Sacred 2

If the game is CPU bound on PS3, does that basically mean it's PPU only?

No, not at all. A lot of graphics processing is done on the GPU. We do skinning, parts of the particle processing, shadowmap filtering and acceleration for the deferred renderer on the SPUs. At the end of the day, I think we have 5 SPUs running for a good 10ms per frame, depending on the number of shadowmaps, lightsources, and characters.
I really wished there would have been a few months more to optimize, but if I average it over the time the project ran, we had maybe two full-time PS3 programmers, for maybe 4 man-years of programmer time.
Still, I should have pushed less for 1080p and worked PPU performance more. Hindsight, 20/20, all that.
 
What are the problems you encounter on the PPu and on the Xenon? LHS?
No, it's more a software architecture problem, really. The game was in development for long enough that it is fundamentally a single-threaded architecture in a multi-threaded world. If you want to parellelize such code, you usually surround chunks of your code with locks an run it on a background thread. That works "fine" on the 360, but is simply technically impossible on Cell, which is why we have more CPU issues on the PS3.
If there was time, you could rewrite those parts to make them runnable on Cell and make them faster on Xenon as well. It is not as if the hardware wasn't fast enough for a game like this.
 
Thanks, T.B. for your candid answers.

Are you working on a patch to fix the gameplay bugs? (If you can confirm without breaking NDA)
 
T.B.'s no longer with them, so unless he's moved again, or they're contracting... ;) It's always good to here about the background to a project. Sacred 2 is another good example of the issues of PS3 development from a traditional x86 background. It was a first take from a PC developer, so it's understandable. And sadly they're not around to get a proper crack at a console game. Do we know if the financial returns on the console ports were worth it? From what little I've heard, Sacred 2 has sold quite well. Which wouldn't be unsurprising. This genre is under-represented, and any title of this ilk should expect the keen fans to buy into it!
 
No, it's more a software architecture problem, really. The game was in development for long enough that it is fundamentally a single-threaded architecture in a multi-threaded world. If you want to parallelize such code, you usually surround chunks of your code with locks an run it on a background thread. That works "fine" on the 360, but is simply technically impossible on Cell, which is why we have more CPU issues on the PS3.
If there was time, you could rewrite those parts to make them runnable on Cell and make them faster on Xenon as well. It is not as if the hardware wasn't fast enough for a game like this.
I get the picture when you and your team started to work on the game a while ago with a potent OoO CPU in mind and face problems latter on.
I remember reading stuffs about these issues I did a search and for those interested this presentation "at least we aren't doing that : real life performance pitfalls"
from Gamefest 2008 is still available on MS site.
OT
Actually there are funny bit in this presentation like:
about Block on thread syn
If (ThisProject.NearEndOfProject (cEpsilon))
{
thisProject.ToOptimizeElsewhere ();
thisteam.CrossFingers ();
engineTeam.QueueTask (RedesignMulticore System);
}
or threading models
All threading models are equal
-----Well... yes and no
----some are much more equal than others
/OT

I hope Sacred2 did correctely I saw it peaks in some European country console charts, I hope it will be enought sale to justify a rework for our joy, past some bug and perf issue the game is a really good (and time consuming) hack slash from what I read here and there. Fixed I think that the game could show legs but I think Ascaron should act pretty fast or the game will die at least in the console realm.
And again thanks for yours comments insights and honesty it's really refreshing in our gamer "PR crushed" world ( :LOL: )
 
Last edited by a moderator:
T.B.'s no longer with them, so unless he's moved again, or they're contracting... ;) It's always good to here about the background to a project. Sacred 2 is another good example of the issues of PS3 development from a traditional x86 background. It was a first take from a PC developer, so it's understandable. And sadly they're not around to get a proper crack at a console game. Do we know if the financial returns on the console ports were worth it? From what little I've heard, Sacred 2 has sold quite well. Which wouldn't be unsurprising. This genre is under-represented, and any title of this ilk should expect the keen fans to buy into it!
I think it's more the "used to fast single threaded OoO CPU" that hurts.
From my understanding and what I read making good use of the xenon or the cell is not that different but try to grow an engine developped for a single thread/core cpu in to a multi thread one won't give you any good results (mostly worse on the PS3). Things has already gotten better as multi core are now standard. Sacred 2 is a huge game development have started a while back as TB hints at this time dealing with mulitcores was not the devs team focus.
 
Well, that is assuming you do the downscale separately and not in combination with that last post-process step. ;)
There are actually some pretty neat things in that renderer. Doing a deferred-pass in 1080p with 13 lightsources and 9 shadowmaps at the speed that we're doing it...
Indeed knowing that the CPU is the bottleneck it's clearer that your team managed something pretty impressive on the graphic side of things.
I hope a rework of the multithread engine is on its way, through expansion that kind of game can show impressive legs. I hope Ascaron will find enough founding to do the job or they will limit themselves to the pc realm. I see nothing coming than can compete with sacred 2 in the console realm, they have their chance to be THE hack&slash of this gen (blizzard is not interessed in competing in the console realm). If they fix the engine soon they will gain the ability to amortize their initial investissement (+ fixing) through years and multiple expansions.
Especially as this kind of economic model (selling expansion as DLC) is pretty profitable as it forces users to buy original games and it limits the impact of second hand market as Ascaron still have the chane to have its share of this market through DLC.
 
I see nothing coming than can compete with sacred 2 in the console realm, they have their chance to be THE hack&slash of this gen.
Genre fans are hoping Snowblind Studios will finally release something for this gen of the level of technical achievements they managed on PS2 (4 player local and online coop, 2x supersampled AA, multiple shadow-casting lights, best-of-class particles, procedural water deformation, fairly varied and balanced classes and gameplay). Current rumours are of a LOTR title.

Oh, another annoying bug I've remembered in Sacred 2 that stil messes me up, the direction of the player-position arrow on the map isn't at all accurate. It's supposed to show the way you are facing, but is broken. Thus you can see your character facing south down a road, you can walk down the road, and then check the map to see you've been travelling north instead of south, though the map still reports you are facing south. Navigation was hard because of this, and now requires reference to the compass and reminding oneself the map direction is meaningless, which is pretty hard when you see on the map that your character is pointing down the road you want to travel down. It's very unintuitive to ignore that information and refer to the compass position. Just another annoying bug that made its way not only past QC and into the game, but also wasn't addressed in the last patch.
 
OK, I'll just reply without quoting, as iPhone cut&paste would drive me mad.
First of all, as Shifty said I am no longer with Ascaron and neither is the rest of the console team. The other console guys started a new studio a few weeks back (Gaming Minds), which is not affiliated with what is left of Ascaron. I cannot comment on wether or not they will be hired to make a patch, as I quite frankly don't know. What I do know is that I have not been contacted. I am, however, off the market till October and they know that. Feel free to speculate wildly. ;)
As for financial returns, I guess it depends on how you fudge the numbers. If all you attribute to the "costs, PS3" is the platform specific costs, I am sure it was hugely profitable. If you split the costs to develop the PC version among the three platforms, it's a lot less pretty. All based on publicly available sales numbers, which is all I have.

Oh, and liolio, thanks for the kind words about the PS3 renderer. It's pretty cool what you can do when you write your own Direct3D. ;)
 
OK, I'll just reply without quoting, as iPhone cut&paste would drive me mad.
First of all, as Shifty said I am no longer with Ascaron and neither is the rest of the console team. The other console guys started a new studio a few weeks back (Gaming Minds), which is not affiliated with what is left of Ascaron. I cannot comment on wether or not they will be hired to make a patch, as I quite frankly don't know. What I do know is that I have not been contacted. I am, however, off the market till October and they know that. Feel free to speculate wildly. ;)
As for financial returns, I guess it depends on how you fudge the numbers. If all you attribute to the "costs, PS3" is the platform specific costs, I am sure it was hugely profitable. If you split the costs to develop the PC version among the three platforms, it's a lot less pretty. All based on publicly available sales numbers, which is all I have.

Oh, and liolio, thanks for the kind words about the PS3 renderer. It's pretty cool what you can do when you write your own Direct3D. ;)


Well it's hard to speculate without knowing what part of the world you're working, however it would be cool to know if you've gone back to the PC or stayed in Uncharted territory. :LOL:


So when you say "PS3 is the platform specific costs, I am sure it was hugely profitable" that means you and the other PS3 dev didn't make the millions $$$ so often quoted as the costs to develop on that platform?
 
Genre fans are hoping Snowblind Studios will finally release something for this gen of the level of technical achievements they managed on PS2 (4 player local and online coop, 2x supersampled AA, multiple shadow-casting lights, best-of-class particles, procedural water deformation, fairly varied and balanced classes and gameplay). Current rumours are of a LOTR title.

At least one of their games was canceled. I'm not sure what they're doing now.

Oh, another annoying bug I've remembered in Sacred 2 that stil messes me up, the direction of the player-position arrow on the map isn't at all accurate. It's supposed to show the way you are facing, but is broken. Thus you can see your character facing south down a road, you can walk down the road, and then check the map to see you've been travelling north instead of south, though the map still reports you are facing south. Navigation was hard because of this, and now requires reference to the compass and reminding oneself the map direction is meaningless, which is pretty hard when you see on the map that your character is pointing down the road you want to travel down. It's very unintuitive to ignore that information and refer to the compass position. Just another annoying bug that made its way not only past QC and into the game, but also wasn't addressed in the last patch.

If I remember correctly, it's just that the map is always fixed and does not rotate. I agree that it's not the most intuitive implementation, but damn you are quite the whiner when it comes to this game.
 
No I think he's having the same problem that intermittently affects the PC version also. The arrow for your character on the map is supposed to rotate with your character. But sometimes it gets stuck pointing North only no matter what direction your character is going.

Regards,
SB
 
No I think he's having the same problem that intermittently affects the PC version also. The arrow for your character on the map is supposed to rotate with your character. But sometimes it gets stuck pointing North only no matter what direction your character is going.

Regards,
SB

Ok, I'll check it out tonight and see if I have the same problem.
 
At least one of their games was canceled. I'm not sure what they're doing now.
Bought by Warner Brothers who have stated they are going whole hog on LOTR games which would fit in with SB's pedigree. It's interesting you say a game has been cancelled, as my theory is they were repostioned after being purchased, dropping whatever they were creating to be published by Activision (before the Blizzard merger), and rejigging it to make a LOTR game. This is a crying shame though. Must be demoralising to work years on a game only to have it pulled and be set onto something else. At least now they have a home and can get down to creating games instead of managing business decisions.

If I remember correctly, it's just that the map is always fixed and does not rotate.
In the large map screen, the player arrow rotates to show the direction the player is facing,. ell, it's supposed to...
I agree that it's not the most intuitive implementation...
No, it's a bug ;)
...but damn you are quite the whiner when it comes to this game.
Yep :devilish:. I was asked for bugs, so I'm giving them. If there weren't so many, I wouldn't have much to say, would I :p. And if they were more whiners on the Ascalon team, we likely wouldn't have these bugs either. Quality needs attention to detail, not an attitude of 'well, they can get by. It's good enough'. Some of these bugs are 5 minute fixes, so why do they still exist after release and a patch? I can understand the pressures of getting a game out, getting some money, etc. but these days bugs should be fixed by patches, and the PS3 patch was firstly late, and secondly fixed very little. Okay, things are tight. Give them a break. Live in hope of later patches fixing things...oh, they are working on DLC instead of fixing the game!
 
No, not at all. A lot of graphics processing is done on the GPU. We do skinning, parts of the particle processing, shadowmap filtering and acceleration for the deferred renderer on the SPUs. At the end of the day, I think we have 5 SPUs running for a good 10ms per frame, depending on the number of shadowmaps, lightsources, and characters.
I really wished there would have been a few months more to optimize, but if I average it over the time the project ran, we had maybe two full-time PS3 programmers, for maybe 4 man-years of programmer time.
Still, I should have pushed less for 1080p and worked PPU performance more. Hindsight, 20/20, all that.

Thanks for this and the other answers, T.B. It's really cool to hear about these things. So you actually went for the SPU GPU assist, and I like the idea of making your own Direct3D renderer which you then power by SPU and GPU combined.

Yeah, those few months of extra polish would have been great, though, especially considering that after the initial whining about bugs and other review niggles, I'm hearing a fair bit of good word from people actually playing the game, including on some of the bigger podcasts out there. Typically they mention it is one of the very rare games their spouses will play with them (always good), and they seem to be enjoying the game despite its flaws. As said above, considering that the chance there will be any form of competition for this game in the short term, I definitely think investing a few more months could pay off over the next two-three years.

Particularly on the PS3, this was quite a popular genre, made successful on the PS2 by the excellent Baldur's Gate and Champions of Norrath games (as we all know).

Genre fans are hoping Snowblind Studios will finally release something for this gen of the level of technical achievements they managed on PS2 (4 player local and online coop, 2x supersampled AA, multiple shadow-casting lights, best-of-class particles, procedural water deformation, fairly varied and balanced classes and gameplay). Current rumours are of a LOTR title.

You forgot Snowblind's version of the mega-texture ... :) Grandmaster should do an interview with these guys and figure out if they came up with it before Carmack. ;)
 
Bought by Warner Brothers who have stated they are going whole hog on LOTR games which would fit in with SB's pedigree. It's interesting you say a game has been cancelled, as my theory is they were repostioned after being purchased, dropping whatever they were creating to be published by Activision (before the Blizzard merger), and rejigging it to make a LOTR game. This is a crying shame though. Must be demoralising to work years on a game only to have it pulled and be set onto something else. At least now they have a home and can get down to creating games instead of managing business decisions.

A friend left there after feeling pretty down about the cancellation and shift in focus.

Yep :devilish:. I was asked for bugs, so I'm giving them. If there weren't so many, I wouldn't have much to say, would I :p. And if they were more whiners on the Ascalon team, we likely wouldn't have these bugs either. Quality needs attention to detail, not an attitude of 'well, they can get by. It's good enough'. Some of these bugs are 5 minute fixes, so why do they still exist after release and a patch? I can understand the pressures of getting a game out, getting some money, etc. but these days bugs should be fixed by patches, and the PS3 patch was firstly late, and secondly fixed very little. Okay, things are tight. Give them a break. Live in hope of later patches fixing things...oh, they are working on DLC instead of fixing the game!

There's no such thing as a five minute bug. Maybe you can identify a bug in five minutes, but then you have to fix it, build it yourself, test it yourself, get it code reviewed, checked-in, have a build submitted to QA, let QA smoke it, etc., etc. For an engineer, I don't think you can really look at any task as taking less than three hours, just because of the focus factor. But, when you bring the rest of the process into it, not to mention the added overhead in releasing a patch on a console platform, small things can become very expensive very quickly.

Typically, what happens is QA will find a bug, reproduce it, and submit it to a database (like DevTrack or Bugzilla) with a classification. Engineering time is allocated to the bug queue based on severity and time. All A-level bugs will need to be fixed before ship, but there are tons of smaller bugs that will never be fixed because there simply isn't time or money.

The fact that Ascaron only barely managed to ship this game before going bankrupt is a testament to how far their back was against the wall and just how much triage had to be done to ship the game.

I remember one project where the lead programmer had over 300 bugs in his queue at ship. He was the lead, so he got a lot of bugs assigned to him by default, and some of them were significant, but none were ship-blockers. I think I had about 20 at that time, but none were severe. Mostly they were little features that got deprioritized.

On another project, my team was roped into fixing bugs for another game that was close to shipping and they were in danger of missing their release date. A lot of them were workflow bugs, to help speed up automated smoke tests and so forth -- interesting that they were put off until the end, but that's the way things are when you get close to the finish line.
 
A little LTTP (well, I did have to import to play it in Australia, sigh). This is an incredibly addictive game! Possibly the most fun I've had on my console in a good long while, actually, with that "just one more hour" vibe working well.

I have a Shadow Warrior as a pure summoner to level 48 or so, and have rolled a bunch of other classes for pure solo play. I'm struggling to find one I adore, though.. the classes just really don't seem to gel with me too well. As an example, I was loving my High Elf (pure caster) until I got to a boss I simply couldn't damage due to high resistances. After a good amount of sighing I gave up and rolled an Inquisitor who feels terribly underpowered, and dies a lot. My twin-bladed Seraphim (for melee goodness) also is taking a long time to hit good decent form... so I'm a little lost. I hate dying in these sorts of games, since I'm not really reckless... when I die I feel like the character/build failed, as opposed to me doing something I shouldn't have. The dryad looks ok but I don't typically like "archer" classes... and the temple guardian is a support character, which isn't too fun for solo play.

Any tips for good build guides or whatnot? Most of the ones I've read on forums are fairly high-level, saying "choose x when you hit 150" leaving you to struggle early on. This wouldn't bother me too much but your choices here are so very critical - and you simply can't respec. If they wanted to patch anything, let me freaking respec my character.

Anyway... definitely worth a purchase, as much as I'm complaining. What classes are you guys really digging?
 
Back
Top