The biggest challenge this is going to have is the bus width of 1024 Data pins. This isn't realizable in a DIMM type architectures we have today. And the number of balls you'd have in a BGA package which would be going on the Graphics cards is extremely high w/ very fine pitch balls, and routing of the data is going to be a nightmare.
I like the ideas, but there is added cost in multiple areas, and I don't see this becoming mainstream for PC memory for a long time, and still a few years out in the scale Hynix is talking about for a long time.
Instead the JEDEC spec has 128b HBM chips spec'd and these will be replacing GDDR5 over the coming years. It's still a routing nightmare and manufacturing of PCB for that fine of ball pitch isn't going to be cheap either. You'll see it in very high end Graphics Cards starting next year, but it becoming comonplace (like GDDR5 is today on graphics cards) is several years out, when that fine of pitch becomes more cost effective.
Results 101 to 119 of 119
Well given that, as of right now in Fall 2014 the PS4 and Xbox One successors are likely still 4 or 5 years down the road, what are the chances a reasonable HBM solution will be found for consoles that need to launch at a $399 price-point ?
Imagine a Grand Theft Auto 7 on PS5 with much smoother and higher fidelity graphics, like this CGI video:
The reason I say this, is because APUs in standard systems (i.e. laptops) are going to require a memory interface w/ replaceable modules a la DIMMs. And I do not believe we are ready for 512b and 1024b memory modules. A DIMM isn't going to handle this, it would need to have BGA interface which means a replaceable memory module are going to be in relatively large claimshell sockets that aren't going to be inexpensive. One might ask why not direct mount the memories. The reason is MOBOs need the memory to be replaceable and allow for wide range of configurations while Graphics cards don't.
Game consoles are a different story as they will be fixed memories. And as is the case w/ the XB1 and PS4 the APU has large DDR3 and GDDR5 wide data interfaces as compared to traditional APUs.
Will we see this on next Gen consoles in 5yrs+ (6yrs+ from the launch of this generation), I think it's highly probably, but not because HBM has become mainstream in PC design, but mainstream in GPU design (similar to what GDDR5 became mainstream in GPU design.)
I still don't think the 4-5yrs from launch of this generation (which is 3-4yrs from now) is realistic for the next generation to launch. Given the peak of innovation in techniques and optimization is several years into the generation a 4-5yr console generation is just too short for Devs to make the most of the generation.
Have we really had systems before where the majority of their processing power has been so flexible? Meaning the GPUs for both systems are highly programmable out side of just graphics work.
- Join Date
- Oct 2014
- Rep Power
The way this generation is going I almost wouldn't mind. I'm pretty disappointed with all the hype the games are receiving only to be let down by mediocre games and greedy practices.
I think that's just the new status quo of mainstream gaming. Lots of hype, lots of money pumped in bribing reviewers and building up an advertising campaign, lots of resulting sales... and the game is decently average.
Then gamers forget and get hyped for the next game.
kind of like destiny.
They start out with some seeming novel ideas with the improved power, but it takes them a 2-3 game cycles before they've figure out how to really get it done good, and what players really want, and turn this into a final product which usually starts coming together by year 3.
Before that, you got some novel ideas, and they Hype train starts running, but the first pass is going to have warts, and people are going to complain about it. Take a look at any game in the first year of generation, and compare it to games of year 3, and then compare that to games of the final year.
Cutting Game console lifes shorter isn't going to improve things, it's actually going to do the opposite as it resets developers back to the beginning of a learning curve before they really got a chance to perfect some of the stuff that they were working on (which will carry forward.)
what is definitely different this time though, and i expected this before and i think it's happening. i said that less people will buy next-gen consoles...i was wrong but what i did get right was that these consoles were going to have a difficult time trying to sell.
so how they fixed that is to start making games that work on last-gen and next-gen platforms...so while we have more games this way, we have less interest once people have the console. also i believe most devs are still using last-gen engines with some modifications.
when the PS2 came out, everyone had to rebuild their engines. even when the PS3 came out. there weren't any cross-gen games. if not most but at least a good amount of games had PS3/360 specific engines that wouldn't work on the PS2.
so now while they did figure out a way to stay sustainable. you have more games out there that don't really look all that next-gen. for the average joe i mean. they don't see a big difference. hardcores do. but we see a lot of things average people don't so makes sense.
i think cross-plat games will become a normal, look at what nintendo is doing next-gen. we already have like 10 different platforms where sometimes games can come out (including phones). it's already happening.
this is going to push the tech less overall but it will be a lot more sustainable for developers and have a broader market.
otherwise, i didn't think developers would've been happy moving next-gen. they barely had a decent amount of people on last-gen consoles. this precisely why so many supposed good games are still last-gen comptatible because it's still a viable source to make sales.
moving forward, their focus is going to be on simple machines that adapt ports better. we can't have alien tech anymore.
Keep in mind, we're just now coming up on the end of year 1. Year 2 we'll have a lot more games focused totally on this gen.
I think it surprised the whole industry how well this generation has sold this early. I think a lot of it has to to do with value for price.
Omar likes this post
the x86 architecture was definitely surprising but it was definitely the right move. i liked it from the reveal.
so it didn't occur to me that they could easily do ports. well, it's just amazing to me how many new indie games come out on the PS4...and they all play perfectly. how $#@!ing awesome is that? it's like buying a steam box RIGHT NOW LOL.
it feels like we couldn't pin point why consoles were good...we just had good games, they weren't the best, they didn't push the envelope but they were fun. now it seems console gaming is really taking off. now we have the same things we had on PCs...like chatting with friends while you're doing whatever you want to do...it's still not multi-tasking in the same level...at all but as far as gaming goes, it's pretty much there.
when i game on my PC, i don't do everything else, i just limit myself to browsing and chatting with friends. consoles got this down pretty well...maybe will never be as far as navigating with a kb/m but it's getting there.
but yeah, definitely the only way to move forward and i can't imagine what it will be next time. it's sad to see physical hardware go...and we may still have a generation left in there but it is slowly becoming a streaming industry...movies are ahead of us (because no processing power is needed i suppose?) and so one can guess it's going there...with PS Now and EA/Ubi opening up their own services.
do you think we could have streaming boxes with unlimited power? that would be incredible.
The Sony-Nvidia GPU partnership was made public in December of 2004:
For PC graphics, Nvidia launched the GeForce 7800 GTX (codenamed 'NV47' aka "G70") in June 2005. The RSX would be based on the NV47 / G70 GPU, but with half the number of ROPs (8 instead of the full 16) and a 128-bit bus to GDDR3 instead of the full 256-bit bus.
The concept of PS3 having two CELL processors was ditched sometime either in 2003 or by mid-late 2004 at the latest. Hense, the SCEI-Nvidia GPU partnership for PS3 announced in Dec 2004. Obviously Sony and Nvidia had been working together for awhile before the official announcement.
Now, by the time PlayStation 3 launched in late 2006 with CELL and RSX, Nvidia already had a whole new generation GPU for PCs on the market, the GTX 8800, codenamed G80. This GPU was well beyond PS3's RSX in features and performance.
In a nutshell, PS3 was released in late 2006 with RSX being a mid-range 2005 GPU.
Last edited by parallax scroll; 10-09-2014 at 22:17.
- Join Date
- Jan 2007
- Northern Ireland
- PSN ID
- Rep Power
Don't need this thread, too early to be talking about this.Trophy-licious!
If its anything like this gen you'll have to wait about a year or so for most of the good games to come out
And the wheel of time keeps rolling on as we (proverbial we) yet again repeat basically the same arguments as to why this generation is different and we should have a shorter time until next generation launches...
I argued against the comments supporting that we'd have a shorter generation of PS3-PS4 and I'm again arguing against the comments now. The same basic tents hold, and that's what game developers need in order to give us the best experiences possible. They are the heart and sole of the industry. Without time to grow and learn in a generation, we get less than stellar games as is often seen in games at the begging of a generation. Shortening the cycle shortens the cycle of Stellar games, and also cuts into profits as it's either more work spent on cross-gen games, or a smaller customer base in that next gen.
So even if we get a whizbang jump in technologies a bit earlier (and the costs get into console range for using it) if you F over the SW Devs and the Customer base by having shorter periods of Stellar games, and SW Devs not making as much money. The SW Devs die and the customers go away. If you don't believe MS and Sony are aware of this, then you're mistaken.
Users Browsing this Thread
There are currently 1 users browsing this thread. (0 members and 1 guests)