HD4850 vs. HD4770 DUEL

[SonDa5]

WILL DO!!!!

How about this weekend?

 

[SonDa5]

Some Peacemaker browser benchmarks with my HD4770s.

 

[theorw]

How about this weekend?

Well as far as i d like to do it i have still some heat issues with my 8400...
I am currently running 4.05ghz stable but i think i can do a 4,4ish for some bencehs...
And i ll stress those 4850s too and see what i get!
I now have a miserable 19k but both cards are @750/1000.
Will keep u updated for any results i have this weekend!

 

[theorw]

Unfortunately,while i managed after 50 or so reboots to boot on 4.45 GHz @1.5 volts,i run 3dmark06 only to reach the 2nd CPU TEST and hear the BIP-BIP before the OVERHEAT SHUTDOWN....
Will try the same thing later tonight when the temps are a bit lower(ambient now @31)
I have a total of 14 fans in my case and they dont seem enough to cool my 8400 successfully...
The gains of the 4,45 were nice though...(from what i noticed on the SM2.0 TESTS.135 fps on start-115fps on finish on proxycon)

 

[Scrizz]

u will lose in '06 seeing as you have a dual core.

 

[SonDa5]

u will lose in '06 seeing as you have a dual core.

Probably. Unless he can do some crazy over clocking.

We can still compare the score at least. 3dMark06 has 3 scores, 2 from GPU 1 from cpu.

 

[1Kurgan1]

Probably. Unless he can do some crazy over clocking.

We can still compare the score at least. 3dMark06 has 3 scores, 2 from GPU 1 from cpu.

Yeah, but the CPU directly (massively) effects the other 2 scores. Im going to use an extreme example, with a AMD 5000_ BE at 3.3ghz and a 4870x2 I was only scoring I think 12k. Then with a PII 720 at 3.7ghz and 4870x2 I was scoring 19.7k. The processor score for the 5000+ was like 2300 where as it was like 5k for the 720. So that accounts for 2700 of the 7700 point difference, but the rest is in the GPU scores.

 

[theorw]

Probably. Unless he can do some crazy over clocking.

We can still compare the score at least. 3dMark06 has 3 scores, 2 from GPU 1 from cpu.

Yeah that was exactly what i was thinking !We eill compare the gpu scores only!!!Agree on that?
But tommorow this will be cos now ny girl is calling!!!

Yeah, but the CPU directly (massively) effects the other 2 scores. Im going to use an extreme example, with a AMD 5000_ BE at 3.3ghz and a 4870x2 I was only scoring I think 12k. Then with a PII 720 at 3.7ghz and 4870x2 I was scoring 19.7k. The processor score for the 5000+ was like 2300 where as it was like 5k for the 720. So that accounts for 2700 of the 7700 point difference, but the rest is in the GPU scores.

Of course the faster processor allows for better scores but that only to an extend.
I mean yeah a prescot will score 8k on my system while 840016k by default but theres a point where the processor doesnt offer THAT much!!!!
Theres a point that when the CPU unlocks all the GPU power only GPU matters then!
I hope that my 8400@4,4-4,6ghz will do that!
And then again there are my 4850s that i will OC some more tomorrow!

 

[1Kurgan1]

Of course the faster processor allows for better scores but that only to an extend.
I mean yeah a prescot will score 8k on my system while 840016k by default but theres a point where the processor doesnt offer THAT much!!!!
Theres a point that when the CPU unlocks all the GPU power only GPU matters then!
I hope that my 8400@4,4-4,6ghz will do that!
And then again there are my 4850s that i will OC some more tomorrow!

Actually when it comes to 3dmark06, there really isn't a benefit wall from a better proc. There are people who got very close (and maybe even broke) 30k on 3870x2's, I'm pretty sure someone here even scores like 28k with a single 3870x2 with a highly clocked Q9650. Where as my PII 720 thats clocked very well and plays any game maxed fine only pushed out 19.7k with a 4870x2. And a 4870x2 is a signifigantly better card than a 3870x2.

 

[theorw]

Actually when it comes to 3dmark06, there really isn't a benefit wall from a better proc. There are people who got very close (and maybe even broke) 30k on 3870x2's, I'm pretty sure someone here even scores like 28k with a single 3870x2 with a highly clocked Q9650. Where as my PII 720 thats clocked very well and plays any game maxed fine only pushed out 19.7k with a 4870x2. And a 4870x2 is a signifigantly better card than a 3870x2.

I am not referring to the overall score but to the GPU scores alone.
The CPU score affects the overall score so the higher the CPU score alone the higher the OVERALL score!
Thats why i only suggested comparing the GPU scores!Get me?

 

[Wile E]

I am not referring to the overall score but to the GPU scores alone.
The CPU score affects the overall score so the higher the CPU score alone the higher the OVERALL score!
Thats why i only suggested comparing the GPU scores!Get me?

Even the gpu scores go higher with a faster cpu. 06 is not very taxing on today's cards, so it's cpu bottlenecked. Raising the speed of the cpu raises the score of EVERY test.

At any rate, all the quad owner would have to do is set the numproc=2 switch in windows, and he can test with a "dual core" as well.

 

[1Kurgan1]

Yep, like Wile E said and I said before, in switching procs I picked up 7.7k points in 3dmark06, but only gained 2.3k on CPU score, meaning switching procs added 5.4k to my GPU scores, thats as much added per GPU scores as the CPU score, and theirs 2 GPU scores that got inflated so it just blows it way up.

 

[theorw]

Even the gpu scores go higher with a faster cpu. 06 is not very taxing on today's cards, so it's cpu bottlenecked. Raising the speed of the cpu raises the score of EVERY test.

At any rate, all the quad owner would have to do is set the numproc=2 switch in windows, and he can test with a "dual core" as well.

Yep, like Wile E said and I said before, in switching procs I picked up 7.7k points in 3dmark06, but only gained 2.3k on CPU score, meaning switching procs added 5.4k to my GPU scores, thats as much added per GPU scores as the CPU score, and theirs 2 GPU scores that got inflated so it just blows it way up.

Well i do the test in my 2core but when we compare these we might actually do the SET CORE thingie and test our cards on even terms with 2CORES RIGHT?

 

[SonDa5]

Well i do the test in my 2core but when we compare these we might actually do the SET CORE thingie and test our cards on even terms with 2CORES RIGHT?

If I can figure out how to set the 2 core thingie I'll do it to be fair.

I'm running Windows 7 so if anyone knows how to do this and can give me a hand in figuring this out it will be appreciated.

 

[theorw]

U can always try setting the AFFINITY on 3dmark only to 2 cores right???

 

[Wile E]

If I can figure out how to set the 2 core thingie I'll do it to be fair.

I'm running Windows 7 so if anyone knows how to do this and can give me a hand in figuring this out it will be appreciated.

Hit the WindowsKey+R to bring up the run box. Type msconfig and hit enter.

Go to the Boot tab, and select Advanced Options

In the advanced options window, check the number of processors box, and select 2 from the drop down.

Now reboot to your dual core machine. Don't forget to change back when you're done tho. lol.

 

[SonDa5]

Now reboot to your dual core machine. Don't forget to change back when you're done tho. lol.

Thanks for the info. I guess I'll have to cripple my machine to be fair.

Not sure how the Q9550 E0 will handle running on 2 cores. Looking forward to seeing how it reacts.

 

[theorw]

I will now try to run a 3dmark on 4,36 without the cpu test.Just GPU in order to avoid crash due to overheating...
Run your quad in 2cores at the same speed and we can compare results!

 

[theorw]

Well SONDA5 my friend we are close!
I managed the following scores this morning with some low ambient temps running SM2 and SM3 separately with my 4850s running at a moderate 750/1000 each.

I think with some more tweaking i ll beat ya

 

[ste2425]

how do i get hold of any 3D benchmark software? do i have to pay for the utility? As id like to share my performance results as im very very happy with my 4850

 

[theorw]

No man!No money required!
Download from futuremark the free edition of 3dmark06 and fire it up!!!

 

[ste2425]

well its downloading now ill be shore to post
i gues my rig will help out coz it'l show how the 4850 performs in a more mid/low range rig

well that my reults, in the little movies my graphics never went under 28 sometimes reach higher then 35, my cpu hower maxed out at 2 fps

 

[theorw]

well its downloading now ill be shore to post
i gues my rig will help out coz it'l show how the 4850 performs in a more mid/low range rig

http://img.techpowerup.org/090907/7777.jpg
well that my reults, in the little movies my graphics never went under 28 sometimes reach higher then 35, my cpu hower maxed out at 2 fps

U are SERIOUSLY CPU BOTTLENECKED my friend!

 

[SonDa5]

I will now try to run a 3dmark on 4,36 without the cpu test.Just GPU in order to avoid crash due to overheating...
Run your quad in 2cores at the same speed and we can compare results!

I'm about to try this. Not sure If I'll be able to run my Q9550 at 4.36GHZ with only 2 cores though. If not I will get it as close as possible.


Stay tuned.

 

[theorw]

Will do!
But did u check out my scores?
And they were achieved with a 2core,not quad!
The difference is only 700points i think in each test.Not THAT much justifying a QUAD being used dont u think?
Anyway tomorrow i am gonna give my 2nd 4850 some pencil too and push her all the way up to 800+MHz and see what i can do there
What clocks are u running yours anyway?