[Semiaccurate] GK104/Kepler/GTX680 Next Week?

[notty22]

If true, I hope the "Turbo Boost" can be disabled for regular overclockers.

I bet it can. In fact I will guess how it works. It will be bios implemented, but changeable only through the Nvidia control panel and will require a re-boot to enable and disable it.

 

[Vesku]

Wish I owned a popcorn company.

*Grabs some popcorn, awaits epic GPU media battle*

Wonder if clock auto-adjusting Kepler SKUs will carry a premium.

 

[Imouto]

I don't understand that "Speed Boost" thingie. Any modern card has several frecuencies in the core and memory for each type of work and load. What's the difference here?

 

[f1sherman]

This Speed Boost" thingie is a function of TDP and the required GPU load

so pretty much like Intel Turbo

 

[sandorski]

Hopefully the changing clocks on the fly doesn't introduce issues.

 

[BallaTheFeared]

It's not a problem with Fermi, my 470's will switch between 405Mhz and whatever overclock speed I have in some games depending on the load.

Sometimes one will be running at 60% load and 405Mhz while the other is at like 30% and 900MHz.

At least when using a fps limiter/vsync.

 

[thilanliyan]

It's not a problem with Fermi, my 470's will switch between 405Mhz and whatever overclock speed I have in some games depending on the load.

I think the bigger issue is memory clocks. IIRC there have been issues with memory clocks when switching between power states. Can't remember if it was AMD, nV, or both that had that problem.

 

[BallaTheFeared]

If this is a turbo core feature that probably wouldn't be an issue though right, as only the core/shader frequency would be in flux?

 

[thilanliyan]

If this is a turbo core feature that probably wouldn't be an issue though right, as only the core/shader frequency would be in flux?

Yep, probably not an issue.

I think a similar thing happens with AMD Powertune already.

 

[Arzachel]

Oh boy, it's not like Nvidia's TDP ratings aren't too misleading already. "Hey guys, it's a 250W part, we promise!*"

*Unless you're thinking of actually, you know, using it.

 

[railven]

Well then, maybe instead of putting a 600W in her rig, I'll go ahead and up that to a 750W. Just in case.

Turbo Boost in our videocards? That is actually a nifty idea. I wonder if it would work like Intel's technology were it idles some of the chip and ramps up the working end, or if it will just do like my old 2004 MSI motherboard with DOT (Dynamic Overclocking Technology) and overvolt the processor and raise the multiplier a few ticks when some arbitrary limit was reached - which I never figured out when.

Bring the new tech on!

 

[MrK6]

Oh boy, it's not like Nvidia's TDP ratings aren't too misleading already. "Hey guys, it's a 250W part, we promise!*"

*Unless you're thinking of actually, you know, using it.

*Desktop and decoding usage only.

 

[f1sherman]

That ship sailed long ago - TDP is under control.
Chip size alone should tell you that.
Then remember Apple is back, and all those laptop wins...

On power envelope alone, AMD will have trouble selling it to PETA Beowulf cluster built by Al Gore

 

[BallaTheFeared]

Something must have made every OEM including Apple sign Nvidia for Kepler


Haters gonna hate though...

 

[SolMiester]

I can understand multi core turbo stepping when only 1 core used, allowing clocking to TDP limit, however if this is just 1 core, why not just clock it as high as possible when in 3D?...
Will there be a threshold of FPS before kick in?. It sounds cool and good on NV for leading GPU hardware again, but it sounds a bit of a gimmick?..

 

[f1sherman]

Oh boy, it's not like Nvidia's TDP ratings aren't too misleading already. "Hey guys, it's a 250W part, we promise!*"

Oh boi...talking about barking at the wrong tree

:whiste:

Speaking of the "Samaritan" demo, Rein noted that when they showed it off last year, it took three Nvidia cards and a massive power supply to run. However, they showed it off the demo again that ran using a new, not yet released Nvidia card and one 200 watt power supply.

http://www.gamesindustry.biz/articles/2012-03-08-gdc-epic-aiming-to-get-samaritan-into-flash

 

[PhoenixEnigma]

I can understand multi core turbo stepping when only 1 core used, allowing clocking to TDP limit, however if this is just 1 core, why not just clock it as high as possible when in 3D?...
Will there be a threshold of FPS before kick in?. It sounds cool and good on NV for leading GPU hardware again, but it sounds a bit of a gimmick?..

Probably similar to the all-cores turbo on CPUs. Not all instructions use the same amount of power, there's usually a gap between maximum specified ambient temperature and actual ambient, probably a couple other factors. If there's a safe way to eat into some of that headroom, why not grab that extra couple FPS?

 

[tviceman]

Oh boi...talking about barking at the wrong tree

:whiste:




http://www.gamesindustry.biz/articles/2012-03-08-gdc-epic-aiming-to-get-samaritan-into-flash

Don't worry, they will happily select to ignore this information.

 

[BallaTheFeared]

Don't worry, they will happily select to ignore this information.

Nvidia card and one 200 watt power supply.

lolwut

 

[railven]

Does that mean that one 200W PSU alone is powering the card or the whole system? If the whole system, this thing must run at negative power!

WTB clarification.

 

[Imouto]

Utter bullshit.

 

[f1sherman]

not at all

it may require some thinking though :whiste:

 

[Vesku]

Must be off on the wattage or referring to one of those secondary GPU only power supplies.

 

[Doougin]

i think its supposed to be it only used 200 watts

 

[Vesku]

Well the CPU would probably be using a third of that or more. That demo doesn't look Celeron friendly.