HD4870 and Crossfire issues...

rkwerchan

Distinguished
May 15, 2009
15
0
18,510
Ok... I'm pretty sure I'm screwing myself over here somewhere... here's my setup:

Abit AB9 QuadGT mobo
(2) 2GB Crucial Ballistix Tracer PC6400
(2) 2GB Samsung PC6400
Intel E8500 Wolfsdale @ stock 3GHz
Thermaltake 850W PSU
(2) Radeon HD4870 cards
Running Windows 7 RC1

I haven't OC'd anything yet, just playing around with stock speeds, but my numbers just aren't adding up. 3DMark Vantage, with one card, scored me at a P7795, with a GPU score of 8488 and CPU of 6261. Add the other card and configure Crossfire and it bumps up to P10147, GPU - 13208, CPU - 5986. Is it me or are these numbers retarded low even for normal clock speeds?
 

rkwerchan

Distinguished
May 15, 2009
15
0
18,510
Really? Have there been issues with 7? I guess I could probably read up on it and figure it out for myself, but while I've got your attention.......

oh, and I love the avatar! I grew up around Stevie... my mom used to date Tommy Shannon back in the day... that was my first concert ever... kick ass!
 

boudy

Distinguished
Mar 13, 2009
498
0
18,790
Well, the drivers are just tweaked Vista drivers, so I could really see some problems with it (Since the drivers arent really made and optimized for Windows 7).
If you have enough space on your hard drive, you could probably dual boot and have Win7 on one partition and (Insert OS of preference here) on the other

oh, and I love the avatar! I grew up around Stevie... my mom used to date Tommy Shannon back in the day... that was my first concert ever... kick ass!

Damn, your so freaking lucky! That's awesome though. I can only dream of something like that.
 

rkwerchan

Distinguished
May 15, 2009
15
0
18,510
Alright... I'm gonna give that a shot and I'll let you know how it works in a bit... thanks for the tip... I wouldn't have thought of that one...

and yeah, it was pretty cool until I got older and still had a hippy mom... it was a sad day in my house when he died...
 

boudy

Distinguished
Mar 13, 2009
498
0
18,790
No problem, I know a few people who are doing it and they say it works fine. Let me know if you are still having trouble.

and yeah, it was pretty cool until I got older and still had a hippy mom

lol that sounds like fun (*sarcasm alert*)

it was a sad day in my house when he died

It was a sad day for everyone, and it still is sad :(
 

rkwerchan

Distinguished
May 15, 2009
15
0
18,510
Ok... clean Vista Ultimate x64 install - same specs with both cards - P10109 - GPU 12812 (that's my badge number at work!) - CPU 6191... so... CPU went up a bit, everything else lowered a bit... so it's not Windows 7... could it be that my motherboard is a little older? I appreciate the help, by the way...

*edit* this was run at a resolution of 1600x1200... I'm gonna try 1280x1024... also, looking at this site almost makes my numbers look normal, but I thought they should be higher... I dunno...

http://www.legitreviews.com/article/734/15/
 
I see nothing wrong with these numbers.
You won't get 2x the benchmark numbers with two cards. A slight drop for the CPU seems reasonable as there's likely some overhead with adding the second card.

Generic benchmarks add up points for a total score. An improvement of 30% when adding a second card doesn't necessarily translate to 30% improvement in all games.

A better way to test is to do some gaming benchmarks such as Crysis. Games optimized for SLI should usually see between 30% and 80% improved frame rates though it varies by game, settings, resolution etc.

Again, your numbers seem fine.
 

rkwerchan

Distinguished
May 15, 2009
15
0
18,510
Thanks photonboy... I've been trying to find a few tests to compare mine too, but haven't had a whole lot of luck... did find one in another thread on this board and I'm pretty close to that one, so I guess I'm satisfied... dumb thing is that I don't even do any computer gaming or anything... I just keep getting good deals on parts so I pick em up and put em on... This is my first build that's been a work in progress for the last year... :heink: