)
- Google Play – Subscribe to our audio podcast directly through Google Play!
- RSS – Subscribe through your regular RSS reader (audio only)
- MP3 – Direct download link to the MP3 file
This episode of the PC Perspective Podcast is sponsored by Casper!
Hosts: Ryan Shrout, Allyn Malventano, Jeremy Hellstrom, and Josh Walrath
Program length: 1:34:57
-
Week in Review:
-
0:51:17 This episode of the PC Perspective Podcast is sponsored by Casper!
-
News items of interest:
-
1:14:25 Honey, I Shrunk The NES
-
1:26:26 Hardware/Software Picks of the Week
-
Ryan: Sapphire Nitro Bot
-
-
Closing/outro
Everyone still going NUTS
Everyone still going NUTS over the async.
“TimeSpy is DISCREDITED for not taking full advantage of hardware features like a real developer would!!!”
“Since now async will make rx480 instantly faster than gtx1080SLI in dual precision multiparralellism async fully hardware compute enabled titles, which will be all of them, how can you all sleep at night?”
(On pillows full of shredded Nvidia-shill cash baby!)
p.s. send me some!
Thanks for the quality programming, homies.
The Red is back in the Black,
The Red is back in the Black, AMD survives too live on when so many were saying otherwise! Zen is on its way, but where is AM4 and any Bristol Ridge desktop SKUs until Zen gets here in large enough numbers from the retail channels.
“Official: AMD now stands for Avoiding Miserable Death”
http://www.theregister.co.uk/2016/07/21/amd_q2_2016_profitable/
3DMark (not so much) “Async”
3DMark (not so much) “Async” compute.
http://www.overclock-and-game.com/news/pc-gaming/50-analyzing-futuremark-time-spy-fiasco
But, yeah, they do benchmarks all their lives, so whatever they say must be the Bible
Nvidia and its friends are
Nvidia and its friends are really pushing the DX11 benchmarking as a diversion from what will be coming to AMD’s Polaris, and older GCN generations once the games and gaming engines are optimized for Vulkan/DX12. Nvidia is having to clock its GPU much higher to get the same relative performance on Vulkan/DX12 optimized games. Just look at the older AMD GCN generation GPU’s boost from Vulkan/DX12 and see that hardware based async-compute is coming into its own. And Polaris even has instruction pre-fetch added to the GPU processor’s hardware to get even more execution resources utilization via of its async shaders, and ACE units and Hardware Schedulers. Nvidia has taken its GPUs and relied on higher clocks to attempt to get its performance metrics up as a way to compensate for its lack of fully in the GPU’s hardware async-compute features. Polaris even has a primitive discard accelerator in its GPU hardware to cull out any unnecessary geometry from wasting GPU resources on AMD Polaris based GPU SKUs.
With all this extra in hardware features AMD’s GPU are going to run a little hotter, and use more power, but that’s only because AMD’s is making more use of its extra execution resources compared to Nvidia GPUs. The games/gaming engine makers will be continuously tweaking their products to get at every last little bit of AMD’s Polaris/GCN in hardware features for games performance just watch the benchmarks for AMDs GCN GPUs improve over the next year. Trust no reviews that declare a definite winner this early in the game with so much change happening in the Graphics APIs, GPU hardware and new 14nm/16nm process node development. That 14nm Samsung/(GF licensed) process node will see improvements in preventing leakage and power usage as it gets more mature, so expect that things will continue to improve for AMD’s products fabricated at 14nm.
Here comes the
Here comes the Server/HPC/Workstation APUs(1) on an Interposer! Expect that there will be consumer versions derived from the Server/HPC/Workstation APUs on an Interposer designs. It’s only a simple step for AMD who already has GPU SKUs on an interposer with HBM to add a Zen cores die to the interposer along with the GPU Die and HBM2 Die stacks.
Its more economical to fab the CPU cores die, and the GPU(Vega/Greenland) die separately and wire them up via a silicon interposer to each other and to the HBM2 die stacks, so it will not be a single monolithic die but a separate dies based APU integrated on a silicon interposer, and AMD is already 2/3rds of the way there with its GPU/HBM on an interposer designs, so adding that Zen cores die on the interposer along with the GPU die, and HBM2 die stacks is the next logical step in the evolutionary process for AMD.
(1)
“AMD mulls a CPU+GPU super-chip in a server reboot”
http://www.pcworld.com/article/3099190/hardware/amd-mulls-a-cpugpu-super-chip-in-a-server-reboot.html