Author Topic: How hot is too hot? Overclocking thoughts... (Read 27,932 times)

wreckedcarzz · « **on:** May 20, 2008, 11:00 PM »

Hey all,

I have always been amazed at the fact that you can speed up parts of a computer via a few simple steps, and "overclock" the speeds that things operate at. I haven't really been able to successfully do this, until a couple days ago. I stumbled across an updated RivaTuner and said "what the heck, why not?" and grabbed it.

I now have my GPU core running at 754MHz (905 max) and my GPU RAM at 900 (1220 max). However, the card runs idle at about 150F (65C). When gaming, I can run most (I am still testing) of my games at max options, including AA and resolution. The card temp seems to barely rise.

Is it OK to have the card operate at this temp? Can I push it harder and see just how far it will go? And are there any compact, external ATi GPU water cooling kits that aren't too pricey?

And on a related note, can BIOS be edited to allow overclocking of the CPU and FSB (and even RAM perhaps)?

-Brandon

Lashiec · « **Reply #1 on:** May 20, 2008, 11:20 PM »

Mmmm, well, usually the card should not run at such high temperature on idle, but a temperature of 65º C on load is quite good. You should throw some GPU stress test to the card to see if the image has artifacts (or if everything goes down) before you can be really sure, but you can push it harder. How hard? That depends, overclocking takes its time, and it's better to proceed in small steps. Unless you're really going to overclock it to the max, I don't see any need for a water cooling kit, you know, GPUs usually do not overclock as good as CPUs, and particularly ATI cards. And GPUs withstand really high temperatures.

Yeah, you can mess with the BIOSes of the cards, but never got there.

Armando · « **Reply #2 on:** May 21, 2008, 12:50 AM »

ERrrrrrr... Don't want to hijack that thread but... my laptop's GPU is often between 60-65º C. Should I be worried ?

wreckedcarzz · « **Reply #3 on:** May 21, 2008, 01:20 AM »

I have RivaTuner managing the fan speed, but I can do it manually and leave it at full blast all the time. Might wear out the fan though. I took the core MHz to 775 with no issues in-game(s) so far, but I'll do some "official" stress testing and artifact scans to make sure before I proceed.

Armando, thats a bit hot for a laptop IMO, my laptop only hits about 100F (I don't know the C level for that, maybe 40ish?) while my sister plays The Sims 2 on it. It isn't exactly an up-to-date card, though.

How big is the fan on it (the external fan)? Make sure its not dusty or anything

Shades · « **Reply #4 on:** May 21, 2008, 01:31 AM »

Well Lashiec, here in Paraguay the ambient temperature is 35 degrees (celcius) or higher for about 60% of the total year. Watercooling is an serious option over here, because moving air that itself is already 35 to 50 degrees hot does not cool the equipment that well.

It is only my fear of leaking watercooling systems that prevent me from installing one. On the other hand, changing all coolers every 6 months because of reliability issues is also a hassle. You cannot get the good brands for a reasonable price here in Paraguay, and I have lost already two mainboards and 5 SATA harddisks.

Deozaan · « **Reply #5 on:** May 21, 2008, 01:32 AM »

Just out curiosity, what is the default speed of your GPU that you're raising it from?

wreckedcarzz · « **Reply #6 on:** May 21, 2008, 11:30 AM »

It is about 400-415 if I remember correctly for the core, and 700-ish for the RAM. I need FPS > texture un/loading speed.

So about a 300MHz overclock for the core, and 200 for the RAM.

EDIT: I just tested out Test Drive Unlimited, max resolution, max detail, High Dynamic Range rendering enabled, no AA - I get about 12-17 FPS driving and about 28-35 in houses/stores. I am going to try and take the core up a bit farther and see if I can pull 20+ FPS.

What HDR is http://en.wikipedia.org/wiki/High_dynamic_range_rendering for anyone interested

tomos · « **Reply #7 on:** May 21, 2008, 12:45 PM »

Quote from: Armando on May 21, 2008, 12:50 AM

ERrrrrrr... Don't want to hijack that thread but... my laptop's GPU is often between 60-65º C. Should I be worried ?

Armando
Tuckndar had success cleaning his fan (dust)

Quote from: TucknDar on March 23, 2008, 09:51 PM

about the temperature again, not that anyones interested, I'm sure... uh, well I think I've got some dust in the fans. Thanks for your mentioning this, or I probably wouldn't've dug around so much. Now I know that the dust can be easily removed from my laptop

back to topic...

you could always pm him or continue off-topic in that thread

(says me, off-topic

)

TucknDar · « **Reply #8 on:** May 21, 2008, 01:24 PM »

Quote from: tomos on May 21, 2008, 12:45 PM

Armando
Tuckndar had success cleaning his fan (dust)
Quote from: TucknDar on March 23, 2008, 09:51 PM
about the temperature again, not that anyones interested, I'm sure... uh, well I think I've got some dust in the fans. Thanks for your mentioning this, or I probably wouldn't've dug around so much. Now I know that the dust can be easily removed from my laptop

back to topic...
you could always pm him or continue off-topic in that thread (says me, off-topic )

Removing the dust made a huge difference on the CPU temp. My laptop was running at about 55C idle, but after I got the dust out of the fan it's now steady at 40C idle. I suck at math, but I believe that's 22% cooler!

Lashiec · « **Reply #9 on:** May 21, 2008, 02:42 PM »

Quote from: Shades on May 21, 2008, 01:31 AM

Well Lashiec, here in Paraguay the ambient temperature is 35 degrees (celcius) or higher for about 60% of the total year. Watercooling is an serious option over here, because moving air that itself is already 35 to 50 degrees hot does not cool the equipment that well.

Well, yeah, in those extreme cases watercooling is something worth considering, but in normal environments it's only really useful for xtreme OC

, air cooling tend to be sufficient for the rest.

wreckedcarzz, you could take a look to more serious air cooling as well for an extra push. Or another card for that matter

Carol Haynes · « **Reply #10 on:** May 21, 2008, 04:02 PM »

You need to be convinced that the readings are true too ...

My system runs incredibly hot - but then the sensors all read very hot within seconds of switching on. My two GPUs (for example) currently read 62C and 80C

These are normal temperatures for these cards and have been ever since they were new. There is loads of air circulating in my system so it isn't a problem of ventilation. I figure the sensors are crap or the nVidia software does a bad job reading it. My system is currently on 24/7 and the temps record as stable at those temperatures and I get the same values as soon as I am able to open the nVidia panel in Windows.

To start with I was worried but now I just live with it - they have been doing it for over a year now so it can't be too bad!

Armando · « **Reply #11 on:** May 21, 2008, 04:56 PM »

Quote from: Carol Haynes on May 21, 2008, 04:02 PM

You need to be convinced that the readings are true too ...

My system runs incredibly hot - but then the sensors all read very hot within seconds of switching on. My two GPUs (for example) currently read 62C and 80C

These are normal temperatures for these cards and have been ever since they were new. There is loads of air circulating in my system so it isn't a problem of ventilation. I figure the sensors are crap or the nVidia software does a bad job reading it. My system is currently on 24/7 and the temps record as stable at those temperatures and I get the same values as soon as I am able to open the nVidia panel in Windows.

To start with I was worried but now I just live with it - they have been doing it for over a year now so it can't be too bad!

I think I'm going to relax too for now... Although cleaning the fans could be an option.
AFAICT the temperatures have been pretty much always the same since I bought my laptop almost 2 years ago. So... Unless Dell sold me an Inspiron with dusty fans...

GPU now reads 67C and i8kfanGUI tells me that it went up to 73C since last reboot...

wreckedcarzz · « **Reply #12 on:** May 21, 2008, 05:59 PM »

I know my temps are accurate as I have software reading it directly from the output the card normally gives to the ATi Catalyst Control Center - I have 3 external (in the PC, but not *in* anything, just around/near) sensors running, between 82 (HDD) and 95 (CPU) watching the heat that isn't cooled by the fan.

NewEgg is having a sale of new ASUS GPUs (nVidia and ATi) that come equipped with their new "Arctic Fansink". The information states that it can decrease the temp by up to 20C vs stock cooling. I might buy one, but it won't be for a few months at least. $:-\$

And on the topic of dust, I have read several reviews of my PC case (it was new when I got it, no reviews) most of them say/complain that is is a mass dust vacuum due to the oversized fans (and enormous fan on the top). According to what I have read, it sucks in so much air that it ends up pulling in dirt and debris with it - ending up being dust. And it is true, I have used 3 cans of compressed air on this, as well as a now-really-disgusting toothbrush, and there is *STILL* dust in it, collecting constantly. But I live in a desert, what do you expect? *Wreckedcarzz wipes dust from mesh covering front fan*

Carol Haynes · « **Reply #13 on:** May 21, 2008, 06:20 PM »

Quote from: wreckedcarzz on May 21, 2008, 05:59 PM

I know my temps are accurate as I have software reading it directly from the output the card normally gives to the ATi Catalyst Control Center

Part of the problem is that these onboard sensors (mobo or graphics cards) aren't always terribly accurate, and they are heavily influenced by components nearby and often poorly calibrated so you get silly temperatures even when cold!

When I bought my current motherboard it was touted as being super cool (and not in the 'cool dude' sense) and includes copper pipes and heat dispersal systems etc. It says you should be able to run it with the minimum of fans etc. Having said that the motherboard sensors report temperatures in the BIOS at switch on that are nearly as hot as my previous motherboard working at full tilt. Either there is something seriously wrong with my mobo (there isn't it runs stable all the time) or the sensors simply report silly values. For example I get a motherboard temperature that is HIGHER than the CPU temperature - which is obviously stupid - but it is just a function of poor callibration and the location of the sensors producing biased and eccentric results.

I think the the important thing is to keep an eye on what your sensors report on idle and under load over a period of time and then monitor what happens and if you suddenly start getting reported temperature climbs you know something is wrong. The raw values are (IMHO) pretty meaningless!

yotta · « **Reply #14 on:** May 21, 2008, 08:36 PM »

well im RLLY woried now, my CPU,RAM and GPU all run at over 90C:S

Deozaan · « **Reply #15 on:** May 21, 2008, 09:07 PM »

Just be careful with how much you overclock your GPU. The GPU itself can withstand high temperatures, but other components on it may not.

I learned that the hard way when my GPU fan stopped spinning and melted off my GPU. The GPU still works, but obviously wouldn't last very long without another fan blowing on it.

Carol Haynes · « **Reply #16 on:** May 21, 2008, 09:33 PM »

Quote from: yotta on May 21, 2008, 08:36 PM

well im RLLY woried now, my CPU,RAM and GPU all run at over 90C:S

Hmmm - despite what I wrote that does seem a bit high

Take the side of your box and blow a large desk fan into it on full power and see if the temperatures drop. Also check to make sure all your fans are running properly.

One other thing you might want to check is that all the fans are installed the correct way round. Generally speaking you want air to be sucked inat the front of the case and all other fans suck air out. Check on the side of the fans for the flow direction arrow.

wreckedcarzz · « **Reply #17 on:** May 22, 2008, 06:54 PM »

I have a fan jury-rigged to cool my PCI/PCIe slots by holding onto the 4 pin power cable that goes to my USB card in my PCIe 1x slot. The bad part is that if the fan does stop on my GPU, and stuff starts melting, it hits not only my $40 USB card, but my $150 X-Fi card and my $110 TV Tuner card.

There is also a fan below the aforementioned cards, on top of the PSU, helping the PSU push air out of the case. But that doesn't do much...

I might take my 2 leftover 120mm fans and see if I can't add them into the setup with some 4 pin -> fan connectors I have. $:-\$

@yotta: Yes, that is way too hot for normal usage - take a fan (you can get one at the local drug store for $10) and blast the highest volume of air you can into the case, and monitor the temps. I used to do this with my undersized eMachines case (everything fit, but it was too small and the heat just kind of snowballed). You might want consider a new case (preferably with a lot of fan holders).

Lashiec · « **Reply #18 on:** May 22, 2008, 11:20 PM »

Quote from: wreckedcarzz on May 22, 2008, 06:54 PM

The bad part is that if the fan does stop on my GPU, and stuff starts melting, it hits not only my $40 USB card, but my $150 X-Fi card and my $110 TV Tuner card.

Buy cheaper cards and use the "deluxe" ones on Sundays

Armando · « **Reply #19 on:** May 23, 2008, 12:42 AM »

Deozaan · « **Reply #20 on:** May 23, 2008, 09:02 AM »

Quote from: wreckedcarzz on May 22, 2008, 06:54 PM

The bad part is that if the fan does stop on my GPU, and stuff starts melting, it hits not only my $40 USB card, but my $150 X-Fi card and my $110 TV Tuner card.

It didn't drip on anything else. The fan was screwed onto the heatsink at 3 points. The plastic melted at those three points and the fan fell off. When I looked in my case it looked like someone had used a hot glue gun to attach my fan to the heat sink. It was sticky melted. Not drippy liquid melted.

wreckedcarzz · « **Reply #21 on:** May 24, 2008, 07:01 AM »

Ah, OK. I had to un-overclock from 799 back to 775, due to game crashes. And I also found out that lower RAM OC = higher FPS. Might take the RAM back to stock and just OC the core.

wreckedcarzz · « **Reply #22 on:** June 12, 2008, 07:44 PM »

Reviving the thread, I have some info that may interest people here at DC that have OEM PC's and want to overclock their CPUs...

NOTE: I did this, I have a stable overclock of 273MHz above stock (it is something, im Celeron D so quiet down C2D/C2Q users

) BUT I TAKE NO RESPONSIBILITY IF YOU FRY YOUR PC.

Guide:
Following the guide here, EXACTLY, you can get a working stable OC: http://www.zuneboards.com/forums/tech-help/18125-overclock-locked-bios-works-branded-pcs-such-dells-gateways-compaqs-ect.html

ClockGen note: If you run into issues and have to reset, and when you try to Apply Selection and the PC freezes, delete ClockGen.exe and re-extract a new copy. It took me 20 minutes and 3 resets to figure this out

My Results:
My core clock STOCK: 3457MHz (3.46GHz)
My core clock OCed: 3730 (3.73GHz)
Total OC for CPU: 273MHz (0.27GHz)

So an approximate 9% or so overclock, not bad for a first time OCer (CPU-wise). I took it up to 3750MHz, but the PC froze and the sound card emitted a loud (and annoying) buzzing noise, so I have found my *top* overclock. Horray!