*** NVIDIA 470/480 Series Thread ***

Quote:

---

Originally Posted by Aedubber

Seems like there is way too many cards burning up ... Looks like I'll have to put a hold on ordering gpu cards .. Might have to stick with the initial plan of running 2 hd5870s .

---

I haven't heard of any cards actually burning up, runninng hotter yes. It is a must that you use a full tower case with either lot of fans or water cool your cards, whether I own ATI or Nvidia water cooling is a pretty good investment for the long haul! You don't have to buy the most expensive water block out there. Each year more and more heat is being generated by GPU's, imagine how cool you could be running without any overclocking. If you live in a warm climate water cooling in the long run isn't a bad idea, the water cooling companies must have a big grin on their faces since the release of Nvidia's GTX 470/480! :D

Well i am watercooled as we speak but what im saying is the cards are dead... Not cool

my cards never get above 80c, running at 800mhz core, 2000mhz mem, with 1.050v, custom fan profile is set in place, 90c will turn the fans to 100%.

Anyone have any opinions regarding OCCT as stability benching?

I recently got my GTX480 under water, and done some benches using Kombustor, Furmark, Vantage, Dirt2, Unigine and OCCT.
Seems like whenever I go above 875 / 2001 on either core/mem - every test except OCCT kinda gives a "true-false" result. Especially Kombustor and Furmark doesn't show any artifacts or instability above like 900/2200, although both Vantage and Unigine show instability and OCCT gives :banana::banana::banana::banana: loads of errors :(

Anything above 850/2001, I've set voltage to 1.2 in AB.

it's called a false positive, and it's because furmark and kombustor only test the core, not memory so I'd suggest turning the memory speed down

Quote:

---

Originally Posted by tentp0le

Anyone have any opinions regarding OCCT as stability benching?

I recently got my GTX480 under water, and done some benches using Kombustor, Furmark, Vantage, Dirt2, Unigine and OCCT.
Seems like whenever I go above 875 / 2001 on either core/mem - every test except OCCT kinda gives a "true-false" result. Especially Kombustor and Furmark doesn't show any artifacts or instability above like 900/2200, although both Vantage and Unigine show instability and OCCT gives :banana::banana::banana::banana: loads of errors :(

Anything above 850/2001, I've set voltage to 1.2 in AB.

---

Looks like you just found three programs that are reliable for showing instability, but Helloworld is correct. Return memory to stock clocks and find your maximum overclock for your GPU and Shaders, memory clock speeds are less important then the GPU and Shaders unless this new GPU is different than the previous. ATI cards though respond well to fast memory which is why they started using GDDR5 in the first place.

Quote:

---

Originally Posted by Systemlord

Looks like you just found three programs that are reliable for showing instability, but Helloworld is correct. Return memory to stock clocks and find your maximum overclock for your GPU and Shaders, memory clock speeds are less important then the GPU and Shaders unless this new GPU is different than the previous. ATI cards though respond well to fast memory which is why they started using GDDR5 in the first place.

---

Hmmm wasn't too much changes for a stable clock throughout all benchies, atm I'm running the card @ 950 / 2150 - which seems rock solid in eveything. In OCCT I ran it for 10 min, and had a total of 36 errors - which isn't bad.

Quote:

---

Originally Posted by tentp0le

Hmmm wasn't too much changes for a stable clock throughout all benchies, atm I'm running the card @ 950 / 2150 - which seems rock solid in eveything. In OCCT I ran it for 10 min, and had a total of 36 errors - which isn't bad.

---

I wonder if OCCT is really seeing errors, it used to be if you can't see artifacting then there were no problem or issues as long as it was stable in games. Could you be so kind and tell me how to make the change in AB 1.6.0 6 Beta to unlock voltage limits as I seem not able to make any changes. :)

weird how if you up the clocks on some modern processors they become more stable, not necessarily stable, but more stable

Quote:

---

Originally Posted by Systemlord

I wonder if OCCT is really seeing errors, it used to be if you can't see artifacting then there were no problem or issues as long as it was stable in games. Could you be so kind and tell me how to make the change in AB 1.6.0 6 Beta to unlock voltage limits as I seem not able to make any changes. :)

---

Ya, I could see the artifacting when it happened quite clear, since the dots are white to red background.
Consider everything below 100 errors for a 10 min test, to be rock solid. Pretty much done every test I can do, and all is good.

As for AB, there is a v1.6.0 Final version out, head over: herel.
Besides the obvious; where you need to go to "settings -> tick "unlock voltage control"". Also I recommend you to flash your card with a modified bios with "max volt" for the core aka v1.213. For most card there is a 70.00.21.00.xx bios out; "xx" is different depending if you have EVGA or another brand of the card.

At least this is how I did it :)

Quote:

---

Originally Posted by tentp0le

Ya, I could see the artifacting when it happened quite clear, since the dots are white to red background.
Consider everything below 100 errors for a 10 min test, to be rock solid. Pretty much done every test I can do, and all is good.

As for AB, there is a v1.6.0 Final version out, head over: herel.
Besides the obvious; where you need to go to "settings -> tick "unlock voltage control"". Also I recommend you to flash your card with a modified bios with "max volt" for the core aka v1.213. For most card there is a 70.00.21.00.xx bios out; "xx" is different depending if you have EVGA or another brand of the card.

At least this is how I did it :)

---

I was told EVGA cards couldn't be flashed with the modded bios. :confused:

And the reason for that being?

I'm sure the EVGA cards can be flashed with newer bios with a higher max voltage. At least people in this thread has done it, posted a couple of pages back. Not sure if it is "FTW edition" users only...

I'm at work atm, but I could give you a better description later.
Take a look here and you'll see the top 4 bios' is the newest for GTX480's.
Revision 70.00.21.00.82 for EVGA's, for me using Zotac the revision is 70.00.21.00.02

Quote:

---

Originally Posted by tentp0le

And the reason for that being?

I'm sure the EVGA cards can be flashed with newer bios with a higher max voltage. At least people in this thread has done it, posted a couple of pages back. Not sure if it is "FTW edition" users only...

I'm at work atm, but I could give you a better description later.
Take a look here and you'll see the top 4 bios' is the newest for GTX480's.
Revision 70.00.21.00.82 for EVGA's, for me using Zotac the revision is 70.00.21.00.02

---

I was told this by someone over at overclockers.net, don't know if there's any truth to it but it came from someone with a lot more experience than myself. My BIOS version is 70.00.21.00.80, I have zero experience with flashing a videocards BIOS and to attempt it without knowing what it is I'm doing is sure to fail! I can't believe there isn't a visual guide with pictures showing people how to flash rather than tell how to flash a cards BIOS! No YouTube video either, am I in the correct dimension? :confused:

Anyone care to enlighten me about the missing temp readouts for VRM on 480's? Did they put FETs without temp sensors on these/forgot to put temp sensors in that area?!

Quote:

---

Originally Posted by p2501

Anyone care to enlighten me about the missing temp readouts for VRM on 480's? Did they put FETs without temp sensors on these/forgot to put temp sensors in that area?!

---

Quoted from EVGA forums:

Quote:

---

GTX 480 VRM indeed has temperature monitoring capability, however it requires _external_ thermal sensor to be connected to one of VRM input pins to provide temperature readings. That pin is not connected to anything on GTX 480, so there cannot be any working VRM temperature monitoring software for GTX 480 neither now nor in future.

---

Wow, okay thank you! Do they also happen to provide info on how one actually would one go on about getting a sensor attached there?!

OCCT GPU
FurMark
KOMBUSTOR
eVGA OC Scanner

What's the best to test the OC on the GTX 400 s ?

Edit**

15mins of OC Scanner

0,912v Idle / 0,961v Load
http://img823.imageshack.us/img823/7495/semttulohy.jpg

Are my cards performing ok ?
http://img810.imageshack.us/img810/1884/heavens21.jpg

Uploaded with ImageShack.us

i7 980X @ 4.43Ghz + 3xGTX 480 @850Mhz @1950Mhz mem

PhysX off!

http://rapidshare.com/files/397764473/Vantage_42389.jpg

Unigine Heaven

http://rapidshare.com/files/39777459...ine_Heaven.jpg

Stone Giant - Avg 242 FPS

1 more quick Vantage run @900Mhz @1.1v

http://rapidshare.com/files/39779365...tage_43295.jpg

http://www.techpowerup.com/124333/GE...GA_Cooler.html

http://www.hkepc.com/forum/viewthrea...age%3D1&page=1

EVGA has a better rear plate cooler that is said to reduce load temps by 7-10C! They could have done better though in the middle section.

had a bit of a scare after trying 1.2v to try get past 950mhz
it was still artifacting at bootup and on desktop after restarting the pc but thankfully after a full shutdown and left unplugged for a while now its ok again

it seems anything over 1.175v and this card crashes out of heaven in the first 10sec with gpu kept under 40c
not sure what vreg is reaching still have the stock plate on it but i doubt temp could be the problem in this case it wouldnt get hot enough in such a short space of time

been playing with stability testing

games seem stable at 900 core 2100 mem 1.125v evga oc scanner also runs fine at this
folding is only stable at 850mhz 1.125v or 800mhz stock v1.05
farcry2 benchmark is the best program i have found for testing memory stability much more than 2100mhz and it starts showing pink splotches

this is just the v as set in afterburner havent tested with a multimeter

I was just wondering, has anyone tested the 400's to see how much memory clocks makes a difference?

e.g. 900MHz/stock memory vs 900MHz/2100MHz memory

"The ENGTX480 ships with 32 shader processors (what Nvidia calls “CUDA cores”) disabled" ---- maximumpc.com

could we enable cuda cores? does cuda cores make a difference?