18 August 2014 - 21 August 2014
Insider: There is a few title's in development that are using world physics or knowledge of nature. Iv seen a few games in development phase. One was a zombie game .. and it uses a real world time. I was shown the world changing over years you could get lost because of the dynamic changes. World physics is not 100% cloud dependent tho. But the online cloud processing was 60ms in test in lab 20ms. They use one of the data move engines to process "pim" information and a form of synthetic object render. There using geometric meshes that do not need to be processed with textures. Parametric. That is cloud based online. But it works in tandem with offline and also by mixing it with procedural geometry synthesized directly on gpu or coprocessors. 2 of the data move engines are specifically design to handle this.. I don't like talking about the cloud stuff in till ms have directly addressed the white elephant in the room. It will do more harm then good. Because people must see that 160ms and 700kb connection will services for world physics. Data packets are streamed in like a video is streaming the information is directly linked to dynamism. Crunchy numbers server side is very important..
An analogy would be that the console architecture knows how to build and display the world and even move it in some degree . But it lacks the amount of control over this world because it is limited by hardware processing. So another hardware links in to tell it how to do it . this hardware that tells it how to do it is doing it via procedural synthesis. Onces the console architecture is given the command it understands the command and will procedural this command intill the next command is given.. cycals these cycals do not need to be address spontaneously hence the low packet information required. The problem with cloud is people thing at this stage that whole games will render in the cloud and beam down.. but that technology is not driven alone by the data center but by the speed of the information. The internet on a global network is not fast enough to keep up.. 2018 will be when full texture stream in will happen but it may not be a visualisation that we see .. :) lets leave this here. Its to early for this level.. but state of decay could be the game .. inless it is the other game from xbox original :) expect a few games @ e3 to use world physics tho ..
Part 41. The AXI bridge. The future of interconnection buss design
Insider: Time for some digging.. this is one aspect or very important aspect to put forward ..Hierarchical..
Its is why so much information has been hard to understand when looking at x1 architecture. The x1 architecture is not dependent of clock speeds. all hardware communication is via the axi bridge/technology .. it is the future of interconnection buss design. There is no single 32/64 bit bus or address space pipeline.. there is a segmentation between cpu/gpu/ram/cach/audio/dsp "simplified terms" acceleration is a simplified term when thinking fsb pc architecture. But its burst axi transaction. The x1 axi bridge is not only on segment. The dram cntr / host guest gpu mmu / host guest io mmu/ are all linked hierarchical. The ps4 can not in any way match this architecture. Because the ps4 buss is connected by the address space for cpu gpu hsa v1. It uses a pc approach to send information around the architecture sony quoted there supercharged pc architecture based around cpu/gpu reading to gddr5 address space. But this will for every be the problem for them they can not gat off or upgate clocks as the cpu up clock will effect gpu not only in power difference but also heat..
The x1 architecture is dark silicon by design. The axi interconnecting can set clocks dependent of memory cpu dsp gpu. Upclock or gating off. The dsp/fpga are using a form of microblaze architecture modified for cach reading and logic. But they are cpu and gpu comput dependent. So with the hierarchical design we are working with a new interconnecting architecture. Ms design this architecture not for today. The traditional cpu/gpu coherent hsa is all ready avaliable but there is still thing locked of because they had to retool the sdks dx12 and future implementation is happening now.
So the whole soc is built within an axi hierarchical architecture and is why early oban designs were a ring buss acceleration design. But the real technology is burst ratio accleration. The whole architecture design is to accelerate the interconnecting but not by using tricks "ps4 gddr5"
Once specialized engines games are released halo5/gow5 your going to be able to make sense of this information. There is a massive amount of programming knowledge going to have to be used to get the most out of x1 architecture as the design is even more difficult then cell but as developers learn more and more gfx fidelity will be available .. ms are masters of software sdk the big sdk update will really see a 4 fold jump in gfx architecture available to devs ... there is going to be a software and hardware revolution later this year. Unfortunately it will take time...
But don't take my word for it so dig.. and remember some technology is still not officially announced. There is enough information for mrc to understand why there is so many op/s why there is two gpu threads. And why cu are split in two. Why axi dsp is clocked different to gpu/cpu and why.. this will help glu the information together. Enjoy and gameon !! :)
Misterx: Wow, that was ace to read. Sure many new indies will sight to ID@xbox after reading this :)
What about future PCs? Will axi tech hit the PC market?
Insider: Its hard to say. But it could be the key to linking apu gpu and math coprocessor. .amd is experimenting with tech.. but pc would need an architectural redesign. .. I think this is where super mobile technology wjll once again prove the difference
/end of part 41/
Insider: Mrx I think its best that we keep away from the topic of kotaku and zoe.. you know I think it sheds a bad light on the dark side of the industry. I also see it as bullying and down right wrong to suggest what people do with there private life to drag it so violent into the lime light . Even tho it was to gain traction in the industry. Its no worse then sony money hating reviews and published propaganda. I hear gaf has been trolling the person involved. Its a shame they cant see through the fog and figure out that behind the screen's and in the back office rooms there pretty much getting it done to them.. while I dont condone unsafe work ethics . I do find it bizarre to see the comments people put up to dignify there moral grounds in society weather gaming or to female anatomy. I find it very disturbing to think these people have no Facit or control of there reality or its environment ... While there is nothing wrong with admiration of ether partys in context but some things are best left to foliage at the bowls of the internet. Neogaf is really starting to truly show it is the pie piper of the Internet even the evil lore has chimed in from his dank sony funded life style of leisure minus the suit any comments that leave figitury conjecture and the soul goal is to belittle the other sex becouse they used the oldest job in the history books to gain. I think the people who forwarded the review should be fired or demoted and an apology should be scripted before all honest is lost to review scores or offical review facits. But the whole 'O' look its a Gggg gg girl gamer has to stop in this industry there is nothing wrong with complimentary conversion.. I know hundreds of female gamers who are tied of the lara croft stereo type that a lot of males carry through life and visualization of the other person when it is a female gamer on the other end of live. But I guess the easy way to get what I am saying is the blog as a whole should rise over this internet garbage.. a gamer is gamer.. Game on.
P.s I have no problem with contemporary art provided it is not posted here to humility or bully. Somethings are good for the Murrell :)
Insider: This whole 1080p Diablo thing going over at gaf is just a bunch of sdf thinking that they can manipulate the direction xbox one is heading now.. 1080p was all ways going to happen.. funny how ms gave them a code and bam there it is glorious 1080p.. then df have there payed sony suits do calibration test on xbox one to say its dropping frams constantly and is the worest version. Fun this why is it that there is more drops that apparently were fixed in the patch for ps4 .. the only thing iv noticed and heard from others is the atmosphere on ps4 is not as smooth .. tricks. I haven't seen or noticed one drop on x1 edition and to me it looks clearer to. Funny this code update :)
Good work to the bloggers. And the hitman he really did a number on n4g and gaf but as there payed money hates its a shame this will all be last months news soon enough.. but the more that people dig and put news out there the more people that join the community the more .. I will verify that yes tomb rader rg is and will remain 100% exclusive.. there just waiting till the p r night mare stops... like one dev said you would think sony is paying these people to riot .. true words have never been said.. crazy to think that the honestly is coming from mrx and the people but ablest ms ant money hating hate campaigns and they ant suits running the show but gamers.. very good things are heading your way guys.. people in charge are going to address some information soon enough.. GAME ON
forza_maniac: Rise of The Tomb Raider is Ours
A FC4 Flag directly presented from a dev to me:D
MIsterx: Yes, New Tomb Raider will never be on PS4 due to its only 1.8TF old gen architecture. If its publushed by Microsoft than...PC only maybe in the future.
Part 42. ESRAM and Bandwidth. No depended of CPU usage as in PS4
Misterx: Another round on why Sony lied about real PS4 GDDR5 speed.
Sony PS4 Effective GPU Bandwidth is 140 GB/s Not 176 GB/s – Disproportionate CPU and GPU Scaling
Read more: http://wccftech.com/sony-ps4-effective-bandwidth-140-gbs-disproportionate-cpu-gpu-scaling/#ixzz3B1gXLCP6
gforce1981: "For me it has been clear at day one, that PS4 memory speed (176) is peak and max. speed in theory and not real world situation.
Other hand MS has told that their target was 150 GB/s speed and they got it running REAL WORLD code.
For me this is old news, but it is interesting that some sites have opened their eyes...
"One out of every eight cycles is a bubble, so that's how you get the combined 204GB/s as the raw peak that we can really achieve over the ESRAM. And then if you say what can you achieve out of an application - we've measured about 140-150GB/s for ESRAM. That's real code running. That's not some diagnostic or some simulation case or something like that. That is real code that is running at that bandwidth. You can add that to the external memory and say that that probably achieves in similar conditions 50-55GB/s and add those two together you're getting in the order of 200GB/s across the main memory and internally. "
MS was honest from the start because in the real professional industry and devs know that 'maximum theoretical' bandwidth is not possible.
That is why MS has said the 140-150GB/s on the esram and the 50-55GB/s on the DDR3. (we know the maximum theoretical bandwidth on DDR3 = 68GB/s) but MS have been honest from the start.
Sony on the other hand has CONTINUOUSLY fed BS to their fan base.
take Killzone MP Resolution for one
and with the bandwidth of the GDDR5
they are not completely lying but they just leave out the fact that the 176GB/s is theoretical and that is ONLY if the GPU is the only thing that uses it.
MS could have just lied and said their bandwidth is 209GB/s esram and 68GB/s DDR3 ram = 277GB/s (using the same as what Sony is saying)
Both memories have separate controllers so it can be used at the same time. Think GPU can use eSRAM and CPU can use DDR3 ram (no need to have conflict)
IN FACT, the ddr3 uses 4 memory controllers and the eSRAM also uses 4 memory controllers.
which this means, less contention again and you can have the CPU using say 2GB for a certain processing and other stuff using the other channels and you don't get CONTENTION.
You have what is called "CONTENTION" which means as there is more demands from stuff like CPU the Actualy bandwidth decreases more than normal because you have CPU & GPU fighting for the same memory.
This is why in the AMD presentation about the xbox one, AMD them selves clearly stated about using only the eSRAM for GPU stuff because it avoids CONTENTION.
This is why games on the PS4 are lacking
*Number of people on screen (because that processing uses CPU, takes away from GPU bandwidth)
*Physics and other things like AI & Simulations (once again used by CPU).
Most games these days barely use multiple cpus. and when they do this is where it is really going to PUNISH the PS4.
That is another reason why Dx12 is such a game changer. and the xbox one was designed to deal with everything being used at the same time.
gforce1981: The Answers are here http://www.altera.com/technology/sy
As SoC developers shifted their focus from faster clocks to more cores, they changed the nature of the memory problem. Instead of one CPU demanding ever more megabytes per second (MBps), now we face many different processors—often many different kinds of processors—all clamoring for access at once. Also, the dominant patterns of memory access have changed. Scientific and commercial data-processing tasks typically involve strong locality of reference, or at worst slowly spooling a very large data set through a relatively compact algorithm. Provisioned with modest local SRAM or cache, a single CPU on such a task could be quite undemanding of main memory..
And here is why the PS4 has diminishing Bandwidth.
DRAM chip designers exploited this simplicity in order to deliver greater density and energy efficiency.
Consequently, DRAMs deliver their best bit rate when asked for large blocks of data in a predictable order—one that allows bank interleaving.
If the SoC deviates from this regular pattern, effective bandwidth of the memory system can drop by an order of magnitude."
Order of MAGNITUDE (this is why the memory drop in the PS4 is more of a drop the more the CPU needs this ram also)
** Remember MS saying its about data flow and data in the right places. and remember the XBOX One has a lot of on chip storage / cache
"New Pattern of Access Unfortunately, the evolution of SoCs has undermined the assumptions of the DRAM designers.
Multi-threading and emerging trends in software design are changing the way individual cores access memory. Multicore processing and the growing importance of hardware acceleration mean that often, many hardware clients will contend for main memory. Both of these trends complicate the simple locality of reference upon which DRAM bandwidth depends.
Multi-threading means, among other things, that when a memory request misses its caches the CPU doesn’t wait: it begins executing a different thread, whose instruction and data areas are likely to be in entirely different physical memory regions than those of the previous thread.
Careful multi-way cache design can help with this problem, but in the end, consecutive DRAM requests are still more likely to reference unrelated areas of memory, even if the individual threads carefully optimized their memory organization. Similarly, multiple cores contending for the same DRAM channel will tend to scramble the order of DRAM accesses."
The idea of multiple DRAM channels is not limited to game systems. Packet-processing SoCs began providing multiple, fully-independent DRAM controllers several years ago. But the strategy has challenges. It imposes even more complexity on memory optimization, as system designers must decide what data structures to map to which channel or controller.
It also, of course, creates the possibility of giving a particularly demanding task its own private DRAM controller, which in some embedded applications can be valuable.
It is pretty self explanatory why the PS4's GDDR5 memory can be hit so bad and why all of their games are very 'minimal' CPU intensive
Well I finally found what I was looking for (YAY)
This is from the Microsoft and AMD Game Developers Presentation! (From MS and AMD) not some no name person
talking about PS4 bandwidth and xbox one.
"The Xbox One’s 32 MiB of ESRAM of general purpose RAM runs at 102 GiB/s (which is said to be ‘sometimes faster in practice’ – and indeed can spike up to a theoretical 204 GB/s).
Microsoft places great emphasis that there’s zero contention, meaning the CPU (for instance) doesn’t access the ESRAM.
This means the Xbox One’s GPU is free to go to town and use the limited resource as it requires. ESRAM “Makes everything better” – a few common uses would be Render Targets, Compute Tasks, Textures and Geometry. "
"Microsoft also provided “the four stages of adoption” when it came to the ESRAM, which seems to indicate how developers should utilize the RAM. The first, would be the allocation of a ‘small number’ of render targets in the ESRAM (Generally is the Depth / Stecil pass, followed by color targets, then other stuff). Clearly a render target is a prime candidate for the ESRAM. This means that a scene is held in memory (usually near completion in the drawing process) but maybe requires say a Pixel Shader to run.
The second memory which was used is given an ‘Alias’ so that it can be reused again and again." *** This saves bandwidth
"The final is Asynchronously DMA resources in/out of the ESRAM, which could well leverage the power of the compute commands the the Xbox One’s move engines.
This would ensure that there’s always a flow of the most important data inside the ESRAM, and would help to optimize its use.
Due to the usage of this while you’re rendering, it’s important for the developer to map their memory and plan everything in advanced – meaning good knowledge of the hardware. Microsoft believe we’ll be seeing developers using this final technique in the ’3rd wave’ of titles, but there are a few second wave games (likely those who’re first party) already using this technique."
and finally this
"The next clue was from Sony’s Playstation 4, whose leaked documents showed that it had a 20GB/s (roughly) bus used for the CPU access.
Clearly there’s a huge performance issue if there’s DRAM contention (for instance, the GPU is trying to gobble up all the bandwidth of the DDR3).
In their document, Microsoft use a play on old BASIC programming to get their point across. 10. Use ESRAM as Much as possible. 20. Leave DRAM for the CPU and DMA. 30. Goto 10."
Sony will be found out that they are feeding BS to their fans.
It is worthwhile looking at this Presentation from MS and AMD!!!
Game on people.
Mistercteam: Dont forget the real deal is esram for compute task and for geometry texture, it is need ALU
like i said 100 times, compute task is not exactly run on GDDR5 or DDR3
it is on vector register
so if they said eSRAM make compute task + geometry + texture better than certainly it is a PIM
Misterx: I think PIM have its own memory for FPGA. ESRAM is just ESRAM and should be used to quickly assemble calculated pixels from varios sources(GPU, PIM) for output.