New Geforce 650 TI Boost

[NaturalViolence]

Not mine but a similar card to my GTX 260:



The largest card that I know of is probably the voodoo5 6000:


3D accelerators from the 90s in general tended to be pretty big:


Most high end cards are pretty big (10-14 inches long at least). The 8800 GTX/Ultra was enormous as was the 9800 GX2 and GTX 280 if you want some recent nvidia examples. Dual GPU AMD cards like the 6990 can get pretty big too.

Sorry I don't really have any images of any of my graphics cards on hand. And I'm not going to go dig them up just to take one.

[Zeikai]

Small update, while the auto overclocking is very nice, getting upwards of 1293 I'm gonna try some regular OC and see how high I can get it.

---

(04-02-2013, 07:02 AM)Zeikai Wrote: Small update, while the auto overclocking is very nice, getting upwards of 1293 I'm gonna try some regular OC and see how high I can get it.

Got the memory clock up to 3400mhz up from 3000mhz. It seems like I can't overclock the core clock though unless the auto is disabled, which I have no idea how to do. Although 1293mhz is probably the highest it runs stable at anyways.

[yodenny]

wtf
lol
why?
i thought the more advance the more compressed in size but i was wrong i felt like this guy
http://www.youtube.com/watch?v=k9QXY80OxS0
ill post this pic anyway

[NaturalViolence]

yodenny Wrote:wtf
lol
why?
i thought the more advance the more compressed in size but i was wrong i felt like this guy

If you sit and think about it for a little bit it makes perfect sense. Processing power is primary constrained by two things: power/heat and transistor count/die size.

Power/heat constrains internal signal frequencies.
Transistor count/die size constrains microarchitecture complexity.

Size fixes both of these problems. More space for large/complex cooling solutions that can dissipate heat faster and large capacitors/VRMs/MOSFETs/etc. to deal with the crazy high current. Also if you run out of transistors the only way to have more transistors in your architecture is to either shrink the transistors, make bigger chips, or split the architecture into multiple chips. Shrinking transistors can only be done about every two years so in the meantime if you want to add more transistors to a microarchitecture you need to make the chips bigger and/or use multiple chips. Obviously this increases the amount of space needed for the chips but on top of that the wiring needed to connect the multiple chips to each other requires an even larger/more complex circuit board.

yodenny Wrote:ill post this pic anyway

Take it down! Take it down! That cabling is awful.

[yodenny]

(04-02-2013, 10:03 AM)NaturalViolence Wrote: Take it down! Take it down! That cabling is awful.

funny thing is i once heard my pc sound like a motorcycle today i thought it was going to go on fire then i realised it was a plug hitting one of the fans smdh

[Zeikai]

So after doing lots of testing yesterday, the highest stable OC I managed to get was a 60mhz core clock oc (above the max boosted) which would be 1293+60 or 1353mhz. Memory clock was OC from 3000mhz to 3400mhz.


So far I'm extremely pleased with the card and for the price it's very good.

[rpglord]

(04-03-2013, 05:16 AM)Zeikai Wrote: So after doing lots of testing yesterday, the highest stable OC I managed to get was a 60mhz core clock oc (above the max boosted) which would be 1293+60 or 1353mhz. Memory clock was OC from 3000mhz to 3400mhz.


So far I'm extremely pleased with the card and for the price it's very good.

If these clocks are indeed correct, you should consider yourself very lucky, you got a golden card, because most kepler cards max around 1200-1250mhz
Whats happening probably is your are reading clocks from Heaven or Valley, and these benchmarks have a bug, and report higher clocks then what they really are.
You should use msi afterburner to monitor your clocks, they are correct there

[Zeikai]

(04-03-2013, 05:27 AM)rpglord Wrote:

(04-03-2013, 05:16 AM)Zeikai Wrote: So after doing lots of testing yesterday, the highest stable OC I managed to get was a 60mhz core clock oc (above the max boosted) which would be 1293+60 or 1353mhz. Memory clock was OC from 3000mhz to 3400mhz.


So far I'm extremely pleased with the card and for the price it's very good.

If these clocks are indeed correct, you should consider yourself very lucky, you got a golden card, because most kepler cards max around 1200-1250mhz
Whats happening probably is your are reading clocks from Heaven or Valley, and these benchmarks have a bug, and report higher clocks then what they really are.
You should use msi afterburner to monitor your clocks, they are correct there

Nope I'm pretty sure these are accurate, the boost on the actual card that automatically overclocks goes up to 1293 on all it's own, tested this with evga precision, and their oc scanner, ran it for 8 hours last night, max temp of 87C and 0 artifacts on furry tessa donut. But I guess I will consider myself lucky if what you say is true.Also I can't even test this with msi afterburner, it isn't supported. Not sure if it's cause it's brand new or if it's because of the evga brand.

I should also add, the card is almost never actually being used that much, I haven't really had a game push the card over 90% power yet, and it can go up to 110%, which at 110% is there this beast overclock would actually kick in. The whole boost management system truly is fantastic.

[NaturalViolence]

Lower end cards with less SPs will OC better if provided the same cooling. This seems normal to me.

[Zeikai]

Card is still performing fantastically with no issues with overheating or the overclocking, just wanted to check and make sure no one else wanted any tests run.