*GTX 260/280 Unofficial TPU! Thread*

[AddSub]

I just picked up another brand new EVGA GTX 260 for only $161 AR/CB. Should be in my hands on Friday. I'm still working on getting my CPU up to 3.8GHz. I can only do 3.72GHz fully stable. I can boot at 4GHz (well most of the time) but lot of completely random BSOD'ing when benchmarking. I will keep tweaking my CPU/RAM until the second GPU arrives. Benchmark results to follow

 

[waxking1]

I hear what you are saying about the current 280's, I'm in the mood to get something, just sold my brand new warranty returned 790i ultra so have some cash, just think that I would be dissapointed in the 280 in as much as I am guessing there would actually be little noticeable difference overclocked as they dont seem to clock as well as most of these 260's, now if I could find a 280 that would do at least 770 on the core, then that would be a bit different!

Any idea when these uprated 55nm 280's are speculated at hitting shelves?

Actually I think the core goes from 756 to 783 with no stops in between. I've seen where watercooled cards have done 783 but I don't want to watercool for various reasons. The current rumor about the 280+ is around 3 weeks from now.

This website has the 280+ and the 350 listed. I'm hoping to get a 280+ and stepup to a
350.

http://austin.net.au/ProductList/tabid/103/Default.aspx?Category=GRAPHICS+CARDS

 

[johnspack]

Wish my 280 wouldn't artifact above 720, and shaders above 1480.. this thing is a pig heheh! And where are those big bang drivers, the last set just sucked....

 

[SplitBURST]

Pushing these cards to the limit.

I wounder, when pushing these cards to the limit with these instantly high clock speed how does it affect the lifespan of the graphic card, is heat the only thing one should be concerned about? (obviously no artifacts should arrive as well) Also is there anyone that have done some voltage mods physically or in bios?

 

[Tatty_Two]

I wounder, when pushing these cards to the limit with these instantly high clock speed how does it affect the lifespan of the graphic card, is heat the only thing one should be concerned about? (obviously no artifacts should arrive as well) Also is there anyone that have done some voltage mods physically or in bios?

It's a lot safer to push a graphics card than it is for a CPU for instance, simply because your motherboard may well give you huge voltage options that would just burn the chip, a graphics card is deliberatly hardware limited to a max voltage, of course if there is a voltage mod for a particular card and you increase the available voltage by lowering the resistance thru the regulators then yes, not only do you void your warranty, it is also highly likely that you will shorten the lifespan of your card.....now if you tend to keep your cards for like over 2 years then you may be affected by this.

As far as BIOS mods go, there is certainly a risk of getting it wrong and getting a bad flash but that is rarely irreversible if you know what you are doing, there is a limit on what you can do voltage wise because of the regulation so it is safer but has less performance impact than a "hard mod".

 

[TonyStark]

My DS3 died on me, so I upgraded my motherboard to Asus P5Q Pro. Crysis benchmark has gone up by 8-9fps. I think the GTX260 was bandwidth limited by the PCI-Express 1.1 slot. It would be nice if someone with a PCI-E 1.1 board could overclock the bus to 110MHz and bench the card to confirm this.


New bench:

Crysis 1680x1050 | DX9 | high | Vista 64-bit
Min FPS: 35.69
Max FPS: 60.59
Average FPS: 49.47

 

[SplitBURST]

It's a lot safer to push a graphics card than it is for a CPU for instance, simply because your motherboard may well give you huge voltage options that would just burn the chip, a graphics card is deliberatly hardware limited to a max voltage, of course if there is a voltage mod for a particular card and you increase the available voltage by lowering the resistance thru the regulators then yes, not only do you void your warranty, it is also highly likely that you will shorten the lifespan of your card.....now if you tend to keep your cards for like over 2 years then you may be affected by this.

As far as BIOS mods go, there is certainly a risk of getting it wrong and getting a bad flash but that is rarely irreversible if you know what you are doing, there is a limit on what you can do voltage wise because of the regulation so it is safer but has less performance impact than a "hard mod".

I did the bios volt mod, am currently at 780/1560/1200 with 790/1580/1200 the PC died! haha so I ingress volt on the PCI-E slot by 2v (I think) still running test on 780/1560/1200 I can smell the electricity in the room its quite disgusting I must say

 

[SplitBURST]

My DS3 died on me, so I upgraded my motherboard to Asus P5Q Pro. Crysis benchmark has gone up by 8-9fps. I think the GTX260 was bandwidth limited by the PCI-Express 1.1 slot. It would be nice if someone with a PCI-E 1.1 board could overclock the bus to 110MHz and bench the card to confirm this.


New bench:

Crysis 1680x1050 | DX9 | high | Vista 64-bit
Min FPS: 35.69
Max FPS: 60.59
Average FPS: 49.47

I have the GA-P35C-DS3R is this also PCI-E 1.1?

 

[waxking1]

I did the bios volt mod, am currently at 780/1560/1200 with 790/1580/1200 the PC died! haha so I ingress volt on the PCI-E slot by 2v (I think) still running test on 780/1560/1200 I can smell the electricity in the room its quite disgusting I must say

790/1580/1200 and 780/1560/1200 are the same thing. With these settings the card is running at 783/1566. The next shader jump is all the way up to 1620. The jump comes when you set the shaders to 1594 so any shader value between 1540 and 1594 is running at 1566. For a shader value of 1566 the max core you can run is 1/2 which is 1583, unless you have found a way around this. If so please share.

 

[SplitBURST]

790/1580/1200 and 780/1560/1200 are the same thing. With these settings the card is running at 783/1566. The next shader jump is all the way up to 1620. The jump comes when you set the shaders to 1594 so any shader value between 1540 and 1594 is running at 1566. For a shader value of 1566 the max core you can run is 1/2 which is 1583, unless you have found a way around this. If so please share.

At first i thought so too. I saw that everest was showing different values then my other programs and behave just like you explain d but...even thue everest said something different the true clock speed was stated in GPU-Z and EVGA Precision I know this because 790/1580/1200 did not work but 780/1560/1200 did! how do u els explain that

 

[waxking1]

At first i thought so too. I saw that everest was showing different values then my other programs and behave just like you explain d but...even thue everest said something different the true clock speed was stated in GPU-Z and EVGA Precision I know this because 790/1580/1200 did not work but 780/1560/1200 did! how do u els explain that

I can't. Maybe something else is at work here. I'll try to do some kind of testing when I get a chance. You very well could be right. Might be easiest if I just ask EVGA.

On another note, what is the difference between the US and Europe as far as power goes. I've never been to Europe but don't you use 230v power?

 

[Kursah]

At first i thought so too. I saw that everest was showing different values then my other programs and behave just like you explain d but...even thue everest said something different the true clock speed was stated in GPU-Z and EVGA Precision I know this because 790/1580/1200 did not work but 780/1560/1200 did! how do u els explain that

1580 might have pushed the shaders up to 1622...it's hard saying. But I know from experience what waxking is stating about the clockspeeds is true. Everest reads what the driver set clocks are, not the hardware set clocks...thus not correct by definition.

It could also be that the OC was not stable enough for enough tests between those clock changes. How did you verify stability? Are you sure it was as stable as you claim at 780/1560? One run is not necessarily enough, sure it's okay if you're doing an OC run to see how far you can go before settling on a set 24/7 stable OC. But if it wasn't entirely stable, you set clocks up in driver that may or may not change in hardware set, and you run again that would explain the stability issues...you could've already been at the brink of stability and problems at the lower OC and just squeeked by for stability.

Keep us posted tho! Interesting results for sure!

 

[Tatty_Two]

At first i thought so too. I saw that everest was showing different values then my other programs and behave just like you explain d but...even thue everest said something different the true clock speed was stated in GPU-Z and EVGA Precision I know this because 790/1580/1200 did not work but 780/1560/1200 did! how do u els explain that

I can....it's a glitch with GPU-Z perhaps? Dont forget also, Overclocking the PCI-E slot on the motherboard also overclocks the card and changes ratio's as in 9600GT, W1z did a large piece on it a while back and I read that the same applies to the 200 series, that may be the factor?

Found the article...................

http://www.techpowerup.com/reviews/NVIDIA/Shady_9600_GT/

On another note, I just posted my 2006 score on HWBot for Team Palit (Palit own Gainward ) and I have noticed that mine is the 2nd fastest 2006 score in there and the highest overclock, thing is, the geezer who is No1 loses in every single test apart from the one SM3 canyon flight test and his SM3 score is therefore so much higher his overall score beats me by about 500 points! well gives me something to aim for I suppose.

Strange thing is, his card is clocked a lot slower than mine and so is his CPU, how does he do that?....any idea's?

http://www.hwbot.org/quickSearch.do?hardwareId=GPU_1355

 

[waxking1]

I did a little more testing and I'm not convinced which clock values are right. Its funny how the clocks interact with each other. For example: If you set your core and shader clocks to 733/1496 the Precision graph reads 734/1512. If you change the clocks to 733/1495 the Precision graph changes to 756/1512. I'm sure there are more of these examples if you looked for them.

Anyway I ran 3dM06 twice Once at 733/1496(734/1512) and Once at 733/1539(734/1512). First values are on the slider and values in parenthesis is on the graph. My 3dm06 score went up almost 100 points when running the clocks higher on the slider. I'm not sure it proves anything, but I'm no longer convinced the graph values are the correct ones. Shouldn't 3dm06 stay almost the same since the graph values were the same for both tests?

 

[SplitBURST]

1580 might have pushed the shaders up to 1622...it's hard saying. But I know from experience what waxking is stating about the clockspeeds is true. Everest reads what the driver set clocks are, not the hardware set clocks...thus not correct by definition.

It could also be that the OC was not stable enough for enough tests between those clock changes. How did you verify stability? Are you sure it was as stable as you claim at 780/1560? One run is not necessarily enough, sure it's okay if you're doing an OC run to see how far you can go before settling on a set 24/7 stable OC. But if it wasn't entirely stable, you set clocks up in driver that may or may not change in hardware set, and you run again that would explain the stability issues...you could've already been at the brink of stability and problems at the lower OC and just squeeked by for stability.

Keep us posted tho! Interesting results for sure!

Note: I will refer to the different clock set by set1 "730/1512/1188" set2 "783/1566/1188" (according to Everest)

Your right! after 1h playing Crysis screen went black and I had to reboot. I got tired of this so I flashed back the original bios. If I cant run at set2 It's not wort it set1 is max I can run without the bios volt mod (and this is STABLE and heat development much lower) not so far from set2. I never got too benchmarking set2 with set1. But what I did do was compare stock settings too set1 and the gain is about 9 FPS on the Crysis demo at all high dx10 1680*1050. In the game itself fps vary from 25-60 and that's the whole reason for me trying to max oc's, under 30 fps in some scenes just blows and even at set2 the fps could go that low! I suspect it's more a driver problem, has to be?

BTW when I now have my default bios back my warranty is legit right? there aint' no way they can see that the bios have been flashed back and fort?

 

[johnspack]

Not likely, only if you did a hardware mod, or if you burnt something!

 

[johnspack]

Well, so far I can only hit 13.3k something in vantage, I'll bet with a oced e8500 I could hit 16! I gotta hurry up and get these darn intel parts...

 

[Tatty_Two]

Note: I will refer to the different clock set by set1 "730/1512/1188" set2 "783/1566/1188" (according to Everest)


BTW when I now have my default bios back my warranty is legit right? there aint' no way they can see that the bios have been flashed back and fort?

Your right, but you have to remember, if the card genuinly fails, you will not have the opportunity to flash the card back and therefore your warranty will be void.

 

[SplitBURST]

Your right, but you have to remember, if the card genuinly fails, you will not have the opportunity to flash the card back and therefore your warranty will be void.

hehe I get that

 

[johnspack]

Well, found an internal build 178.15 that fixes all of the problems of the 177.98-178.13 bugged drivers. Digital vibrance back and working, multi card support fixed. Just lower performance than the 177.92s. Here big bang! here boy....http://downloads.guru3d.com/GeForce-ForceWare-178.15-XP-64-bit-download-2088.html you will see the links for the rest of the oses there. So want to take a poll as to how long for 180 series to show up?

 

[Kursah]

I'll give em a shot for sure! My rig is unhooked atm, setting up and tuning a build for a friend. Just popped an 8800GTS in it, running the .13's issue free thus far in Vista x64, but I think I'll find the 177.83's and go with those...for me at least, they seemed to be the most stable and consistent. I'm just hoping the 180 series come out sooner than later and have some good tweaks in them!

As-far-as the poll goes...I think NV is gonna string it out at this point...or bam it could happen tomorrow...really I'm dissapointed they weren't released in sept, that could be a good thing tho...maybe they found issues and these later 178 drivers are just a variation of the 180's...and the true 180's will be worth the wait...hard saying tho.

Keep us posted john!

 

[johnspack]

I'm still finding the 177.92s are by far the best performers still...

 

[PaulieG]

Thought I'd share how my 260gtx is doing....

 

[johnspack]

I need my e8500, here money money!

 

[Kursah]

Nice results man! Seems like a damn good combo! I do miss my ol' Q6600 here and there, but it needed way too much voltage for my likings at higher than 3.6/1.42 lol. Was a good chip for what it was.

I'll get back into this thread once this build I'm currently using is completed and off to my buddy (Friday night is the plans).

I've gotta put my 216 GTX260 through the runners soon! Still shocked that it is only running 1.06v for Extra clocks compared to my 192 GTX260's 1.12v. But it was completely stable at FTW clocks thus far and that's about as-far-as it's been taken.

Again Paul, nice work! Thanks for submitting your results!