It looks like Microsoft has released a comparison article about the Xbox 360 and PS3 specs and how they compare. While I doubt that any AnandTech reader would even begin to take a Microsoft-authored article about Xbox 360 to be unbiased, I have received a number of emails asking me to analyze the comparison and provide some direction as to what is real and what isn't.
I've spent a good portion of this weekend doing just that, and I'm hoping to start publishing my findings either here as blog entries or as a series of Insider Stories on the front page beginning tomorrow.
If you have any specific questions about the analysis done by Microsoft (you can read it here) or any other spec-related questions, post them here and I'll try my best to address them in the article.
Take care.
I've spent a good portion of this weekend doing just that, and I'm hoping to start publishing my findings either here as blog entries or as a series of Insider Stories on the front page beginning tomorrow.
If you have any specific questions about the analysis done by Microsoft (you can read it here) or any other spec-related questions, post them here and I'll try my best to address them in the article.
Take care.
Facebook
Twitter
RSS
32 Comments
View All Comments
Creathir - Wednesday, May 25, 2005 - link
Where is the skepticism of Sony's numbers? Everyone speaks of them as gold...Just playing devil's advocate. I really find is sickening in how Sony is always looked upon as this "credible source"... these are the same people that brought us MiniDisk and BetaMax... they also have always inflated their numbers considerably. I do like Microsoft, I won't be ashamed to admit it, but I don't take their numbers at face value either. I look to Anand and such sources for a non-biased view. However, many other sites are talking of Sony's numbers as if they are straight from God's mouth... and the "war" is already over. I hate to burst people's bubble, but I hardly believe this to be the case...
As I’ve always said, the true winner here will be the gamer… only good things will come out of stiff competition.
clauzii - Tuesday, May 24, 2005 - link
Ups - THIS article:http://www.realworldtech.com/page.cfm?ArticleID=RW...
clauzii - Tuesday, May 24, 2005 - link
This CELL-article is quite indepth on the core structure - quite technical though ;)JoeS - Tuesday, May 24, 2005 - link
Anand,In the PS3 DocOc demo (http://media.ps3.ign.com/articles/615/615000/vids_... Sony showed off an impressive demo combining satellite imagery with depth mapping which generated earth’s terrain for simulated flight in real-time. It was composed entirely within the Cell processor. The RSX GPU was not used at all.
Does this provide any insight into a comparison of each systems CPU potential?
Does this provide any insight into the potential of the Cell/RSX combination versus the X360’s CPU/R500 combination?
nowayout99 - Tuesday, May 24, 2005 - link
MS spinning the media? I'm shocked. The content has been pretty well disected as containing half-truths, assumptions, opinions, taking things out of context, and trumpeting hollow victories (i.e. total system bandwidth and "general" processing power)eric - Tuesday, May 24, 2005 - link
Hello,I am looking forward to your analysis. A few things you might touch on.
1) I assume a multi-threaded game might have separate threads for things like path finding, collision detection, AI, physics, etc. Would you expect, in general, a cell SPE or one of the xCPU hardware threads to perform better for such work?
2) How much computation/bandwidth will be taken
up in the RSX if it performs 4xAA?
3)Given the information you have available, which GPU do you think is superior?
4)Both Sony and MS presented maximum performance
figures. Which system do you expect to be able to closer achieve their respective limits?
Thanks,
Peter Kelly - Tuesday, May 24, 2005 - link
My biggest complaints with the article were:1) X360 CPU = 3x PS3 CPU for general purpose performance - can you really just throw away the SPEs for "general purpose" tasks (whatever they are?). There's integer performance on there too, at least. I was under the impression that the SPEs could run general purpose code, even if it wasn't optimised for it - assuming your "general purpose task" can be split over multiple cores, a X360 core would have to be 3x faster than a SPE in order to even be as fast as the PS3 GPU at that task (you could match off one PPE against one X360 core pretty fairly, I think, leaving 6/7 SPEs to face off against 2 X360 cores).
2) The bandwidth comparison - as others have pointed out, surely the comparison is flawed if over 90% of your bandwidth is to ~2% of memory? Versus 100% of bandwidth to 100% of ram? Counting edram bandwidth like this - especially if it's just between the edram's own logic and its memory - puts us on a slippery slope where we could be counting cache bandwidth or SPE bandwidth, I think.
3) The GPU comparison - seems to be based completely on assumptions about the RSX architecture in the absence of details? Seems to ignore scalar performance on the nvidia side in his Gflop calculation? Assumes that nvidia's shader op figure includes all the things he adds into Xenos'?
I'm glad Anand his having a look at this! So much FUD, imo. One thing I'd also like to see him touch on, if possible, is the processing requirements of games - if speeding up certain areas gets you a disproportionate speedup overall. I was under the impression that physics took a MASSIVE proportion of CPU time in a game - surely if you can even just speed up that, you'll see a big benefit overall (isn't this why we have physx on the PC side)? MS is talking in nebulous terms about "general purpose" processing - but what exactly does that break down to? Game logic? Do we really need 3 cores for just game logic? I guess I'm wondering - who's right about what types of processing need optimising?
clauzii - Tuesday, May 24, 2005 - link
Is one of the beautiful things of the SPEs not that the 256KB actually can act as both cache and memory, ie. SPEs containing cache instructions??"Game programmers do not want to spread their code over eight processors, especially when seven of the processors are poorly suited for general purpose programming. Evenly distributing game code across eight processors is extremely difficult."
Isn´t this statement dependent on how good development tools the amount of ´control´ these tools give/have on the SPEs??
dwell - Tuesday, May 24, 2005 - link
I'd love to hear your interpretation of MS document Anand. I started reading beyond3d's take on the document when it first surfaced on majornelson.com. I can't pretend to understand the content at their level, hopefully you will break it down for the layn00b.blup23 - Tuesday, May 24, 2005 - link
Someone did a little analysis of MS analysis, if anyone's interested:http://forums.gamespot.com/gamespot/show_messages....