Sign in with
Sign up | Sign in
Your question
Closed

Crossfire underperforming.. help very much appreciated!

Last response: in Graphics & Displays
Share
January 14, 2012 12:22:08 PM

Edit: SOLVED YEAAEAAEAAAA WOEWOEKWOKEOWKOEJIRJOEARJ
SOLUTION: was starting to think i had to blow some cash, u guys hecka saved me!
did all the steps you guys mentioned (saving reformat for last… that was a pain) but it has me having a nerdgasm right now LOL. With/or above the comparisons I gave. the 140 upgrade on the 5870 was well worth it!

ok, dfusco, completely right about the overhead. Did a reformat, installed the drivers + cap2, followed the steps on the cata page. need4speeds and megamanx00 - moved the graphic cards to the right places, some websites showed some improvements on this as well, but man, def worked here.

the comparison vitornob mentioned was interesting. Def will OC later when I feel the need, but as of now ALL games are maxed and playing like butter. Metro 2033, BF3, Dead Island, crysis and older games as well, saw some smexyyyy improvements. All maxed, AAs and all, it's so pretty and smooth. Even that insanely long benchmark for RE5 saw improvements, and i'm sure it'll be better when i OC down the road. having FPS in the 150s I'm not one to complain. lol will save this benchmarking tool when I OC just to record the difference. According to forums on this specific processor it can be pushed pretty far, probably end up seeing as little as +5 to +10 fps and every extra bit is nice.

My 3dmark vantage scores rose CONSIDERABLY. Compared with the diff options I have tried over the week rose min from 1k.

Also just the accumulated random crap and changes that happened without my knowledge over time was probably the biggest enemy here. PC no longer when opening certain applications and whatnot am I forced into the basic mode (I had done it for so long I just kinda forgot it happened and accepted it/happens all the time on school vista/w7 comps lol). I think the only unanswered question I had was running things in 2d mode… I still don't get what you mean/how to keep things running in 3d mode when clocking for even better scores? Would be even better to have more improvements. As of now my expectations have been met and surpassed in some aspects woo hoo. you guys have a great day. pretty beast now

thanks a lot guys, that gave me more heartache than it should have. I need to start backing things up more often so it won't take me so long next time haha

MUCH LOVE! UPVOTES FOR U GENIUSES! oh the simple things ;) 
double update:

all listed specs and placement of every item into mobo are exactly. the. same. His setup outperforms mine by 2,000. (matching with gaming website's benchmarks)

it's like he spends a dollar for a buffet and I can only receive a soda because the air we breath is bottlenecking me. haha this is the best metaphor I could make. The exact same thing being used is bottlenecking me. This is why I thought I had hardware issues or messed up a simple step during setup, or over 2 years a software issue has come up and plagued me.

i think this is a big difference between two exactly. same. computers. i simply understand bigger numbers = better and 2000 seems like a big difference to me.


Update: sorry, I must have been unclear =(.

I've built the exact. same. rig. as a friend. and HIS rig outperforms mine by 2,000 on the gpu spectrum on 3dmarkvantage and his gaming fps were consistently higher than mine. (same cpu) so the bottleneck people are mentioning is strange since he's doing just fine with the same cpu.

This is why I believed a setting or simple procedure I don't understand/haven't done has been overlooked... I did find out I made the mistake of placing the 5870 second instead of first in the PCI slot. That is fixed.

Both built the rig 2 years ago and upgraded to crossfire two weeks ago.

Hi everyone. I've been a lurker on Tomshardware for a while now, and I really appreciate the community here =) Me and my friend have the same setups, and mine has been constantly underperforming and has me worried...

Specs:
Windows 7 64
Monitor: HANNspree 32'' 1080p (1920x1080) 16:10
Thermaltake TR2 600W
Western Digital Caviar Green 1TB 5400 RPM
OCZ Gold 6GB (3x2GB)
MSI 790FX-GD70 AM3 AMD 790FC ATX AMD Mobo
AMD Phenom II X2 550 Black Edition Callisto 3.1 GHz Socket AM3 80W Dual-Core
(Unlocked it to quad core)
Full case with 3 fans, keeps cards around 40 idle, 70 gaming
Sapphire Radeon HD 5850 1GB (slot 1)

Under 3DMark Vantage, Performance Preset my 5850x5870 got a score of 16189. About 2,000 less than 5850x2 (http://www.tomshardware.com/reviews/radeon-hd-5850,2433...) but according to other sites it should be about the same.

Using Resident Evil 5's benchmark test I'm getting 129.7fps@1920x1080 and 146.7fps@ 1280x1024 whereas http://www.legitreviews.com/article/1107/4/ is getting 161.7

Through Fraps:

In Metro 2033 my single 5850 played at max with max AA around 12-20fps (averaging at 18). Crossfired I'm getting around 30 average.

BF3 saw slight improvements, around +7fps

Crysis 2 saw slight improvements as well: ~+10fps

random note (I don't understand if this means anything): through GPU-Z the GPU Core Clock and Memory Clock are far higher on tab 1 "5800 series" with 1440 shaders (I assume the 5850) than the card with 1600 shaders (5870)
5850: 400MHZ Core Clock, 1000 MHZ Memory Clock
5870: 157MHZ Core Clock 300MHZ Memory Clock

I was definitely expecting more and am boggled by my 3dMark Vantage and RE5 benchmarks. My friend has the exact same setup (cpu and all) and his scores are much better than mine.


thanks so much for anyone's help! I'm worried I have a dud video card or something...

Best solution

January 14, 2012 12:35:35 PM

If you use your gaming rig as your everyday PC it will be running all sorts of things a testing rig on a tech site will not. It will run slower because of the extra overhead. Your scores actually look pretty good.
Share
January 14, 2012 12:54:00 PM

Your pulling over 510w and I doubt that psu puts out a true 600w. I don't think it'll last long.

Your running a 5400rpm hd, that's gotta hurt performance.

Your also taking a performance hit by running your memory in single channel mode rather than the optimal dual channel mode.
Score
0
Related resources
January 14, 2012 3:46:39 PM

dfusco said:
If you use your gaming rig as your everyday PC it will be running all sorts of things a testing rig on a tech site will not. It will run slower because of the extra overhead. Your scores actually look pretty good.


hm... Actually my friend (same PC, higher benchmarks) reformatted a while ago (as opposed to myself, it's been over 1.5 years). I will try reformatting and testing things again ASAP.

To Geekapproved: Could the lack of power be a cause of this? My friend with the same rig and power supply so far has had no issues...

as for the 5400rpm, my friend with the same gaming rig is on the same hard drive yet with much higher benchmarks than me.

Oh and.... I'm a noob to this... what do you mean by dual channel mode? where did I set the rig up improperly I would like to fix this.

hm so far I need to look into reformatting and setting my memory into dual channel mode. I'm wondering if that's what made such a big difference between two exact same rigs. Thanks for the replies so far, still looking for suggestions/answers if anyone else would like to contribute, thank you all!
Score
0
January 14, 2012 4:15:38 PM

Put the faster 5870 card in your first pcie slot so that it is the lead or master card.

-Some games don't use crossfire much and mainly use the first card.
-You can overclock the cards separate. Set the 5850 at the overclock settings that tested ok before.
-Underclock the 5870 a bit so that the clocks are closer.

If you check the speed when you are just running the 2d desktop, the cards will go to slower 2d clocks to save power.
Score
0
January 14, 2012 5:00:03 PM

need4speeds said:
Put the faster 5870 card in your first pcie slot so that it is the lead or master card.

-Some games don't use crossfire much and mainly use the first card.
-You can overclock the cards separate. Set the 5850 at the overclock settings that tested ok before.
-Underclock the 5870 a bit so that the clocks are closer.

If you check the speed when you are just running the 2d desktop, the cards will go to slower 2d clocks to save power.


-o wow I will change their placement tonight!
-I was planning on OCing the 5850 closer to 5870 instead of under clocking... I need to do more research on this as I hear just using overdrive and sliding switches to the right isn't the best way. I am looking into MSI Afterburner as soon as I feel safe that there isn't something wrong with any hardware and do way more research as I'm worried about messing something else up now =/
-As for the 2d desktop, do you mean Windows 7 Basic? Every time I start 3dmark vantage I get, "The color scheme has been changed to windows 7 basic: a running program isn't compatible with certain visual elements of windows." and then things become smaller/more simple/sharper looking, not all zoomed in like I normally have it on the 32 inch tv/monitor.
Score
0
January 14, 2012 5:16:16 PM

Place the 5870 in the first slot and connect your monitors to it so it will be the Master card. It will help to OC the 5850 to be closer to the 5870. Also keep in mind that you won't get the same scores in a test rig using an overclocked i-7 920 or better and two 5850s simply because your CPU will limit your cards in most synthetic benchmarks.
Score
0
January 14, 2012 5:49:05 PM

You know that you might have some CPU bottlenecking there. Usually your CPU would bottleneck SLI/Crossfire systems.

Just look at Resident Evil 5 results:
1280x1024 vs 1920x1080 (60% more pixels), going from 146,7fps to 129.7fps? (11,6% lower fps only)
There's some great CPU bottleneck there.

First solution: crazy CPU OC
Second, expensive solution: get a Sandy-bridge

Second solution better IMO, but first one is easier and cheaper :) 
Nice OC for you!!

EDIT: I have two slower cards than yours. So it would be an interesting comparative. I'm gonna download the Resident Evil 5 benchmark, put all possible candy-eye. And then do some benchmark. If I have higher scores than yours it would indicate that you some serious bottleneck problems. Since your faster cards isn't delivering their power due to CPU constrains.
Score
0
January 15, 2012 3:36:22 PM

sl33pyh34d said:
hm... Actually my friend (same PC, higher benchmarks) reformatted a while ago (as opposed to myself, it's been over 1.5 years). I will try reformatting and testing things again ASAP.

To Geekapproved: Could the lack of power be a cause of this? My friend with the same rig and power supply so far has had no issues...

as for the 5400rpm, my friend with the same gaming rig is on the same hard drive yet with much higher benchmarks than me.

Oh and.... I'm a noob to this... what do you mean by dual channel mode? where did I set the rig up improperly I would like to fix this.

hm so far I need to look into reformatting and setting my memory into dual channel mode. I'm wondering if that's what made such a big difference between two exact same rigs. Thanks for the replies so far, still looking for suggestions/answers if anyone else would like to contribute, thank you all!



So you and your friend both built crossfire gaming machines with 5400rpm drives, mismatched video cards, underpowered psu's and incorrect memory configurations......and you want to know why it's underperforming........What were you guys thinking???
Score
0
January 15, 2012 3:43:52 PM

geekapproved said:
So you and your friend both built crossfire gaming machines with 5400rpm drives, mismatched video cards, underpowered psu's and incorrect memory configurations......and you want to know why it's underperforming........What were you guys thinking???


hmm i must have been unclear.

same rig, same everything. His rig is living up to standards. Mine is not. about 2,000 points higher.

CPU and power bottlenecking only affecting me when he has the exact same rig? I don't understand.

I've put 5870 in slot 1 now. Wondering about what you mean on the 2d desktop, how do I change that?

Beyond this I suppose a reformat is needed.


double update: Ah, I read back through my posts and think my ideas were unclear, I apologize.

all listed specs and placement of every item into mobo are exactly the same. His setup outperforms mine by 2,000.

it's like he spends a dollar for a buffet and I can only receive a soda because the air we breath is bottlenecking me. haha this is the best metaphor I could find. The exact same thing being used is bottlenecking me. This is why I thought I had hardware issues or messed up a simple step during setup, or over 2 years a software issue has plagued me.

Score
0
January 15, 2012 4:10:33 PM

dfusco: Will reformat as soon as I can. I'm sure that'll help a bit, ty =) His school did make him buy a laptop so his school life and gaming life are separate. I pretty much do everything on the gaming pc.

need4speeds: You've pointed out something I overlooked! didn't see i didn't put the 5750 first. I'm sure this will help ty!

Need4speeds: What do you mean 2d desktop? How do I switch to 3d desktop when testing?

Vitornob: Ah, the resolution thing makes sense, ty. Still wondering about the large difference between his luck and mine tho.

thanks guys I'm sure this will help, still wondering about the difference. See my above post if I was unclear between my comparisons.
Score
0
January 15, 2012 8:14:21 PM

Next time you decide to build a gaming machine, talk to someone that has a clue about gaming machines before you buy your parts. Just sayin.
Score
0
January 16, 2012 1:59:15 AM

Best answer selected by sl33pyh34d.
Score
0
!