12-15 IIRC
I read some articles about 7nm being called 10nm by Intel. Is that rubbish ?
Predict: Next gen console tech (9th iteration and 10th iteration edition) [2014 - 2017]
- Thread starter Shifty Geezer
- Start date
- Status
I read some articles about 7nm being called 10nm by Intel. Is that rubbish ?
The smallest feature size as an indication of the node performance is getting crazy. There seems to be that much of a difference where intel 10nm is similar to others 7nm.
And then we have 7nm EUV which is going to be 20% smaller and 10% faster... but it's still 7nm.
I don't know but I think 7nm+ is worth waiting a year if they can sort it out. Supposedly they are still having issues getting the frickin lazer beams to reach death ray levels of power.
I thought you already knew?
I was mistaken and I was curious to know what the size of my mistake was. After all the topic is next gen hw and what's the point to post into thread if it's not related to next gen hw?
You were picking a fight with a strawman.
Intel's 10nm is roughly following its cadence, which is improving transistor and interconnect density in step with each of its nodes.
7nm is following the foundry industry's cadence, which has stretched the mostly traditional transistor and interconnect density improvements over two nodes as of late.
It's been dominated by marketing for years at this point, and Intel in prior nodes did opt to skip scaling metal at least once. Everyone blames everybody else for starting the marketing wars, but nobody's numbers have been reflective of any specific or consistent feature for a long time.
Since Intel has opted to keep scaling its metal pitch more consistently, there are various measures where its 10nm and foundry 7nm processes trade blows or are somewhat close.
In terms of expectations for the consoles, one thing I recall is that the prior generation was introduced at 90nm. It would last through the 65nm, 40nm, and 28nm transitions before the current gen appeared. In terms of raw performance, the current gen was something like 6-8x the prior generation, which wasn't quite the extent of the prior one's improvement. The mid-gen consoles most clearly exceed the mark set by the prior generational shift, at 16nm.
If people are counting on 7nm to be the next generation's node, that is closer to 2 full node transitions, since that is hiding the comparatively unpopular 20nm and 10nm nodes that don't scale transistors and metal pitch equally.
The 7nm node might be akin to two slightly better than average node transitions, but it might be hard to expect that matching 3 full node jumps--when 3 jumps was a wee bit short of the usual trends last time.
Based on the max power consumption of both Pro and XBX (respectively 155W and 180W), can we expect similar max power consumption for PS5 and XB2 ? Particularly if the cooling tech used in XBX is also used on those future machines ?
And if so, if PS5 is hitting 160W-180W max, how much tflops at 7nm can we reasonably expect ?
And if so, if PS5 is hitting 160W-180W max, how much tflops at 7nm can we reasonably expect ?
180W on 7nm should give you at least 7000 horsepower.
Tkumpathenurpahl
Veteran
Enough for Lots of Ducks?
I hope they'll launch the same year, because it's going to be as epic as 2013.Its the consensus that PS5 will be the next console released, but is it beyond MS to release the next Xbox before PS5?
I mean 2 mid gen upgrade = 1 next gen jump?
Sony is in a position to launch earlier for a few reasons: Their mid gen was 2016 so a reasonable lifespan is at least one year sooner than MS. The Pro's 4.2TF means the tech allowing a significant improvement is also 1 to 2 years sooner (it's maybe 1.3x to 1.4x average gpu improvement per year, but slowing down). Analysts claim sony didn't put a UHD BR because they intended to put it only when adoption rate of 4K TVs reached at least 50%, which is soon.
But there is a good possibility sony might plan to wait a while, considering MS cannot recover any significant market share until next gen. They would milk the ps4 and pro for a few more years, maybe losing a year head start to MS, or launching at the same time. The pressure to launch earlier is on the challenger, but it's a double edge sword since launching later allows better hardware specs. Sony launching the Pro a year earlier was a very weird move.
Really? I think they tried to force to board to their advantage, while they had the upper hand.
I mean, we are all giving MS well deserved credit for their BC/FC with the XBX. If Sony were to make the a big jump like MS did, they might not have been able to run PS4 software as is. And I assume that Sony does not have similar capability in the BC/FC department at the moment. And if that is the case, they might have ended up releasing a PS4P+ that might have had interop issues with PS4 software.
All of this is pure speculation of course. But I do wonder how Sony can counter MS and their BC solution. And even more puzzled if the BC capability actually does matter.
Another question does the XBX BC stuff work for indie titles also?
Darn, I think I should have split this into different thread..... lazy
//Edit : Added question about correct thread.
Rikimaru
Veteran
I bet Sony could launch 6TF machine in 2016 too. PS4 Pro even has features X does not. It's a matter of price. They clearly intended to price it $399. Even after year X is $100 more expensive.
I think you've mixed it up a bit, IIRC Intel's 10nm is about similar size as competitors 7nm. The reality is that all those numbers are pure marketing these days and Intel just pretties up their numbers less than the competition
It is not like one console is 499 dollars/euros and the other 399 dollars/euros... The big surprise is why Microsoft launched one year later...
Suppose the Pro was clocked like the X, it would be about 5.5TF which was sony's second choice. They said it was lowered to 4.2 because of costs. That would still require other techniques for improved efficiency and maybe the lowest cost gddr5x to keep it on 256bits. But still allow their hardware-level compatibility which seems to be more important to Sony than MS. There's no way the dropped 5.5 design was a different architecture.
I imagined doubling the Pro GPU the same way, adding yet more advanced features, like they did with dr-fp16 and bufferID, and have significant power reductions which AMD might be focussing on. 16GB Gddr6 256bits, then clock 10% higher than the X on 7nm+. 200W. 12TF. 400mm2. Something like that.
I imagined doubling the Pro GPU the same way, adding yet more advanced features, like they did with dr-fp16 and bufferID, and have significant power reductions which AMD might be focussing on. 16GB Gddr6 256bits, then clock 10% higher than the X on 7nm+. 200W. 12TF. 400mm2. Something like that.
Nearly 37% higher clocked GPU doesn't come free for yields.
I don't follow.
?
Last edited:
Source for this?..
- Status