Yesterday’s Gamelab had Mark Cerny as a special guest. Cerny is the lead architect on the PlayStation 4 and is one of the important figures behind the development of Sony’s next gen console.
At one point during his presentation he talks about how he came to the conclusion that the PS4 did not needed a Cell processor but something more standard i.e. the x86 architecture.
“After the PlayStation 3 shipped, the hardware team immediately started working on reducing the manufacturing costs of the console. And in 2007 they also began a post modern analysis of PlayStation 3. What had worked and what had’nt, with the goal of beginning of the basic direction to take with the next console,” he said during the presentation.
He talks about how he had two ways in approaching the CPU for the PS4. Either make the Cell more powerful or go for the x86 architecture.
“In other words this was the initial step in creation of PlayStation 4. And for the first time in Sony Computer Entertainment’s history, this process was inclusive and collaborative, the ICE team, the other tools, the other technology teams were invited to participate. Now the most obvious path though was to continue using CELL. Though the learning curve was steep, it was clear that mastery of SPUs was leading to some amazing titles like Uncharted: Drake’s Fortune, so it was an definitely was an option to continue using CELL. Perhaps even enhancing it to make it more powerful or easier to use. But there were other options too, we can move to a more conventional architecture with just a CPU and a GPU. If we went with that approach, choosing a CPU will be a very big deal because that will determine which vendor we work with and thereby determine pretty much every aspect of our project. Timeline, business structure, development cost in a way. So there were two main possibilities for the CPU, the power PC architecture that we used in the PS3 and the CELL, and the x86 architecture used in pretty much every modern PC.”
“The conventional wisdom as expressed by the number of first party game programmers was that the x86 architecture was unusable in a games console. Now I took this very seriously and thought that a console needs to be competitive with a much more powerful PC and if game programmers need straight forward power PC architecture to make that happen. I understand. I just wondered whether this is true, cause the potential design space for the PlayStation 4 was going to pretty limited for using the x86 CPU.”
He then talks about how he spent his holidays researching 30 years of x86’s history and finally concluded that this is something that programmers can embrace.
“So I spent my 2007 holiday in researching the 30 some year history of the x86, all the way from its creation in the 1970’s to the recent enhancements. My conclusion was that the conventional wisdom had been correct but progressive enhancements by Intel and AMD have finally resulted into something console programmers can embrace. And then I started thinking. I just sacrificed my holiday to investigate some philosophical points of a console that I am not really assigned to be working on and won’t be releasing for another 5 years. That’s passion, that’s enthusiasm and may be I should consider working on this project more deeply.”
He then talks about his skill sets and expertise and how he met Shuhei Yoshida and Masa Chitani to eventually become the lead system architect on the PlayStation 4.
Quite a story, right? What do you guys think?
The 49 minute long presentation can be seen here.
Let us know your thoughts in the comments section below.