• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Coolest & Hottest Cards: Benchmarking with a difference

555BUK

555BUK

555BUK

Soldato
Joined
10 Feb 2007
Posts
3,469
Rather than benchmarking performance, I thought why not change things and benchmark heat/coolness instead. It may help people when researching which card to buy.

Rules
1). Any Air-Cooled unmodified card is welcome.
2). Card must be ran at stock settings (clocks, fan etc).
3). Any case is welcome, but please keep the case closed and fans at your normal settings.
4). Heaven 4.0 Benchmark will be used for testing. This should simulate a current demanding game.
5). MSI Afterburner should be used for measuring temps.

Downloads
Heaven 4.0 Benchmark
MSI Afterburner

Instructions
1). Start Afterburner, ensuring all card settings are at stock. Display Hardware Monitor and leave running in the background.
2). Open Heaven 4.0 and select the following settings:-
Preset: Custom
API: Dx11
Quality: Ultra
Tessellation: Extreme
Stereo 3D: Disabled
Multi-Mon: Disabled
Anti-Alias: x8
Fullscreen: Yes/tick
Resolution: 1920x1080
3). Run Heaven 4.0 for at least 10 minutes. Just leave looping, no need to hit the benchmark button.
4). Quit Heaven when time is up.
5). Post card details as per my example below.
6). Post screenshot of Afterburner Hardware Monitor. Hover over peak temps with mouse to list all values and press Alt+Prt Scr key to capture.

My Results
Card: HIS HD 7950 IceQ BOOST
Default Clock Speed: Core 925MHz, Mem 5000MHz
Idle Temp: 32deg
Load Temp: 68deg
Idle Fan: 30%
Load Fan: 47%
Case and cooling: Silverstone TJ08 MATX Case with single 180mm fan on low speed setting
Subjective Noise Comment: Very quiet at idle, audible but unintrusive at load.

nlwv.jpg


Please be honest.
 
Last edited:
It's not very scientific, nor is the noise levels as that is dependent on case, case location and your own hearing.

Plus no way to guarantee people are setting Heaven correctly :(
 
RedvGreen said:
It's not very scientific, nor is the noise levels as that is dependent on case, case location and your own hearing.

Plus no way to guarantee people are setting Heaven correctly :(
Click to expand...
Spoil-sport:p. It's not meant to be scientific, just real-world examples. Everyone mates their cards to different components within different environments. The exercise is to just obtain a pretty rough idea. Perhaps the easiest measure of cooling efficiency is the difference between idle and load temps.
 
Last edited:
too many variables it would take many many many samples to get any decent results you wouldnt even get a pretty rough idea.

You would have to include, is it close to a radiator, is it near a window, is it near a door, whats the rear exhaust space like, PC case on the floor, PC case on a desk etc etc.
 
I have got to be in with a shout on this, I have two HD 5970s in quadfire. Unfortunately I won't get time for the next few days.
 
Theres too many variables to make this conclusive. Every aib partner has different fans, shrouds and speed profiles.
Also its crap idea about stock settings, for example my 7950 uses a boost bios,
at default its 800/1250 and boosts to 925/1250 and throttles every 2 secs as it uses 1.18v.

I of course increase the power limit and it'll hold 925, but then I also run my card at 1.05v when at 925/1600.
Or my when gaming 1175/1600 at 1.2v
 
11bntvl.png


Card: HIS 7770 iCooler (Arctic s1 plus with phobya g-silent 12 @ ~500rpm)
Default Clock Speed: Core 1000MHz, Mem 1125MHz
Idle Temp: 25
Load Temp: 40
Idle Fan: -
Load Fan: -
Subjective Noise Comment: Cant hear it, except coil whine. booo
 
spixelspixel said:
Card: HIS 7770 iCooler (Arctic s1 plus with phobya g-silent 12 @ ~500rpm)
Default Clock Speed: Core 1000MHz, Mem 1125MHz
Idle Temp: 25
Load Temp: 40
Subjective Noise Comment: Cant hear it, except coil whine. booo
Click to expand...
That is cool. Just 15deg difference between idle and load will be very hard to beat.

...but it's not a stock card.
 
Last edited:
So many people complaining, but there are so many overclocking threads where people bench cards within different systems within different environments and showing little or no proof of stability other than passing a 2-5min benchmark. What's the big difference between posting a one-off system benchmark and one-off max temp reading?

All I am asking to see is examples for what temps peoples cards reach at stock settings within the same random systems and environments as they use for posting benchmarks.

Granted, the only scientific bit is that the same benchmark running same settings is used to gather the temps over a minimum duration.

Lets see some 780's, 290X's etc. If only I still had my GTX 480.
 
Wanted to see what my chiller was like, doesn't matter about ambient temps for me. Though they were about 20c.

Chiller set to 11c and load temps of 16-17c. Final temp of 18c because the bleeding chiller turned off halfway through the second loop! :D

I'm thinking a 290X would love my chiller.




Uploaded with ImageShack.us
 
Think this is a great idea at the very least you could get an idea on the best or worst case scenario for the given benchmark. With all of these different custom cards I'm always interested to see what the load temp would be like cause I can only keep my 8600gt for so long.

Would be much more useful than the benchmark posts that have zero relevance to me but I'm sure it's great for people that are into benchmarking. Anything that could help potential buyers even a little bit should be encouraged not shot down. Get your 780's and 290x on this there's going to be a lot of new 780 owners over the next few days that's for sure.
 
Last edited:
It's not really comparable to benchmark threads.
They're just balls to the wall, artifacts and all, let's see what you can squeeze out of your card scoreboards.
Stability isn't claimed or expected unless stated.

This thread could be useful as a very rough guide, but again, there are so many variables involved here.
 
I have added a "Case and cooling" mini description to the list of useful items within the original post. I don't want to over-complicate this for everyone, but I would like the info to be useful, even if not scientifically nailed down to insignificant margins of error.

For me at least it will provide an idea of what to expect from each card type, and many people value cool running as much as overclockability. I am not knocking benchmarking and overclocking, for which I have created and posted within the odd thread or two. Just trying to add to them.
 
Last edited:
Ok 555 here we go then. Like I said this test isn't very conclusive as there's so many variables. But anyway I'll provide you some results.

Ambient temp was 26c

I think my card is a possible reflection towards the 290x.
If the 290x came with a lower voltage and all those cores on the 28nm process are stable then maybe we'd see 5-10c less temps for the same clocks. If we can reduce the load temp by 15-20c then an overclock of 150-200mhz on air with comfortable temps should be quite easily achieved. But yes the stockcooler is pretty poor, just like the 6990 was, but it's still going to be a hot card even with a decent air cooler, it's going to be a fine balance between noise and cfm performance.

Right my gcard is a powercolor http://www.powercolor.com/us/products_features.asp?id=455
My Results
Card: Powercolor twin fan HD 7950 BOOST V5E
Default Clock Speed: Core 850MHz, boost 925 Mem 5000MHz

In completely stock mode it will run at 925/1250 using 1.18v and auto adjust fan.
It will throttle at 48c and the temperature results and benchmark reflect this.


The above is unfair thermal test as the card is throttling also the performance from the throttling is a 5fps hit.
So here is a true standard rep of my gcard




925 stock 1.18v throttling to 850
30-62c 23-50% fan
Fans over 46% are just about audible

925 stock 1.18v 20% power limit
31-67c 23-54% fan
Fans over 50% are getting close to my personal limit. But not particularly loud.

What it should have been
925 1.05v 20% power
31-57c 23-46%

925 1.05v 20% power
31-55C 51% fixed

And just to indicate the excessive 1.18v voltage allows the core to achieve a stable 1125mhz.
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom