MS: "Xbox 360 More Powerful than PS3"

[Bobbler]

Red Cloak (the poster in that gamespot link) has a history of posting things like that on many different forums. You'd think he was on MS' payroll for the amount of FUD he spreads. It's almost amazing...

What's worse is he doesn't seem to understand anything, even when told/shown (with proof) straight out what he says is BS he still goes on his merry way and continues to spread it.

 

[j^aws]

System Memory Configuration & Bus System (Break Down):

PS3: Memory Configurations & Bus Systems
256MB XDR System Ram: 25.6GB/s
256MB GDDR3 VRAM: 22.4GB/s
CELL to RSX: 35GB/s (20GB/s write + 15GB/s read)

Yep.

CELL EIB: 300GB/s peak

IIRC, the EIB is clocked at half, 1.6 Ghz, at 96 bytes/cycle (128 B/cycle burst) -> 154 GB/sec (205 GB/sec)

CELL FlexIO Bus Bandwidth: 76.8 GB/s (44.8 GB/s outbound, 32 GB/s inbound)

That's the ISSCC 2005 disclosed figure. For PS3, it's 40 GB/sec (35 CELL-RSX, 5 I/O).

RSX to Memory: 48GB/s Effective (can access both XDR & VRAM simultaneously)

For peak access to 'external RAM'. Also for RSX, 22.4 GDDR + 35 FlexIO ~ 57 GB/sec

South Bridge: 5GB/s (2.5GB/s upstream + 2.5GB/s downstream)

Yep.

Xbox 360: Memory Configurations & Bus Systems[/B
512MB GDDR3 Unified Ram Design (Xenos to Ram): 22.4 GB/s

Yep. Xenon+Xenos to 'external' RAM.

Xenon Internal Bus: 1.5GB/s

Not sure. Probably missed this somewhere?

Xenon to Xenos: 21.6 GB/s (10.8GB/s read + 10.8GBs write)

Has a caveat. As mentioned earlier, when accessing GDDR at 22.4 GB/sec, only 10.8 GB/sec Xenon_L2-Xenos available.

Xenos to EDRAM: 32GB/s
10MB EDRAM Internal Logic: 256 GB/s
South Bridge: 1GB (500MB upstream + 500MB downstream)

Yep.

 

[jvd]

http://www.gamespot.com/forums/show_msgs.php?topic_id=24056891

:smile:

well come one , its the same as ms getting hit with launch some launch titles lacking fsaa .


Sony was all big at thier e3 press confrence talking about fp32 hdr let alone fp16 hdr . Now devs are using int 8 to get more speed .

 

[Lysander]

RSX to Memory: 48GB/s Effective (can access both XDR & VRAM simultaneously)

This is unrealistic, not effective; you just added both BW (xdr, gddr3) together.

I will do the same for 360gpu: 22.4 GB/s gddr3 + 32 GB/s to edram = 54.4 GB/s

 

["Nerve-Damage"]

the problem mckmass is he put effective numbers for the ps3 and real numbers for the xbox 360 .

Makes it look better than it actually is

All the systems numbers that I listed are either effective or peak rates. Real-world performance will always dictate final performance. So Cells EIB peak performance of 300GB/s may only be 210GB/s (or less) of real-word performance depending on the intensity of the data that’s being executed across the channels. Or the Cell FlexIO of 76.8GB/s of peak bus bandwidth will of course drop in overall real-world performance as stated with the Cell EIB. The only thing (in theory) high peak rates are suppose too guarantee is that there’s enough real-world bandwidth (at least 75-80%) there for task to perform efficiently without to much latency (i.e. slow down). You’ll never aspect a game like “pongâ€￾ too rapidly deplete or topple the bandwidth; however a game like Heavenly Sword that’s a whole other story.


In the end PS3 has higher tolerance for real-world performance than Xbox 360 if you choose to accept the truth or not. My Xbox 360 e-penis isn’t hurting because another system has better specs (on paper), only time will tell (CES or Sony Feb-Conference) if the PS3 lives up to those specs.

 

[jvd]

Never Jawed has already corrected your numbers. You should go read what he said

 

["Nerve-Damage"]

This is unrealistic, not effective; you just added both BW (xdr, gddr3) together.

I will do the same for 360gpu: 22.4 GB/s gddr3 + 32 GB/s to edram = 54.4 GB/s

Effective or peak rates versus real-world performance are two different things. Which I do understand!! That doesn’t negate the fact the RSX has access “simultaneouslyâ€￾ to both of them (XDR & GDDR3); be it a physical hardware link or software allocated data link. And what you did above was kind of lame, since the 10MB EDRAM already sits on the GPU die and the EDRAM memory allocation (10MB part anyway) is only there to support the resolution 1280x720p. And yes I do know the EDRAM supports the supposed free Anti Aliasing and other post processing effects.

 

[hadareud]

http://www.gamespot.com/forums/show_msgs.php?topic_id=24056891

:smile:

Gotta love the internet

 

[ralexand]

I'm confused. Is Int8 considered hdr lighting? I thought RGB was int8.

 

["Nerve-Damage"]

IIRC, the EIB is clocked at half, 1.6 Ghz, at 96 bytes/cycle (128 B/cycle burst) -> 154 GB/sec (205 GB/sec)

This one I will agree upon (for now anyway).

That's the ISSCC 2005 disclosed figure. For PS3, it's 40 GB/sec (35 CELL-RSX, 5 I/O).

Actually you’re quite wrong here…

The flexIO provides 12 channels at 6.2GB/s per channel, hence the 76.8GB/s of peak bandwidth performance. The FlexIO besides connecting the CPU to GPU, and CPU to South Bridge, also provides individual channels between the CPU and Blu-Ray Drive, Blu-Ray Drive to South Bridge, Hard Drive to CPU, Hard Drive to South Bridge, and the other channels are split between the South Bridge and other IO devices such as Bluetooth, ECT. Yes, the Cell has a physical connection (FlexIO channel) between it and the Blu-Ray drive and Hard Drive (Data streaming purposes versus going through the South Bridge). You don’t have to believe me, but know that I know it’s true.

And for comparison sake the PS2 has a 2,560bit data bus system and a peak 48GB/sec bandwidth rate among its GPU, CPU, and other IO devices. The current FlexIO is only 16-bit data wide.

Facts:
PS1: 132 MB/s peak bandwidth
PS2: 48GB/s peak bandwidth (thousand times more than PS1)
PS3: 76.8GB/s of peak bandwidth (little more than 30% gain over the PS2)

Edit: And your figure of 40GB/s only includes the CPU to GPU and CPU to South Bridge which totals 40GB/s already. You seem too have forgotten there are actually FlexIO channels that link the IO devices to the south bridge. So my question to you is; where are the other bandwidth data (numbers) that are allocated for linking the Blu-Ray drive, Hard Drive, Bluetooth, and ECT just too the South Bridge. In other words you only factored in the CPU to GPU and CPU to South Bridge connections. You didn't factor in none of the connections that lye between the IO and South Bridge. And if you chose to believe, the ones between the CPU and two main IO devices (Blue-Ray and Hard Drive). Remember you said 40GB/s

 

[mckmas8808]

So, "Nerve-Damage" you are telling me that the RSX can read from the system memory while at the sametime the CELL reads from the same system memory?

And this can happen with the RSX accessing its own memory giving the RSX 48GB/s of total read and write speed for graphics?

If this is true the PS3 doesn't seem so BW straved as people make it out to be.

 

[j^aws]

Actually you’re quite wrong here…

Nope.

The flexIO provides 12 channels at 6.2GB/s per channel, hence the 76.8GB/s of peak bandwidth performance. The FlexIO besides connecting the CPU to GPU, and CPU to South Bridge, also provides individual channels between the CPU and Blu-Ray Drive, Blu-Ray Drive to South Bridge, Hard Drive to CPU, Hard Drive to South Bridge, and the other channels are split between the South Bridge and other IO devices such as Bluetooth, ECT. Yes, the Cell has a physical connection (FlexIO channel) between it and the Blu-Ray drive and Hard Drive (Data streaming purposes versus going through the South Bridge). You don’t have to believe me, but know that I know it’s true.

FlexIO is a chip-to-chip interconnect bus. Your numbers above are still referring to the ISCC 2005 numbers, like they were referring to 4.6 GHz CELL clocks. The FlexIO B/W for PS3 are,

CELL<->RSX 35 GB/sec (20+15)
CELL<->I/O South Bridge 5 GB/sec (2.5+2.5)

And for comparison sake the PS2 has a 2,560bit data bus system and a peak 48GB/sec bandwidth rate among its GPU, CPU, and other IO devices. The current FlexIO is only 16-bit data wide.

Facts:
PS1: 132 MB/s peak bandwidth
PS2: 48GB/s peak bandwidth (thousand times more than PS1)
PS3: 76.8GB/s of peak bandwidth (little more than 30% gain over the PS2)

You're not comparing like-for-like here.

Edit: And your figure of 40GB/s only includes the CPU to GPU and CPU to South Bridge which totals 40GB/s already. You seem too have forgotten there are actually FlexIO channels that link the IO devices to the south bridge. So my question to you is; where are the other bandwidth data (numbers) that are allocated for linking the Blu-Ray drive, Hard Drive, Bluetooth, and ECT just too the South Bridge. In other words you only factored in the CPU to GPU and CPU to South Bridge connections. You didn't factor in none of the connections that lye between the IO and South Bridge. And if you chose to believe, the ones between the CPU and two main IO devices (Blue-Ray and Hard Drive). Remember you said 40GB/s

Devises like the HDD, Blu-ray, network, basically I/O etc, connect to the South Bridge chip. This chip then connects to CELL via FlexIO.

http://www.vgpub.com/screenshots/2374.html

The 40 GB/sec number for FlexIO includes I/O southbridge and CELL-RSX.

 

["Nerve-Damage"]

So, "Nerve-Damage" you are telling me that the RSX can read from the system memory while at the sametime the CELL reads from the same system memory?

Do I believe it?

Yes!

It’s really no different than Nvidias up in coming Turbo Memory/Caching System…IMO.

But if there is an actual physical link between the RSX & XDR memory module, that would reduce any possible latency problems that might occur if there was only a software driven allocated link.

And this can happen with the RSX accessing its own memory giving the RSX 48GB/s of total read and write speed for graphics?

Real-world performance wise; I say at least 80% percent of that number. Like I said effective and real-world are miles away where graphics are concerned.

If this is true the PS3 doesn't seem so BW straved as people make it out to be.

I’m human just like anyone else…if I make a mistake then I will concede that I did. Anyhow, PS3 isn’t that alien (technological wise) compared to the PS2 hardware.

 

["Nerve-Damage"]

The 40 GB/sec number for FlexIO includes I/O southbridge and CELL-RSX.

Dude, you actually proved my point!?!

You’re correct with the 40GB/s between the Cell & RSX and Cell & South Bridge combination.

But there are physical connections (FlexIO channels) between the South Bridge and IO devices. How do you think the South Bridge receives or communicates with the other IO devices (through empty space)? There is a physical connection among the IO devices to the South Bridge. There is a data rate carried among them…….

You see the orange/yellow arrows between the Blu-Ray Drive and Hard Drive connecting them to the South Bridge. Those are channels! You’re only stating the data rate between the CELL & Southbridge and Cell & RSX which total 40GB/s.

Devises like the HDD, Blu-ray, network, basically I/O etc, connect to the South Bridge chip. This chip then connects to CELL via FlexIO.

But how are they connected my friend?

 

[j^aws]

Dude, you actually proved my point!?!

You’re correct with the 40GB/s between the Cell & RSX and Cell & South Bridge combination.

But there are physical connections (FlexIO channels) between the South Bridge and IO devices. How do you think the South Bridge receives its data from the IO devices (through empty space)? There is a physical connection among the IO devices to the South Bridge. There is a data rate carried among them…….

You see the orange/yellow arrows between the Blu-Ray Drive and Hard Drive connecting them to the South Bridge. Those are channels! You’re only stating the data rate between the CELL & Southbridge and Cell & RSX which total 40GB/s.

You're confused. The SB talks to CELL via FlexIO. The FlexIO is an interconnect bus FROM CELL TO RSX, and FROM CELL to SB. The HDD, Blu-ray etc. connect to SB with their own busses and DO NOT touch FlexIO. Again, FlexIO connects CHIPS, i.e RSX and SB.

 

["Nerve-Damage"]

You're confused. The SB talks to CELL via FlexIO. The FlexIO is an interconnect bus FROM CELL TO RSX, and FROM CELL to SB. The HDD, Blu-ray etc. connect to SB with their own busses and DO NOT touch FlexIO. Again, FlexIO connects CHIPS, i.e RSX and SB.

See thats where you are wrong....

There is only one bus system in the PS3, while the XDR and GDDR have there own specific bus links. Sony stated all IO devices are interconnected by the FlexIO bus system, thus reducing latency seen with multiple bus systems.

Give me a sec, I will find the article……..

 

[j^aws]

See thats where you are wrong....

There is only one bus system in the PS3, while the XDR and GDDR have there own specific bus links. Sony stated all IO devices are interconnected by the FlexIO bus system, thus reducing latency seen with multiple bus systems.

Give me a sec, I will find the article……..

Just look at a PC architecture. They have USB busses, SCSI busses, IDE busses etc, which connect to the SB. The SB and GPU communicate with the CPU on a PCI-E bus...

 

[scificube]

Well...I only have a Sony HDTV. It works so I gues that's a good experience I guess.


Moderator: Post edited because of user stupidity...

Well thanks...I actually tried to edit it but for some reason I couldn't. Glad it's official I'm stupid now

 

["Nerve-Damage"]

Just look at a PC architecture. They have USB busses, SCSI busses, IDE busses etc, which connect to the SB. The SB and GPU communicate with the CPU on a PCI-E bus...

Hence the reason why Sony didn’t go that route………

Link:

The Cell design includes an I/O subsystem called FlexIO. FlexIO is very very fast. FlexIO allows multiple cell chips to be directly connected with no "glue chips". FlexIO will replace all the legacy and current PC IO standards such as PCI AGP ATA USB 1394 serial parallel with one standard. This will eliminate much of the clutter, confusion, and complexity of the PC. FlexIO will greatly simplify the OS.

November 28, 2005

Link:

Monday, 08 August 2005

FlexIO:

The FlexIO is similar to the regular system bus in a system. It handles the data transfer requirements between the Cell and the rest of the system. The Cell has two configurable FlexIO interfaces that are each 48-bits wide with signalling rates in the 6.4GHz band. It provides a mind numbing 76.8GB/s data transfer rate which is roughly 10 times of what’s provided by the Athlon 64s. However, this is split into two parts. The data, which is coming in and is intended for the Cell processor (Coherent Data) and the data that is meant to go from the processor to the rest of the system (called incoherent data).

Anyhow, we would not be surprised if, in PS3, a large chunk of this bandwidth is utilized between the GPU (NVIDIA supplied) and the Cell. In the larger scheme of things, it could also be shared by the I/O devices for some fairly serious connectivity.

Link:

The connection between the Cell chip and the rest of the system will be handled by the proprietary ‘FlexIO’ bus, providing a daunting 76.8GB/s maximum bandwidth. This huge number is broken up considerably by the fact that the ingoing and outgoing data handling capacity of this bus is split into two parts; ‘coherent’ data traffic intended for transmission to other Cell processors and ‘non-coherent’ traffic intended for the rest of the system as in a typical ‘front side bus’ arrangement.

Link:

It is logical to assume that much of the CELL's massive FlexIO bandwidth will be devoted to its interface with the NVIDIA-supplied GPU. As mentioned above, the XDR memory interface uses a 64-bit wide XDR interface with 3.2 Gb/s signalling to provide a peak bandwidth of 25.6 GB/s. The interface width of the DRAMs themselves is actually programmable so the usage of certain DRAMs may not imply anything as to the width of the actual interface and visa-versa. The FlexIO signalling rate is 6.4 Gb/s, by contrast. CELL's nominal supply voltage is stated as 1.0V at 90 nm, which equates to about 3.6 GHz according to this slide from ISSCC. Since the PPE is an in-order core, it could implement vertical threading not unlike that found in Niagara. This would mean the core would switch threads on cache misses, etc. to maximize performance. However, other documents make references to the threading capability as "SMT."

Link:

The Rambus specs are impressive. For memory, the Cell uses Rambus' XDR (extreme data rate), which can provide a total bandwidth of 25.6GB per second. For I/O, the Cell uses FlexIO to pass information outside of the processor.

According to Rambus, FlexIO has a maximum bandwidth of 76.8GBps, giving the Cell's bus a theoretical bandwidth total of more than 100 GBps bandwidth. No PC processor on the market comes close to that performance.

This bandwidth, together with the potential for high gigaflops speed, could make the Cell a graphics powerhouse—which is why Sony is betting the future of the PlayStation platform on it.

Lets say for the sake of argument that I’m wrong with all my information.

Where is Sony statement or an article stating that the PS3 total FlexIO is only 40GB/s of peak bandwidth. Or better yet an article stating the connections between the South Bridge and IO devices are only simple bus interfaces.

Because all my data points to the FlexIO being the main interconnect among all the IO devices. Which makes sense...too improve bandwidth performance, versus using PC variant connections between the South Bridge and IO devices? Especially a System taunting high bandwidth data streaming from the Blu-Ray drive.

Because I remember specifically reading or hearing Ken Kutagari stating that the PS3 FlexIO provides 70+ GB/s of data transfer during one of his press conference.

 

[j^aws]

...
Lets say for the sake of argument that I’m wrong with all my information.

Yep, you're wrong.

Where is Sony statement or an article stating that the PS3 total FlexIO is only 40GB/s of peak bandwidth. Or better yet an article stating the connections between the South Bridge and IO devices are only simple bus interfaces.

I've shown you TWO block diagrams from E3 2005.

Because all my data points to the FlexIO being the main interconnect among all the IO devices. Which makes sense...too improve bandwidth performance, versus using PC variant connections between the South Bridge and IO devices? Especially a System taunting high bandwidth data streaming from the Blu-Ray drive.

Again, the I/O devices DO NOT directly connect to CELL. They connect to SB. It's the SB that connects to CELL with FlexIO. The 6 UNIVERSAL_SERIAL_BUS devices connect to the SB, with their own busses. Your links have confused what FlexIO is. It's the equivalent of PCI-E. The PCI-E bus isn't going to replace USB nor is FlexIO, but they work together. This is basic stuff.

Because I remember specifically reading or hearing Ken Kutagari stating that the PS3 FlexIO provides 70+ GB/s of data transfer during one of his press conference.

Lets use another angle. COMMON SENSE.

I say 40 GB/sec. You say 76 GB/sec.

76-40 = 36 GB/sec

There's no way that the HDD, network, Blu-ray, USB devices etc. are going to need MORE bandwidth (36 GB/sec) than CELL-RSX (35 GB/sec). I'm repeating myself now, so you can believe what you want if that makes you happy...