Sign in with
Sign up | Sign in
Your question
Solved

Radeon 5850 OC settings

Last response: in Overclocking
Share
March 14, 2011 3:32:57 AM

Hello friends at Toms, I’m trying to OC my video card a Powercolor Radeon 5850.

Method 1: I increase the clock cycles in increments of 5 to 10 in CCC and I run Furmark after each edit then test with the heaven benchmark. When I reach a point where it fails, I reset the clock back 5 to 10 and that should be my sweet spot. Right?

Method 2: Let the “Catalyst Control Center” auto tune the card. I did this first thinking I could get near the best OC for the card then tweek it some.

The default settings are clock 760/ mem 1050.
the "Auto tune" in CCC ramps it up to 865/1215 but the card freezes.

I’ve managed to manually get up to 805/1050 and it runs stable at 810/1050 it crashes. I’m confused why the auto tune would set it so fast that it crashes.
This is the first computer I built myself and I'm kinda new to this stuff.
Thanks for your time.


AsRock x58-xtreme mobo
Intel Core i7 930 Mhz
CORSAIR DOMINATOR 6GB (3x2GB) 240-Pin SDRAM DDR3 1600 (PC3 12800) CMP6GX3M3A1600C8
CM 690 case
CM 750 watt ps
CM Hyper N520
PowerColor Radeon 5850 gpu
Asus CD ROM
M-audio Studiophile AV 40 speakers

More about : radeon 5850 settings

March 22, 2011 12:07:09 AM

davcon said:
Use this to oc your gpu
http://event.msi.com/vga/afterburner/download.htm



I actually have that installed. Since I posted the question I've become aware of another OC tool for the GPU called the "AMD GPU Clock TooL". It's actually supported by AMD, where the MSI one is not. I just can't find any documentation specific to it.

Do you have any good articles I might find helpful? I'm pretty noob to this stuff.

Thanks for your help.
m
0
l
Related resources
a b U Graphics card
a c 183 K Overclocking
March 22, 2011 3:41:24 AM

gerry410 said:
I actually have that installed. Since I posted the question I've become aware of another OC tool for the GPU called the "AMD GPU Clock TooL". It's actually supported by AMD, where the MSI one is not. I just can't find any documentation specific to it.

Do you have any good articles I might find helpful? I'm pretty noob to this stuff.

Thanks for your help.

I use that tool as well then set-up a profile in Afterburner.
It's trial and error until you find compatible settings for your particular rig.
Run the Kombuster for stability.
BSOD's indicate improper settings.
Just keep your temps in check and experiment.
m
0
l
March 23, 2011 1:03:31 AM

davcon said:
I use that tool as well then set-up a profile in Afterburner.
It's trial and error until you find compatible settings for your particular rig.
Run the Kombuster for stability.
BSOD's indicate improper settings.
Just keep your temps in check and experiment.


Perhaps you can add a little light on this for me. The stock clock for the GPU is C760/M1050.
I have it running stable @ 800/ 1050, a modest increase. If I go up to 810 / 1050 I get the BSOD when I run the Unigene Heaven benchmark. So i suppose the best I can get is 800/1050 right?

Heres the clunker. If I use the CCC auto tune feature with the radeon software it pumps it up to 865/1215. This will crash. By why does the cards own native software clock it so high?
Thanks
m
0
l
a b U Graphics card
a c 183 K Overclocking
March 23, 2011 3:24:54 AM

gerry410 said:
Perhaps you can add a little light on this for me. The stock clock for the GPU is C760/M1050.
I have it running stable @ 800/ 1050, a modest increase. If I go up to 810 / 1050 I get the BSOD when I run the Unigene Heaven benchmark. So i suppose the best I can get is 800/1050 right?

Heres the clunker. If I use the CCC auto tune feature with the radeon software it pumps it up to 865/1215. This will crash. By why does the cards own native software clock it so high?
Thanks

Your GPU Clock is good but i would suspect the memory clock is way too low 1050.
My 5850 is stock 765/1125.
I've had it 850/1200 stable.
865/1215 is too much imo with or without an overvolt.
Try for 800/1175 or something realistic without having to overvolt.
m
0
l
March 30, 2011 12:08:25 PM

davcon said:
Your GPU Clock is good but i would suspect the memory clock is way too low 1050.
My 5850 is stock 765/1125.
I've had it 850/1200 stable.
865/1215 is too much imo with or without an overvolt.
Try for 800/1175 or something realistic without having to overvolt.


Hi Dav, I've got it to 800/1070. If I increase the mem clk to 1075/1080 it fails. I don't understand why I can't get the mem clk higher. Any ideas?
m
0
l
May 10, 2011 9:09:47 PM

unless you overvolt by an insane amount you won't hurt the card. a safe place to start is the stock voltage and clocks of a 5870. If your card can achieve that I would be happy. Make sure to watch your temps and up the fan speed if necessary. I have my 5850 @ 875/1200 @ 1.15v easy & safe,

BTW i use Sapphire Trixx software to OC
m
0
l
May 20, 2011 12:09:49 AM

Well I'm happy to report that I have it running @ 810/1100 and so far stable. I use FurMark and Unigne Heaven benchmark for testing. i I will continue the increase in small bits till i reach the sweet spot.

Do you guys know anywhere I can get some documentation about the GPU clock tool? I understand its officially supported by AMD but theres nothing on the web site.
Thanks
m
0
l
a b K Overclocking
May 26, 2011 7:44:51 PM

I use a gaming profile that sets my card to 950/1200 at 1.2v every time I play a demanding game.Your oc is too low imo.You don't want the gpu to be rock stable 24/7 like a cpu,you just need to be able to play your games without artifacts and crashes.So,I would really suggest pushing the clocks and voltage a bit more to get a larger performance gain.
m
0
l

Best solution

a c 125 U Graphics card
a c 100 K Overclocking
May 26, 2011 10:15:14 PM

First off, it's odd that your CCC settings allows that high of clocks. I guess it's not a reference model cuz my CCC limit was 775/1150 before I edited my BIOS.

Anyway, all cards are going to behave differently. IMO OC the Mem first in 50mhz incremets initially... it'll crash. Go back to last stable and increase by 10mhz. It's important to keep the card hot. Unfortunately sometimes FurMark might be stable for 10 mins and then a game crashes... you just have to keep in mind that your OC maybe be causing the crash, so dropping another 5mhz might stabilize things. Also when OCing the memory keep a keen eye on the FPS - be sure to use AA (2x or 4x). If the FPS drops at higher mem speeds, it means you're close to the max speed it will do without crashing, but it's faster than it can handle stabily and results in lower FPS.

As for GPU clock, same idea but run 0xAA. You can increase voltage if temps stay lowish but IMO run GPUZ if you do and keep an eye on the VRM temps (if you have those sensors). I've had stable core temps with VRM hitting 120C and causing a crash wen running 1.25V.

Anyway I have my 5850s at stock volts, 870mhz core, 1200mhz mem. I find even just 1210mhz mem causes issues in games or benches. I've had my core up to 985mhz stable at 1.225V, but the FPS increase is so minor there's no point IMO. We can't increase memory voltage, and the memory bandwidth is often the choke point.

For reference: http://blogs.amd.com/play/2011/05/25/why-graphics-matte...

"While memory bandwidth doesn’t get as much credit for overall performance, as say the frequency of the part, it is often just as important. The best analogy I have heard for how to describe memory bandwidth is to think of a highway. There are two things that determine how fast you can travel in your car on a given highway. One is the maximum speed limit. For our comparison, think of this as being comparable to the frame buffer size or amount of memory that comes with your graphics card. In some cases, the maximum speed limit (or frame buffer in our case) determines how fast you can travel. That said, just like with a highway, the bandwidth of that highway (think number of lanes) is just as important as the speed limit. If you are traveling really early in the morning, or really late at night, the number of lanes are not that important, as there are usually not a lot of travelers at that time. The minute you change that to morning or evening rush hour, the whole equation changes. At that point, the number of lanes, or bandwidth of the highway, has more to do with your travel speed than the speed limit does. The more lanes there are, the more traffic can flow through, and the faster you go."

To see this in action, simply run your GPU in FurMark with AA on and off.
I did this, and recorded settings and results. I didn't want to go through with 6 more tests to fill out the middle points because the trend is clear. Test was run full screen, 1080p, 4xMSAA in Furmark. You can see that core speed hardly affects FPS while mem speed shows all the increase. I don't have the data points, but with 0xAA, mem speed had zero effect on FPS, but core speed showed all the gains.

(X (vertical) is framerate, Y (horizontal) is core speed, and Z (depth) is mem speed)

Share
June 13, 2011 10:43:38 PM

wolfram23 said:
First off, it's odd that your CCC settings allows that high of clocks. I guess it's not a reference model cuz my CCC limit was 775/1150 before I edited my BIOS.

Anyway, all cards are going to behave differently. IMO OC the Mem first in 50mhz incremets initially... it'll crash. Go back to last stable and increase by 10mhz. It's important to keep the card hot. Unfortunately sometimes FurMark might be stable for 10 mins and then a game crashes... you just have to keep in mind that your OC maybe be causing the crash, so dropping another 5mhz might stabilize things. Also when OCing the memory keep a keen eye on the FPS - be sure to use AA (2x or 4x). If the FPS drops at higher mem speeds, it means you're close to the max speed it will do without crashing, but it's faster than it can handle stabily and results in lower FPS.

As for GPU clock, same idea but run 0xAA. You can increase voltage if temps stay lowish but IMO run GPUZ if you do and keep an eye on the VRM temps (if you have those sensors). I've had stable core temps with VRM hitting 120C and causing a crash wen running 1.25V.

Anyway I have my 5850s at stock volts, 870mhz core, 1200mhz mem. I find even just 1210mhz mem causes issues in games or benches. I've had my core up to 985mhz stable at 1.225V, but the FPS increase is so minor there's no point IMO. We can't increase memory voltage, and the memory bandwidth is often the choke point.

For reference: http://blogs.amd.com/play/2011/05/25/why-graphics-matte...

"While memory bandwidth doesn’t get as much credit for overall performance, as say the frequency of the part, it is often just as important. The best analogy I have heard for how to describe memory bandwidth is to think of a highway. There are two things that determine how fast you can travel in your car on a given highway. One is the maximum speed limit. For our comparison, think of this as being comparable to the frame buffer size or amount of memory that comes with your graphics card. In some cases, the maximum speed limit (or frame buffer in our case) determines how fast you can travel. That said, just like with a highway, the bandwidth of that highway (think number of lanes) is just as important as the speed limit. If you are traveling really early in the morning, or really late at night, the number of lanes are not that important, as there are usually not a lot of travelers at that time. The minute you change that to morning or evening rush hour, the whole equation changes. At that point, the number of lanes, or bandwidth of the highway, has more to do with your travel speed than the speed limit does. The more lanes there are, the more traffic can flow through, and the faster you go."

To see this in action, simply run your GPU in FurMark with AA on and off.
I did this, and recorded settings and results. I didn't want to go through with 6 more tests to fill out the middle points because the trend is clear. Test was run full screen, 1080p, 4xMSAA in Furmark. You can see that core speed hardly affects FPS while mem speed shows all the increase. I don't have the data points, but with 0xAA, mem speed had zero effect on FPS, but core speed showed all the gains.

(X (vertical) is framerate, Y (horizontal) is core speed, and Z (depth) is mem speed)

http://i746.photobucket.com/albums/xx103/Wolfram23/memandcorespeedvsfps.png


Thanks for your time and help wolf. I've got it running 805/1150 and its stable. I will keep these settings for now.Have a great summer.
m
0
l
June 13, 2011 10:43:57 PM

Best answer selected by gerry410.
m
0
l
July 8, 2011 6:26:14 PM

Hi folks sorry if this is in the wrong place but would really like a little info please,I to use overdrive and where it says High Performance GPU clock settings I have it at 780MHZ and the High Performance Memory clock settings are set at 1050MHZ.
I tested Custom Clocks and passed but at the bottom where it says Current Values my GPU clock is at 157MHZ and Memory Clock is at 300MHZ.

I am using a Radeon 5850,Phenom II X4 955BE 3.2MHZ

Please somebody help as I am a little confused as to what my actual values are.

Many thanks in advance,Michael.
m
0
l
a c 125 U Graphics card
a c 100 K Overclocking
July 9, 2011 6:04:10 AM

michael gentleman said:
Hi folks sorry if this is in the wrong place but would really like a little info please,I to use overdrive and where it says High Performance GPU clock settings I have it at 780MHZ and the High Performance Memory clock settings are set at 1050MHZ.
I tested Custom Clocks and passed but at the bottom where it says Current Values my GPU clock is at 157MHZ and Memory Clock is at 300MHZ.

I am using a Radeon 5850,Phenom II X4 955BE 3.2MHZ

Please somebody help as I am a little confused as to what my actual values are.

Many thanks in advance,Michael.



157/300 are the idle clocks. It automatically ramps up to full clocks when you play games.
m
0
l
!