BFBC2: considering upgrading

[Porter_]

i saw none

266.35 beta

...as in, i saw no performance gain

 

[Jodiuh]

Ah, bummer.

 

[Porter_]

Ah, bummer.

should be remedied next week

 

[Jodiuh]

What and where did you read? Could be very helpful for some folks.

 

[Porter_]

guess i was a little cryptic, i'm upgrading to SB next week.

 

[SHiZNiLTi]

I don't know how alot of you guys are testing your frames, but anyone running a fermi card with anything less than an i series chip while claiming to have high minimum frames makes no sense.

My rig runs the game at pretty much medium settings with aniso and aa set at 1 and I get down to the 20's on panama canal when theres smoke or a lot of action going on. My 24" runs at 1920x1200.

Maps with less foilage run great, but panama will show you your system sucks if you run fermi with a 2+ year older chip.

I am probably going to snag a sandy bridge in the next week or two to finally be able to play this game properly.

False^

I rarely ever drop below 50fps in a full server no matter whats going on around me with my Q9650.

This is with with a gtx570 which is getting returned soon, the 5870 I tested gets an around 7-10 more fps in heavy action like this using the below settings.

GPU usage is around 50-75% peaks to 89% during large in your face explosions.

CPU usage is a constant 82%.

Full server Panama Canal
1600x1200
4xAA
16xAF
All HIGH settings
HBAO off

scroll to the right -------------------------------------->

This is with (3) tanks in very close range to me in a fire fight and one used smoke along with lots of dust from the map...

 

[tlbig10]

I see the culprit. WINAMP!

 

[SHiZNiLTi]

I see the culprit. WINAMP!

rofl, I try to stay away from anything AOL related, but WINAMP FTW!!! I use it because I can set it to use my onboard sound which is optical out going to a harmon kardon receiver. This way I can have music playing on surround sound and have my X-Fi card dedicated to in game sound Blends amazing with open ear HD595s

 

[Jodiuh]

AND xfire. It's like looking into a time capsule. Got Roger Wilco on there too buddy?

New graph!! 470 run definitely had more combat in it and that would explain the crazy variations. Also, I never heard 470 while gaming.

 

[Porter_]

interesting, thanks for sharing Jodiuh

 

[Az Syndicate]

I have a 460 OC a bit and a Q9650 and my 460 never had a problem hitting full speeds. Jod can confirm this.

 

[Jodiuh]

Wish there was a way to snap a screenshot if frames went below a threshold...60 in this instance. Love to know what the heck happened a minute in. :/

Edit: Regarding Syn's setup...IIRC, you ran 1920 x 1200, 32x CSAA, HBAO on, and usage tho higher than ours, was not pegged. However, his config flew past both mine and another Fermi/C2Q setup...it was maddening. I remember dipping into the low 20's that day w/ low settings and no AA. Would be nice to see a usage graph from you w/ 1920 4xAA, High detail, and HBAO off.

 

[Philosoraptor]

You really listen to Coldplay while you play BC2?

 

[Jodiuh]

This is why I box out w/ paint. Can't have you guys finding out I listen to Britney Spear's "Stronger" on repeat while gaming.

 

[pokevitek]

Fun facts time....

Fun fact #1 - running facebook in background while playing BC2 = deacrease of 60% of performance (IE 8)....

fun fact #2 - it seems my crashing problems was coused by microphone This game can be sometimes really strange....

 

[SHiZNiLTi]

You really listen to Coldplay while you play BC2?

lol yep, That remix by Stuart Price aka Thin White Duke is great! He is an awesome electronic musician, his remix Coldplay - Talk (Thin White Duke Remix) is pure audiophile eargasmic. Lately I've been listening to alot of music that has been electronically remixed or dubstep remixed depending on the song... pretty much anything with good SQ with a decent bass line for my Velodyne DD-15 Digital Drive Series 1250W Class D (3000W Peak) subwoofer under my PC desk and no I did not spend the $5699 msrp on it, got it free from a friend of the family The only negative thing about this setup is that at full tilt my mouse will jump a 1/4" off the mouse pad lol.

This is why I box out w/ paint. Can't have you guys finding out I listen to Britney Spear's "Stronger" on repeat while gaming.

Just the fact that you know the name of a Britney Spear's song scares me lol

How you liking the new 470 so far Jod?

 

[ObiDamnKenobi]

So my MSI 6950 is in place, but only briefly tested unfortunately. Had less than an hour last night to try it in BC2 with different settings. Putting all settings at 19x12 and max looks awesome, but framrates are a bit low.

Honestly I'm wondering if I need to do some tweaking, my FPS seemed to be way lower than it should be Fraps on my G15 display showed around 30 fps, with dips below 20. The AnandTech test showed 62 fps average at this setting. But I think I had AA and AF higher, and I turned HBAO off but I've noticed I need to exit the game before it takes effect (?). I also didn't mess with AA/AF settings in CCC, which I think is set to "always on" and might take resources. I'm assuming this should be set to "application settings" right?

Need to do a proper Fraps recording tonight and see what average I get. And then consider whether I want to do the 6970 unlock

ps: §·H·ï·Z·N·ï·L·T·ï, sorry I didn't do the benchmark with my 260 you asked for. But I'd messed with the clock and memory clock on that card like crazy and I'm not even sure it was properly stable. I would probably have gotten ~15 fps at anything above 16x10..

 

[Italiano86]

False^

I rarely ever drop below 50fps in a full server no matter whats going on around me with my Q9650.

This is with with a gtx570 which is getting returned soon, the 5870 I tested gets an around 7-10 more fps in heavy action like this using the below settings.

GPU usage is around 50-75% peaks to 89% during large in your face explosions.

CPU usage is a constant 82%.

Full server Panama Canal
1600x1200
4xAA
16xAF
All HIGH settings
HBAO off

Nice argument, but you also happen to own one of the most powerful c2q's which is comparable to an oc'd i5. You are a minority as far as the c2q chips go, and those with the entry to mid range c2d or c2q will not have optimal performance in game.

Also, if you were pushing a higher resolution, I'm sure your gpu would be taxed a bit more, and the usage would scale up. That data tells me your cpu is dealing more than your gpu at that resolution. I think your setup would really show it's stuff if you were running 1920 to 2560. You would probably have minimum frames in the 40's to low 50's.

 

[SHiZNiLTi]

Nice argument, but you also happen to own one of the most powerful c2q's which is comparable to an oc'd i5. You are a minority as far as the c2q chips go, and those with the entry to mid range c2d or c2q will not have optimal performance in game.

Also, if you were pushing a higher resolution, I'm sure your gpu would be taxed a bit more, and the usage would scale up. That data tells me your cpu is dealing more than your gpu at that resolution. I think your setup would really show it's stuff if you were running 1920 to 2560. You would probably have minimum frames in the 40's to low 50's.

Yep I'm very pleased with the upgrade decisions that I made.

I originally had a E8500@4ghz and a gtx260, I would often drop to low 30's and that was in DX9 with shadows off and medium/high settings!

I'm very happy being able to just pop in a Q9650($250) and 5870($200) and be able to run highest settings and not have any issues with low fps and rarely dropping below 60fps.

Total upgrade cost was only $250 after I sold my E8500 and gtx260

 

[Az Syndicate]

I basically did the same upgrade last year (2010)

I got a Q9650 to replace a E8400. Total cost to upgrade was $80 selling old E8400
I got a 460 to replace the 260 Total cost was like $50 after rebates and selling old 260

Once again, I don't have the GPU slowdowns that other people seem to have.

 

[Jodiuh]

How you liking the new 470 so far Jod?

Well...it thinks it's a 460 w/ its .912 VID...so...

...own one of the most powerful c2q's which is comparable to an oc'd i5.

A 3.6 Ghz Q9550 w/ an 850 Mhz 460 dipped into the 30's. An i5 760 @ stock speeds w/ the same 460 never went below 60. Yorkfield might be one of the better C2Q's, but as far as Nvidia and Fermi's concerned, it's garbage.

...E8500@4ghz and a gtx260...

Played many a L4D session on that setup (E8400). Very cool, very quiet, very fast. I bet Crysis would have been pretty swift w/ reasonable settings too. Though, I'm still holding out for the GTX 990 4 GB and an i13 octa to play it properly.

Once again, I don't have the GPU slowdowns that other people seem to have.

Prove it!!

 

[SHiZNiLTi]

Played many a L4D session on that setup (E8400). Very cool, very quiet, very fast. I bet Crysis would have been pretty swift w/ reasonable settings too. Though, I'm still holding out for the GTX 990 4 GB and an i13 octa to play it properly.

ROFL!!!

 

[nobody_here]

Well...it thinks it's a 460 w/ its .912 VID...so...

my MSI 460GTX 1Gb runs .962 on Perf. Level 3.....

 

[Jodiuh]

ROFL!!!

I remember going from a P35/88GTX to a 680i/SLI88GT for that game and it ran like crap...still bitter. Terrible, terrible board/chipset that was.

my MSI 460GTX 1Gb runs .962 on Perf. Level 3.....

I finally got lucky and struck gold after a lame Q9550 (3.6Ghz), i5 760(18millionvolts), and GTX 460 (1.018 VID).

 

[Gilbert_pwns]

Yes. A quad core would be the biggest improvement. I was doing probably 20fps or so at best on low settings with my E4400 Dual core. Got a 965 Deneb Phenom quadcore and the difference is obvious. I can play on all high settings now and this is with a 4870.

 

[DualOwn]

crank up the settings to all high with alot of aa and af, you should see higher gpu usage.

 

[Jodiuh]

And even lower framerate.

 

[nobody_here]

And even lower framerate.

Sine the thread is pretty kaput how do you like your Fascinate? Most of my posts here are on mine, waiting to finally get Froyo but I'm thinking our odds are better we will see Gingerbread and skip Froyo altogether.

 

[Jodiuh]

I've been through just about every android phone on Verizon and the Fas obliterates them all. And now that I've got screen filter to dim Mr. Oled, I'm no longer going blind late at night. Froyo, hah! Your funny.

tlbig10 should be in here any minute with 5850 graphs.

Edit: Shiz!!! Edit ur frigging picture explosion. I feel like I'm panning a spreadsheet on my phone.

 

[tlbig10]

Going to do my best to keep this organized, but here comes a lot of graphs and data!

Installed a 5850 this evening and have been rerunning all my tests. I pitted the 5850 against the 4850 since the 460 was not working properly and wouldn't be a fair test for it. Here are the results and boy was I clueless on just how important your CPU is in BC2:

4850 vs. 5850 - Medium detail, 1xAA:

4850 cpu and gpu usage:

5850 cpu and gpu usage:

You'll notice this medium detail setting is far more CPU-intensive for the 5850 and therefore is not working very hard despite a maxed CPU - aka CPU is the bottleneck. The 4850 is pegged while the CPU is ~80% - good balance IMO.

From here on out I won't post any 4850 cpu/gpu screen shots since there is no change.

4850 vs. 5850 - High detail, 1xAA:

5850 cpu and gpu usage:

More CPU bottleneck here, but the GPU is starting to get some more use. Results from medium to high detail are nearly identical on the 5850, but the 4850 is starting to fall behind as the GPU runs out of steam.

4850 vs. 5850 - High detail, 2xAA:

5850 cpu and gpu usage:

Same story, but 5850 GPU is inching closer to being the bottleneck. 4850 is not playable at this setting per my standard.

4850 vs. 5850 - High detail, 4xAA:

5850 cpu and gpu usage:

Ah, now we're finally maxing the 5850. Full CPU and GPU usage here.


4850 vs. 5850 - High detail, 8xAA:

5850 cpu and gpu usage:

Now we're finally starting to see the 5850 stress itself and not be able to maintain the 60fps average. Yes, it was brutal playing BC2 with the 4850 at these settings, but i trudged through it so we can see the differences!

I'll be installing a Phenom II X4 945 on Saturday, so it'll be interesting to see what it'll do to my lower setting tests where the CPU was the bottleneck. Goal is to reach 50+ minimum framerate and hopefully an extra core and 200 more mhz will make that happen.

 

[Jodiuh]

I don't have enough resolution to process this information.

Edit: Ok, got through it w/ a whole lot of scrolling. Would be nice to see if you were maxing the 4850's 512 MB buffer via Afterburner in the 4x AA run. Sure seems that way. Hopefully w/ some more speed, a forth core, and some OC on that 5850, you can get where you need to be. MLAA works on Cypress, right?

Also, thanks for all that. Lots of work I'm sure.

 

[nobody_here]

What program is being used to create those graphs?

 

[samuelmorris]

Dunno about the rendering, but the data can be obtained simply by using fraps' benchmark utility, which excel can work with.

 

[Nosenuggets45]

Instead of starting a new thread, I'm going to post in here. I'm running at all high settings except 4xaa, low shadows, HBAO off, and medium for overgrowth, undergrowth, and water. Generally I get a pretty solid 60-70fps, but I do get dips into the 40s. Why is this happening, and why is my GPU/CPU usage not maxed?

 

[tlbig10]

I'll take a stab here and say you appear to be GPU limited with your settings since you aren't maxing all 4 cores. You can incrementally add more detail/AA to put more load on the GPU, and if your min/avg is the same, then your CPU is to blame. If adding a little more detail/AA reduces min/avg, then you're GPU limited.

Perhaps you should work your way backwards and see if the frames improve. Take some GPU load off and allow the CPU to work harder.

 

[pokevitek]

Well, about CPU, quad on 4 GHz should really be enough. More than enough. My Core i5 760, OC at 4.2 GHz has three cores at about 50% and last maxed out. Only way I was able to max CPU load was default 2.8 GHz frequency. With no Turbo Boost. Since you are running quad at 4 GHz, I think you sould have lower CPU usage.

For your GPU, usage not beeing maxed, this is really strange. Once again, your CPU is FINE.

For the dips into 40, I have HD 6850, with medium OC. I would guess I am very near HD 6870. I played at 1680x1050 with All High, HSAO on, 4xAA. I got average FPS around 60, with dips down to about 35 in really intensive situations. These dips are normal thing in this game.

This made me play with 1AA, and I disabled Bloom, since to me its worthless. This gave me another boost. Now I play with 75 AVG, still getting drops down to about 38.

Very IMPORTANT thing about this game is, although its not as bad as someone says, its not good either. And not just performance-wise. For example, new patch brought me few seconds freeze few second after map loads, no matter if I am in class selection or I jump to action really fast....

 

[ObiDamnKenobi]

interesting stuff guys, thanks.
I found out why my new 6950 was only giving me ~30 fps at 19x12; BC2 only maxes one CPU core! Found some others complaining about this too. Supposedly fixed in an early patch, but alt+tabing clearly showed otherwise on my system. Setting core affinity in task manager I got it to use 4 cores, and got around 55 FPS avg with 19x12, high settings 4xAA. Still seems a bit low though, as the 5850 above got around the same, or better with similar settings Perhaps time to consider unlocking the extra shaders, and OC-ing a bit..

Also annoying that I have to alt-tab and change core affinity every time I run the game! Have to investigate this a bit more, considering turning off Hyper threading and see.

 

[-Sn1PeR-]

interesting stuff guys, thanks.
I found out why my new 6950 was only giving me ~30 fps at 19x12; BC2 only maxes one CPU core! Found some others complaining about this too. Supposedly fixed in an early patch, but alt+tabing clearly showed otherwise on my system. Setting core affinity in task manager I got it to use 4 cores, and got around 55 FPS avg with 19x12, high settings 4xAA. Still seems a bit low though, as the 5850 above got around the same, or better with similar settings Perhaps time to consider unlocking the extra shaders, and OC-ing a bit..

Also annoying that I have to alt-tab and change core affinity every time I run the game! Have to investigate this a bit more, considering turning off Hyper threading and see.

Are you not running a fully-patched version of the game? You shouldn't have to change the core affinity manually What OS?

 

[ObiDamnKenobi]

Are you not running a fully-patched version of the game? You shouldn't have to change the core affinity manually What OS?

Steam version! Checked the update settings there to, and the whole "integrity of game cache". Running Vista 64bit.
No idea why this wouldn't work on my system, perhaps I'll try a reinstall of the game.

 

[travbrad]

For what it's worth my framerates in BC2 almost doubled going from an OCed HD4830 to the GTX460 in my sig (with the same CPU as you). The GTX260 is sort of in between those cards though so it's hard to say for sure. These numbers are min/avg for 4830/GTX460:

1080p 4xAA16xAF, Highest, HBAO enabled: 17/28 21/50
1080p 1xAA16xAF, Highest, HBAO enabled: 19/44 24/50

BUT...when I turn off AA I become more CPU limited. Here's a nice chart showing CPU scaling (notice how the quads annihilate the duals/tri-cores ): http://techreport.com/articles.x/20188/7

You would probably benefit from both a CPU and GPU upgrade (although that's not easy on the wallet ). Probably more from a CPU upgrade for BC2, although that's also more expensive.


P.S. I'm sure it's already been answered by now after 12 pages, but that's my 2 cents