• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Merry Xmas Pottsey

“As said before, thats all well and good but i have yet to see a "high physics" game which looks as good as the current titles who dont use them,”
Are we talking about graphics or physics? If its physics then the PPU games do have better looking physics then the current none physic titles. It sounds to me like you don’t like the graphics in PPU games so are saying the physics are not worth while from the PPU. But the graphics have nothing to do with the PPU.




“THAT is the issue, even the link you posted looks pretty poor for the reasons previously highlighted”
I don’t agree with the reasons highlighted as they are wrong. The physics are far better then what the CPU does. You say the debris looked just like games today. Then show me that much debris that doesn’t fade away and has as much interaction all done on the CPU. It’s the amount of physics that’s impressive.
Unless you prove otherwise what you say is wrong. I don’t believe the CPU can do those physics without massive slowdown.





“Everything maxed so my game was crawling, so if your theory was correct the CPU should be getting hammered, however”
Why would you CPU get hammered from physics in a low physics screen? Apply the high physics ini tweak, set off a nuke then show me the CPU not getting hammered.





“So at presant physics cards are not worth buying because any games using them do not look particularly impressive”
So UT3 doesn’t look particularly impressive? The speed increase form the PPU isn’t worth while?
 
Just to brush this whole "Crysis CPU limitation" bull aside once and for all, here are the graphs for my Q6600 @ 3.2GHz running a loop of Crytek's own CPU benchmark. The benchmark as most of you know involves *a lot* of blowing up vehicles and buildings etc. with rockets and all sorts.

8f3n4sj.png


Granted it's not a nuke going off, which Pottsey will undoubtedly exclaim, but it does show that Crysis is pretty much 99% GPU-limited and that even physics performance in this particular title is completely dependant on the CPU waiting on the GPU.

My CPU isnot breaking a sweat, so a dedicated PPU would be useless in Crysis even if the support was there.
 
Ulfhedjinn said:
I don't think he's trying to say that Crysis supports PPUs, that would be nuts. :p

He's trying to say that Crysis runs so slow, particularly when a lot of physics is going on, because of the CPU. Nothing to do with the game being super GPU-limited apparently. :o
Click to expand...

I know very little of Crysis :o - Got the game, only played the demo.

Lol I agree. With the load shown above it's not looking like it's taxing the CPU with nothing left to run physics which would explain the low FPS. That's not what I see from above as well as numerous other posts that it's not taking quad core as expected, is it even utilizing dual core 100%?.

I played the Crysis demo with the GTX and the 2900Pro. Wasn't really impressed with High although playable. Just reminded me of a vamped up Far Cry with nothing too special to show. Now Timeshift looks great and runs like silk. That's why I know nothing about Crysis (Get the PPU and test it :o oops). I've bought the game (Crysis) but if Conroe systems are finding it difficult then I'm going to feel the burn just that little bit more. Crysis is for my next card. Tiomeshift is fun and all the other titles run well so I'm just leaving this game for now :(.
 
“Crysis is heavily GPU limited at this time, the devs and nvidia have acknowledge an issue with performance on high end gfx cards”
That’s true. But during physics when the GPU is doing no work the game becomes CPU limited. Have a light easy screen for the GPU. Apply normal physics everything’s ok. Apply high physics with the ini tweak with the nuke and the game grinds to a halt. Proving its CPU limited when it comes to physic.





“Just to brush this whole "Crysis CPU limitation" bull aside once and for all, here are the graphs for my Q6600 @ 3.2GHz running a loop of Crytek's own CPU benchmark.”
Doesn’t count as the hign end physics I am talking about are not in the benchmark.

I fully admit Crysis is GPU limited. But its also CPU limited during high physics.
 
Pottsey said:
I fully admit Crysis is GPU limited. But its also CPU limited during high physics.
Click to expand...
No it's not, did you even look at my performance graphs? :o

If it was CPU limited you'd see nearly 100% load pretty much constantly on at least one core, but not only does the game support quad-core, it almost never reaches maximum load on any of them.

I knew you'd just whinge about a nuke not being in the benchmark, and I really wish I could be bothered getting the sandbox set up with a nuke going off but I really can't. I don't even know how to use the thing.
 
Last edited:
Pottsey said:
“Crysis is heavily GPU limited at this time, the devs and nvidia have acknowledge an issue with performance on high end gfx cards”
That’s true. But during physics when the GPU is doing no work the game becomes CPU limited. Have a light easy screen for the GPU. Apply normal physics everything’s ok. Apply high physics with the ini tweak with the nuke and the game grinds to a halt. Proving its CPU limited when it comes to physic.
Click to expand...


No it's a performance issue with the games engine on a) lack of high end GFX utlisation and b) lack of multicore utilisation but B is to be expected since the GPU isn't being used efficiently the CPU regardless of core numbers will sit around doing nothing. Fix the GPU limitation in the engine and the CPU utilisation will magically jump.

The developers have acknowledged and have said a patch will be out to fix this, on top of that nvidia are releasing new drivers to further improve performance and to which said they said themselves that every day they work on it they are seeing improvements.
 
“No it's not, did you even look at my performance graphs? ”
I knew you'd just whinge about a nuke not being in the benchmark,”
Yes I looked at your graph. The graph that had no high end physics so wouldn’t be CPU limited. I am not whingeing, your being unfair and unreasonable. You ran a low physics benchmark which we expect to have low CPU usage. You then use that benchmark do demonstrate how the game isn’t CPU limited during high physics usage. See my problem? You test is unfair and doesn’t prove you right.





“Fix the GPU limitation in the engine and the CPU utilisation will magically jump.”
That’s not needed. Turn you back to the high graphics screen and set the nuke of behind you with an simple screen in front of your. You’re now not GPU limited. Yet the CPU spikes. A game can be both CPU and GPU limited at the same time.
 
Pottsey said:
“No it's not, did you even look at my performance graphs? ”
I knew you'd just whinge about a nuke not being in the benchmark,”
Yes I looked at your graph. The graph that had no high end physics so wouldn’t be CPU limited. I am not whingeing, your being unfair and unreasonable. You ran a low physics benchmark which we expect to have low CPU usage. You then use that benchmark do demonstrate how the game isn’t CPU limited during high physics usage. See my problem? You test is unfair and doesn’t prove you right.
Click to expand...
Dude, honestly, how am I being unfair and unreasonable?

How is the test not fair? It's Crytek's own CPU benchmark that's built into the game, keywords there. You can rest assured that the GPU benchmarks utterly thrashes my PC.

Honestly if you're going to take a completely fair test and ignore it like that, calling it unfair and other daft things, then there is absolutely no hope for you. If you won't admit that just one game does excellent physics that are only held back by the GPU, just one damn game, then you're not a man of facts to me anymore Pottsey, and you were one of the best.



Edit: I got into a bit of a huge rant about how uber Pottsey used to be but I cut it shorter. :p
 
Last edited:
“Dude, honestly, how am I being unfair and unreasonable?
How is the test not fair? It's Crytek's own CPU benchmark that's built into the game, keywords there. You can rest assured that the GPU benchmarks utterly thrashes my PC.”
I said the game is CPU limited during high physics usage with the tweaked ini. There are no high physics usage in that benchmark so you wouldn’t expect to see the game become CPU limited.

You are running a low physics usage benchmark as proof the game isn’t CPU limited during high physics usage that’s unfair.

You ran a CPU benchmark without high physics. Any results you get do not prove the game is not CPU limited during high physics as you didn’t have any high physics in the benchmark.

Ulfhedjinn please tell me you understand me know. I don’t know how else to explain this. Perhaps its just late and sleep is needed. Saying I am not a man of facts anymore is harsh. What I said above is true and fair. Your test is unfair and doesnt prove my wrong.

EDIT:
Try this way. The game is only CPU limited during high physics. Your test had no high physics. So it shouldn’t have been CPU limited. So the results you got are what we expect with low CPU usage. Do your test again with the ini tweak and high physics then I will back down and agree if the CPU usage stays low.
 
Last edited:
Pottsey said:
I said the game is CPU limited during high physics usage with the tweaked ini. There are no high physics usage in that benchmark so you wouldn’t expect to see the game become CPU limited.

You are running a low physics usage benchmark as proof the game isn’t CPU limited during high physics usage that’s unfair.
Click to expand...
Define "high physics." There's tons of explosions, debris, collapsing walls, pieces of exploding cars etc. If anything I can see more physics in this benchmark than I see in a video of something like G.R.A.W. What makes it less than "high physics"? The fact there's no running water or a cloth blowing in the wind? :confused:

Pottsey said:
Saying I am not a man of facts anymore is harsh. What I said above is true and fair. Your test is unfair and doesnt prove my wrong.
Click to expand...
That probably was far too harsh actually, but I don't think you're aware of the respect some people around here have for you in the way that you usually debate. You're seriously not at your best in this thread, particularly when you go calling fair tests "unfair" just because the result is not in your favour.
 
Last edited:
“Crysis is GPU limited, it has been known for the past week or 2 now and it is being addressed.End of story.“
No its not end of story. Why is this so hard to understand? Crysis is GPU limited. When high end physics active its both GPU and CPU limited.

You can make Crysis not GPU limited but CPU limited with high end physics.
 
As far as I understand if you run Crysis and only change the settings available in the menu then it is GPU limited.

However the default blast radius for the Nuke is very small which is limiting the physical calculations that the CPU is performing. I have posted a link which shows how to change the blast radius of the nuke so that the explosion actually effects the trees and buildings around it. I have also posted a link to a video which shows the fps crawling when these settings are applied. This indicates that when you turn up the level of physics being applied the cpu starts to struggle.

I can't carry out the tests myself as I don't have Crysis and my PC is getting a bit dated.

Could someone else please try these settings and see what happens to CPU usage and the FPS.

Ulfhedjinn, perhaps you could do this as you have a quad core?
 
Marc Fraser said:
Ulfhedjinn, perhaps you could do this as you have a quad core?
Click to expand...
I would actually be glad to but I have absolutely no clue how to set up the sandbox and stuff to generate a nuke going off, and I don't have any saves with one as I don't think I'm that far into the game yet. Only time I've seen one going off is in multiplayer and I don't want to go messing with .cfg files there in case PunkBuster thows a fit.

I'm sure someone else with a quad who knows how to do it will show up.

Just out of curiosity though, what makes the nuke in Crysis "high physics." I still don't understand what seperates it from normal physics.
 
“Define "high physics." There's tons of explosions, debris, collapsing walls, pieces of cars exploding and all sorts.”
High end physics have explosions that are bigger then 5 meters. Lots of objects moving all at once, cloth, liquids, soft metal. Yes there are tons of explosions but not all going off at once and all the explosions have a very small explosions’ radius. Stand just outside the radius and no physics are applied.

1 collapsing wall or house or a few objects in a small area are not high end physics.

1 house area collapsing with a few objects less then 100 = low end physics

10+ house’s size area collapsing with 1000’s if not 10,000 bits = high end physics.

A nuke going off with a 10 metre physics radius and not blowing down buildings and tress = low end physics. This is Crysis default.

A nuke going off with a massive radius far bigger then 10 metres knocking down building and tress.= high end physics, needs the ini tweak.



EDIT:
“You're seriously not at your best in this thread, particularly when you go calling fair tests "unfair" just because the result is not in your favour.”
You still don’t understand how it’s unfair? It’s got nothing to do with the results. Look your test has no high end physics in it. How can it be a fair test to show the game isn’t CPU limited during high end physics when it doesn’t test high end physics?
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom