Overclocking 5750 (mobility 5850) - Results inside

henrikrox

macrumors 65816
Original poster
Feb 3, 2010
1,218
2
EDIT; Post #9, #12, #21 is updated with more benchmarks / more clocks, let me know of other games to test


So i decided to do a light overclock of the 5850 today.

My game of benchmark is Starcraft 2. I have used a singleplayer mission with alot of stuff going on in the beginning. I have used fraps to find the avg, max and minimum fps of the first 60 second of that mission.

TEST 1: (STOCK CLOCK) (628 core clock / 1000 memory clock)
Frames: 2926
Time (ms): 60 000

FPS
Min Max, Avg
41, 61, 48.767

Since the test only lasted 60 seconds the temperature only reached 58 degrees Celsius. However im using fans at 1800 1800 1600 rpm, while they are at stock 1000 1000 1000 rpm.

TEST 2: (700 core clock / 1050 memory clock)
Frames: 3328
Time (ms): 60 000

FPS:
Min, Max, Avg
46, 68, 55.467

Remember this is a very light over clock but the difference is about 7-8 fps which is good. One thing weirds me out, that the temp still is the same. Okay i didnt bencmark long but it still hovers around 60 degrees. Fans are still the same as test 1.

So thats one test out of the way. The gpu didnt break a sweat.

Now begins gpu stress testing to notice how far this card can go.

also for those are interested gpu-z temp readings are as attachments so you can check out the temps and clocks.
 

Attachments

henrikrox

macrumors 65816
Original poster
Feb 3, 2010
1,218
2
good to see the progress. You and saschke are some of the first ones to post up info regarding OCing 5850. So far so good. Saschke pushed it a little further but i'm sure you guys have a lot of room with fan controls added.

https://forums.macrumors.com/threads/986620/
Yes, there is alot of room left, i mean with the overclock i did the fans are still running at 20% of what that are capable off, so it can be pushed. That i know
 
Comment

Felias

macrumors member
Jul 29, 2010
74
0
Germany
Groovy, you're one of the first who is doing this officially. promising results already, i am anxious to see what's possible (should be waaaay more)
 
Comment

TMRaven

macrumors 68020
Nov 5, 2009
2,099
1
Play the ai for one game, then come back and use your player cam to get very accurate fraps benchmarks. You can adjust the fraps benchmark to last as long as you want.
 
Comment

henrikrox

macrumors 65816
Original poster
Feb 3, 2010
1,218
2
Didnt think of that TMraven, awesome tip, its late here in norway now, so ill do it when i wake up in a couple of hours.

Going to a third clocking test aswell,

so stock, overclock (mild) and overclock (medium)

will be exciting to see.
 
Comment

henrikrox

macrumors 65816
Original poster
Feb 3, 2010
1,218
2
5 min fraps benchmarks test.

Took a game vs the computer.

Watched the replay, and fast forward to about 5 min into the game before beginning benchmark. And i used player cam, so the benchmark is exactly the same when running stock and OC.

All details are on high, and some ultra, all at native resolution. i have the i5 quad core, 4gb ram.

BENCHMARK 1 (STOCK CLOCK - 628 core clock - 1000 memory clock)

Frames, Time (ms), Min, Max, Avg
3673, 300000, 46, 64, 58.217

BENCHMARK 2 (700 core clock - 1050 memory clock)

Frames, Time (ms), Min, Max, Avg
3939, 300000, 54, 76, 65.650

Kinda awesome considering how little it has been clocked, and the potential it has.

Im Loving it

Edit, i have tried with saschke clocks and the results are under

BENCHMARK 3 - 750 core clock - 1075 memory clock)

Frames, Time (ms), Min, Max, Avg
4067, 300000, 57, 78, 68.783

We are now seeing a 10-13 fps difference over stock clocks.

Which is great, temps never exceeded 71 degrees. they were running at 2000rpm / 2000rpm / 1500rpm. Still not making any loud noises. Just a hum in the background.


BTW:

If you have any other requests for games, please let me know, ill do my best to cover them and see what the difference with STOCK / OC is at those games. Sorry for my crap english today, i kinda stayed up all night and didnt go to bed
 
Comment

TMRaven

macrumors 68020
Nov 5, 2009
2,099
1
What's the max temp you get for gpu when odd fan is maxed, and you've been playing starcraft2 for a good hour or two?
 
Comment

henrikrox

macrumors 65816
Original poster
Feb 3, 2010
1,218
2
Never hadd the ODD fan maxed, far from it, havnt seen the need for it yet, when playing at OC clocks before, i played about 2 hours of campaign, max temp was 78.

The ODD fan has been around 30-40% max.

However i just feel its weird how cool this computer is. I mean, the cpu is now idling at 27 degrees. Thats kinda insane. didnt know the i5 was so much more cooler then the i7. Also the gpu is idling at 40-45 which is also kinda nice.
 
Comment

henrikrox

macrumors 65816
Original poster
Feb 3, 2010
1,218
2
One last benchmark :p lol this is hard to put away

Anyways here is the 3DMark Vantage peformance between stock and oc clocks for those who are into that.

I just have the trial version, but still you get the idea.

here are some pictures

STOCK CLOCK (628 / 1000)



GPU SCORE: 7224
GPU TEMP (max): 61 degrees celsius



OC CLOCK (750 / 1075)



GPU SCORE: 8384
GPU TEMP (max): 68 degrees celsius


Over 1000 score difference :), while watching the gpu demos, the oc ran much more smoother, and was pushing alot more fps. and i still cant believe those temps i get by increasing the fans just a bit. Check Macfan in the picture to see what im running at, and what the potential is.

Macfan is awesome to control temps, real temp for cpu and gpu-z for clock speed / gpu temp.
 
Comment

vandozza

macrumors 6502a
Jun 14, 2006
612
0
Australia
I'm guessing this is also possible on the 5670?

I'd like to try this in an i3 with the 5670. Could you link me to the tools that you're using. I've got Windows 7 64bit running in BootCamp.
 
Comment

GyroFX

macrumors 6502
Jan 14, 2002
389
2
Los Angeles and NorCal
very promising results. There doesn't seem to be a huge difference between your clocks and saschke's clocks. Maybe somewhere in between will be quit decent
 
Comment

KennyW

macrumors newbie
Sep 5, 2007
13
0
Any overclocking tools under OSX ?

Is there any visual difference with the 10-13fps increase for the games played ?
 
Comment

saschke

macrumors 6502
Feb 2, 2008
382
28
Very nice work Henrik!
Somehow MacFan isn't running for me. Will try later...

I think a OC to 750 / 1075 is perfectly fine, we just have to gauge how far we
can go without getting graphic errors :)

@ADMIN: Can we merge this thread and that one!?


EDIT:
Just to let you guys know - I've been running some tests with the OC settings (750/1075) yesterday. I'm really not easy to please framerate-wise so I'd like to list how some games run according to my perception. I never activate AA or AF:

Crysis Warhead: 1920*1080 -> GAMER -> PERFECTLY SMOOTH /// FULL2560 -> GAMER -> STILL ALLRIGHT
Modern Warfare 2: FULL2560 -> MAXXED OUT -> PERFECTLY SMOOTH
StarCraft 2: 1920*1080 -> MAXXED OUT -> PERFECTLY SMOOTH /// FULL2560 -> MAXXED OUT except shaders to high -> PERFECTLY SMOOTH
Bad Company 2: 1920*1080 -> MAXXED OUT -> PERFECTLY SMOOTH
PES2010: FULL2560 -> MAXXED OUT -> PERFECTLY SMOOTH
Splinter Cell Conviction: 1920*1080 -> MAXXED OUT -> PERFECTLY SMOOTH /// FULL2560 -> MAXXED OUT -> STILL ALLRIGHT

hope this helps..
 
Comment

henrikrox

macrumors 65816
Original poster
Feb 3, 2010
1,218
2
Very nice work Henrik!
Somehow MacFan isn't running for me. Will try later...

I think a OC to 750 / 1075 is perfectly fine, we just have to gauge how far we
can go without getting graphic errors :)

@ADMIN: Can we merge this thread and that one!?


EDIT:
Just to let you guys know - I've been running some tests with the OC settings (750/1075) yesterday. I'm really not easy to please framerate-wise so I'd like to list how some games run according to my perception. I never activate AA or AF:

Crysis Warhead: 1920*1080 -> GAMER -> PERFECTLY SMOOTH /// FULL2560 -> GAMER -> STILL ALLRIGHT
Modern Warfare 2: FULL2560 -> MAXXED OUT -> PERFECTLY SMOOTH
StarCraft 2: 1920*1080 -> MAXXED OUT -> PERFECTLY SMOOTH /// FULL2560 -> MAXXED OUT except shaders to high -> PERFECTLY SMOOTH
Bad Company 2: 1920*1080 -> MAXXED OUT -> PERFECTLY SMOOTH
PES2010: FULL2560 -> MAXXED OUT -> PERFECTLY SMOOTH
Splinter Cell Conviction: 1920*1080 -> MAXXED OUT -> PERFECTLY SMOOTH /// FULL2560 -> MAXXED OUT -> STILL ALLRIGHT

hope this helps..
Macfan is CMD, so the files you downloaded wont do anything. You have to start cmd as admin, then navigate to the macfan folder using cd and cd.. commands.
When you are in the macfan folder, just type Macfanx64.exe and see the fans you have, and what they are running at, and what their minimum and maximum rpm is then change accordingly. To change the rpm, just type macfanx64.exe 2000 2000 1500. That sets fan 1 & 2 to 2000 rpm, and fan 3 to 1500rpm. See the pictures for better explanation

Great work with the games you tested, its really a big difference playing games with stock clock and OC clock. Nice to see crysis running great. And yes, im very picky about fps aswell, and i really do notice the 15+ fps boost it gives to starcraft 2 in large battles etc. its just great.
i think 750 / 1075 is the sweet spot. I think 800 / 1100 is easily achievable aswell. But then the fan needs to run higher and the imac becomes a bit noisy.
However we got to remember that this cards is made for laptops, and the imac has so much better cooling then a cramp laptop. So its much safer to overclock

Btw, anyone remember the name of the GPU stress testing tool, i can only remember prime, but thats for cpu.
 

Attachments

Comment

jpjandrade

macrumors member
Mar 17, 2010
49
0
Just want to thank you guys for doing this. It's very promising results. Really interested in more data!
 
Comment

henrikrox

macrumors 65816
Original poster
Feb 3, 2010
1,218
2
So i got to stress test the GPU at stock and OC clocks. Here are the results


Things to note:
* The fans are running at the exact same speed when doing stress testing at stock and OC.
* The temps maxed after about 8-9 minuets, then stayed at that level through the rest of the testing period. Therefor i didnt see the point on letting it run for hours. Since the temp didnt change after the 9 min mark.
* Notice the difference in max,min,avg fps on the two pictures.
* Notice how small the temp difference is.
* Fans are not running high, and its not noisy, which means there is alot of room to play with.
* I will test Crysis benchmark tool later tonight / tomorrow. To really get the idea what the overclock does to games that require a lot of GPU power.
* Next time, i will do 3 tests, with one being even higher then 750/1075. Probably around the 800/1100 mark.


GPU stress test 1 - Stock clocks (628 / 1000)

Runtime: About 30 min



FPS:
Min: 21
Max: 58
Avg: 38

Max Temp: 70,5 C


GPU stress test 2 - OC clocks (750 / 1075)

Runtime: About 30 min



FPS:
Min: 30
Max: 67
Avg: 44

Max Temp: 74 C
 
Comment

saschke

macrumors 6502
Feb 2, 2008
382
28
Great work Henrik! I will do some more testing tonight, too.
Regarding MacFan - thanks for the explanation, maybe this will help others. I've done exactly that the first time I've tried to set fan speeds. Didn't work. Anyhow, I will try again later tonight :)
 
Comment

Snowcake

macrumors regular
May 18, 2010
187
0
You know that the mobile 5850 is based on the desktop 5770 which operate on 850 MHz, so technically you should reach 850 MHz no problem.

The question is the cooling.
 
Comment

Similar threads

  • Mcc_alexander
0
Replies
0
Views
389
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.