Benchmarking the Native 10.6.7 ATI Drivers- Submit Yours!

[tonymacx86]

 

[InActive8618]

My 5770 scores with cpu oc @3,36):

10.6.7 regular ati kexts: 38,55
10.6.7 MBP at kexts: 39,38

Slightly better but having problems with geekbench.



Just for the record, with my native cpu clocks (@2,8):

10.6.6 regular ati kexts: 32-33

 

[vvlad]

Hi

10.6.7 MBP drivers: 36.21
10.6.6 regular kext: 35.66

Tests run on cinebench 11.5 64 bit, with xfx radeon 5850 black edition, [email protected], 8GB RAM.

 

[Say Heading]

Gigabyte SilentCell 5770 - i7 2600

 

[tonymacx86]

Interesting- it seems like these drivers are working better with SB cpus. Could be that the tests are flawed/cpu biased...

 

[justr]

Interesting- it seems like these drivers are working better with SB cpus. Could be that the tests are flawed/cpu biased...

That's what I was wondering. Any chance that SB is just that much more efficient at feeding data to the GPU?

 

[grafixidea]

the opposite here...

10.6.6 -> 33.80
10.6.7 + MBP drivers -> 31.XX !!!!

I feel that I did something wrong

Slightly better but having problems with geekbench.

Yes, I had this also before rolling back to 10.6.6 via clean install. What is the problem with Geekbench? anybody knows???

Going to test updating to 10.6.7 and MBP ATI drivers again but after getting another HDD so I can make tests on separate install.

 

[samuel77]

My computer seems slightly faster with the new kernel and drivers.

cinebench 11.5 scores:

New drivers and kernel
open gl 40.97
cpu 6,63

Old drivers and kernel
open gl 38,4
cpu 6.14

asus p8p67 intel i7 2600
kingston 1600x
saphire radeon 5770

 

[justr]

My computer seems slightly faster with the new kernel and drivers.

cinebench 11.5 scores:

New drivers and kernel
open gl 40.97
cpu 6,63

Old drivers and kernel
open gl 38,4
cpu 6.14

asus p8p67 intel i7 2600
kingston 1600x
saphire radeon 5770

Why I don't understand is why Sandy Bridge systems are showing much higher GPU scores than Bloomfield systems... but lower CPU scores!

For instance, on 4.0 Ghz i7-950 w/ Gigabyte 5770, I get ~6.94 and 32.32 respectively for CPU and GPU scores using the old drivers. OK, I'm overclocked moderately, but I'd still expect a 2600 to crush my results in the CPU area.

And given that the 2600 actually scores significantly lower (6.14 vs. my 6.94), I wouldn't expect that same SB system with an almost identical GPU to post GPU scores that are nearly 30% higher than mine.

What's going on? Are SB CPU results not accurate? Are SB CPUs/architecture that much more efficient at feeding the GPU? Is Cinebench flawed in some significant way?

 

[Implied]

Interesting- it seems like these drivers are working better with SB cpus. Could be that the tests are flawed/cpu biased...

Could it be due to the Built-in Graphics on the Sandy Bridge CPUs? Maybe they somehow add some FPS to the Benchmarks? Like some sort of interlacing between the 2 GPUs was added or something?