Sense, this makes none... Crysis fails me

[TridenT]

Here are the results... I had this all in order and hit run benchmark. Let it go and come back and here is what I get...

I used the Crysis Benchmarking Tool, Crysis 1.21. Vista 64 Ultimate. (Ran in DX10 at 32-bit, 64-bit crashes I think)

System:
Asus Rampage Formula X48 (Stock bios, not 403)
Crucial Ballistix Tracer DDR2-800 ram 4x1GB
Two HIS Radeon HD 4850 512MB Video cards in Crossfire (8.6 Hotfix drivers)
Intel Core 2 Duo E7200 CPU at 2.53Ghz(or whatever stock is)
1TB Western Digital 5400-7200PRM Green Power HDD

RESULTS:
Run #1- DX10 800x600 AA=No AA, 32 bit test, Quality: Medium ~~ Last Average FPS: 39.75

Run #2- DX10 1024x768 AA=No AA, 32 bit test, Quality: Medium ~~ Last Average FPS: 40.62

Run #3- DX10 1152x864 AA=No AA, 32 bit test, Quality: Medium ~~ Last Average FPS: 40.49

Run #4- DX10 1280x720 AA=No AA, 32 bit test, Quality: Medium ~~ Last Average FPS: 36.68

Run #5- DX10 1280x960 AA=No AA, 32 bit test, Quality: Medium ~~ Last Average FPS: 40.49

Run #6- DX10 1280x1024 AA=No AA, 32 bit test, Quality: Medium ~~ Last Average FPS: 40.93

Run #7- DX10 1400x960 AA=No AA, 32 bit test, Quality: Medium ~~ Last Average FPS: 39.20

Run #8- DX10 1400x1050 AA=No AA, 32 bit test, Quality: Medium ~~ Last Average FPS: 40.06

Run #9- DX10 1600x1200 AA=No AA, 32 bit test, Quality: Medium ~~ Last Average FPS: 39.68

Run #10- DX10 1680x1050 AA=No AA, 32 bit test, Quality: Medium ~~ Last Average FPS: 37.39

Run #11- DX10 1900x1200 AA=No AA, 32 bit test, Quality: Medium ~~ Last Average FPS: 37.12

Run #12- DX10 1920x1080 AA=No AA, 32 bit test, Quality: Medium ~~ Last Average FPS: 35.61

So.. First off, why am I seeing BETTER frame rates at HIGHER RESOLUTIONS?

 

[videogames101]

CPU limited or something?

 

[LittleNemoNES]

CPU limited. Your CPU needs to be faster to take advantage of your Xfire config. OC that sucker (e7200)

 

[TridenT]

Originally posted by: gersson
CPU limited. Your CPU needs to be faster to take advantage of your Xfire config. OC that sucker (e7200)

Would if my motherboard could do it. -_- It fails every benchmark at 333mhz x 9.5, and just because it overvolts itself AND causes its own errors... Very lame. Not the CPUs fault or the memory either, I've tested those. It's just the board, it's crap.

 

[Zenoth]

You're seeing better frame rates at higher resolution because at those resolutions the GPU it the one working more, at lower resolutions you become much more CPU-dependent. Indeed, your CPU is slow for a Crossfire setup. You might need around 3.0Ghz to 3.4Ghz to take a much better (perhaps still not a full one) advantage of your Crossfire setup. And keep in mind that Crysis is already CPU-intensive game, especially during firefights where lots of physics kicks in (for example just check the CPU benchmark).

 

[TridenT]

Originally posted by: Zenoth
You're seeing better frame rates at higher resolution because at those resolutions the GPU it the one working more, at lower resolutions you become much more CPU-dependent. Indeed, your CPU is slow for a Crossfire setup. You might need around 3.0Ghz to 3.4Ghz to take a much better (perhaps still not a full one) advantage of your Crossfire setup. And keep in mind that Crysis is already CPU-intensive game, especially during firefights where lots of physics kicks in (for example just check the CPU benchmark).

What is disappointing though is that even at 437.9 MHz on the DRAM Freq I still get two seconds into Orthos and it fails. 5-5-5-18-2T. It's SO STUPID! It's not even fair how crappy this board is. I know the ram isn't actually at fault either because it does perfect in other boards at higher frequencies and lower timings.

 

[videogames101]

That board OCs amazingly well. Try disabling some of the auto-oc crap. If it always fail, RMA it, because that board is an OCing dream.

 

[TridenT]

Originally posted by: videogames101
That board OCs amazingly well. Try disabling some of the auto-oc crap. If it always fail, RMA it, because that board is an OCing dream.

Probably too late to RMA. I don't know, does anyone want to give me a full setup of settings to try out? Maybe I am just doing something wrong. (Although I highly doubt it)

 

[Zenoth]

If you RAM fails it may need more voltage. There's also a possibility that you'll need more northbridge voltage (MCH), or FSB voltage.

 

[TridenT]

Is 1.63V safe for the NB on this board? It hits around 55C. CPU hits around eep... Just hit a record high. >_> 72C. Might want to put it down, hehe... yeah... just errored at 6 minutes 30 sec. Test 4, 4000 Lucas-Lehmer iterations of M19174369 using 1024K FFT length.
FATAL ERROR: Final result was CF2660D8, expected: 3247CE82.
Hardware failure detected, consult stress.txt file.
Torture Test ran 6 minutes 31 seconds - 1 errors, 0 warnings.
Execution halted.

Where is stress.txt, I can never find it. Funny how my CPU idles at 30C but at load it climbs to 70C... Probably too high on voltage, it's at 1.30V or so. (Can run stock steady at around 1.12V) Frustrating! Impossible to OC this thing I swear.

 

[duragezic]

You might want to head over to the CPU/Overclocking forum and read up (lots of other people with a similar configuration as you). There should be some advice or guides on determining what component is holding you back. Overclocking is a bit more complicated these days so you'll want to take the right steps and only be stressing one aspect. i.e. with the Athlon 64/X2, you can test for max HTT (like FSB) by using the lowest CPU multiplier, most relaxed/slowest RAM settings, then increase till you get failures. I'm guessing the C2D might have a similar process. You could also get some advice on your temps and see if those are the issue. Stock coolers tend to be better these days than in the past but you can only go so far with them.

 

[TridenT]

Originally posted by: duragezic
You might want to head over to the CPU/Overclocking forum and read up (lots of other people with a similar configuration as you). There should be some advice or guides on determining what component is holding you back. Overclocking is a bit more complicated these days so you'll want to take the right steps and only be stressing one aspect. i.e. with the Athlon 64/X2, you can test for max HTT (like FSB) by using the lowest CPU multiplier, most relaxed/slowest RAM settings, then increase till you get failures. I'm guessing the C2D might have a similar process. You could also get some advice on your temps and see if those are the issue. Stock coolers tend to be better these days than in the past but you can only go so far with them.

Yeah, I have OC'd the C2Ds before. I had unreal overclocks with my E6300 about one-two years ago. I could manage putting it onto lower than stock voltage and manage 3.6GHz easily. I couldn't manage higher OC because my board would just cut out. (P5B and 680i)

 

[40sTheme]

Uh.... you could probably hit 3.2GHz on that thing with stock voltage. 1.3V is outrageous if the stock is 1.12V. If it's even a SEMI well binned chip, you'll hit pretty good clocks at stock voltage. You don't need to worry about NB voltage until you hit really high speeds.

 

[SirJangly]

Hmm, off topic, but this explains something. I run at 1440x900 on a phenom 9500/8800gt, and always wondered why performance is usually better with everything maxed out

 

[BassBomb]

why medium when you have crossfire?

 

[TridenT]

Originally posted by: BassBomb
why medium when you have crossfire?

Crossfire doesn't scale in Crysis.

Also, the game is a lot better with better frame rates than with lower ones.. I rather have lower quality but higher fps.. Otherwise I can't do shit.

 

[error8]

You're not overclocking it right. That mobo should take it very far. I'm at 3,9 ghz with my E7200 on a cheapo MSI P35, with the worst 4 modules of Kingmax value ram. Maybe you need some bios update or something.