4850 overclocking gains in GRID: 775/1003 vs 625/993

[dug777]

Core clock increase of 24%, mem of 1%.

Q6600@stock/4GB DDR2/P5Q PRO/Asus 4850 512mb, different tracks from the demo each time (so you can only compare each one to its pair).

Was just done doing the replay function, so each 'run' is indentical for each clock speed (any way to make the replays longer?).

Everything maxed (seems to revert to 'custom' when I change the res (1600x1200 75Hz), no vysnc, latest cats (high quality mode, edge-detect AA + Adaptive - Performance)

1600x1200 8xAA
oc'd
Frames Time (ms) Min Max Avg
461 10140 35 54 45.464

stock
Frames Time (ms) Min Max Avg
436 10625 30 49 41.035

(11% on avg)

1600x1200 4xAA

oc'd
Frames Time (ms) Min Max Avg
457 8758 44 65 52.181

stock
Frames Time (ms) Min Max Avg
449 9577 39 59 46.883

(11% on avg)

1600x1200 2xAA

oc'd
Frames Time (ms) Min Max Avg
587 10181 45 72 57.656
stock
Frames Time (ms) Min Max Avg
546 10346 41 66 52.774

(9% on avg)

Has a nice increase on minimums, seems odd that I got a 16% improvement for the 8xAA run, but only 10% for 2xAA. I would have thought mem limitations would have limited the gains for the higher AA run more...

 

[AzN]

Grid has biggest impact when you increase shader. I wish ATI cards could overclock shader seperately.

 

[Tempered81]

Nice results, dug

 

[Phew]

I don't know how CPU-bound GRID is, but could it be that your CPU at only 2.4 GHz is holding back that overclocked 4850? Easy way to check would be to OC the Q6600 and redo some of the tests.

 

[dug777]

Originally posted by: Phew
I don't know how CPU-bound GRID is, but could it be that your CPU at only 2.4 GHz is holding back that overclocked 4850? Easy way to check would be to OC the Q6600 and redo some of the tests.

All runs below at 3Ghz (333x9):

1600x1200 4xAA:
Oc'd GPU
Frames Time (ms) Min Max Avg
568 10176 44 64 55.818
stock GPU
Frames Time (ms) Min Max Avg
516 10226 39 60 50.46

No real change, looks like I'm firmly GPU limited at that res/AA level. (ugh, my board is running 1.392vcore for auto at that speed 😛)

 

[toyota]

Originally posted by: dug777

Originally posted by: Phew
I don't know how CPU-bound GRID is, but could it be that your CPU at only 2.4 GHz is holding back that overclocked 4850? Easy way to check would be to OC the Q6600 and redo some of the tests.

Went up to 3Ghz (333x9):

1600x1200 4xAA:
Oc'd
Frames Time (ms) Min Max Avg
568 10176 44 64 55.818
stock
Frames Time (ms) Min Max Avg
516 10226 39 60 50.46

No real change, looks like I'm firmly GPU limited at that res/AA level. (ugh, my board is running 1.392vcore for auto at that speed 😛)

so you overclock your gpu and those 10% results are good but the 10% you got from overclocking dont mean anything?

 

[dug777]

Originally posted by: toyota

Originally posted by: dug777

Originally posted by: Phew
I don't know how CPU-bound GRID is, but could it be that your CPU at only 2.4 GHz is holding back that overclocked 4850? Easy way to check would be to OC the Q6600 and redo some of the tests.

Went up to 3Ghz (333x9):

1600x1200 4xAA:
Oc'd
Frames Time (ms) Min Max Avg
568 10176 44 64 55.818
stock
Frames Time (ms) Min Max Avg
516 10226 39 60 50.46

No real change, looks like I'm firmly GPU limited at that res/AA level. (ugh, my board is running 1.392vcore for auto at that speed 😛)

so you overclock your gpu and those 10% results are good but the 10% you got from overclocking dont mean anything?

😱 I do apologise but I don't follow you 😱

With my CPU at 3Ghz, I saw the same 11% improvement in average FPS from oc'ing the GPU. So the CPU overclock doesn't seem to have allowed my GPU oc to 'strech its' legs' and open up more of a gap over the stock clock run.

Since I didn't use the same track or replay to compare this 4xAA run to my first 4xAA run, I can't really validly compare them, but on the face of it it does appear that the average has risen quite significantly as a result of my CPU oc. Is that where you are coming from?

 

[vj8usa]

Originally posted by: toyota

Originally posted by: dug777

Originally posted by: Phew
I don't know how CPU-bound GRID is, but could it be that your CPU at only 2.4 GHz is holding back that overclocked 4850? Easy way to check would be to OC the Q6600 and redo some of the tests.

Went up to 3Ghz (333x9):

1600x1200 4xAA:
Oc'd
Frames Time (ms) Min Max Avg
568 10176 44 64 55.818
stock
Frames Time (ms) Min Max Avg
516 10226 39 60 50.46

No real change, looks like I'm firmly GPU limited at that res/AA level. (ugh, my board is running 1.392vcore for auto at that speed 😛)

so you overclock your gpu and those 10% results are good but the 10% you got from overclocking dont mean anything?

Doesn't seem to be quite a 10% gain from the CPU OC. The average and max went up, but the min framerate didn't budge.

 

[toyota]

Originally posted by: vj8usa

Originally posted by: toyota

Originally posted by: dug777

Originally posted by: Phew
I don't know how CPU-bound GRID is, but could it be that your CPU at only 2.4 GHz is holding back that overclocked 4850? Easy way to check would be to OC the Q6600 and redo some of the tests.

Went up to 3Ghz (333x9):

1600x1200 4xAA:
Oc'd
Frames Time (ms) Min Max Avg
568 10176 44 64 55.818
stock
Frames Time (ms) Min Max Avg
516 10226 39 60 50.46

No real change, looks like I'm firmly GPU limited at that res/AA level. (ugh, my board is running 1.392vcore for auto at that speed 😛)

so you overclock your gpu and those 10% results are good but the 10% you got from overclocking dont mean anything?

Doesn't seem to be quite a 10% gain from the CPU OC. The average and max went up, but the min framerate didn't budge.

what? 39-44 is a 13% gain

 

[dug777]

Originally posted by: toyota

Originally posted by: vj8usa

Originally posted by: toyota

Originally posted by: dug777

Originally posted by: Phew
I don't know how CPU-bound GRID is, but could it be that your CPU at only 2.4 GHz is holding back that overclocked 4850? Easy way to check would be to OC the Q6600 and redo some of the tests.

Went up to 3Ghz (333x9):

1600x1200 4xAA:
Oc'd
Frames Time (ms) Min Max Avg
568 10176 44 64 55.818
stock
Frames Time (ms) Min Max Avg
516 10226 39 60 50.46

No real change, looks like I'm firmly GPU limited at that res/AA level. (ugh, my board is running 1.392vcore for auto at that speed 😛)

so you overclock your gpu and those 10% results are good but the 10% you got from overclocking dont mean anything?

Doesn't seem to be quite a 10% gain from the CPU OC. The average and max went up, but the min framerate didn't budge.

what? 39-44 is a 13% gain

He's looking to my first run (39 and 44 minimums).

Like I said, I can't really compare the two since they're not the same track or replay, but on the face of it the avg improved, but not the min and barely the max.

 

[AzN]

Originally posted by: dug777

Originally posted by: Phew
I don't know how CPU-bound GRID is, but could it be that your CPU at only 2.4 GHz is holding back that overclocked 4850? Easy way to check would be to OC the Q6600 and redo some of the tests.

Went up to 3Ghz (333x9):

1600x1200 4xAA:
Oc'd
Frames Time (ms) Min Max Avg
568 10176 44 64 55.818
stock
Frames Time (ms) Min Max Avg
516 10226 39 60 50.46

No real change, looks like I'm firmly GPU limited at that res/AA level. (ugh, my board is running 1.392vcore for auto at that speed 😛)

At 1600x1200 4xAA it becomes more GPU limited than CPU limited. At least with 4850.

 

[vj8usa]

Originally posted by: toyota
what? 39-44 is a 13% gain

I think you're misinterpreting the numbers (or maybe I am). How I see it, the "oc'd" and "stock" values that you quoted are BOTH with the 3GHz CPU. The ones in the very first post in this thread are both with the CPU at stock. Both sets of data have 39/44 as the min FPS.

 

[dug777]

Originally posted by: vj8usa

Originally posted by: toyota
what? 39-44 is a 13% gain

I think you're misinterpreting the numbers (or maybe I am). How I see it, the "oc'd" and "stock" values that you quoted are BOTH with the 3GHz CPU. The ones in the very first post in this thread are both with the CPU at stock. Both sets of data have 39/44 as the min FPS.

You are correct, apologies for the extreme confusion I'm creating here!

 

[AzN]

Originally posted by: vj8usa

Originally posted by: toyota
what? 39-44 is a 13% gain

I think you're misinterpreting the numbers (or maybe I am). How I see it, the "oc'd" and "stock" values that you quoted are BOTH with the 3GHz CPU. The ones in the very first post in this thread are both with the CPU at stock. Both sets of data have 39/44 as the min FPS.

Don't mind toyota. He thinks CPU makes huge difference at GPU limited situations. He thought Crysis would run high setting on 8600gt if he has faster CPU.

 

[MustangSVT]

Originally posted by: Azn

Originally posted by: vj8usa

Originally posted by: toyota
what? 39-44 is a 13% gain

I think you're misinterpreting the numbers (or maybe I am). How I see it, the "oc'd" and "stock" values that you quoted are BOTH with the 3GHz CPU. The ones in the very first post in this thread are both with the CPU at stock. Both sets of data have 39/44 as the min FPS.

Don't mind toyota. He thinks CPU makes huge difference at GPU limited situations. He thought Crysis would run high setting on 8600gt if he has faster CPU.

ditto, ignore xxxxxs (insert your choice of word there )

 

[toyota]

Originally posted by: Azn

Originally posted by: vj8usa

Originally posted by: toyota
what? 39-44 is a 13% gain

I think you're misinterpreting the numbers (or maybe I am). How I see it, the "oc'd" and "stock" values that you quoted are BOTH with the 3GHz CPU. The ones in the very first post in this thread are both with the CPU at stock. Both sets of data have 39/44 as the min FPS.

Don't mind toyota. He thinks CPU makes huge difference at GPU limited situations. He thought Crysis would run high setting on 8600gt if he has faster CPU.

there you go again with your lying BS, trolling and usual flamebaiting but everybody already knows what kind of person you are.


my comment to him was based on his own graph. go back and read his post because it does look like he had a gain from oc on the cpu but its hard to really tell exactly what he is saying.

 

[vj8usa]

dug, did you try OCing your 4850's RAM further? I'd be interested to see how much of an increase you get from OCing RAM vs OCing GPU.
Or were you not stable beyond the 1% OC?

 

[dug777]

Originally posted by: vj8usa
dug, did you try OCing your 4850's RAM further? I'd be interested to see how much of an increase you get from OCing RAM vs OCing GPU.
Or were you not stable beyond the 1% OC?

I think I'm good up to about 1023, will give it a try and let you know.

I always seem to end up with cards that do insane oc's on either the memory or the core (6600GT did an insane core oc', but mem wouldn't budge, 6800GS did an insane mem oc but core would only get somewhere equivalent to a 6800U, this one does a utterly insane core oc but mem barely budges), but never both. Can't have it all I suppose.

 

[dug777]

Originally posted by: toyota

Originally posted by: Azn

Originally posted by: vj8usa

Originally posted by: toyota
what? 39-44 is a 13% gain

I think you're misinterpreting the numbers (or maybe I am). How I see it, the "oc'd" and "stock" values that you quoted are BOTH with the 3GHz CPU. The ones in the very first post in this thread are both with the CPU at stock. Both sets of data have 39/44 as the min FPS.

Don't mind toyota. He thinks CPU makes huge difference at GPU limited situations. He thought Crysis would run high setting on 8600gt if he has faster CPU.

there you go again with your lying BS, trolling and usual flamebaiting but everybody already knows what kind of person you are.


my comment to him was based on his own graph. go back and read his post because it does look like he had a gain from oc on the cpu but its hard to really tell exactly what he is saying.

It wasn't clear 😱

To confirm, initial post is all stock, latter post is all 3Ghz, and 'oc'd' and 'stock' in the tabular format refer to the GPU.

 

[toyota]

Originally posted by: dug777

Originally posted by: toyota

Originally posted by: Azn

Originally posted by: vj8usa

Originally posted by: toyota
what? 39-44 is a 13% gain

I think you're misinterpreting the numbers (or maybe I am). How I see it, the "oc'd" and "stock" values that you quoted are BOTH with the 3GHz CPU. The ones in the very first post in this thread are both with the CPU at stock. Both sets of data have 39/44 as the min FPS.

Don't mind toyota. He thinks CPU makes huge difference at GPU limited situations. He thought Crysis would run high setting on 8600gt if he has faster CPU.

there you go again with your lying BS, trolling and usual flamebaiting but everybody already knows what kind of person you are.


my comment to him was based on his own graph. go back and read his post because it does look like he had a gain from oc on the cpu but its hard to really tell exactly what he is saying.

It wasn't clear 😱

To confirm, initial post is all stock, latter post is all 3Ghz, and 'oc'd' and 'stock' in the tabular format refer to the GPU.

well when you said the following : "Went up to 3Ghz (333x9):

1600x1200 4xAA:
Oc'd
Frames Time (ms) Min Max Avg
568 10176 44 64 55.818
stock
Frames Time (ms) Min Max Avg
516 10226 39 60 50.46
No real change, looks like I'm firmly GPU limited at that res/AA level. (ugh, my board is running 1.392vcore for auto at that speed. "

I thought that those were just 2 runs comparing the cpu speed. thats the reason I asked why you said no change. anyway it gave Azn something to troll about so at least he is happy.

 

[angry hampster]

Originally posted by: Azn
Grid has biggest impact when you increase shader. I wish ATI cards could overclock shader seperately.

Is this true of all ATI cards and with all overclocking utilities? I am upgrading my 8800GT to an HD4870 this week, and have no experience overclocking ATI cards. Will I still be able to use Rivatuner?

 

[dug777]

CPU is back at stock.

1600x1200 4xAA:

stock
Frames Time (ms) Min Max Avg
527 10526 41 56 50.067

625/1023
Frames Time (ms) Min Max Avg
532 10576 41 56 50.303

625/1063
Frames Time (ms) Min Max Avg
539 10555 42 57 51.066

775/1063
Frames Time (ms) Min Max Avg
592 10506 44 63 56.349

Different replay again.

 

[AzN]

Originally posted by: toyota

Originally posted by: Azn

Originally posted by: vj8usa

Originally posted by: toyota
what? 39-44 is a 13% gain

I think you're misinterpreting the numbers (or maybe I am). How I see it, the "oc'd" and "stock" values that you quoted are BOTH with the 3GHz CPU. The ones in the very first post in this thread are both with the CPU at stock. Both sets of data have 39/44 as the min FPS.

Don't mind toyota. He thinks CPU makes huge difference at GPU limited situations. He thought Crysis would run high setting on 8600gt if he has faster CPU.

there you go again with your lying BS, trolling and usual flamebaiting but everybody already knows what kind of person you are.


my comment to him was based on his own graph. go back and read his post because it does look like he had a gain from oc on the cpu but its hard to really tell exactly what he is saying.

You can call me a liar all you like but I'm just quoting what you said. :disgust:

Here you go... http://forums.anandtech.com/me..._key=y&keyword1=crysis

Originally posted by: toyota
high settings for non core2 cpu users? NO. medium settings at 1024x768 is playable until after the core level then its all low for the snow and carrier levels. with a good core2 duo you could probably put most of the settings on high at 1024x768. regardless of what people "think" Crysis is very cpu demanding below 1280x1024.

 

[toyota]

Originally posted by: Azn

Originally posted by: toyota

Originally posted by: Azn

Originally posted by: vj8usa

Originally posted by: toyota
what? 39-44 is a 13% gain

I think you're misinterpreting the numbers (or maybe I am). How I see it, the "oc'd" and "stock" values that you quoted are BOTH with the 3GHz CPU. The ones in the very first post in this thread are both with the CPU at stock. Both sets of data have 39/44 as the min FPS.

Don't mind toyota. He thinks CPU makes huge difference at GPU limited situations. He thought Crysis would run high setting on 8600gt if he has faster CPU.

there you go again with your lying BS, trolling and usual flamebaiting but everybody already knows what kind of person you are.


my comment to him was based on his own graph. go back and read his post because it does look like he had a gain from oc on the cpu but its hard to really tell exactly what he is saying.

I'm just quoting what you said. I can search the thread where you said that if you'd like. :laugh:

you already tried that over on techreport and you know that wasnt exactly what I said. that was out of context and you know it. I was saying that at 1024x768 a core 2 duo would help if you wanted to turn some settings up and I wasnt meaning with just using an 8600gt.


we can also show that real mature pm that you sent me that is real though.



now back on topic I never said anything about his cpu and this game other than to question his graph.

 

[AzN]

Originally posted by: toyota

Originally posted by: Azn

Originally posted by: toyota

Originally posted by: Azn

Originally posted by: vj8usa

Originally posted by: toyota
what? 39-44 is a 13% gain

I think you're misinterpreting the numbers (or maybe I am). How I see it, the "oc'd" and "stock" values that you quoted are BOTH with the 3GHz CPU. The ones in the very first post in this thread are both with the CPU at stock. Both sets of data have 39/44 as the min FPS.

Don't mind toyota. He thinks CPU makes huge difference at GPU limited situations. He thought Crysis would run high setting on 8600gt if he has faster CPU.

there you go again with your lying BS, trolling and usual flamebaiting but everybody already knows what kind of person you are.


my comment to him was based on his own graph. go back and read his post because it does look like he had a gain from oc on the cpu but its hard to really tell exactly what he is saying.

I'm just quoting what you said. I can search the thread where you said that if you'd like. :laugh:

you already tried that over on techreport and you know that wasnt exactly what I said. we can also show that real mature pm that you sent me that is real though.


now back on topic I never said anything about his cpu and this game other than to question his graph.

Oh really? This is what you said about 8600gt playing crysis high detail..

high settings for non core2 cpu users? NO. medium settings at 1024x768 is playable until after the core level then its all low for the snow and carrier levels. with a good core2 duo you could probably put most of the settings on high at 1024x768. regardless of what people "think" Crysis is very cpu demanding below 1280x1024.

You want some more? :laugh:

Originally posted by: toyota
IT IS CPU LIMITED AT 1280X1024 OR BELOW. your own link shows that a fast core2 cpu will nearly DOUBLE the framerate over an AMD cpu at 1024x768. below 1280 with my comp I would get a bigger increase in fps by going with a fast core2 cpu than going with a faster card and thats a fact. I used an 8800gt for two weeks and it only made 5 fps difference(from 31 to 36fps) at 1024 over my 8600gt.