Drakrami said:
barneystinson69 said:
3 years ago $399= 1.8Tflops. And for $1500, 6Tflops is the biggest rip-off. Even pre-made PCs are far more powerful for $1500. But ok...
|
I dunno what you are smoking... a GTX 980 TI, 5.7 Tflops, costs $650 at the very least. And that's just one graphics card alone... Seems like you know very little and even too lazy to google. Why bother continue arguing?
|
I think the hair on my head is starting to turn grey with the blatant abuse of flops as some kind of performance metric.
You are both being highly inaccurate.
An nVidia GPU with 6 Teraflops is not going to be representative or even comparative of an AMD GPU in the consoles at 6 Teraflops.
Nor are flops a complete representation of a systems actual gaming performance anyway.
And nor are Flops tied to any kind of pricing/costing model.
mutantsushi said:
Scorpio is the next gen of Xbox, it is not retaining same basic architecture as Xbone, e.g. memory, and certainly CPU will change as well> Sony's keeping all those elements the same with simple upclock meant optimization target for devs was really just one platform. Xbone's ESRAM approach was a dud, and they aren't going to hamstring themselves by sticking to it... Meaning it will be a new platform.
|
We have no idea what the specifications for Scorpio actually is, Microsoft will likely retain the same basic Architecture as the regular Xbox One, but will go with a better memory subsystem, more modern GCN blocks and possibly a more modern x86 chip, that doesn't mean backwards compatability will be broken though.
As for the eSRAM, eSRAM is a fantastic technology, when used and implemented correctly.
The issue with Microsoft however was that they implemented eSRAM at the expense of compute units, that doesn't mean the technology is bad however.
And honestly, I hope they go with an off-chip solution like eDRAM to be used as a CPU's L4 cache, reduce power consumption, allow for more seamless Xbox One backwards compatability and tiled based rendering approaches.
Intrinsic said:
Yes higher clock= more heat.
|
Not exactly.
If you have a higher clockrate, but more aggressive binning, you can run at a lower voltage, so you can have both a higher clock and less heat.
Scorpio launching a year later means that Global Foundries/TSMC will be able to get more intimate with 16nm/14nm and make refinements to their processes that should translate to less leakage, so clockrates can be increased without any increase in power consumption or the need for more aggressive binning.
Intrinsic said:
PS4 is going with the minimum for stability and to reduce costs. AMD can Run the GPUs at a higher clock cause they know they are charging more for just the GPU. If they wanted to clock those GPUs at say 1.2TF as the minimum clock meaning that all GPUs are guaranteed 5.5TF, then it means they will take GPUs that can be over clocked to 1.3GHz and still remain stable for prolonged periods. That in turn reduces they amount of usable GPUs they can get from a wafer. Which drives up the price.
|
Whilst for the most part you are accurate, in many instances to get more workable chips, you just need to be more aggressive with the voltages or more conservative with the clocks.
But there are a ton of more factors involved, you can actually choose different types of transisters to use which have their own power characteristics, changing the layout of the GPU can help reduce leakage, patterning improvements and other refinements can actually make a larger, higher clocked chip, be cheaper and use less energy than another chip.
However, the current situation with the PS4/Neo/4k/Pro is not going to be representative of chips coming out of the fabs in a years time or define what Scorpio is going to be from a chip perspective.
And you can bet Sony will take advantage of those process advantages as well when they happen, with a silicon respin of the PS4's SoC.
Heck we don't even know what Fab they are using, there is a difference between TSMC's 16nm and Global Foundries 14nm for example... And AMD has historically used both. (But mostly only TSMC for Semi-custom designs oddly enough.)
Intrinsic said:
Unless the scorpio is using something other than Polaris 10, it's not going to hit 6TF. 5TF most likely but not 6TF. And for it to even hit that five they would need to seperate the cpu and GPU meaning they won't go with an APU design anymore. The heat from a CPU just won't allow you run the GPU at too high a clock in an APU design.
|
I think a clockrate of 1300mhz is more than feasible in a years time once 14nm/16nm matures.
But a wider design based around something like a cut-down Vega is probably going to be more likely.
There are a few technology's that AMD has used on the Desktop which I am unsure if the consoles used which can save energy and reduce heat... Known as the Resonant Clock Mesh, even then there are improvements being done with that technology.
CGI-Quality said:
Also, try not to quote much more than 3 posts in one reply.
|
Why is this the first time I am hearing of that rule? I have been doing it since forever. D: