*** Official Radeon HD5800 Series Overclocking Thread ***

[Gaul]

DEF BIOS

5870 @ 975/1250

[Caped Crusader]

Anyone having problems with 2D voltages being upped to 1.6ish when using MSI Afterburner?

I had to uninstall the damn thing. Has to be a program bug.

This guy Jonny Manc has exactly the same problem: http://forums.guru3d.com/showthread.php?t=306019

[fornowagain]

Don't worry the card has overvoltage protection, you won't kill if it says its 1.6v, its not.

[Caped Crusader]

Don't worry the card has overvoltage protection, you won't kill if it says its 1.6v, its not.

I hope you're sure about that, else I may have voided mine's warranty or something like that... the VRM temp was in the upper 70's

[fornowagain]

There is, same as the 4870x2. Tried all I could to get that over 1.37 volts and it couldn't be done via software.

Maybe someone with a DMM and the back off the card will give the actual value, but its there.

Got some Info:

http://forums.anandtech.com/messagev...=y&STARTPAGE=2

Good news for the 5870
157/300 .95v idle
400/900 .95v/1.063 blu-ray, UVD
400/1200 1.063v 2d multi-display eyefinity
600/900 1.063v OVP/OCP initiate throttle
850/1200 1.162v 3d performance

And, loco just did 1200mhz core / 1250mhz mem on Tri-Fire using 1.375v thru afterburner (LN2 cooled). He says that > 1.375v doesn't give him any higher mhz, so he thinks that OCP must exist somewhere around 1.375v.

http://blogs.amd.com/play/2009/09/22...the-community/

With the ATI Radeon HD 5800 series we’ve implemented a hardware-level overvolt protection scheme where a signal from the regulators can be fed into the GPU directly and the GPU can take action if the regulators indicate they are operating out of their specification. In the unlikely event that such a scenario happens, rather than the board turning off, the GPU is designed to clock down to get the regulators back into a normal operating zone and then clock back up when they have done so.

[To(V)bo Co(V)bo]

Anyone else think that Ati might have initially deperfomanced the 5 series at launch on purpose to seemingly show lower performance numbers. Then after a month or two there will be some new driver that increases the performance 20-30% across the board. Maybe they did this to fool nvidia into thinking the cards are slower, and then bam gt300 killers in desquiese. I wouldnt be surprised, these things should be performing way faster considering no crossfire overhead. Or these drivers are very badly unoptimised.

[mR Yellow]

Hi Guys

I've been following this thread for some time now.
My Sapphire 5870 arrived and i did some testing.

I ran the VANTAGE on performance setting.
GPU P11900 and CPU P12600.

Surely i should be getting higher than this? Thats the same score i got with my 4890??? Any ideas?

PC specs:
ASUS M3A78-T
PhenonII 940BE @ 3.6ghz
8gb DDR2 800mhz
850w Corsair PSU

I've made sure i don't have AA on...so i'm pretty stumped here with this low score.

BTW I've ran a quick OC (GPU 900 and MEM 1250 def volts) with MSI afterburner and score stays relatively the same.

[Monstru]

Oh boy, I finally managed to finish the article about HD 5850. Now I need to go and relax a while untill I touch these cards again for some serious OC!

[dan7777]

nice review LAB 501 . seems like from that review the gtx 295 is still leading apart from hawx bench. ....http://lab501.ro/placi-video/ati-rad...micul-gigant/6

[Theorw]

Here is a question:In the msi afterburner i didnt see any vMEM increase option.
So i am guessing the vMEM will increase only by soldering???

[Chickenfeed]

Hi Guys

I've been following this thread for some time now.
My Sapphire 5870 arrived and i did some testing.

I ran the VANTAGE on performance setting.
GPU P11900 and CPU P12600.

Run it at 1680x1050 at stock clocks at the Extreme preset ( or better yet run it on the default Extreme preset if you have the display for it ) I am curious as well as you are not the first person to get around 12000gpu on the P setting with a Phenom II. This should at least tell us if its being bottlenecked ( as 1280x1024 is still very cpu dependent and P mode doesn't use anti aliasing )

[mR Yellow]

Run it at 1680x1050 at stock clocks at the Extreme preset ( or better yet run it on the default Extreme preset if you have the display for it ) I am curious as well as you are not the first person to get around 12000gpu on the P setting with a Phenom II. This should at least tell us if its being bottlenecked ( as 1280x1024 is still very cpu dependent and P mode doesn't use anti aliasing )

I get GPU 6xxx in vantage extreme.

[dan7777]

I get GPU 6xxx in vantage extreme.

this is what you should be getting...http://www.facepunch.com/showthread.php?t=811221 around 8xxx maybe bottleneck ???

[mR Yellow]

this is what you should be getting...http://www.facepunch.com/showthread.php?t=811221 around 8xxx

Yes, does extreme setting eliminate CPU bottle neck? I can't see the guys system specs for an apples to apples comparison.

[Oese]

test if u are using pciw 2.0 16x

besides that, i experienced serious slower figures with pcie link aspm (some power saving) enabled in bios. and you should enable catalyst AI..

[mR Yellow]

test if u are using pciw 2.0 16x

besides that, i experienced serious slower figures with pcie link aspm (some power saving) enabled in bios. and you should enable catalyst AI..

I've checked PCI-E speed using GPUZ and it is running x16.

I will have a look at bios setting when i get home from work.

[Monstru]

nice review LAB 501 . seems like from that review the gtx 295 is still leading apart from hawx bench. ....http://lab501.ro/placi-video/ati-rad...micul-gigant/6

Thank you Dan. Yes, in terms of absolute performance for one card, the GTX295 still has the edge. The HD 5870 offers a ton of advantages though (single GPU, low power consumption, lower noise level, lower temps, lower price, etc), beeing a very elegant high-end solution. And HD 5850...I don't even have to say it. I think it's more bang for the buck as the HD 4850 was at it's time

[zalbard]

Did anyone bother changing TIM on their 5870? Is there any point?

[Monstru]

I don't think you will gain much from that, gone are the times when videocards would come with white goo betwen the GPU and the heatsink.

[Dennet]

Did anyone bother changing TIM on their 5870? Is there any point?

I took my 5850 apart last night to check the memory modules. I also changed the standard thermalgrease for Arctic Silver 5. It lowered my temps after 5 minutes of Furmark from 73 degrees Celcius to 71 on stock clocks.

I guess it's up to you if you think it's worth the trouble.

[zalbard]

I took my 5850 apart last night to check the memory modules. I also changed the standard thermalgrease for Arctic Silver 5. It lowered my temps after 5 minutes of Furmark from 73 degrees Celcius to 71 on stock clocks.

I guess it's up to you if you think it's worth the trouble.

Looks it then!
MX2 or MX3 could probably gain 1C more.

[Astennu]

That would be nice.... But it also shows the stock paste is not bad at all.

I recently switched to MX-3. Its the best i found so far. MX-2 is also good. After that i would use AS5.

[NanoBullet]

I know i'm not in the best place for asking this but LIGHTMAN could you do me the favor and tell me what's your GPU Score in 3DMark Vantage performance with the 5870 stock clocks?

I would truly appreciate it.

[Funky]

Got the XFX 5870 past 1Ghz.

Stock cooling. Just changed the thermal paste by MX2 (didnt really do anything for me, no change in temps ).

Orb : http://service.futuremark.com/compare?3dmv=1474328

Asus bios.

[pr@$r1g]

hey shripad i have seen ur review man it was just awesome man ,i cant resist that

having fun with ur xfxhd5870

great job !!!