I though that game had the terrible draw distance which is the exact opposite of Lair?
SPU usage in games
- Thread starter betan
- Start date
Having completed N3, I can say that the massive army scenes were a bit of a joke, you initialy had large numbers on screen but as soon as the action started you ended up fighting maybe ten different npc's whilst all around you there were little dummies doing the same thing over and over again. Also if you snook round the game in different ways you'd meet blocks of soldiers etc that just stood there doing nothing. All in all absolutely nothing to brag about.
For those that are questioning Jobe's remark of using 7 SPE's. For those that have played the Warhawk beta (before the patch was applied) you may have noticed that the game would consistently crash (I won't bore you with the details, but it had to do with a bug in the low level core code running on the SPE's) and, thus, hang the PS3. Normally, the PS3 can recover from game crashes by returning control to the O/S, but when a game explicitly takes over the O/S SPE (7th) like Warhawk does the PS3 cannot recover from a game crash and your forced to perform a cold start on your PS3.
Blade47167
Newcomer
So is that why it started happening even more often when the 1.8 update was released? Did they clean up the OS enough that it no longer needed the 7th SPU? After 1.8 even with a few updates the game was still very sensitive and seemed to lock up quite a bit
Hopefully they get that worked out.I think most would agree that "percentage power" quotes have no scientific value.
I interpret them as how many SPUs are used, and even though that doesn't really tell anything about efficiency, it is still interesting how PS3 games are evolving.
Do you have any reference for the bug? Please bore us with the details.
In any case it doesn't really tell anything about how many SPUs are used.
No scientific value, but as long as you trust the source not to be playing PR games, you can take them as the instinctive response of a master tradesman. When you work very closely in an industry for a long time, you get a general feeling that usually leads to pretty accurate estimations. From being in the business for so long, they'd have an inclining, a cop's hunch, as to how well they're doing on Cell.
I am having difficulty calling "1/3 of PS3 power" an accurate estimate even with all the world's experience. And even if Cell is "hunchable", I don't see any definition of "power" yet alone a meaningful and more importantly measurable one.
Also 1/3 power claims coming from developers are suspiciously consistent with Sony's president's "1/5 power" claim for first generation PS3 games.
A third the power isn't accurate as in you can comfortably claim it's a third of what's available, but it should be a measure of how much room they think they have to go. That is, they know they've got lots of leg room, and they're not going to have to squeeze extra performance from somewhere. And likewise, they appreciate the cap on available performance and know they're not going to be able to multiple everything they've got in games by 10x.
As for power, it's a measure of what they can do with their games. This isn't a console v console metric, but a console on it's own metric, so it's measured according to its own standards. Of all the data they can shift and numbers they can crunch on it, they'd be looking to triple that in future.
Suspiciously consistent? Could that be because they're all measuring the same thing and getting the same result? If I have a piece of string, and my boss estimates it as 12" long, and I estimate it's about a foot, does that mean I'm just copying my boss, or we're both looking at the same piece of string and using our own experiences of length, coming to the same estimate?
That's probably the most unfortunate drawback with going exclusive as a developer, whether first or second party. People assume conspiratorially that every thought that passes through your head is implanted by the console owner. Why is it so questionable that these estimates are in the same ballpark? Do you think developers have already maxxed out PS3, mastered Cell, and nothing will improve much? That where Sony lie that there's loads of headroom, the truth is far more restricted, but no-one will openly admit it?
I wanted to create a thread to keep track of how SPUs are used in games.
From Q&A with Dylan Jobe of Warhawk[Official PS Blog] :
"4. What did you do on this game that you couldn’t do on another platform?
It’s hard to answer this and not sound like a gratuitous SONY sales pitch
Although I would say it’s the sum-total of all of our natural phenomenon in the game. Our clouds, procedural water, atmospheric scattering, terrain, etc. All of this stuff runs in parallel on all 7 SPUs simultaneously every frame – I’m still not sure if the game community is giving enough credit to just how fast the SPUs really are."
Of course the number 7 is a little puzzling.
I didnt know that they needed 7 SPUs to do this.
I guess it must be the volumetric clouds that demands a lot
Or maybe it is the fire effects
Anyways it seems to be a fun game.
Yes, I am sure Sony's President have the same "sixth sense" with Naughty Dog/Incognito developers and made an accurate power estimate that put Resistance and Genji 2 at the same level.
I mean if you want to believe that, be my guest. I will remain skeptical for the time being.
Though I have no idea how difficult to render those procedurally, I think some real life rain clouds look similarly blurry on the edges.
In any case, Warhawk's menu background clouds look realtime, are above and beyond anything out there, pretty much photo-realistic and I really don't like that term.
I assume they are realtime because I haven't seen any repetition or compression artifacts.
Sarcasm aside, don't think so.
With match-making and more dedicated servers, it may very well be the best multiplayer experience out there, IMO of course.
Let me tell you this way, when the beta first hit it was crashing so regularly some had to restart their PS3s every 5-10 minutes and completing a match was almost impossible. That didn't stop anybody.
So what exactly is your view? That ND+Incognito are using 80% of PS3's performance, games won't improve a great deal, and when they said 30% they were doing what they told by corporate HQ?
Claiming a power utilization percentage from nowhere is pretty much PR shit.
Of course, the point of those claims is not really educating public on numbers, but really conveying the same idea I happen to believe, that is first/second even third party Cell utilization will improve significantly over time. Obviously they think putting a number out there makes it more convincing.
Currently all we can have is some rough SPU numbers.
Once for example Lair 2 comes out and does unbelievably better progressive mesh along with other tasks, we can safely say Lair 1 sucked at Cell usage and Factor 5 is getting better and better.
When that time comes, the other generic PR numbers will start disappearing or will be replaced by measurable ones (polygons count, character count etc.)
Surely the edge tools give feedback on how much as a percentage the Cell's are being used/unused. Also PS2 has similar software for developers to give them feedback on CPU usage.
Obviously 30 percent or even 50 percent is an estimate on where, either they think they are (educated guess) or a gauge in what something like the egde tool is telling them (or a combination of both). Also people should bear in mind that with these percentages/estimates that is based on current code and usage.
They might also be referring to code usage to. I.E. that given the learning experiance on the PS3 if they had to do it all again the code might run faster (as in everything the more you do the better you get to know your subject and the quicker and more efficient you become at the task).
While benchmarks have hit about 98 percent of theoretical maximum (gflops) of the Cell. I would be astonished if the games are leveraging anywhere near that level (anywhere near 80 percent). Simply due to the fact that games are a much more complex beast (an understatment) than doing matrix tables or FFT's. Many of these benchmarks also state that if they refined their code the could probably get a better performance than stated (for example Barry Minor's Julia Ray Tracing Fractol).
Obviously 30 percent or even 50 percent is an estimate on where, either they think they are (educated guess) or a gauge in what something like the egde tool is telling them (or a combination of both). Also people should bear in mind that with these percentages/estimates that is based on current code and usage.
They might also be referring to code usage to. I.E. that given the learning experiance on the PS3 if they had to do it all again the code might run faster (as in everything the more you do the better you get to know your subject and the quicker and more efficient you become at the task).
While benchmarks have hit about 98 percent of theoretical maximum (gflops) of the Cell. I would be astonished if the games are leveraging anywhere near that level (anywhere near 80 percent). Simply due to the fact that games are a much more complex beast (an understatment) than doing matrix tables or FFT's. Many of these benchmarks also state that if they refined their code the could probably get a better performance than stated (for example Barry Minor's Julia Ray Tracing Fractol).
Betan don't put it past devs who code for this architecture to be able to ballpark the level of Cell they feel they are using and paint that usage in percentage terms; with access to profiling tools and a good sense for the methodologies, it's not too difficult for someone to get a sense of what they could be doing differently to extract more performance from the architecture, or have a sense of what theoretical headroom might be.
Obviously the metric is arbitrary, but that doesn't make it a metric without value.
Obviously the metric is arbitrary, but that doesn't make it a metric without value.
Yep, they are called profilers.
I didn't want to go into that, but theoretical maximum would be a reasonable (not necessarily meaningful) definition of "full power" . In that regard do you think Uncharted is using Cell around 30 percent of max?
Getting a sense is different though.
While I think every programmer can have some rough idea of bottlenecks or weak points of its code, for better or worse expected improvements and reality rarely fits in the same range.
Even for a single core single thread Risc-like architecture, its not easy to come up with numbers because the software itself has its own difficult to measure efficiency metrics, independent of its efficient usage of the hardware.
If you really want to quantify things best you can do is to look at operation count and compare it to maximum, but clearly that's not what they are doing. Of course operation count by itself never shows how efficient you are.
No argument there. The question is whether those numbers are influenced by PR or not.
However they're coming up with these numbers, it clearly isn't tied to the number of SPUs used.
To take it to an extreme, tou could be using all SPUs and not be using a lot of Cell's power. It depends how much you're stressing each SPU. Warhawk, for example, simultaneously claims to be using all 7 SPUs, and around a third of the Cell's power. Whatever about how the latter is measured, the two statements aren't incompatible with each other.
Similar threads
- Locked