Gearing up for CrossFire

[DL126]

Looks to me as though when using a dual slot video card that a 2nd one ain't gonna fit in the next slot.
Is that an optical illusion???

I'd hate to shell out for another card and then the damn thing not fit in there.

[IRQ Conflict]

Depends on the cooler, It should squeeze in there ok though.

[DL126]

Hate to drag up an old thread, but it's a bit related so I figured I'd add to this one.

Going through CrossFire Hell as of late, I've been doing LOTS of reading.
Some of this stuff you guys probably already know, but I'll post it for those who might not.

During all this reading I've done, I learned that the PCI Express slots and CrossFire are controlled by the Intel MCH.
So in trying to figure out why CrossFire didn't seem to be working, I started going backwards.
Looking in Device Manager, I discovered that even though I had downloaded and ran the latest Intel INF package, they weren't actually getting installed.
All of my Intel embedded devices listed a generic Microsoft supplied driver dated 6/1/2006.

Ran the installed package again using the -A flag to extract all the drivers.
Then went through Device Manager updating all of the aforementioned devices one at a time.
They now all show Intel supplied drivers with presumably the latest dates.

Did it fix CrossFire??
Damned if I know yet ...
I've screwed with this crap 'til my eyes are spinning around.

I'll mess with it later when my nerves have settled and curiosity once again reaches the point I can no longer stand it.

All of this was done in Vista x64.
I haven't even started on XP i386 yet.

[DL126]

Nope.
Didn't fix it.
In Vista nor XP.

Think I'm done with Cross Fire.
Need any 3870's???

[IRQ Conflict]

Did you uninstall the catalyst software first? Drivers ans CCC.

Do that. Then use Driver Cleaner to clean up anything related to ATI. Install Intel MB support software reboot then install the 8.1 Catalysts/AVIVO.

[DL126]

Did you uninstall the catalyst software first? Drivers ans CCC.

Do that. Then use Driver Cleaner to clean up anything related to ATI. Install Intel MB support software reboot then install the 8.1 Catalysts/AVIVO.

Done.
Even went a step further ... ran DC Pro & Driver Sweeper both.
Then went into the registry looking for ATI entries manually. None existed.
Manually check HDD for ATI directories ... none existed.
Then installed latest CCC package ... again.
No difference.
Lightsmark 2007 score of 193 with CrossFire on or off.
Same fps either way.

Interesting note ...
Lightsmark 2007 score in XP 32 bit was 246 using single card.
Dropped about 10 fps in CrossFire mode.

Side note ...
I'm running Vista Build 6001 (the public RC of SP1) ...
Wonder if that would make a difference.

[IRQ Conflict]

Hmmm I think lightsmark doesn't work with CF, I think iirc. My score stays at ~105.8

Try a couple of passes with 3dmark?

[DL126]

Done.
Used 3DMark06 on all default settings.

CrossFire off = 9735
CrossFire on = 13266

Guess that means that it's working.

[IRQ Conflict]

Thats what it means.

[DL126]

Same tests run under XP Pro 32 bit vs. the Vista 64 previous runs.

Crossfire On = 12596
CrossFire Off = 10342

[Zertz]

Same tests run under XP Pro 32 bit vs. the Vista 64 previous runs.

Crossfire On = 12596
CrossFire Off = 10342

That's interesting, Vista x64 results are actually quite a bit better

[IRQ Conflict]

huh? look again....

[DL126]

I didn't do any screen shots, but I did print the results.
And since it seems that a few are taking notice of them, reckon I'd better clarify the results.
I don't have a freaking clue what some of this stuff means, I just wanted to play Crysis on faster frame rates.
Well that didn't happen, but now you guys have me running benchmark crap.
10+ years working on puters and I've never run any benchmarks 'til now.
You have corrupted me forever!

Vista Ultimate 64 bit Build 6001

Crossfire on = 13266
SM 2.0 Score - 5805
SM 3.0 Score - 7331
CPU Score - 2541

Return to Proxycon - 47.54 FPS
Firefly Forest - 49.215 FPS

CPU 1 Red Valley 0.804 FPS
CPU 2 Red Valley 1.285 FPS

Canyon Flight - 83.224 FPS
Deep Freeze - 63.389 FPS

Crossfire off = 9735
SM 2.0 Score - 4219
SM 3.0 Score - 4375
CPU Score - 2543

Return to Proxycon - 32.063 FPS
Firefly Forest - 38.247 FPS

CPU 1 Red Valley 0.804 FPS
CPU 2 Red Valley 1.288 FPS

Canyon Flight - 41.94 FPS
Deep Freeze - 45.562 FPS

XP Pro 32 bit SP2

Crossfire on = 12596
SM 2.0 Score - 5470
SM 3.0 Score - 6870
CPU Score - 2471

Return to Proxycon - 44.432 FPS
Firefly Forest - 46.74 FPS

CPU 1 Red Valley 0.786 FPS
CPU 2 Red Valley 1.243 FPS

Canyon Flight - 80.9 FPS
Deep Freeze - 56.503 FPS

Crossfire off = 10342
SM 2.0 Score - 4463
SM 3.0 Score - 4914
CPU Score - 2482

Return to Proxycon - 34.75 FPS
Firefly Forest - 39.637 FPS

CPU 1 Red Valley 0.785 FPS
CPU 2 Red Valley 1.255 FPS

Canyon Flight - 45.503 FPS
Deep Freeze - 52.775 FPS

[IRQ Conflict]

Heh, your first test showed Vista being slower than XP? no?

Oh well....Games suck! Benchmarks are the way to play!

[DL126]

Oh well....Games suck! Benchmarks are the way to play!

Obviously I'm no expert in the benchmarking scene ...
But from what I've seen thus far, hardware and software can be "tweaked" to possibly give "desired" results.
I'm no programmer mind you, but I've seen and read enough to be convinced that benchmark results can, at least to some extent, be "tailored" to the tester's taste.
A couple of examples:
1) Apop's tests with Crysis and CrossFire were done with the Demo version of the game.
Reading around Crysis forums, you'll find that all sorts of programming code was changed between the Demo and final release.
Now this has NOTHING to do with Apop ... just a possibility of why his framerates were so different from my own.
2) I posted some screenshots in the Lightsmark 2007 thread. Yes those were actual results.
But what I didn't tell you in that post is that running that test multiple times produced multiple scores.
Those were the best ones. So in this particular case, technically I "manipulated" the scores to achieve more desirable results.


Therefore, I'll continue to rely on what my eyes tell me rather than some score produced by any given test.
However, running the 3DMark06 did serve the purpose of telling me that CrossFire actually was working.
It's just disappointing that it didn't work for the reason I bought the hardware.

And with this post, this thread has begun to wander off topic.
So I'll stop here.
Thanks for all your help IRQ!