I'm going to break all this down as it will be easier.
| torok said: Second, I don't seem to understand the PCI-E bandwidht problem and the massive performace hit that coping data from CPU to GPU creates. When you work with, let's say, CUDA, sometimes you forget about doing some work that would be 200x faster on the GPU just because the amount of time to pass and get back the date would take longer than doing it slowly on the CPU. |
Glad you cleared up the point of you not understanding the PCI-E bandwidth problem.
Some datasets aren't terribly bandwidth or latency sensitive, some are, the ones that are reside next to the GPU's GDDR5 Ram, this would range from textures to geometry data for the tessellators to sheer number crunching compute.
| torok said: An unified memory architecture will allow PS4 do physics calculations that won't be on PC versions simply because it can jut get the data and calculate on GPU without data passing costs. You have data, it's on CPUs memory space (DDR3), then you must pass it to the GPU memory space (GDDR5). Now try it using PCI-E and see if it looks fast for anything. I can already garantee you, if the data is small, it doesn't matter if the GPU does the operation 300x faster, you will have to use the CPU. |
Well, the proof is in the pudding in regards to Physics calculations.
If you take the Unreal 4 engine tech demo, the PC was able to have far more partical physics calculations than the Playstation 4, Physics isn't terribly bandwidth sensitive, it can be done completely on a PC's GPU and reside next to the GPU's GDDR5 memory.
Heck Ageia's first Physics card didn't even use PCI-E Express and only had 128Mb of GDDR3 Ram on a 128bit interface which is a testiment to how lean Physics really is on bandwidth requirements.
Asus eventually released a PCI-E variant but that was still only PCI-E 1x 3.0.
The limiting factor with Physics is and has been compute resources, it can be stupidly parallel hence why GPU's are well suited to that sort of processing.
| torok said: Second, the "memory wall". If you don't believe, reasearch on Google or ask a HPC specialist. This is the single reason that supercomputing migrated form the traditional supercomputer PRAM model for the distributed model we see on clusters. And is one of the primary reasons for Cloud Computing. More machines, more memory bandwidth. On old supercomputers (current ones are actually clusters) you increaser the core count and tried your best to find a better memory technology, but that race was lost long ago. |
You're talking servers and super computers, that's your problem, consoles and the Desktop PC are far removed from that.
I still stand by that we are far from a memory wall in the PC space, I have yet to encounter such a thing and I have one of the fastest consumer CPU's that money can buy, I saw gains from having my 6 cores/12 threads @ 3.2ghz to 4.8ghz, I saw gains moving from AMD's 6 core Phenom 2 x6 to an AMD FX 8-core 8120 @ 4.8ghz all with the same DRAM speeds.
Granted my Core i7 has Quad-Channel DDR3 that will help somewhat in some scenarios, but even with half the bandwidth the differences were neglible, Intel's Predictors are fantastic.
Plus, next year with Haswell-E we will have DDR4.
| torok said: Cache is there to help, not to solve. It all depends if the ammount of data you need will fit on it or not. If it won't, you will have to refill it anyway. Having more cores only makes it worse. And cores only help if the code you are running is optimized to use them. A lot of games are 32-bit executables, just to show how they aren't that worried about optimization now (3GB RAM limit for the executable). |
Cache helps to solve the bandwidth and latency deficit of having the CPU to grab data from system memory.
L4 cache like in Intels Iris Pro is allot more flexible in that regard as it has 128Mb of the stuff.
As for the 3Gb memory limit, that's partly true, most games are starting to support 64bit now anyway, heck FarCry did back when the Playstation 2 was in it's prime.
| torok said: Now, on PS4, you know what you have under the hood. The developer knows exactly. You can even predict and interfere in the way things will be on cache to gain performance. You will have games using heavy parallel processing in 8-cores plus offloading some physics to GPU (SPH it's a good example). |
I agree with the first part.
But you are completely wrong on the second.
Here it is in bold so that it sinks in...
The Playstation 4 and Xbox One do NOT have 8 cores dedicated to gaming, they reserve a core or two for the OS and other tasks.
A PC can and does offload game rendering and Physics calculations to GPU's, no questions asked, at 4k or greater resolutions with all the settings on max, something the Playstation 4 could never ever hope to possibly achieve.
Here is why: The Playstation 4 doesn't have enough compute resources.
However with that in mind, I am willing to eat my hat if you can get a Playstation 4 to play Battlefield 4 in Eyefinity at 7680x1440 with everything on Ultra and achieve 60fps if you really think it's the be-all and end-all of platforms it certainly could do that right? (Currently it's 900P and High-settings. - GOOD LUCK!)
| torok said:
And the GPU part it's even worse, it's much more harder to write code that runs calculation well on any GPU than on a specific one, simply because the compilers suck (I mean CUDA tools and OpenCL tools). For the CPU part, you have amazing compilers (partially thanks to Intel works on the 70s and 80s with matematicians). And besides that, it is currently a mess. NVidia and AMD can't create a common GPU API/Framework for doing calculations and we have to watch OpenCL AMD GPUs and CUDA NVidia GPUs (they work with OpenCL too, but NVidia only creates tools for CUDA and they are far better. Current dev tools for AMD GPUs simply suck terribly bad compared to CUDA). And that will make one hell of a difference. |
There are other alternatives other than Cuda (Only locked to nVidia) and OpenCL. This is the PC, you can make your own if you felt inclined.
Mantle is coming it will be a game changer.
Whatever nuances that multi-platform developers make for console is going to translate into real gains for AMD's Graphics Core Next GPU's on the PC.
| torok said: I like MSI and Gigabyte MB, the Gigabyte 990FXA-UD5 is a good one |
Don't get the Gigabyte 990FX-UD5, it's a horrible motherboard with bad voltage regulation and vdroop. - Personal experiences with over a dozen boards.
The UD7 wasn't much better, the UD3 is downright crap because of the lack of mosfet and VRM cooling.
And AM3+ is a bad investment anyway, Socket FM2+ is the platform to go for as that's AMD's focus as Steamroller is cometh to that socket, whist AM3+ isn't getting any updates for at-least untill 2015.
In-fact I can't recommend Intel enough because AMD has woefull single threaded performance and are extremely power hungry.
--::{PC Gaming Master Race}::--
