-
Hey Guest. Check out your NeoGAF Wrapped 2025 results here!
ARS Technica 360 Article - Part 2 now up
- Thread starter 3rdman
- Start date
Razoric said:
Vince can sometimes come across as agressive, but the points he makes are usually based on more than just speculation.
And gofreak's recent tech posts have been as enjoyable to read as Pana's recently. Again, some commonsense to the melee can't be bad.
Simple fact it, there is little information out there on PS3 architecture. Anything discussing specifics is currently conjecture (such as massive hits, DMA scheduling clashes etc).
And there have been two major articles quoted recently - this one and Major Nelsons, that seem to use spurious reasoning and often contradict each other in places.
monkeymagic
Banned
SteveMeister said:
The majority of videogame consumers aren't too concerned with an online service.
And by majority I mean the 65M or so that bought a PS2 over an XBox.
It's perfect for games like 1st and 3rd person shooters though which is why I suspect it's so popular on XBox since the majority of its games seem to be of that ilk.
DarienA said:
Thats MS' weakness IMO - over-reliance on XBL.
Reality is that about 2% of the overall console userbase is an XBL subscriber. Thats with MS pushing with all its might.
They need to make sure they don't take their eye off the ball with Xbox 360, otherwise all they are doing is a nice big public focus group for PS4's online service...
Lord Error
Insane For Sony
Interesting article, I'd like to know the credibility of the guy who wrote it.
One point that stuck out is his assuredness how SPEs on Cell will not be usable for physics calulations. I think that goes directly against what people who are supposed to be programming physics engines for Cell are saying. Ageia/Novodex people (authors of the PPU chip which is an architecture that shares some important simillarities with Cell) seem pretty sure they will make their physics API work really fast on Cell. I think the same kind of statement came from guys from Epic who want to include Novodex API in their U3 engine.
I think part of the problem is that Novodex/PPU is right now the only multithreaded physics engine, but it also happens to be really good. So, I'm not sure if it's that this type of architecture doesn't lend itself to physics calculations or it's more of a stuck-up oldschool thinking that needs to change while the actual architecture is good.
One point that stuck out is his assuredness how SPEs on Cell will not be usable for physics calulations. I think that goes directly against what people who are supposed to be programming physics engines for Cell are saying. Ageia/Novodex people (authors of the PPU chip which is an architecture that shares some important simillarities with Cell) seem pretty sure they will make their physics API work really fast on Cell. I think the same kind of statement came from guys from Epic who want to include Novodex API in their U3 engine.
I think part of the problem is that Novodex/PPU is right now the only multithreaded physics engine, but it also happens to be really good. So, I'm not sure if it's that this type of architecture doesn't lend itself to physics calculations or it's more of a stuck-up oldschool thinking that needs to change while the actual architecture is good.
Razoric said:
I'll give you Vince (I have yet to see home post anything contradictory to the Sony PR, he's like Major Nelson for Sony), but gofreak has been pretty level headed about everything being thrown around. It's pretty hard to seperate speculation from fact when these debates are being thrown around, especially because so much about the CELL and it's performance in certain tasks is unknown.
How about some context?
Remember those researchers that build a supercomputer out of 500 PS2s networked? Basically they needed to code (in linux) all their research stuff by taking advantage of the vector processors.
So they broke everything down into vector form to get the speed needed.
You'll simply need to do the same next gen. Take advantage of the massive power at your disposal, and not moan if your lazily ported x86 code runs slowly.
Remember those researchers that build a supercomputer out of 500 PS2s networked? Basically they needed to code (in linux) all their research stuff by taking advantage of the vector processors.
So they broke everything down into vector form to get the speed needed.
You'll simply need to do the same next gen. Take advantage of the massive power at your disposal, and not moan if your lazily ported x86 code runs slowly.
Marconelly said:
Is it the only one? Havok are on X360 and PS3, and I'm sure I heard them talking about exploiting the multi-core architectures of the new consoles. Havok generally has always had a much better rep than Novadex, and I'm sure their high standards will remain going forward. Novadex has really only risen to prominence out of its association with the Physx chip, but as engines go, Havok is tough to beat, AFAIK.
sonycowboy
Member
SteveMeister said:
And what does Microsoft have?
1) First Party titles (Sony seems to have a substantial edge in 3rd party exclusives currently)
2) Live
This seems like an odd statement.
Are you of the belief that there is no power differential between the two systems? And I don't mean theoretical. We've heard numerous rumors that the PS3 is more powerful than the 360. So either you don't believe the hype (from a real-world application standpoint) or you don't think it'll make a difference next generation.
You're saying that Sony only has 2 things going for it?
1) Exclusive titles (both first party and the major third party franchises)
2) Brand Name
but what about?
3) Power advantage
4) Blu-Ray (using movies as a way into the living room and more storage)
monkeymagic
Banned
sonycowboy said:
I don't think the inclusion of Blu-Ray can be underestimated.
The potential market for it over the next 5-7 years across Japan, US and EU is probably 10x that of an online service (that is even if Sony don't have their 'Playstation World' up and running for PS3 release).
midnightguy
Member
I am no techie, but it seemed clear to me that this Ars article was not very good. I'd like to see a counter-article to appear that refutes some of the things said.
Live isn't that big of a factor now, in terms of userbase, but broadband penetration is simply bound to get bigger and bigger, and Live is MS simply having a long term vision. I think the next 5-10 years will see a significant growth in broadband penetration, therefore increasing the importance of a reliable online service.
Panajev2001a
GAF's Pleasant Genius
Rastex,
judge me the way you want, attack ad-hominem do not scare me. Attack what I say, it should be much easier
.
Yes, I do think aaaa0 has more programming experience than I do (which does not mean I haven't done shit at all though), but if you expect me not to reason and abandon conversations when the "god" speaketh, you are obviously thinking wrong.
judge me the way you want, attack ad-hominem do not scare me. Attack what I say, it should be much easier
.Yes, I do think aaaa0 has more programming experience than I do (which does not mean I haven't done shit at all though), but if you expect me not to reason and abandon conversations when the "god" speaketh, you are obviously thinking wrong.
monkeymagic
Banned
midnightguy said:
I've been reading Ars for the last 4-5 years and Hannibal (the author) is no slouch.
That said, most of what he has written about CELL with regards to its physics/AI abilities has been conflicted by developers working with it such as Epic Games.
IJoel said:
It will also have an impact in the early adopters, since at least for the moment it's pretty much the hard-core set that does Live. If there are several killer Live games at launch then most are likely to follow quickly. (PD0, PGR3, Madden, etc.)
Live should be a good 2.5m by then; not huge but it could contribute to a strong first quarter or two.
monkeymagic said:
Agreed, I haven't so much of a problem with his hardware analysis as some of the conclusions or strong suggestions he makes regarding consequences for certain workloads. He's right, though, if you take "normal", "traditional" code from a desktop processor or whatever, and run it on either of these processors, it's likely not going to fly. I guess people didn't expect that to hold for X360's CPU anywhere near as much as for Cell, but there've been rumblings for a bit about that from various people so perhaps it's not so surprising..
sonycowboy
Member
IJoel said:
I expect broadband to be HUGE this generation. Over 60% of internet acess is now broadband and it's expected to be over 80% by then end of 2006. SBC just announced they'll be offering DSL for $14.99/month, so the ISP wars may be coming back.
I also expect Sony to know this and to have a substantial online service this generation. And I think the basics of Live willl be there. One username, friends, matchmaking, etc. You know the Live 1.0 stuff. I'd also expect them to offer some alternative services and infrastructure that Microsoft doesn't. I don't expect them to outLive Microsoft by any means, but I think it's clear with the hardware (3 Gigabit ports, WiFi, detatchable HDD) that online is an important part of Sony's PS3 plans.
seismologist
Banned
I didn't realize alot of this about Xenon. I'm starting to think it will struggle at running straight PC ports. And I'm having doubts that X360 will be an a decent replacement for a high end PC.
Gofreak, Pana,
My comments weren't meant as attacks, though I guess re-reading what I said it did come out harsher than I intended. Hell, I'm a student as well. My point is that there is a huge wealth of knowledge that comes from writing game-specific code over a number of years that you just can't get anywhere else.
My comments weren't meant as attacks, though I guess re-reading what I said it did come out harsher than I intended. Hell, I'm a student as well. My point is that there is a huge wealth of knowledge that comes from writing game-specific code over a number of years that you just can't get anywhere else.
Hannibal missed the mark with this article. I know that sounds wierd, considering the first part was good, and I like endorsed him as a good source for info. But a lot of the stuff he said in this part of the article goes against what we've heard from devs online and in interviews. Not to mention it would contradict the multi-million dollar decisions both Sony and MS made with their chip designs. It's one thing to have Sony or MS wrong. But both of them? I don't know about that. Physics should be very much an important part of both machines. And there was some talk about Xenos possibly getting into the act.... AI and game control are another story, but both machines should be quite capable in their current configurations. I'd like to see some more devs respond to this, b/c it just seems like Hannibal is approaching this particular situation with the wrong mindset. PEACE.
gofreak said:
Yep, Havok has already announced their support and moving of their entire line of products to PlayStation3 and claim they will "optimize" them, for whatever that's worth: Havok PR
Hah. If I were to see Sony doing the same as MS did with Major Nelson, I would comment in defense of their design. The closest they've come is Ken Kutaragi's comment about a samurai or some shit... which is more of a WTF-kinda comment; I refuse to post about who is more like Jet Li and who is more like... say... Chuck Norris.
I've already stated that the X360's GPU is quite elegent, but it would seem that post-E3 the majority of argument is against the PlayStation3... hmm.And, unfortunatly, I don't get paid. LOL.
Panajev said:
I suppose that's what happens when you're out like Lothario instead of working
shpankey said:
SPEs aren't good for branch intensive code. You can run it on a SPE, but you'd want to avoid branches where possible.
Hannibal mentions "branch-intensive" physics and AI code etc. won't run very well on X360 or PS3's CPUs.
A lot of physics code isn't branch intensive or doesn't necessarily need to be. Parts that are need to be treated carefully - that's where you "do a little jig".
AI is a fuzzier point, it really depends on what you're doing - there's lots of different ways of going about AI.
Hannibal left the suggestion that X360/Cell etc. wouldn't be good at physics and "non-graphics" code but I've trouble agreeing with that. Maybe the branch-intensive aspects wouldn't, but that's a different point, and wouldn't be entirely compatible with the suggestion that all non-graphics code wouldn't work very well. If AGEIA or Epic or whomever says that Cell will be great for physics, I'd trust them on that. It was one of the target workloads for Cell.
monkeymagic
Banned
Razoric said:
Was GoW running on any kind of X360 hardware?
Because it looked light years beyond everything else that was supposed to be running on 1/3 alpha dev kits.
shpankey said:
In his defense though, he did make mention that to get the most out of this chip would require more work from the devs...this is where XNA comes in and (hopefully) it provides the tools to make these games sing.
I get the impression that many of those who are arguing it about this stuff are looking at it from a PC point of view and (like you said) aren't thinking outside the box.
Panajev2001a
GAF's Pleasant Genius
YellowAce said:
WAVES HANDS FRANTICALLY!!!!!!!!!!!!!!!!!!!!!!!
Please god of bi-CMOS engineers, hook this lowly Kutaragi servant up
.sonycowboy said:
Not only that, but broadband over powerline (BPL) may become more of a widespread reality very soon. A company in Manassas, VA is already offering commercial service for $40/month @ 3.5Mbit down; and 1Mbit up. Their residential service is 300-500Kbps and it only costs $29/mnth. The real boon will be to all of the people who live out in the hinterlands who have no access to cable or dsl access. Broadband over satellite is not really a viable alternative ATM because latencies are so high.
The point is, we're finally starting to see the some benefits from the 1996 Telecommunications overhaul. We have all of these various media companies from different sectors: cable, telephone, and power utilities who will ultimately be vying for the same $$.
Ain't competition good.
As far as Sony's online plans go, they have to implement a Live-like service. As well, they would be stupid NOT to charge for it. Sony could use the revenue stream right now; and getting that reliable monthly fee from X-million subscribers is awfully compelling.
Too many people short-sell this aspect of gaming because of where it currently stands. To all those lads who doubt, look to the future.
Spectral Glider
Member
So let's see, to sum up:
Games may look awesome, but maybe not as awesome as some demos. Or maybe they will later on, just not at launch.
Gameplay may suffer, or it may not seeing as programmers are working with closed architecture
Gee, thanks a whole fucking lot, "tech experts."
Games may look awesome, but maybe not as awesome as some demos. Or maybe they will later on, just not at launch.
Gameplay may suffer, or it may not seeing as programmers are working with closed architecture
Gee, thanks a whole fucking lot, "tech experts."
Panajev2001a
GAF's Pleasant Genius
rastex said:
I have done what I can
: GBA (HAM programming library), PS2Linux (SPS2 based development for a TOOL like experience [this ad was brought to you courtesy of Sauce and Sparky]), some OpenGL for the PC... I am thinking of picking up C# and doing managed DirectX to get some experience using DirectX 9 features, but my heart waits for a Linux kit-or similar for PlayStation 3 so I'd better use Cg for Shaders-writing than Microsoft's HLSL.True, working in the industry you pick up a few things first hand, but even in the sidelines you might be lucky enough to get to chat with such smart and friendly people and learn a few things
. The good thing of having an educational background, doing your own research, programming and discussing what you learn helps you understand and not only remember by heart.CrimsonSkies
Member
Agreed.
seismologist
Banned
You have to remember that Alot of programmers look at these things from a pretty high level and are used to relying on a compiler for optimizations.
It's like driving an automatic vs. stick shift.
sure there will be middleware to handle low level coding if needed but the top devs will want to take advantage of the added flexibility.
from what they said at E3, they do. Any 3rd party who wants on their network can be, and all games, not just ones on Sony's network, need to allow for friends lists and communication type features. They haven't shown actual GUI's or anything, but then the system is still a ways further off than X360.
As a long time Ars member (6 years now, though I haven't been there in a while), the articles are always written by qualified, knowledgable people. For this article though, I'm a bit questionable myself, but thats because a) the system analysis is such a spec sheet reliant article and b) Ars is largely a PC site and I don't think he's looking at either system from the perspective of a system or game engine programmer who'll actually be working on it.
Like the PS2 last generation, both of these systems will require a lot more specific coding than your normal PC program. Its going to really throw people for a loop, and I think it'll drive even more teams towards middleware. Epic has to be drooling now that EA has alienated so many companies away from Renderware.
sonycowboy
Member
:lol :lol :lolPanajev2001a said:WAVES HANDS FRANTICALLY!!!!!!!!!!!!!!!!!!!!!!!
Please god of bi-CMOS engineers, hook this lowly Kutaragi servant up
Pseudo judo
Member
Drek said:
Link?
I haven't seen anything at all regarding what Sony intends to do with PS3 online. The most I've seen is a PR rep saying it will be almost as comprehensive as Live.
anotheriori
Member
sounds like ps2 all over again, basically it can be a bitch to make ____ work the way you want (or used to doing), but it can be done. Like any hardware, it can be made to shine (nights/vf2, mgs3 etc) you just have to work with what you've got. The best games will utilize the system's strengths and avoid its weaknesses.
basically what i got out of it:
x360/ps3 good at graphics (ps3 > x360 by some margin)
x360/ps3 bad at physics/ai/game logic (x360 > ps3 by some margin)
physics/ai/game logic made to run 'like graphics' will probably be the solution.
kinda like how ps2 forced people to stream stuff.
basically what i got out of it:
x360/ps3 good at graphics (ps3 > x360 by some margin)
x360/ps3 bad at physics/ai/game logic (x360 > ps3 by some margin)
physics/ai/game logic made to run 'like graphics' will probably be the solution.
kinda like how ps2 forced people to stream stuff.