OC'ing HD5870 + random questions

[Brutality]

New theory towards reboot problem... after OC'in the graphics card a bit I decided to reboot and it failed to reboot, like when OC'ing the CPU.... could it be the motherboard getting too hot?

 

EDIT:

Sounds very much like a heat issue, because I just made some tests;

Furmark for 20 mins at 850/1260 - wait for gpu temps to get to 60 before trying to reboot - won't properly reboot

Furmark for 20 mins at 850/1260 - wait for gpu temps to get to 35/35 before trying to reboot - no reboot problems

 

So either the GPU or the MB or getting too hot during reboot.

 

Now I actually haven't tested.. but it could also be the GPU being shut off during reboot, because the computer keeps running, the monitor is just "shut-off" after the shutdown process... I've got headphones so I actually haven't heard for a windows start up sound...

Anyways is it ok to OC and then wait for things to cool down before rebooting or does the system HAVE to be able to reboot right after stress tests?

 

EDIT2:

Aren't there any software to monitor MB temps?

 

Many thanks in advance - Brutality

Edited December 2, 2010 by Brutality

[sack_patrol]

You can monitor the Mobo temps using HWMonitor or Lavalys Everest.

 

If you can't get a stable overclock over 900/1290 at least, I don't think it's worth overclocking at all. The performance won't be noticeable, and you'll only shorten ur GPUs life.

Edited December 2, 2010 by sack_patrol

[dihartnell]

Edit 1 - Warm is all i allow them to be. Try a voltage in between 2.2 and 2.3.

Edit 2 - I usually test for around 15 minutes initially. test memory and GPU overclocks separately, leab mem at stock and raise Core first until it artifacts. Then put core at stock and raise mem untill it artifacts/crashes. then you will know what the max is for each of these and can try for a combned overclock.

Edit 3 - Max temps were about 90 in the stress testing. Goes nowhere that high in gaming. I normally run mine at 900/1250. Any 5870 should handle that but if it doesn't its more likely to be memory thats holding you back in my opinion, so lower that.

Edit 4 - leave mem at stock and raise core first.

Edit 5 - HWMonitor will give you temps for mobo, cpu, cores and gpu. How much airflow does your case have? The not restarting properly sounds like a failed overclock on the CPU/ram etc rather than GPU. Check temps with HWmonitor.

 

Have you updated the bios of your GPU to latest revision?

[Brutality]

Edit 1 - Warm is all i allow them to be. Try a voltage in between 2.2 and 2.3.

Edit 2 - I usually test for around 15 minutes initially. test memory and GPU overclocks separately, leab mem at stock and raise Core first until it artifacts. Then put core at stock and raise mem untill it artifacts/crashes. then you will know what the max is for each of these and can try for a combned overclock.

Edit 3 - Max temps were about 90 in the stress testing. Goes nowhere that high in gaming. I normally run mine at 900/1250. Any 5870 should handle that but if it doesn't its more likely to be memory thats holding you back in my opinion, so lower that.

Edit 4 - leave mem at stock and raise core first.

Edit 5 - HWMonitor will give you temps for mobo, cpu, cores and gpu. How much airflow does your case have? The not restarting properly sounds like a failed overclock on the CPU/ram etc rather than GPU. Check temps with HWmonitor.

 

Have you updated the bios of your GPU to latest revision?

 

How do you update GPU BIOS?

 

BTW read the post below the one you quoted, I figured something out about the failed reboot

 

Also: currently finding max memory OC (following the guide I posted a link to)

Edited December 2, 2010 by Brutality

[dihartnell]

How do you update GPU BIOS?

 

 

 

 

The vendor usually gives you a tool to do that. I've never owned XFX but Asus and Gigabyte etc alow you to download the latest bios and give you a tool to update your card. XFX probably the same.

 

Put your GPU/MEM back to stock and rerun those tests. If you still get reboot problems you know its not the GPU overclocks. Are you overclocking the rest of the system?

 

In my experience if stop the stress tests or reboot, the temps goes down within milliseconds so it doesn't seem likely to me you need to do a cooldown before rebooting. HWmonitor will show that. No rebooting cleanly could be either a failed overclock or perhaps some device plugged nto the system (USB ports). I once had an issue with a cheapo external USB drive and I currently have an isuse that my reboot fails if I leave my MS X8 mouse plugged in. (no idea why) but if i take it out until I reach post then I can plug back in and all fine.

Edited December 2, 2010 by dihartnell

[Brutality]

The vendor usually gives you a tool to do that. I've never owned XFX but Asus and Gigabyte etc alow you to download the latest bios and give you a tool to update your card. XFX probably the same.

 

Put your GPU/MEM back to stock and rerun those tests. If you still get reboot problems you know its not the GPU overclocks. Are you overclocking the rest of the system?

 

In my experience if stop the stress tests or reboot, the temps goes down within milliseconds so it doesn't seem likely to me you need to do a cooldown before rebooting. HWmonitor will show that. No rebooting cleanly could be either a failed overclock or perhaps some device plugged nto the system (USB ports). I once had an issue with a cheapo external USB drive and I currently have an isuse that my reboot fails if I leave my MS X8 mouse plugged in. (no idea why) but if i take it out until I reach post then I can plug back in and all fine.

 

Okay, but as I said I tried with stock GPU settings and it rebooted fine, but as soon as I start OC'ing the VRAM (prolly same for gpu core) I need to wait some minutes before it will reboot succesfully, i.e. cooldown, I don't see what this has got to do with my CPU/RAM OC, but here are my BIOS settings when I OC'ed the CPU

 

3923.5MHz - Max temp: 49*C through 24 hours hours Prime95 Blend 2x Torture

Multiplier: 9.5

FSB Frequency: 413

PCIE Frequency: 105

DRAM Frequency: DDR2-1036MHz

Timings: 5-5-5-15

CPU Voltage: 1.35

CPU PLL Voltage: 1.5

FSB Termination Voltage: 1.3

DRAM Voltage: 2.2

Load-Line Calibration: Enabled

Spectrums, EIST, C1E and C3-6: Disabled

Rest on Auto

 

Another thing... I read through the guide (http://tech.icrontic.com/article/overclocking-the-radeon-hd-5850/) and stumbled across "We strongly suggest that you do not allow your 5850 to run above 85°C for an extended period of time."

Now I now my card isn' a 5850, and that this is written under the Over-Volting section (which I won't do), but still, when I am doing Furmark and finding the max mem OC, it's running at 84-87*C...

(Currently doing the 6 hour stability test - 850/1290 gave me a VPU, so going with 850/1280)

 

PS. getting XFX bios update is not easy... I need to register my card and I can't find the seriel number, nor do I remember the date I bought it or where >_<

Why don't they just have a download's section and let me update my bios :S

 

NVM found it... but uhm... they don't have any bios updates for my gpu... (see attachment) I've tried selecting all OS'es in the second drop down menu.. no results

they only got driver and manual downloads (looked through everything now)

 

Regarding airflow... see attachment 2

 

Regarding temps, see attachment 3, 4 and 5 (3 = idle temps | 4 = like 1 min after stress test temps | 5 = when I reboot (I could prolly wait a little less time...) and answer this question

 

What are the max temps? I mean, what temperature shouldn't [insert temp section] not get above?

Also, are any of the voltages bad, or is anything in general bad in this HWMonitor =P

 

UPDATE:

 

After doing some tests, I have found that I only get the reboot problem if the System temp (in HWMonitor) is above 38*C.

Opinions?

btw with the help of Furmark I have OC'd my 5870 to 890/1280 (or well, I got a VPU at 895/1280 (well not exactly a VPU, but a rather thick grey lined bar at top of screen), but I don't know if it's best at 890/1280 or reduce mem clocks and then it's prolly possible to increase core? - gonna try some Heaven 2.1 now =)

 

EDIT:

No artifacts or problems in Heaven 2.1 =)

Since I have used my global quota of picture uploading I will just type in the max temps I had during Heaven 2.1 and benchmarking in it;

 

MSI Afterburner:

GPU temp: 75*C

 

Core Temp:

Core #0: 42*C | Core #1: 36*C

 

HWMonitor:

System: 44*C

CPU: 26*C

AUX: 36*C

Core #0: 41*C

Core #1: 33*C

Assembly (HDD): 22*C

Air Flow (HDD): 21*C

Radeon HD 5870: 76*C | 61*C

 

@Radeon HD 5870 - I guess it's core temp / vram temp

Are the temps too high? keep in mind it's Heaven 2.1 and not a game , so I assume the temps should be higher?

 

PS. Is there some temp monitoring program that's compatible with Logitech G15 keyboard? Now I know coretemp is, i.e. I can see temps on my lcd screen on the keyboard, but I wanna see gpu and mobo temps too while in full screen mode, if possible

Edited December 3, 2010 by Brutality

[Brutality]

So I played some Elder Scrolls IV for an hour and suddenly the game freezes, I ALT-TAB and get the tray-pop up: "bla bla display driver has crashed but bla bla recovered"

I'm not sure if this classifies as an VPU, but either way I decreased mem OC by 5, so I'm running 890/1275 now

 

These are the max temps during gameplay: (890/1280)

 

MSI Afterburner:

GPU temp: 81*C

 

Core Temp:

Core #0: 41*C | Core #1: 41*C

 

HWMonitor:

System: 46*C

CPU: 28*C

AUX: 32*C

Core #0: 41*C

Core #1: 37*C

Assembly (HDD): 23*C

Air Flow (HDD): 21*C

Radeon HD 5870: 81*C | 67*C

 

Too high temps?

 

EDIT

 

These are temps with stock settings (850/1200)

 

MSI Afterburner:

GPU temp: 78*C

 

Core Temp:

Core #0: 43*C | Core #1: 39*C

 

HWMonitor:

System: 46*C

CPU: 31*C

AUX: 38*C

Core #0: 44*C

Core #1: 37*C

Assembly (HDD): 23*C

Air Flow (HDD): 22*C

Radeon HD 5870: 78*C | 65*C

Edited December 4, 2010 by Brutality