Pascal when?
Images are sometimes not shown due to bandwidth/network limitations. Refreshing the page usually helps.
You are currently reading a thread in /g/ - Technology
You are currently reading a thread in /g/ - Technology
2016 or 2017?
Will it even matter for the average consumer?
>>51275204
but
>muh energy consumption
>muh vram
>muh floating point
>>51275186
2018
>>51275186
cant wait for the massive disappointement once everyone realize they wont use HBM for the very first pascal cards.
>>51275236
No way will it be pushed back that far. I'm guessing late 2016/early 2017.
>>51275186
>when
Q2 2016 is official word on when mass-production starts. Expect 1080 and lower-numbered cards.
Titan likely won't be out till a few months afterwards with the Ti being at least 3 months after the Titan just like this gen.
>will it matter
Dunno, the Fury has shown us the HBM is a scam, so all that's left is to bet on generational improvements to the architecture.
CUDA users will probably see the biggest boost though, rather than gamers.
>>51275244
1080 maybe, 1060/70 probably not.
2016 obviously. what i'm worried about is if they release pascal so late into the year that the 1080 TI gets pushed into 2017.
>>51275299
Because that matters how?
>>51275248
>the Fury has shown us the HBM is a scam
HBM significantly improved the energy efficiency of Fiji compared to other GCN cards. How is that a scam?
If you mean in performance, bandwidth is not a scarce resource right now, so we won't immediately see any jumps in performance just yet. But don't tell me you expect us to keep 200~300 GB/s GDDR5 setups forever. The energy efficiency is a much more important aspect of HBM.
>>51275248
HBM isn't a scam, it uses less power and less physical space. This means that it can be used in more applications.
Not to mention devs can start utilizing that huge memory bandwidth.
What I'm hoping for, and it's far off for sure, is graphene circuits and batteries. Just graphene products in general.
We are going to see products skyrocketing in complexity, design, build quality and efficiency once they develop a way to get high quality pure graphene produced.
>>51275303
i don't feel like waiting ~6 months for the pascal TI card to release. the sooner the better.
>>51275327
>and less physical space
The day will come when every flagship GPU will fit into my Sugo case, instead of the retarded 30 cm bricks we have today.
>>51275248
>>51275262
Do you guys really think they'll call it the 1080, or is that just what you're calling it for now?
>>51275327
I like the way you think. Elaborate on that. Where in the industry do you see graphene solutions skyrocketing? In mobile computing and smartphones? Because that's what I feel.
>>51275365
>guys, I have a 9800 GT, should I get the 10800 GT next year?
They'll probably change their naming scheme again.
>>51275365
Obviously, yeah, but it's more than that. It only lets water through so it'll make amazing filters, it's also going to be used in manufacturing of screens, touch screens, likely also things like pans.
It is stretchy, it transfers electricity at ~92% efficiency, it can bend without breaking, it's basically a wonder material.
The difficulty is in manufacturing. It's hard to ensure purity, especially when trying to make more than a little bit.
The spot this is going to shine, though, is batteries. It's a super conductor and an amazing capacitor.
A few seconds of charge give you 10s of minutes of usage, so in theory your cell phone battery could get a drop in replacement that's the same physical dimensions, charges in a minute to full, and fully discharges (depending on device) after a week of heavy usage.
Electric cars would finally be a viable alternative, likely getting hundreds of kilometers out of much smaller and lighter batteries that charge in as little as an hour.
This will also allow for light, flexible and efficient solar panels. This, combined with the electric car, could allow for a car that can drive for many many kilometers before ever looking at a plug.
>>51275341
>Waaaah I don't want to wait
Grow up
>>51275406
>why do they call it the gtx 1080?
>because you approach it in the store, do a 1080, and buy it anyways
>>51275244
and still won't see a vast decline in performance for it.
http://www.gputechconf.com/
APRIL 4-7, 2016 SILICON VALLEY
PREPARE YOUR BODIES
>>51275456
I love it. So what's stopping us from using it tomorrow? Procurement? You make it sound like a godsend. What are the setbacks in your opinion?
>>51276222
I'm just hype for a gtx 1080M with HBM 2.0
Due to the high energy efficiency of HBM it seems only natural for nvidia to put it into laptops. I hope at least the top of the line gpu will get it. I am literally waiting and postponing a purchase because of the tremendous advantage an HBM card would have compared to today's gddr5 in laptops
Hope nvidia delivers and doesn't postpone the release.
>>51276751
With the efficiency of Pascal and HBM, I'm hoping that there will no longer have to be M versions at all. Nvidia have already showed some enthusiasm by putting GTX 980s in some laptops, and said how every generation the performance gap between the M and standard version is closing. It's down to very little now, Pascal could be the one.
>>51276312
I'm not >>51275456 but most likely its still in development + a huge facor: cost.
>>51275186
They better have it out by end of 2016, or AMD will blast them the fuck out with arctic islands
>>51276998
AMD will be lucky to stay in the race through price/performance.
>>51277007
>implying AMD doesn't already BTFO regarding price/performance
>>51277007
>>51277053
can AMD even survive until end of 2016? Shit is looking real dire for them. Not even just "lol Sony is gonna go bankrupt" type meme shit. They are legitimately in a red hot danger zone, I heard they only have 700 million cash reserves left.
>>51277063
you'd better hope so or your next GTX card will start at $3000
>>51277079
>>51277063
by the time amd goes bankrupt Intel IGPs will be good enough for medium/high settings gaming on new titles. Fuck nvidia. I mean intel are scum too, but they still have a lot of incentive to continue making their IGPs a lot better
>>51275358
They kept getting larger, and now they're getting smaller.
>>51277063
they've spent the last several years getting their shit together and making a long string of correct decisions (most of which we have yet to see, otherwise nobody would be having this conversation). all they must do is lift the lid to any investor that happens to pass by and dazzle them.
they'd obviously prefer to avoid that unless is becomes strict necessity.
>>51277007
all APIs are poised to swing heavily in AMD's favor. they banked hard on this happening much earlier, which is why we saw all the rebrand cards, but now it's finally happening.
>>51277216
>they've spent the last several years getting their shit together and making a long string of correct decisions (most of which we have yet to see, otherwise nobody would be having this conversation). all they must do is lift the lid to any investor that happens to pass by and dazzle them.
i think someone finally built a full HGS system as proof of concept.
if it all goes well it could be the next amd64
>>51277099
Intel is worse than nvidia though, at least amd cards work on gameworks, Intel literally paid OEMs to not use AMD, costing amd a shit tonne of money and paying back less than a fraction of it.
>>51277233
HSA, i'm mashing my damn keyboard now
>>51277247
HSA is fucking amazing.
>>51276312
The European union has a 10 year project to take graphene from laboratories to industry uasage. Better hope it works out well if you want them 100 GHz processors.
>>51275186
I still dont understand
Since Im plan to buy a mobo now, I dunno if its worth spended for a new one for Pascal due to NVLINK who "might" not be compatible with any current gen mobo
Any news if you need to change mobo for Pascal high end
>>51275893
Dude just walk straight after doing a 1080 and you'll be out of the store.
