By using this site, you agree to our Privacy Policy and our Terms of Use. Close

Pemalite said:

The Playstation 4's GPU from the beginning was not a direct clone of Pitcairn, it had some customizations such as the ACE units.
The Xbox One might take a similar approach though, so a direct quad-drupling of most of the chip is a possibility... It's hard to say which direction they will go at the moment.
Vega will be more clock-rate optimized than Polaris, which microsoft could take advantage of, so they could get away with a smaller chip but higher clock speeds.

The console APU's are all semi custom to some degree yes. All good points though.

Pemalite said:

They likely didn't start their half-gen upgrade untill well after Sony, so that explains part of the delay.

Possibly, but I find that unlikely considering PS4 is the more powerful console and decided it was time for an upgrade 3 years into the gen. Maybe PS heard rumours that XB was shooting for 4 years and decided to get Pro out asap for a head start on Scorpio. Maybe XB heard rumours that PS was shooting for 3 years and wanted to wait and see what Pro had hardware wise so they could answer accordingly. Maybe XB just wants a Pro with slightly higher Tflops so they can use that power marketing to their advantage this time. Maybe XB waited purposely so they could start "next gen" early again and get a jump on PS5. So many possibilities.

Pemalite said:

There is some potential improvements that we know of already, such as the L2 Cache (IF it is for the CPU and not the GPU that is.)
An increase of 4x is nothing to sneeze at.
The Xbox One and Playstation 4 have 4Mb of total L2 Cache (In a 2x2Mb configuration) and the Xbox One could be quad-drupling that number to 16Mb or 2x8Mb, that should result in some decent performance gains for the CPU without touching on clock rates.
The reason why we might as well not expect an Excavator based Carrizo or Zen is cost. 
Those chips eat up a ton of transistors that could be spent on the GPU.

True. Die space isn't infinite. The leaked cache improvements would surely help Jag. Since PS and XB had Jag at 28nm and decided to have it made on 16nm for both Slims and improved on Pro, it could very well be beefed up even more for Scorpio and would save on space and cost. Phil and his "balanced" quote make it sound like it should be (Carrizo) Excavator or Ryzen, but what XB says right now seems to be just whatever everybody wants to hear so...

Pemalite said:

Basically a Halo product.
AMD and nVidia don't release high-end GPU's like Fury and Titan because they move a ton of hardware and make a ton of cash... They sell them because it is a Halo product, people look at Benchmarks and see AMD or nVidia on top and think that company is better overall... And that can boost the sales of other GPU's.

Depends on your definition of halo I guess (no XB pun intended).

The problem I have with this, is that we're talking consoles and not PC's. There's good reason why a true halo console like the PS3, had such poor sales early on. Price and lack of games (cell). It's also the reason why PS4, as well as Pro, were $399 at launch, and have lots of games (x86).

AMD and NVIDIA can sell miniscule amounts of Fury and Titan (to draw attention) because they sell lots and lots of 480's and 1060's etc. For XB to try this with Scorpio would never work because they would have to rely on the One S gaining serious marketshare from PS4, which isn't going to happen. Scorpio has to hit $399-$499 and has to do it's share of making up the gap on PS4 as well as Pro (or start completely fresh as next gen).

I would say their best chance is for XB to do the same thing AMD just did to NVIDIA. Don't try and go head to head with your best and worst. Let your competition come out with their best (Pascal/Pro) while you come out with your worst (Polaris/One S), then a year later come out with your new best (Vega/Scorpio) and try to gain some marketshare by being the most powerful (for a period of time). It worked for the 360 (minus the red ring launch rush).