• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Ordered another 980ti

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

mackerel

Member
Joined
Mar 7, 2008
As I said before, if prices drop enough I wanted a 980 Ti to play with SLI along my existing one. They just have... got a new one on order now priced between cheapest 1060 6GB and 1070.

Snag is... there doesn't seem to be a full cover water block for it. I haven't decided yet if I'll go universal, or not bother at all as it is the Gigabyte with massive triple fan cooler. I'm not sure I'll actually use it for gaming... more interested in it for SLI benching, and after that probably best to move it to another PC and use it for distributed computing. With "only" a 750W PSU I'm not sure I could run SLI OC'd 980Ti anyway... and I'm not shopping for another PSU!
 

Soulcatcher668

Member
Joined
Jan 2, 2012
Sadly, prices never drop like that in Canada. It very difficult justifying buying a second 980ti for more money than a better performing 10 series card. :(
 

EarthDog

Gulper Nozzle Co-Owner
Joined
Dec 15, 2008
Location
Buckeyes!
You can run it just fine on a quality 750W PSU (you dont list your PSU). So long as you are not modding the BIOS on the cards. Do the math. :)
 
OP
mackerel

mackerel

Member
Joined
Mar 7, 2008
I don't recall exactly how much power a single card can take under load. At reference 250W, two of gives 500W, leaving 250W for everything else, that's ok. If it nudges closer to 300W per card, that only leaves 150W for everything else which is a bit tight as the CPU could be 100W+ if OC hard...

The PSU is a Corsair HXi so I do have real time power monitoring and can keep an eye on it before something goes wrong.

As for 9xx vs 10xx... I debated over if I should get this. Well, the other option I had would be to SLI 1070's. But that would cost more and my existing 1070 is not easily extracted from a mini-ITX build. 980Ti is still competitive against a 1070 overall, only costing a bit more in running costs from higher power usage.
 

EarthDog

Gulper Nozzle Co-Owner
Joined
Dec 15, 2008
Location
Buckeyes!
Its a 250W card reference or factory overclocked really. Its only on rare samples you will ever see a stock card hit the TDP (100% Power in MSI AB).

Let me put it this way... I ran an 295x2 (500W) and a 5820K overclocked to 4.5GHz+ on a 750W PSU. It wasn't until I overclocked that card to the limit and pushed the CPU more that I ended up using 750W.

Now, if you swap BIOS, remove power limits and and more than the factory voltage allows, then yes, you would want more. :)
 
Last edited:
OP
mackerel

mackerel

Member
Joined
Mar 7, 2008
I was just looking into this... between pcper and Tom's Hardware, the highest in game average load for a reference 980 Ti was 233.5W at Tom's. Let's assume an easy overclock of 1400 MHz core, compared to stock boost of 1076 MHz, that's a 30% increase. Without touching voltages, you'd expect power consumption to be linear, so that takes it to around 300W.

I had forgotten about the power limits... what's the standard max without bios mod? That would seem likely to be the limiting factor. Now I think about it more, when playing with OC after adding the watercooling previously, I'm not sure if I saw the card hit whatever power limit there was. My notes suggest I could only set 110% on the Asus, which if that's based off rated TDP would be 275W, x2 = 550W, leaving 200W for everything else... I can get away with that.
 

EarthDog

Gulper Nozzle Co-Owner
Joined
Dec 15, 2008
Location
Buckeyes!
Without touching voltages, you'd expect power consumption to be linear, so that takes it to around 300W.
Power limits would be obliterated easily if that is the way it worked. It doesn't. I am sure you can see in your notes from past overclocking that when you were pretty close to fully overclocked, you would just barely be touching the 110% power limit.

Typically default BIOS are set to TDP... so your math at the end there is correct.
 
OP
mackerel

mackerel

Member
Joined
Mar 7, 2008
Ok, I made one incorrect assumption, that total power was purely dependant on core clock. Of course, we have ram and other parts separate to that, so the increase would be less so. I just winged it on the assumption the core would be the bulk of the power. I'm not aware if anyone has shown how much the ram or anything else takes compared to the core.
 

EarthDog

Gulper Nozzle Co-Owner
Joined
Dec 15, 2008
Location
Buckeyes!
Ram is a pittance compared to the core.

It does not scale linearly with clockspeeds is the underlying point. :)
 
OP
mackerel

mackerel

Member
Joined
Mar 7, 2008
A core doing work proportional to clock, should scale with clock in a simplified ideal case. Where is may deviate is if the higher current at higher clock causes more voltage droop, it may be less than expected. Conversely, higher heat from higher clocks could lead to decreased efficiency, and increase power disproportionately. My 280X has always been one step away from thermal runaway. I suppose even for the core, you could argue the memory controller parts, and PCIe side parts, do not have to scale with the GPU core itself, so that may take some edge off it. Any other mechanisms I might have overlooked for now?

Don't make me actually measure this... :)
 

JrClocker

AKA: JrMiyagi
Joined
Sep 25, 2015
I'm running on my main rig:

i7-5820 K
Titan X Pascal (250 W TDP..120% on power slider)
980 Ti Hybrid (250 W TDP...110% on power slider)

With my CPU drawing about 80 W, the Titan at 110% draw, 980 Ti at 105% draw...I'm pulling 775 W from the wall.

My power supply is gold rated, so figure 90% efficiency...puts my power supply load at 697 W (with everything...fans...drives...and a minor load on the processor).

A 750 W power supply is cutting it pretty darn close for 2 980 Ti cards (with 250 W TDP)...850 W should be fine though.
 

EarthDog

Gulper Nozzle Co-Owner
Joined
Dec 15, 2008
Location
Buckeyes!
A core doing work proportional to clock, should scale with clock in a simplified ideal case. Where is may deviate is if the higher current at higher clock causes more voltage droop, it may be less than expected. Conversely, higher heat from higher clocks could lead to decreased efficiency, and increase power disproportionately. My 280X has always been one step away from thermal runaway. I suppose even for the core, you could argue the memory controller parts, and PCIe side parts, do not have to scale with the GPU core itself, so that may take some edge off it. Any other mechanisms I might have overlooked for now?

Don't make me actually measure this... :)
Your choice. I can only speak to experience in that we are not seeing linear scaling. Even with adding voltage, you can reach those clockspeeds and still not hit 110% give or take several %. I don't know why it works that way, but it does. Otherwise, as I said earlier, anyone that overclocked would slam into these meager power limits MUCH quicker than we are now. I can overclock and overvolt the 980Ti I have to the moon and not use 300W on the stock BIOS.

- - - Updated - - -

I'm running on my main rig:

i7-5820 K
Titan X Pascal (250 W TDP..120% on power slider)
980 Ti Hybrid (250 W TDP...110% on power slider)

With my CPU drawing about 80 W, the Titan at 110% draw, 980 Ti at 105% draw...I'm pulling 775 W from the wall.

My power supply is gold rated, so figure 90% efficiency...puts my power supply load at 697 W (with everything...fans...drives...and a minor load on the processor).

A 750 W power supply is cutting it pretty darn close for 2 980 Ti cards (with 250 W TDP)...850 W should be fine though.
This is a good data point for a 'worst case scenario'... almost. I don't think mack [email protected] or crunch numbers on his GPU. So in gaming and such, it will be using less power than his by a fair margin.

@ Jr - It would be great information to know what your ACTUAL power draw is (power % in MSI AB) on the cards. What you set the power slider to doesn't mean much of anything by itself. With your overclocks, are you hitting the power limit when folding?
 

JrClocker

AKA: JrMiyagi
Joined
Sep 25, 2015
@ Jr - It would be great information to know what your ACTUAL power draw is (power % in MSI AB) on the cards. What you set the power slider to doesn't mean much of anything by itself. With your overclocks, are you hitting the power limit when folding?

I did:

With my CPU drawing about 80 W, the Titan at 110% draw, 980 Ti at 105% draw...I'm pulling 775 W from the wall.

You need more coffee! :D

[email protected] gets my power draw up there to these numbers.

When I fold on the 980 Ti and run Heaven benchmark on the Titan, the power draw on the Titan hits the 120% limit...and my power draw from the wall starts to hit over 800W.

I was doing this last night and was seeing my UPS (i.e. power draw from wall) hitting 875+ W from the wall. It's only a 950 W rated UPS, so I started to get worried. I unplugged two of my monitors from the battery side of the UPS (had all 3 running off the battery side of the UPS), and my UPS power draw from the wall dropped 50 W. (Makes sense as they are 25 W monitors).

So, my 775 W power draw from the wall is 800 W from the UPS, minus 25 W for the 1 monitor I have running on there.
 
OP
mackerel

mackerel

Member
Joined
Mar 7, 2008
I don't think mack [email protected] or crunch numbers on his GPU. So in gaming and such, it will be using less power than his by a fair margin.

I don't crunch on GPU (yet) since it is still far too hot to do so, but give it another month or so and I'll start becoming active on PrimeGrid's AP27 project as my first target to collect credits in. I kinda want to do more folding too, and might have another fight with the client some time. I was running only CPU on a PrimeGrid 5 day challenge that ended yesterday, and it was painful near the end with air temps by my computer desk at 32C+ (90F+).

In power terms, I wouldn't crunch on both 980Ti in the same box anyway and will move the 2nd one into a dedicated cruncher once I'm done with SLI benching. I don't feel a need to SLI game on it given the limitations of SLI.
 

EarthDog

Gulper Nozzle Co-Owner
Joined
Dec 15, 2008
Location
Buckeyes!
Missed that Jr!

Just as a reference... played some BF4 for a while on the 980Ti I have...see pic below. After overclocking to 1477 MHz core (+85), 8022 Memory (+200) no voltage changes, I average around 82/83% versus the 79/80 before. With a BIOS set like mine at 300W, that is around 240 and 252W respectively. That translates to about 3.7% increase in power use overall for a +85(5.8%)/+200(10.4%) clock increase. I didn't have my KaW plugged in sadly... As always, YMMV. :)

pwr.jpg
 
OP
mackerel

mackerel

Member
Joined
Mar 7, 2008
Code:
OC	core    TDP %	PSU out	GPU pow
-50	1127	86	282.9	214.9
0	1177	88	288.7	220.1
50	1227	90	292.9	224.7
100	1277	91.9	298.5	230
150	1327	93.9	304.6	234.4
200	1377	95.7	307.5	239.1
250	1427	97.5	313.2	243.5

Ok, I did the testing to see how things scale giving the above table. I ran Heaven and looked at the TDP % in GPU-Z, and power supply output in Corsair Link. It soon became obvious this wasn't going to work as the scene changed in Heaven, the power changed significantly also. So I stopped it where it was, near the dragon status and I moved the camera a bit to look at the statue, and left it like that for the rest of the testing. Variations became much smaller then.

In this scene, my current 980Ti is shown as voltage limited and the core voltage didn't budge from 1199mV for the whole test. I also didn't touch the ram at all, leaving it at 3505 reported.

I varied the clock from -50 to +250 MHz in 50 MHz steps, around the nominal clock of 1177. To do some averaging of the values, I fired up hwinfo64 which samples and shows an average value also. After changing the clock, I waited a few seconds then reset hwinfo64. Then I waited at least 30 seconds before writing down the average shown in the table above.

The reported TDP % went up 2, 2, 1.9, 2, 1.8, 1.8 points. Averaging out overall just over 1.9%/50MHz. Fairly linear, but is it dropping slightly on the top end? I would speculate at this point, since I kept the memory clock static, might that start to limit as the core increased? It may be more useful if I could have measured GPU FLOPS vs power.

Similarly for the PSU power, it went up 5.8, 4.2, 5.6, 6.1, 2.9, 5.7. More variability here, as the PSU reported values moved around a lot more. Still, there seems to be a trend of about 5W/50MHz overall.

The reported GPU power seems to be derived from the TDP %, as it seems to be converted at a rate of 100% = 250W. There is a minor variation which is likely due to a combination of rounding error and possibly samples taken at slightly different times.

For fun I also fired up FurMark. Reported TDP was around 99% (with limit at 100%), and PSU power around 320W. Relaxing power limit to 110%, TDP went up to 108% (271W) and PSU was just over 345W. I noted the clocks and voltages in this state were lower than for heaven. By stressing different parts of the GPU, we can't compare directly across different loads. It is probable that the slope will vary according to the load, until such time it hits one of its limits. In other words, there is no fixed relation between core clock and TDP, as it will depend on the load. However for a constant load, the two do appear proportional if not under some other limit.

Back to my earlier question, it would seem regardless of the load I throw at it, the maximum power will be limited to 275W after OC without bios mod. With two such GPUs together taking up to 550W, that leaves 200W for use elsewhere. I wonder how likely it is to reach 275W per card outside of FurMark...