More RAM seems to improve FPS smoothness in BFBC2

for anyone else who reads this thread - has a capable PC with 4 GB of memory and high end graphics and has min/avg something like 15min/30avg, but you want 45min/60avg, just turn off 32xaa

aa just needs as beefy a GPU as possible - no other hardware will get you there if your GPU is the bottleneck, which it is when you are running 32xaa. Even a superclocked i7 won't help you at that point. And the whole point of AA is to remove a particular kind of rendering artifact, that is basically gone at approx. 4xaa-8xaa. After that, you are just sinking performance for no visual quality change.

 

have you even read XxuserxX 's comments? He says exactly what I've been saying, and the studies and benchmarks, along with others testimony of extra RAM holding the average fps steady and not taking too many drastic low framerates when you got many things on screen like explosions, particle effects, 15 animated guys running around on screen etc.

Here's one of the posts he made which kinda explains this.

"What do you not understand about "average" FPS? Frame rates jump around quite a bit. You really should read an article on Windows 7 memory management. I have been PC gaming since the early 90s and I am a network engineer. I know what I am talking about." (XxUserxX)

and...

"Where are you getting this 5 frames from? Noone has ran any benchmarks or provided any real sample data. Max FPS is not the end all be all, 100 FPS means nothing when a explosion or large firefight cause your PC to go to 35 FPS because your PC is constantly acessing the pagefile. There is more to it than getting a better video card. My system is very tuned and I benchmark better than my cowokers who have better rigs than I do." (XxUserxX)

also...

"Also that article used Vistax64 as a test OS. Vistax64 and Win7x64 do not have the same memory managment.
PDC 2009 Pre-Conference: Windows 7 Memory Management

At his lecture at the PDC's "Windows 7 Developer Boot Camp", Landy Wang showcased many of the memory management changes done in Windows 7 and Windows Server 2008 R2 kernel, one of which is highlighted below. These changes, on average, lead to performance increase of up to two times in memory constrained systems – for free. All you have to do is move to the new operating system.
Removing the PFN Lock

The Page Frame Number (PFN) lock, up to Windows Vista (including), is a single lock in charge of almost all operations regarding page allocations in Windows memory, and as such it is a huge bottleneck when it comes to memory intensive applications and systems. For example, on a machine with 128 cores, SQL server reaches 88% contention over this lock, meaning that almost every second operation is stalled and takes precious time.

This lock has been removed and replaced with finer-grained locks on an individual memory page basis, achieving an up to x15 performance increase on a 32 processor machine. Note that this improvement comes for free to the developer who only needs to swap to the new OS.

Enjoy! If you want more info look it up yourself." (XxUserxX)

and this thread was back in March before I even realized this through the studies and benchmarks.

here's the link again for those who want to read the thread and study Windows 7 management

GameSpot Forums - Battlefield: Bad Company 2 - Does 64-bit give you more performance over 32-bit?

he's got a link on page 2 that links to Windows 7 developer bootcamp, which I'm watching now..

 

you understand that explosions and intense firefights don't cause slowdowns because of hdd access or memory access. those are done on the graphics card. it has to do a ton of processing to push those effects to the screen. the hdd isn't involved when an explosion happens on screen.

in games where the hdd is accessed during the game, having more memory would only alleviate the problem in the specific circumstance where the game was coded to load extra assets into memory AND you didn't have enough memory.

BC2 is a case where assets are loaded into memory and, if you have 4GB, you have enough. It doesn't need any more memory to load up and be ready for explosions, firefights, etc.

You *almost* had it when you listened to what the developer who MADE THE GAME had to say, and then you lost it when you decided that maybe xXuserXx from rando-forums.net understood it better.

So close. What a shame.

 

No. But would love to have a look at those so-called studies and benchmarks.

This.

 

Theres only one thing that makes bc2 going to stars: a good quad cpu. Not 16gb,8,6 of ram, as 4gb is enough for it.
The game performs good even on dual cores,thats for sure,but the quad performance ia another story.

And btw, im sick of peoples complaining about low fps with settings like 32xAA,16xAF maxing out the game with every setting possible on a mobile card like gtx460m or gtx285m while running at 1080p.
These peoples needs to understand what huge hit to performance those settings makes. Those options should be enabled only on high end desktop pcs with a proper high end card.
Also, i understand the biotching about framerate,but lets be clear: the frostbyte engine used in bc2 is born on consoles and is made to run at 30fps.
This means that if your setup runs the game somewhere from 30 to 70 then its perfectly fine.
This is not COD where the prerogative in the engine is 60fps on every platform.
Personally my xps m1730 in sig runs bc2 maxed out @ 1200p with a light 2xAA, with an avg of 40 fps,max 70, min 35. Perfectly enjoyable and playable if you ask. If i start pumping AA then goodbye avg 40 fps, and its pretty normal in pc gaming.
Just my 2 cents with added rant

 

Right.

These are the settings I use and seem the best combination of performance to visual effects.

 

Aw, Hell No! I got the latest in computer technology, and I am not turning down my settings for a game that's 1 year older than my tech. My Nvidia GTX 460m is not the bottle neck. It generates ideal framerates, because it's capable of doing so. I suspect the bottle neck would be RAM, because with the new graphics driver it increased my framerates even more, it's just the dipping of the minimum framerates that is probably the textures and effects that could utilize more memory. This seems to make sense...

 

And Crysis is three years older than a lot of people's tech, and they can't max that out, either.

While it's new tech, don't overestimate what mobile GPUs can do. In any case involving sub-par framerates, nine times out of ten I will call the GPU the bottleneck, not the system RAM.

In the end, choose between eye candy or FPS. Unless it's an older game (or a lightweight one), you're unlikely to max out both.

 

Mate without offence you should then learn how shaders, memory bandwidth and gpu bit capacity means.
Because if you want eyecandy AND fps, better you get a desktop with a good GPU, because a laptop cant do miracles.

The GTX460m is a good card, with 192 shaders, but just 192bit of bandwidht. Its capable, but i guess that if you want to max out BC2 you have to lower your resolution to something like 720p.

Mem that my sli 9800m GTX (gtx260m, basically) can do 1200p maxed out (4xAA) with framerate from 30 to 70 using DX10.
But thats a sli 112 shaders card (112x2) , both high end with 256bit of bandwidth. That should makes you understand your bottleneck.

 

dude... you're gone. good luck.

 

I used to have the very first ever mass produced SLI laptop, the Toshiba x205 sli1. At the time this baby was the Bully of the block as far as gaming went, but when it got stolen, I didn't want to settle for anything less than SLI, although I didn't have the funds. I really needed another computer ASAP, and since I had no issues and loved the performance and customer service of Toshiba, I didn't hesitate to buy the newly released Toshiba X305 q701, although I was bummed it was just a single 9700m gts and not SLI.

I was pleasantly surprised when I got home and installed my games, especially Gears of War on the new X305 q701, because compared to my previous X205 with dual 8600m gts, the single 9700m gts actually gave more smooth performance and higher framerates. Now that I think about it, in the cutscenes on Gears of War, with everything set to max, the SLI setup would skip many frames and maybe generate 10 fps, but the single 9700m gts would play the same cutscenes maxed out at motion picture fluidness of 26 fps.

Thinking about this issue of RAM and looking back to this experience of my previous notebooks, and how it handled max graphics, I'm thinking the 9700m gts pushed out more fluid fps because of the RAM and setup of my X305 q71, as opposed to the older dual 8600m gts which was on par with the horse power of the 9700m gts.

So what I'm trying to say is that your 9800mgtx in sli can't be compared to my Qosmio X505 Q892 and the GTX 460m, I'm betting the single GTX 460m will have a significant edge over your dual 9800m gtx, and I'm suspecting it has to do with architecture and RAM. Also since I have only 4gb of RAM, I'm not seeing the full potential of what this notebook and GPU can do graphics wise...

 

The worse performance in sli its not a matter of the "hardware" but of software, in this case driver.
Never had a problem running GOW in my sli setup, maxing out it. Like i never experienced low fps or stuttering, but i did in the past with some games. Turns out that a good driver and a good sli profile keeps things on par.

Naturally a sinlge card, decent and discrete, is hassle free compared to a sli setup. But from the benchmarks im seeing around about gtx460m, no, theres no way it performs better than my sli setup.

I doubt it can run Cod:BO at 1200p maxed out, with 8xAA,16xAF while keeping an average of 45fps.

Anyway good luck with your tests, get 4gb more and let us know how it goes

 

9800m gtx sli > 460m

good luck.

 

Ugh, I'm sorry I ever said that I noticed a difference in framerate. Everything masterchef341 and kingpinzero are correct. In BFBC2 I did notice something, but so far nothing measurable as far as framerates.

Either way jacob808 - you will not see any amount of performance increase because of RAM, and if for whatever reason you do, it will be marginal, as in a couple percent. Mobile CPU's have their limits, primarily with higher resolutions.

 

I'm going to close this thread; the original topic is starting to get drawn out at this point and there's not much sense in continuing.

Feel free to start another thread if you still have questions about this . . .