But as you know amount of system RAM isn't on it's own a particularly useful reference for overall gaming capability.
A high end gaming PC at the end of 2005 may well have had 2GB (4x console) system memory while by the end of this year it could be argued that only 16GB (2x console) will be high end gaming PC 'standard'. So in terms of system RAM capacity then yes the new consoles look good compared to your typical gaming PC in relation to last generations position but that's were the favourable comparison ends.
For example look at the compute capabilities of the 360 compared to the highest end PC's of 2005 and then do the same with the new consoles using todays hardware (which itself will be either superceded of very close to being superceded by the time those consoles actually launch). As a less used example the peak SIMD throughput of the fastest CPU available when the 360 launched (AthlonX2 4800+) was merely half of Xenon while a mid range Haswell i5 based CPU today has well over 4x the theoretical SIMD throughput of the Jaguar modules in the new consoles.
Similar results can be seen by comparing other metrics like shader throughput (Xenos matched the fastest GPU available at it's launch while Orbis barely registers over a third as fast as the fastest PC GPU available today 4-5 months before it's launch), tecture throughput and arguably memory bandwidth.
This isn't really evidence of anything other than the lack of effort that went into optimising those ports for the new console architecture. It's not like those eary ports that struggled haven't been far exceeded today by games running on that same hardware.
It's June 2013, the consoles don't launch until at least September likely later. 4 months before the Xbox 360 launched you couldn't get a GPU with 512MB of memory. They were only available a matter of weeks ahead of the 360 in the two highest end GPU's available. Yet today there are 3 models available sporting 6GB memory (more than enough by most accounts to match the consoles on graphics memory useage) and 4 months from now when 4Gb memory chips become more available there's no reason to expect 8 GB (and even higher) memory configuration to be available too.
Why be so specific about the implementation? Why not ask what GPU was available in 2005 that had the same DX9+ featureset as Xenos? Or unified shaders? Or 256GB/s bandwidth for the framebuffer? Or a unified memory architecture for that matter (one of the key advantages that is often pointed to as the new consoles big advantage over similarly specced PC's and yet it's an implementation detail that the last 2 generations of Xbox already followed)? I'd say the HSA nature of the new consoles is well behind the technical bar set by the 360 with those features in 2005.
Besides, 8 big desktop PC cores aren't required to match the 8 low power laptop and tablet targetted cores powering the consoles. 4 piledrivers running close to 4Ghz like you'd get in a decent model Trinity would be plenty for that and you can get that for sub $150. By the time the consoles launch Kaveri will be filling that price bracket with yet more powerful cores and full HSA compliance. Coupled with a decent discrete GPU it should be capable of everything the new consoles can achieve via the CPU-GPU links on the APU alone leaving a massive dedicated GPU to deal with the graphics work. Haswell ditto which in it's highest end configuration has similar peak compute throughput on the CPU die alone (including the IGP) as the new Xbox has in it's GPU and CPU combined.
Just image trying to compare the around 310 GFLOP's peak throughput of Xenon + Xenos in 2005 to the 38.4 GFLOP Ahlon X2 as I'm able to do with Haswell today months before the new consoles launch...
I don't follow laptop hardware too closely but it should be easy to create a laptop with far more power than the PS4. I'm over egging it I'm sure but a quad Haswell, 16GB DDR3 and a 780m (around 80% of a desktop 680). Should slaughter any console. Expecting slicker CPU-GPU communication to make up for such a huge power gulf is wishful thinking IMO. Expecially with the likes of Kaveri and Haswell muddying the waters with their own IGP's which could be put to use in much the same was as the console GPU's for low latency GPGPU work. - or Nvidias upcoming Maxwell GPU which may have CPU cores on die with the GPU.
From day one with the exception of a few PC centric games (quake 4 being the main example) the last generation consoles ran cross platform titles as well as the absolute best PC's - so for all intents, better than 99.99% of the PC's available at the time. If they had opportunity to develop from there thanks to devs getting to grips with their new architectures then that just meant those consoles could continue to keep up with the PC's progression for some time further.
This time the new consoles come out of the gate well behind the fastest PC's on fairly well understood and easy to optimise archiutectures. So not only will mainstream enthusiast level GPU's likely still offer similar or better results than the brand new consoles but the consoles will have far less room to improve this generation compared to last. That doesn't sound like a benefit to me.
Here we go again....
