The RSX suffers from latency when reading or writing to the XDRAM.

Latency and a bandwidth bottleneck are two different things.

Final-Fan said: sieanr said: Final-Fan said: sieanr said: Final-Fan said:Is it accurate? YES OR NO. IF NO, EXPLAIN WHY. Is it misleading? You've already answered "Yes." EXPLAIN WHY. Yes, its accurate - I think that pic is from a microsoft slide. 10gbs both ways is by no means a bottleneck. I don't feel like explaining now, so maybe someone else can pick things up. Both systems have their share of problems and flaws in their designs. You don't have to explain (to me, now, at least) why 10.8gbs is not a bottleneck; but you certainly do have to explain why, if it is not, 20gbs/15gbs is a bottleneck in the PS3. That just doesn't seem to make any sense at all. Who said the speed between the CPU and GPU was a bottleneck in the PS3? 256bit split bus in the RSX, ring connector in cell, RSX to XDRAM - those are some of the potential bottlenecks in the PS3 I was talking about. Xbox 360: 512 RAM-->(22.4 GB/s)-->GPU; and 512 RAM-->(22.4 GB/s)-->GPU-->(10.8 GB/s)-->CPU OTHER WAY GPU-->(22.4 GB/s)-->512 RAM CPU-->(10.8 GB/s)-->GPU-->(22.4 GB/s)-->512 RAM

Considering the CPU and GPU share the same bus, any bandwidth the CPU uses limits the bandwidth available to the GPU. So for example if the CPU is using 10 GB per second, there´s only 12.4 GB per second bandwidth left for the GPU.

sieanr said: The RSX suffers from latency when reading or writing to the XDRAM. Latency and a bandwidth bottleneck are two different things.

I think this has been extensively covered already.

- The Xenon CPU has to get all its RAM through the GPU, using a shared bus to access the slower GDDR3 memory.

- The Cell is connected directly to the much faster XDR RAM.

- The Xenos GPU has to share the bus to the GDDR3 memory with the CPU, limiting available bandwidth.

- The RSX is directly connected to the GDDR3 memory, but is allowed to access the faster XDR ram through the Cell's EIB as well. I guess the later option will not be used much, 256 MB should in normal situations be sufficient. Also the default harddrive could function as a provider of virtual memory, if somehow 512 MB of system RAM wouldn´t be enough (a PS3 advantage, because of the default harddrive), but I guess if using the Blu-Ray drive to stream data (so smaller parts of game data needs to be loaded into memory) the harddrive could rather serve for caching.

To quote my earlier post:

'For most stuff, the memory directly connected to the RSX will be used, but there is enough bandwidth to use the XDR memory as well. I've heard of developers using the XDR as additional texture memory as it not only increases texture memory it also increases texture bandwidth. The latency will increase but RSX this isn't a big problem for GPUs and RSX has bigger than usual caches anyway.

I've also heard of developers writing compressed textures directly from the SPEs to the RSX, this bypasses memory altogether so so effectively reduces latency while increasing bandwidth.'

Reading back some posts I guess, the RAM confusion you complained about was already covered.

sieanr, the first picture in your posting is totally wrong. It would be like including the 300GB per second EIB internal CPU bandwidth which also connects the SPEs to their local memory stores.

With regard to the second image, maybe these developer quotes makes you understand its (non-) relevance:

"and it's a total non-issue. You never, ever need to access that memory from the Cell - I can think of some useful debugging things you might do with that access in the testing stage, but that's about it. In fact, on the PS2 you couldn't access that memory from the CPU at all, and it was never really a problem!"

"I can see a couple of reasons why you might want to use it," another developer told us, "but really, they're pretty obscure, and you could probably do them on the RSX anyway, since it's quite flexible. Besides, if you really need to access video memory from the Cell, you can use the RSX to copy it over into main memory really quickly - it's all there on the slide."

"I doubt a single person in the room batted an eyelid when they showed that slide," continued the first source. "It's exactly what we'd expect, and the bits that we actually need to use to make games are perfectly fast."

IMO both images have been abused by XBox 360 fans far too often.

Platform propaganda and misinformation is nothing new, in the distant past PC MSDOS owners and even some PC stores spokespersons claimed that the Amiga was only useful for playing games.

Many stated a GUI is for someone who cannot use the CLI (The Amiga both had a GUI as well as a far more powerful CLI). Many more simultaneous colors (4096 on Amiga, 2 on PC and Mac) were mainly useful for games and some very specialized uses. Of course today we know companies like NASA used them for telemetry, Disney used the machine to animate the Lion King, Dinosaur and many others, Terminator 2 morphing was done on the Amiga, Babylon 5 and Seaquest special effects were done on Amiga, US and Israeli Air Force used them to train pilots using Amiga based flight simulators. The Amiga for quite a while had the most advanced fully multitasking office applications (copy and pasting images from a paint program or maybe a colorful spreadsheet pie chart into a word processor was all already possible in the 80s) and most advanced and multi-functional artistic programs, visual as well audio related ones (no wonder as the Amiga offered advanced stereo sound while the Mac was silent and PCs could only beep), etc. But still the Amiga was never that great of a success in the States (in Europe the platform did much better, with marketshares in some well educated countries like Sweden up to 90%!). Main reasons I think, mass PC advertising and mass misinformation.

I really hope some of the PS3's potential will not be wasted due to similar misinformation campaigns, the whole internet seems filled by PS3 misinformation and IMO pretty much faked reports of Doom and Gloom...

PS3 vs XBox 360 / Review by a music producer:

´As a music producer, the difference between good audio and bad audio can make or break my day so I tend to be very anal when comparing sources. First up was the 360 and to be honest I was a little disappointed, if not altogether surprised at the quality of the sound.

A couple of hundred quid will get you a very decent standalone CD player these days, but my beef with the 360 is that the audio sounds almost like an afterthought. That's kind of okay when you're playing games because your attention is drawn away from the 360's lack of sonic depth, but even when playing my test CD I found it a bit flat and unrewarding and the high end seemed to be out of step with the mid frequencies, giving the overall sound a synthetic, over-processed quality, a bit like a cheap MP3 player.

With the PS3 on the other hand, I have to say I was very impressed with the way it dealt with audio in general. As with the 360, when you're in mid fire fight, you don't really have time to stop and smell the roses, so I stuck in my favourite test CD (The Night Fly, by Donald Fagan, since you ask) and within a few seconds it was obvious that something very special was happening. It was like colonic irrigation for your ears - every guitar note seemed to hang in its own space, drums and percussion tell you exactly were they are in the stereo image and as for the bottom end, let's just say I don't think the people that live downstairs are going to send me a Christmas card this year!

The PS3's CD sound impressed me enough to compare it with my £500 Sony CD player that I use for work. Now this was a real revelation, because the PS3 beat it hands down in every way, eclipsing the fine detail delivered by my standalone player with a sound that seemed more natural and open overall.´

Full article http://www.avreview.co.uk/news/article/mps/UAN/1375/v/3/sp/332702698872332111368

Also in their opinion, you get what you pay for. And from an AV perspective, the PS3 offers more value for your bucks.

sieanr said: The RSX suffers from latency when reading or writing to the XDRAM. Latency and a bandwidth bottleneck are two different things.

source plz (for latency in PS3 AND 360, showing that RSX is significantly worse -- and it has to be trustworthy numbers)

On another note, how you you answer MikeB's point that the entire of any potential latency issue between the XDR and RSX is secondary at best, since it has its own memory to draw on?  

Hey mikeB, have you leanred what effective clock means yet?

Funny that I posted things that are inaccurate or FUD. Maybe that has something to do with what I wrote next to those diagrams? Maybe that is some sort of criticism of what you've done in this thread?

Could you tell me how much faster the XDR is than GDDR3? I'd just like to know if you know what you're talking about. Not to mention XDR latency is only on par with GDDR3.

Besides, the XDR offers about 14% more bandwidth than the GDDR3 bank in the PS3, not significant enough to warrant a "much faster" imho.



True. However, the EDRAM takes a significant amount of strain off the GDDR, more or less enought to make up for bandwidth lost due to the CPU using the RAM as well.



Too bad the RSX has just a 128bit bus to memory.

Practically speaking, the HD would only be used to stream textures as its just to slow to be virtual texture ram. This is something that can be done from disk. Not to mention 12xDVD average read speeds are faster than Bluray.

For the RSX to access the cell, it must go through to memory controllers and two serial busses. That incures one hell of a latency penalty.



Actually, that picture doesn't incllude the cross type EIB connector in the Xenon.

Speaking of which, a funny thing about the Cell is that Ring Connector. See, the ring connector isn't as fast as the Cross type connector in the Xenos or Core 2 Due, not to mention it has a latency penalty - especially if you start operating near peak transfer rates.



Just incase you dont get it, thats the point.

Both sides had stupid specs they hoist up to claim one system is superior to the other. A good example of this is how you mentioned floating point performance. FP has very little impact on the performance of a console.



LOL



This will be a bitch to find, but common sense will tell you going trough two memory controllers adds a lot of latency.

But funny you ask for me to source things while MikeBs claims go unsourced.



I thought this thread was about the performance of nearly dead consoles, not dead systems no one cares about.

But hey, good thing Sony has never waged a misinformation campaign with the PS3s spes. AMIRITE?

Most of what he has said is biased, based on bad sources or no sources at all, or completely founded on false assumptions, but you gotta hand it to Mike, he's defended those false assumptions and biased manipulations to the death in this thread.