http://www.digitimes.com/news/a20160408PD205.html
>Nvidia is ready to announce its Maxwell-based Pascal graphics cards at Computex 2016 from May 31-June 4, with graphics card players including Asustek Computer, Gigabyte Technology and Micro-Star International (MSI) showcasing their reference board products, according to sources from graphics card players.
IT'S HAPPENING
>Maxwell-based Pascal graphic cards
Lol what
>>53929476
>maxwell based
LUL Kappa KappaPride Kreygasm EleGiggle WutFace desu senpai
>>53929500
Looks like Digitimes derped there, but it should be Pascal GPUs and Computex is the best time to launch consumer products
>>53929476
>Maxwell-based Pascal graphic cards
>>53929518
>implying Pascal isn't just Maxwell with DP64 units, some cache tweaking, and HBM/GDDR5X capability
>>53929541
NOPE.AVI
Pascal has Compute Preemption, Maxwell does not
https://devblogs.nvidia.com/parallelforall/inside-pascal/
>GP100 extends GPU addressing capabilities to enable 49-bit virtual addressing. This is large enough to cover the 48-bit virtual address spaces of modern CPUs, as well as the GPU’s own memory. Therefore, GP100 Unified Memory allows programs to access the full address spaces of all CPUs and GPUs in the system as a single virtual address space, unlimited by the physical memory size of any one processor.
Pascal has 49bit virtual addressing, Maxwell does not
Pascal has a lot of features that is not in Maxwell, but keep telling yourself the lies if it makes you feel better
>>53929588
OK, so Pascal can see all memory in a system if it were so coded, hadn't AMD had this with their HSA implementations for some years now?
In any case I'm not seeing evidence of hardware based task dispatching which is fundamentally any different than Maxwell, and as well your chosen quote has nothing to do with compute preemption.
Nvidia are pushing for a radical departure from their current design, and they are also pushing for the largest possible chip from the get go. Everyone sane has some worry on their mind, with thoughts of Fermi creeping back.
>>53929588
Still no true asynchronous compute tho... it's all still context switching between doing compute work and graphics work, it's just going to be a little better at guessing when to do it.
Virtual addressing isn't going to change much if anything either.
They had to spend a LOT of transistors just to add back in proper FP64 capabilities and the wattage for the card has increased vastly compared to last generation.
I'm smelling another Fermi on hand, if not worse since now AMD has a good architecture that can do everything Nvidia's doing already but with a smaller footprint, smaller transistor cost, and smaller power draw.
>>53929720
GCN is a nothing but a HOUSEFIRE
>>53929731
max consumption is a meaningless metric, average over time is what matters.
max consumption happens in very short microsecond-scale spikes. roughly the same amount of times, the GPU also is drawing it's minimum.
max is absolutely meaningless.
>>53929757
>Maxwell can overclock to hell
>Fiji can undervolt like mad
What do
AMD is finally dead
>>53930083
>m-muh 5FPS
>m-muh 50 watts
They're for retards and poorfags, respectively
>http://www.digitimes.com/news/a20160408PD205.html
>Meanwhile, AMD has prepared Polaris-based GPUs to compete against Nvidia's Pascal; however, the GPUs will be released later than Nvidia's Pascal and therefore the graphics card players' third-quarter performance will mainly be driven by demand for their Nvidia products.
AMD, why????
>>53930658
There is no launch date for any Pascal or Polaris consumer cards yet.
Only rumors of a June/July launch.
>>53929731
>what is Polaris
>>53930102
quite the opposite
