*Does the same* Except I use an actual joystick, not a control pad. It really gives you a clear and clean control advantage over those trying to fly a plane with a mouse and/or keyboard. In the helicopters, the advantage isn't that much more, but it's more fun :smile:
*spin-off* idTech Related Discussion
- Thread starter Rangers
- Start date
*Does the same* Except I use an actual joystick, not a control pad. It really gives you a clear and clean control advantage over those trying to fly a plane with a mouse and/or keyboard. In the helicopters, the advantage isn't that much more, but it's more fun :smile:
Silent_Buddha
Legend
Well, other than the unified shaders, edram, and other changes, yes, it's "somewhat" similar.
But it's a far larger deviation from R520 than RSX is from G70.
Which just makes all the hoo-hah about Carmack being unable to come to grips with the PS3 just amusing. As it's the part he has the most familiarity with and the most knowledge about that is currently having speed problems.
And the part that is the most console centric and he's the least familiar with he's had no problems getting good speed out of.
Regards,
SB
It's about the overall integration. The developers have to learn the SPU part (including the memory access/DMA characteristics) before they can get great performance out of PS3, RSX or no RSX. People are very familiar with using HDD, memory, and optical discs too. Doesn't mean they can put together an optimized PS3 solution just because they know how to program individual parts.
Also, it may not be Carmack who's doing the PS3 programming. And they probably need more than one dude.
Also, it may not be Carmack who's doing the PS3 programming. And they probably need more than one dude.
Silent_Buddha
Legend
That's exactly it, he's had no problems with the Cell CPU. Getting greater performance out of it than the X360 CPU. So it can easily be inferred that he's getting good use out of the SPU's as I think it's generally agreed that the general purpose unit of the Cell is rather less powerful than the 3 cores on the X360 CPU. In other words, he has to be using the SPU's fairly well to be getting greater perf out of Cell.
It's ONLY the GPU that he's having problems getting speed out of for ID Tech 5. And apparently just one aspect of the GPU. Nothing else about PS3 (from his comments) is causing him any problems.
Regards,
SB
He has problems with Cell. He keeps b*tching about it, especially regarding its memory access, size and the asymmetric parallelism. It is not an easy platform to work on. However, he has no problem with the theoretically higher specs.
Getting good use out of SPU is different from getting sufficient level of use and performance out of SPU (and the PS3 memory framework) for specific tasks. He stated explicitly that they still have to do more offloading.
Check his old interviews about Cell.
They might have solved some of the problems now, but he definitely has (past) issues with Cell and the memory policies. At the end of the day, if he can get 60fps working on PS3, does that mean that he has no problem with the GPU and Cell at all then ?L. Scofield
Veteran
Complaining about something and actually having problems with it are different things.
Do we know how well RAGE performs on a PC with a 7800 and 256 MBs VRAM?
How do we know iD are fine extracting performance from Cell? Carmack has grumbled about it, and according to this recent misplaced, out-of-context, poorly clarified quote, they're getting the same performance from Cell as Xenon, which we know shows isn't good use of Cell. This is also iD's first ever attempt to design for Cell. Are you expecting them to have mastery over it in contrast to every other developer out there?
I'm not even sure what the argument is now! iD are making efficient use of PS3 because we know they are great at using the Cell and the GPU is even closer to a PC part than the CPU is, so they must have even better mastery of the GPU, yet still RAGE is 20 fps on PS3, proving the console is The Pants?!
Probably around 20-30 fps
Is being the pants good or bad? I thought pants were good.
Silent_Buddha
Legend
Well, I was taking from his comment clarifying the statements made with regards to the article that they are getting better speed (with regards to Rage needs) out of Cell than they are currently out of Xenon. And that the problem holding back rasterization speed on the PS3 is limited to the RSX.
Regards,
SB
Well... they are having problems meeting their own goals for the PS3 platform today.
He also mentioned the need to offload more stuff to the SPUs.
Carmack doesn't strike me as a whiny guy despite all his comments. To be fair, Cell is different and requires explicit memory handling. And it's not uncommon for people to have problems with it. They say they can meet their target... so just give them time and we'll find out. Again, Carmack may not be programming PS3 personally.
More bluntly, it is the norm for multiplatform developers who have a responsibility to make the best product possible on many platforms within a limited timeframe. The PS3 version of these games, be it GPU or CPU issues, has routinely been an issue for developers to get parity performance out of the PlayStation3 platform within the budgets publishers alot. Carmack's statements in 2006 before the PS3 launched have been a pretty accurate summary of the results this generation (multiplatform developers struggling to get parity, with parity and above only being really tapped in general by internal studios):
Fast forward, his references to more theoretical power should be understood in this light. For a multiplatform developer you need to ask (a) how much extra effort to get a functional build? (b) how much extra effort to get parity? (c) how much to get that extra amount and if this route (d) is this going to force negative decisions on the other builts.
Every project is different with different bottlenecks, but Carmack really nailed the general issue way back in May 2006 as a general industry trend to follow. I wish Carmack would have put it bluntly, "There is theoretical performance, and then there is realistic expectations of performance within given budgets." Theoretical specs are worthless. Thankfully id Software believes that they have plenty of time to bring their PS3 build to parity--which is very important to them as the goal of IDTECH5 is middleware.
Alas, all of this isn't so different from what Gabe Newell said long ago as well, "The PC and the 360 are just more straightforward. We can focus on what we want to do, which is make game experiences, instead of sweating bullets over obscure architectural decisions they make with their platform. [...] I didn’t come into this business in the 90s because of some technical fetish. I came in because I wanted to give people experiences that made them have fun."
I can see this perspective as a business owner with a vision, as every week you have a programmer spending trying to manuever around the hardware to extract enough performance to get the build up to snuff is a week that developer cannot add features that actually make the game fun.
When the top 2 selling PS3 titles are multiplatform (GTA4, CoD4) and 5 of the top 10 (Assassin's Creed, RE5, CoD:WaW) the perspective of the multiplatform dev is pretty relevant to how "fast" hardware is and its relative merits in the business environment.
I think the post mortem of this generation won't be so much, "How multiplatform developers were able to tap the extra potential of the PS3" but "How much effort it took to get parity." Talk of "theoretically more power" will be squarely tempered with "the performance developers could extract within budgets." Significantly different things, especially for an established industry with legacy code. But even that said people starting with fresh products still have to hire folks and are at the end of the day forced to produce a product, not benchmarks. These will be lessons applied by any company wishing to compete in a market where multiplatform development is significant to their strategy.
Haven't unified shaders already made it into more recent (i.e., more advanced) PC GPUs already?Well, other than the unified shaders, edram, and other changes, yes, it's "somewhat" similar.
Also, EDRAM is a boon to developers, an extra luxury and performance boost, no matter how you slice it. The fact that it's "different" from tradition PC architectures is irrelevant; there's no technical downside to its presence, as opposed to the RSX not having any of it at all.
Yes, the Xenos is more powerful than the R520. That's precisely the "larger deviation" you're talking about.
Did you have seen firsts multi on the ps3 (but also some last productions) ?
On the ps3 is very rilevant. However I think is ot and to persist in this way hasn't much sense. I suspect the purpose of this argument is only one but maybe I'm too malicious....
Last edited by a moderator:
If the developers can build on top of previous PS3 code base and experiences, increasingly more developers and middleware providers should be able to exploit the PS3 goodness. It won't happen quickly. I remember nAo or DeanoC mentioned this evolution in a series of stages in the Heavenly Sword days.
People were skeptical about Blu-ray at first. Now some are exploiting it. With the advancement of first party titles like KZ2 and Uncharted 2, the multi-platform developers will also benefit from tech sharing.
EDIT:
For Post-Mortem, I'd be disappointed if "360 and PS3 feature parity" is the only thing aimed and highlighted. There are a lot more to learn this gen, including new techniques/practices and the potential to enter new segments/markets.
Two problems: (1) the target is moving as the competition (360, PC) also see improvements (see titles like FM3 versus FM2) and (2) game dev cycles are typically short and the window for making a meaningful impact during the prime part of the generation is narrow.
Operative word: Some. Having something bigger, faster, and better is always a positive if all things are equal. The problem is Blu-Ray wasn't "all things being even." Technical, market adoption, and cost negatives aside (which is sweeping a lot aside) from the practical standpoint content costs money and next gen content creation cost more right out of the gate. DVD9 is a limiting factor but looking at multiplatform titles it is far more often than not the PS3 with the larger optical medium and standard HDD that has been found lacking in multiplatform titles.
So while some exploit Blu-Ray, most have been frustrated by other memory issues (like RAM).
Like most things it isn't whether something has merits to some, but if it is the right decision for the current market. Per the point of my post (multiplatform development) the reality is content costs money and you are going to have to compress your content anyhow to get it quickly into memory. A large, expensive storage medium doesn't solve a lot of the problems many multiplatform developers have.
If it has, it isn't showing up on screen.
Rage may be an example, ironically, where Blu-Ray benefits the PS3 version. There are always exceptions (e.g. Burnout), but these serve to re-inforce the general rule, not negate it.
And with Halo 3 and the NXE MS did the same thing in terms of technology deployment.
As the generation slugs on all developers will get better (better code base and practices, more experiences staffed, tech sharing, better performing middleware, etc) but to this point (2009) its hasn't been a panecea to cure the PS3 ills. It has eased the issue (improved quality, fewer delayed and cancelled projects, etc) but it isn't a magic bullet.
The fact multiplatform titles continue to lag and developers still express the frustration and effort required to get near parity quality products to market (with all the tech sharing, Sony aid, better legacy code and practices, more experienced staff, exploiting and designing more with the PS3 in mind) is pretty indicative that paper specs don't really translate directly to complex projects with millions of lines of code that must be ran on multiple platforms.
If no one knew how many processors were in these devices or how many math ops they could do a second or how much bandwidth they had and were judging the software by what they see on screen and what multiplatform developers have to say, I don't think we would see so many threads with the tone they have.
Whatever way we cut it (you are going to disagree) what isn't disagreeable is that Carmack was extremely accurate in 2005&2006 in predicting the issues multiplatform developers would face with the PS3. It could all be their fault, too, but it doesn't change the fact that Carmack actually knew what he was talking about--which indicates that he shouldn't be so quickly slagged off when he makes comments about how some 3rd parties feel at this point.
Well, that wasn't really the context or the absoluteness of my point was it ...
But whatever!Silent_Buddha
Legend
I'm not sure how much tech sharing will help Multiplatform devs.
After all, the goal of multiplatform is to basically have one engine that can serve multiple platforms. UE3, ID Tech 5, and Cryengine 3 for example. Or various in house engines like Capcom/UBIsoft/etc. have.
How much effort will be put into specialized efforts for the PS3? Rather than just trying to get PS3 to be approximately similar to X360/PC?
It appears to me it'll always be, implement new features on X360/PC...then make PS3 feature compatible if able.
Regards,
SB
After all, the goal of multiplatform is to basically have one engine that can serve multiple platforms. UE3, ID Tech 5, and Cryengine 3 for example. Or various in house engines like Capcom/UBIsoft/etc. have.
How much effort will be put into specialized efforts for the PS3? Rather than just trying to get PS3 to be approximately similar to X360/PC?
It appears to me it'll always be, implement new features on X360/PC...then make PS3 feature compatible if able.
Regards,
SB
People will get more ambitious as they move along, hence the target is moving. FM is an exclusive title like Uncharted 2. As far as exclusive titles go, I don't think PS3 is lagging behind at all in terms of visual.
It's a 10 year cycle. We are still a long way off from tapping the potential of both consoles. As long as experiences and code base are "stackable", people will get better at managing and exploiting the extra PS3 advantages.
Even Carmack's team trying to farm off more jobs to the SPUs is exploiting such advantage as we speak. When they get ambitious for the next rev, they can consider offloading yet more to the SPUs as a natural next step. They are already on their way to optimize for PS3.
In this case, Id ran into DVD9's limitation. Ask them why they need 2-3 DVDs. Carmack has also openly acknowledged that Blu-ray has tangible benefits. where economical and technical points are considered. So it is possible for developers to take advantage of PS3's strength.
As for other aspects, it's a long learning cycle. Not all will make it due to the learning curve, but the best studios will attempt.
... and learn to manage it over time. That's the key point. I think the devs have come a long long way since the beginning.
It is still a strength that a cross platform developer has taken advantage of despite all the negative points you highlighted.
... because the game is not out yet ? Plus the advantage is not only presentation. No disc swapping, plus cheaper and more flexible for developers to publish large content when they feel like it.
Sure ! But the headroom is there and is getting used 3 years in, contrary to blanket statement about PS3 advantages not getting exploited this entire gen.
We already knew multi-platform developers will lag behind first parties since day 1. People were talking about it right here in this forum. Again, the point is even the cross platform developers will exploit PS3's advantages over time.
Who disagreed with that notion ? I have cited the same thing when PS3 first launch. It is common sense.
EDIT:
It includes source code, SDKs and tools.
Not strictly. PS3 developers need to offload stuff to SPUs more aggressively. They are being forced to do aggressive multi-threading and specialized code for PS3 even though they want to use the same code base for both.
Plus Blu-ray's performance characteristics is different. So they'd need to implement mandatory install or no-install. Those are all different and potentially extra work.
Similar threads
