• 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.

670 lack of headroom

Dangerous_Dave

Dangerous_Dave

Dangerous_Dave

Associate
Joined
21 Sep 2010
Posts
455
Hi all,

Been tweaking my EVGA GTX670 but it seems to be throttling on a relatively low overclock (+120) due to exceeding the board power limit. It hits 130% power using EVGA OC Scanner furry stress thing with 96% GPU usage, and is therefore throttling as this is above the 122% max I've set.

Am I doing something wrong?
 
Power consumption/target can be directly related to vrm temps, I would say the card is running too hot over the vrm area.
 
Hello Dave,

to me this do also sound like the card is running too hot. Which model do you exactly have? Have you tried to set up a custom fan profile with EVGA Precision. If you raise the fanspeed earlier then the Auto settings this can often help to keep components little cooler.
Which clocks do you reach when you have set the overclock?
Did you also overclock the RAM? Mostly this is the bottleneck when overclocking 600 Series cards.
 
pgi947 said:
Depends on the card for what % my 670 windforce at 1350mhz was around 85% on air around 70% on water.
Click to expand...

Am I misinterpreting what the board power is? I thought it was %TDP, which I assume is the percent of the thermal energy given off when the GPU is running at 100% load at stock clocks (915MHz I think). If that were the case then with a ~30% total overclock (including boost) at ~100% load should give a board power of ~130%.

But then you wouldn't get 85% @ almost 50% overclock, so maybe I'm misunderstanding.
 
they aren't directly related, the power slider is just a number, just put it on max and then forget about changing it or reading what it's running at

my cards report a lower % on higher clocks, which doesn't really make a whole lot of sense

what boost clock are you seeing in game / benchmark?
that is the only important factor, +130 is a fairly decent OC, but what actual boost clock do you get from that?

the EVGA furry stress thing is also not a very good one to use for actual stress / stability testing as I got totally different results and a lower offset stable in actual gaming, use something like heaven in windowed mode for live tinkering, let it run the full benchmark test as a first off and then backed up with a 1 hour BF3 session - quite a few overclocks I managed Heaven stable that would fall over after 20-30 minutes in BF3
 
Last edited:
if your gpu offset is +120 you should be happy, I don't see what the problem is. The max I can get is +80 and that's with a modded bios @1.2V.
 
My understanding was the % represents the power being used at a given clock speed, if you hit 100% you need to increase it (though it should be safe enough to leave it maxed out all the time).

The relation of power consumption or what ever we want to call the % and temps is vrm temps not core temps, cooler vrms are more efficient from what I can make out, when well cooled the % should lower as did my % moving from air to water. I hope that makes sense as my head hurts a little after tying to type that on a phone :p

Agree that +120 is good, especially if your stock boost was 1050+
 
andybird123 said:
they aren't directly related, the power slider is just a number, just put it on max and then forget about changing it or reading what it's running at

my cards report a lower % on higher clocks, which doesn't really make a whole lot of sense

what boost clock are you seeing in game / benchmark?
that is the only important factor, +130 is a fairly decent OC, but what actual boost clock do you get from that?

the EVGA furry stress thing is also not a very good one to use for actual stress / stability testing as I got totally different results and a lower offset stable in actual gaming, use something like heaven in windowed mode for live tinkering, let it run the full benchmark test as a first off and then backed up with a 1 hour BF3 session - quite a few overclocks I managed Heaven stable that would fall over after 20-30 minutes in BF3
Click to expand...

Good advice :)
 
I think the standard boost is 1110 or something similar. I'm up to somewhere just over 1200 full boost - but I didn't think this seemed like a remarkable overclock. I'm sure I've read people frequently getting ~1300, and of people being disappointed with ~1250.

I understand how VRM temps could indirectly affect power usage - hot things have higher resistance and other adverse electrical effects. Seemed like people were saying the card uses some formula involving the VRM temp to calculate power %, but this makes more sense now.

So are we basically saying that somewhere around 1230 full boost is a good overclock on a 670?
 
Dangerous_Dave said:
So are we basically saying that somewhere around 1230 full boost is a good overclock on a 670?
Click to expand...
That's knocking around GTX 680 stock speeds in terms of benchmarks, it's not bad really.

I've had 3 Gigabyte WF3 GTX 670's now, first one did 1280MHz, second one bought off pgi947 did 1340MHz on my system, and the third one a 4GB version did 1230MHz maximum boost, with only +100 max stable on the memory.
 
Dangerous_Dave said:
Gives an average of ~1285 which is about what I thought

Didn't you have board power issues with your 1340 card?
Click to expand...
Never had any power issues with any of my cards. The only issue I had with the first two cards was heat, due to the card design pushing the warm air through the top into the case, because I had a crap Corsair 650D case - needs a side window fan really, it didn't get rid of heat good enough, so the cards throttled. Plus it was like Barbados in my room due to the warm weather.
 
Dangerous_Dave said:
Interesting to hear you guys talk about heat - doesn't the card automatically start throttling above 70C?
Click to expand...
My first two cards would throttle as they reached 70°C, but this was in May/June time - hotter weather, plus the Corsair 650D case, not very good case for cooling really - better performing cases out there for cooling.

The third - a 4GB card had a beefier heatsink/vrm design, took up 3 slots, with 20°C ambient temp, the maximum card temp was 55°C in my HAF X case I've switched to.
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom