PDA

View Full Version : More X360/PS3 comparison.


jeffool
05-24-2005, 03:54 AM
With the fervor flying about the comparison of X360 and PS3 specifications recently, I've followed a few links and came up with something I found rather interesting. Jerome is a software engineer at Microsoft with a blog and in a recent post (http://spaces.msn.com/members/jeromeontech/Blog/cns!1p-jTa0g3LlkmfZHpjXBlqXw!298.entry) he corrected Major Nelson's recent X360 v. PS3 charts.
Major Nelson did a great job in his article (http://www.majornelson.com/2005/05/20/xbox-360-vs-ps3-part-1-of-4/). I just really thought that the charts for general CPU performance and memory bandwidth comparisons in his article were not fair. Everything else was very accurate, and you should read the rest at his site (http://www.majornelson.com/).Of course Jerome's a heck of a guy. Someone in the comments pointed out an error Jerome made, and he posted (http://spaces.msn.com/members/jeromeontech/Blog/cns!1p-jTa0g3LlkmfZHpjXBlqXw!351.entry) that on his front page as well.

bapenguin
05-24-2005, 04:19 AM
This is interesting.With the criticisms I must also praise your article. You did not spend much time commenting on the fact that the PS3 will come with 19308 Bluetooth channels, 1093801 usb ports, etc. This is very wise, you must be aware of the fact that Sony does not intend to ship the final product with that many ports, it would not be cost efficient. Sony merely needs to express to the user what the system is 'capable of' and then feel the water to see what users actually desire. The final version is going to ship with 4 Bluetooth controller channels (with the option of 4 headsets), 2 usb ports, 1 hdmi output port, and a single Ethernet port. All of the other designs were mere concepts, that would otherwise drive the cost of goods for the console way into the red :)

I wonder how true it is...

bapenguin
05-24-2005, 04:55 AM
Kinda on this topic, I just notice that in the Anandtech (http://www.anandtech.com/tradeshows/showdoc.aspx?i=2423&p=2) article that ATI states their GPU will support 1080p w/ 4xAA at no penalty. ATI did clarify that although Microsoft isn't targetting 1080p (1920 x 1080) as a resolution for games, their GPU would be able to handle the resolution with 4X AA enabled at no performance penalty.

Morratut
05-24-2005, 05:12 AM
I just can't wait to see real life in game screens and videos now.All this speculation is making my head hurt.

Borys
05-24-2005, 05:21 AM
Nevermind.

bapenguin
05-24-2005, 05:22 AM
USB ports cost like 0.01$ each that's not really a problem.

I think the Bluetooth/HDMI thing is more interesting. They were touting support for 7 controllers...but it looks like it MAY only support 4.

Hewie
05-24-2005, 05:28 AM
I think the Bluetooth/HDMI thing is more interesting. They were touting support for 7 controllers...but it looks like it MAY only support 4.

I believe it supports 7 devices: 4 contollers and 3 over devices (ie. dvd remote, web cam, etc.)

...I could be wrong though.

Vandenh
05-24-2005, 05:29 AM
>4X AA enabled at no performance penalty

This is a big thing. If PS3 has to do this with a cost the 360 might well end up being the superior graphics console. Will people be able to see the 4x AA difference? Time will tell...
Anyway.. brilliant ATI design.

crackeriah
05-24-2005, 05:50 AM
>This is a big thing. If PS3 has to do this with a cost the 360 might well end up being the superior graphics console. Will people be able to see the 4x AA difference?
Good point Vandenh. This won't make a huge difference so long as people have Poor-Man's Anti-Aliasing enabled (i.e. a crappy standard def TV), but in the "HD era" it will make a big difference.

TRiLoGY
05-24-2005, 06:19 AM
Good old ATI!

I've got an ATI Radeon X800 PE in my pc.. I find it hard to spot the difference between the different AA Ranges though!.. EG 2048x1536 2xAA looks virtually the same as 8xAA..

PacerDawn
05-24-2005, 06:49 AM
I think the best part of this is that he is a Microsoft employee with a gmail e-mail address:

If you want to contact me directly, feel free to e-mail me at my public mail address, jeromeholman at gmail.com

Oops...

kokyunage
05-24-2005, 06:52 AM
Kinda on this topic, I just notice that in the Anandtech article that ATI states their GPU will support 1080p w/ 4xAA at no penalty.

Quote:
ATI did clarify that although Microsoft isn't targetting 1080p (1920 x 1080) as a resolution for games, their GPU would be able to handle the resolution with 4X AA enabled at no performance penalty.

I don't believe that. In the same article they stated that when the 10 Megs of dedicated ram wasn’t enough for anti aliasing there is an efficient algorithm to compensate for that. So, to me that sounds that in high HD resolutions they need to swap into the system ram for anti aliasing. No matter how efficient theeir algorithm is, there will still be slowdown.

WileE.Coyte
05-24-2005, 07:03 AM
I've come to the realization that there are about 10 people on the planet that can understand ,Microsoft, and Sony math. Four are employed by Sony and four are employed by Microsoft, and the other 2 are laughing their ass off, watching us trying to figure it out.

Rommel
05-24-2005, 08:17 AM
Gotta give the guy credit though. He is not shy about saying who he is and who he works for. More than some in the industry who accept payment can boast.

I for one am not excepting any numbers about which system is stronger at this point. I understand how the 360 works and I can appreciate how powerful it is. The cell is a horse of a different color. I am not willing to decide if it is a true break through in computing yet (Though the design concept probably is) until I see a working unit being taped and measured by a unbiased third party. Thusly, I am staking no beliefs in which system is powerful except to say at the very least - they will be stronger than the Revolution.

RandomViolence
05-24-2005, 09:11 AM
I don't believe that. In the same article they stated that when the 10 Megs of dedicated ram wasn’t enough for anti aliasing there is an efficient algorithm to compensate for that. So, to me that sounds that in high HD resolutions they need to swap into the system ram for anti aliasing. No matter how efficient theeir algorithm is, there will still be slowdown.

I was under the impression that the 10 megs of EDRAM was just enough for 1080i running 4xAA. Please correct me if I'm talking out of my ass, but doesn't 1080i have to render a full frame the same as 1080p, so if the system can do 1080i with 4xAA at no hit, shouldn't it be pretty much exactly the same to do it at 1080p?

kokyunage
05-24-2005, 10:24 AM
I was under the impression that the 10 megs of EDRAM was just enough for 1080i running 4xAA. Please correct me if I'm talking out of my ass, but doesn't 1080i have to render a full frame the same as 1080p, so if the system can do 1080i with 4xAA at no hit, shouldn't it be pretty much exactly the same to do it at 1080p?

I was under the same impression until I read the article. It seems like there are situations when the 10megs isn't enough:

ATI has also implemented a number of intelligent algorithms on the daughter die to handle situations where you need more memory than the 10MB of DRAM on-die (anandtech (http://www.anandtech.com/tradeshows/showdoc.aspx?i=2423&p=2) )

Tyrant
05-24-2005, 01:08 PM
I think the best part of this is that he is a Microsoft employee with a gmail e-mail address:



Oops...

Or maybe, just maybe, he wanted people to contact him at his PUBLIC email address, that wasn't an internal microsoft email addy.

bobbler
05-24-2005, 01:19 PM
I was under the impression that the 10 megs of EDRAM was just enough for 1080i running 4xAA. Please correct me if I'm talking out of my ass, but doesn't 1080i have to render a full frame the same as 1080p, so if the system can do 1080i with 4xAA at no hit, shouldn't it be pretty much exactly the same to do it at 1080p?

The 10mb of eDram oddly enough fits a 480p frame perfectly (I think). Any higher resolution requires tiling (a few passes). There is a performance hit, but its only ~5% (supposedly) or so at worst (one of the ATI guys said that) so they say its free (it may as well be if its sub 5% at worst).

chechenepiphany
05-24-2005, 01:20 PM
Or maybe, just maybe, he wanted people to contact him at his PUBLIC email address, that wasn't an internal microsoft email addy.
just maybe

PacerDawn
05-24-2005, 01:39 PM
Microsoft Hotmail and Google Gmail are in competition. When a Microsoft employee chooses Gmail over Hotmail for any reason, that speaks to people about the quality of service of Hotmail (as in, it's not as good as google, otherwise, why wouldn't he have a hotmail addy?) As an employee of a company, as long as you are identifying yourself as an employee of that company, you are representing that company, whether you are on your off hours or not. Microsoft could see this as a conflict of interest (a Microsoft employee endorsing a competing product) and fire him.

Hence the "oops".

Achilles
05-24-2005, 04:00 PM
Plenty of MS employees have a private Gmail account. They have it because of all the server space that it provides. They won't fire them. You may be shocked to know that some MS employees even run Linux on their home computers or mod their Xboxs and the company doesn't care. It's a pretty easy going place really, I've got no idea why people think they're such a tyrant.

mister_slim
05-24-2005, 09:26 PM
I still don't think that memory bandwidth should be counted, but I can't be bothered to find the right article. Bleargh...

bobbler
05-24-2005, 10:22 PM
I still don't think that memory bandwidth should be counted, but I can't be bothered to find the right article. Bleargh...

Well it shouldn't, they aren't beans.

The 10mb eDram is cache, for all intents and purposes (as far as the CPU sees it, it can't directly read the eDram). So counting it towards general bandwidth doesn't make sense in the least. Its akin to counting the memory bandwidth inside the each SPE in the cell (it has 256k sram per SPE that can oddly be accessed by anything, so it would be even more valid than counting the eDram -- Theres like 400gb/s bandwidth for each of the blocks of sram, and its very low latency).

Its marketing FUD. Even adding general bandwidth doesn't tell you much -- you can have 1gb/s read and 47gb/s write for all that article tells us, which would suck just as much as 5gb/s read/write for many applications. It doesn't tell you anything useful, especially not when its twisted.

mister_slim
05-25-2005, 07:05 PM
I'm just going to call you Mr. Encyclopedia. Thanks.