Overclock.net banner

Crossfire help

1K views 23 replies 6 participants last post by  Sir Amik Vase 
#1 ·
Seeing as prices on the 5850 are coming down now (~£80 on ebay for a used one) im thinking of going CF.
In the next few days im going to be upgrading to -
i5-2500k (Getting it to 4.8ghz i hope!)
Asus P8P67 LE

My current PSU is hyper type-M 580w.
Would i be able to run a crossfire rig with that CPU and mobo without a hitch or should i upgrade to a 6970/gtx 580?

Thanks,
Amik.

ps. My first day back here in 5 months!
 
See less See more
1
#20 ·
Quote:


Originally Posted by Nixuz
View Post

As I just posted in another thread, 16x-4x results in a loss of about 5% on the 2nd card.
It's in no way worth buying a new mobo for, unless you are just really anal about getting the absolute most performance possible.
http://www.techpowerup.com/reviews/A...caling/25.html

Those aren't even cfx results.

I can tell you from firsthand experience that the 4x slot is a handicap when running cfx.
Granted that most of the time without an fps counter you wouldn't notice.

I'll use BC2 as an example, I experienced many noticeable slowdowns at 16x/4x, 8x/8x exhibited no such slowdowns.

The 4x card is always 10-20% bottlenecked.

Attachment 194525

Attachment 194526

Attachment 194527


 
#21 ·
Quote:
Originally Posted by grunion;12303110
Those aren't even cfx results.

I can tell you from firsthand experience that the 4x slot is a handicap when running cfx.
Granted that most of the time without an fps counter you wouldn't notice.

I'll use BC2 as an example, I experienced many noticeable slowdowns at 16x/4x, 8x/8x exhibited no such slowdowns.

The 4x card is always 10-20% bottlenecked.

View attachment 194525

View attachment 194526

View attachment 194527
Unfortunately, hard evidence over the 16x-4x doesn't show the "slowdowns" that you noticed, on a number of applications.
http://www.overclock.net/nvidia/819348-16x-16x-vs-16x-4x-gtx.html
http://www.hardocp.com/article/2010/08/25/gtx_480_sli_pcie_bandwidth_perf_x16x16_vs_x4x4/
Unless SLI has figured out some magical way of scaling over 4x slots that AMD doesn't know about.
1282682382IabXi1rEcG_1_4_l.gif
 
#22 ·
Quote:
Originally Posted by Nixuz;12328206
Unfortunately, hard evidence over the 16x-4x doesn't show the "slowdowns" that you noticed, on a number of applications.
http://www.overclock.net/nvidia/819348-16x-16x-vs-16x-4x-gtx.html
http://www.hardocp.com/article/2010/08/25/gtx_480_sli_pcie_bandwidth_perf_x16x16_vs_x4x4/
Unless SLI has figured out some magical way of scaling over 4x slots that AMD doesn't know about.
1282682382IabXi1rEcG_1_4_l.gif
Post all the "hard evidence" you want, I know what I experienced, the bandwidth discrepancy can be unkind to my cards.
Compared to h220wnzj00 numbers, I saw twice the % loss that he experienced.

Vantage 36098 vs 32009, gpu score, total isn't relevant.
JC2 103 vs 92
Those are the only 2 test we have in common.
But like I said not noticeable for the most part unless you look at numbers, but BC2 was the exception.

It just ridiculous to tell a person to go ahead and get a 16x/4x board when they plan on running crossfire.
 
#23 ·
It's ridiculous to get a 16x/4x board, based upon your personal observations versus several sites showing that the performance difference is, at most, 5%?
I'm not saying that you didn't see what you claim to have seen, but you have to admit that it would be a stretch to dismiss several well respected sites conclusions based upon one persons negative personal experience.
Anyway, I expect we will continue to see more tests of this nature as more people, like myself, upgrade with the 16x/4x boards that they have handy.
As soon as AM3+ boards are available for a decent price, I'll try to get my hands on a 16x/16x and use everything else from my current setup, (barring RAM of course), to see what, if any, discrepancies I can reproduce.
 
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top