FireKillerGR - GeForce GT 730 GDDR5 @ 2210/1600MHz - 3620 marks 3DMark - Fire Strike

[Guest k.pisti]

Just my 2 cents: The frequency increase to this level is a "chase" on its own. So it is a bugged result no matter you did it on purpose or not. 99% this is gonna get removed. You could/should spend the last few minutes of this stage to upload a 100% valid result if you wish to surely keep your place.

[FireKillerGR]

When I am talking about chase I am talking about chasing in order to get a "bugged" run, which I didnt do.

BUT if you are talking about the general rule, then yes, overclocking is "chasing" the performance inrceasement (oh well we are @ hwbot.org).

 

And no, I have no time to re-bench at that point and the results I posted are at least 2-3 days old.

[Guest k.pisti]

No need to be sarcastic with me or use false reasoning. You uploaded a bugged result. It doesn't matter if you wanted it to run buggy, or it happened because of instable oc or whatever the real reason is. Score is going to get removed 99% due to the fact you uploaded a bugged result. It is that simple.

(I'm somewhat moderating this challenge so I know what I'm talking dude)

To avoid getting removed and to be fair with other competitors you should stick to uploading 100% valid results.

[phil]

No need to be sarcastic with me or use false reasoning. You uploaded a bugged result. It doesn't matter if you wanted it to run buggy, or it happened because of instable oc or whatever the real reason is. Score is going to get removed 99% due to the fact you uploaded a bugged result. It is that simple.

(I'm somewhat moderating this challenge so I know what I'm talking dude)

To avoid getting removed and to be fair with other competitors you should stick to uploading 100% valid results.

 

I don't see any sarcasm on FireKillerGR's post but with your attitude I will become sarcastic to you.Who are you and where did you came from and most important why could you be moderating a challenge like this.

For your information he has uploaded a valid result http://hwbot.org/submission/2870790 just to help you with moderating.

[phil]

Also just to help you a little bit more with moderating....a bugged run is called a run that the score is random and can't be repeated.This and the other "bugged runs"including demac's run doesn't seem so random and are very relevent with the gpu clocks.

http://hwbot.org/submission/2872608

http://hwbot.org/submission/2872094

Also have in mind that "bugged runs" happen only in one of the subtests of a benchmark.In the above scores you will see that all subtests have same boost.

I don't know how you can call it(it's something new to hwbot and almost sure never seen before) but it's not a bugged run and only hwbot crew can take a dessision about those runs.

Edited May 26, 2015 by phil

[Demac]

No Phil i think you are wrong here.I have uploaded 4 results in FS two not bugged and two bugged all at the same clocks.It seems to me that the bug happens due to high voltage and relatively low temp.For me when i run at -160 no bug occurs but when i start the GT1 at -150 the thing at my video happens.Also i have plenty of verification links and screenshots that i have not uploaded (they are at staff's disposal at any time) that have only one subtest bugged and it was due to artifacting and not rendering the correct amount of pixels at the specific subtest.And yes these runs are of course frequency related because there are same pixels left for rendering so the higher the gpu clocks the higher the score.I had no intention of uploading any bugged run untli i saw FK uploading his and i clearly stated that in my opinion it is bugged (thus and the video).My 2c and i have nothing personal with you guys at all,on the contrary you helped me HUGELLY for the bios to run Catzilla efficiently!!!

[FireKillerGR]

No Phil i think you are wrong here.I have uploaded 4 results in FS two not bugged and two bugged all at the same clocks.It seems to me that the bug happens due to high voltage and relatively low temp.For me when i run at -160 no bug occurs but when i start the GT1 at -150 the thing at my video happens.Also i have plenty of verification links and screenshots that i have not uploaded (they are at staff's disposal at any time) that have only one subtest bugged and it was due to artifacting and not rendering the correct amount of pixels at the specific subtest.And yes these runs are of course frequency related because there are same pixels left for rendering so the higher the gpu clocks the higher the score.I had no intention of uploading any bugged run untli i saw FK uploading his and i clearly stated that in my opinion it is bugged (thus and the video).My 2c and i have nothing personal with you guys at all,on the contrary you helped me HUGELLY for the bios to run Catzilla efficiently!!!

 

Actually I think that the truth is somewhere between the two sides.

On the one hand, yeah the score boost is actually obvious but on the other hand this happens every single time (!) when we run the card at 2150~ + with temp & voltage used @2.09 GHz. In other words, it doesnt happen every 1 run out of 5. Also the scaling from 2150 to 2210 MHz would be considered as normal (between two "bugged" runs).

 

Furthermore, commenting on what you said about rendering pixels I think whats happening is as legit as LOD can be. With LOD you change actually the "difficulty" (quality) of the gonna-be-rendered frames which is almost the same (we dont know for sure what happens at these clocks - at least I dont) with what happens here.

Now, for the guys who will say that LOD doesnt make such impact; please re-consider by thinking about 8xxx/9xxx NV series on 3D01 Nature (if I remember correctly) when you would earn at least 200 FPS (from 450-500 default to 700-750 FPS on the same clocks) by applying LOD.

 

Last but not least, from my side its also up to FM, as its getting paid for each license. In other words, in that cost, "security" against bugs & cheats is included and as far as I can see my score is invalid based on benchmark owner's rules.

 

To avoid getting removed and to be fair with other competitors you should stick to uploading 100% valid results.

 

On this part, I dont know you and you dont know me. Which means that you can also work for NASA, CSI, FBI or w/e you want.

But IF you are a mod (or mod assistant, etc.) I think it would be better to go through PM, email or facebook (as Christian did) and dont post stuff like "you will removed if your dont do what I say" & "So it is a bugged result no matter". IMO you handled the communication part terribly as you insulted someone (even if that someone doesnt really care about your online insults and how much you fail to deal with controversial submissions) online which could hurt his own rep. Cheers

Edited May 27, 2015 by FireKillerGR

[TerraRaptor]

Common sense says that score growth can not be higher than change in clocks. So if you raise, say, gpu freq +10% (2000 to 2200) the score can not get +20% higher - it must be counted a bug as the growth is expected to be equal or less than 10%.

[FireKillerGR]

Common sense says that score growth can not be higher than change in clocks. So if you raise, say, gpu freq +10% (2000 to 2200) the score can not get +20% higher - it must be counted a bug as the growth is expected to be equal or less than 10%.

 

I would NOT disagree but as I mentioned earlier a very close example is LOD on 01 as it changes the graphics (quality) of the frames which will be rendered and can give more than 20% fps increase without touching the clocks.

[TerraRaptor]

well man, LOD was accepted by community a long time ago and it was a battle those days if it is legit or not. LOD is rather an exception - wireframe hacks and manipulations with textures is forbidden since the beginning.

[Splave]

Firekiller when you run is it just looking like a black screen with score bar underneath? this is how it looks for me when I get it to bug to 3k at 1575/1600. I think it happens when very high volts + instability for me 1.55v on water

[FireKillerGR]

Firekiller when you run is it just looking like a black screen with score bar underneath? this is how it looks for me when I get it to bug to 3k at 1575/1600. I think it happens when very high volts + instability for me 1.55v on water

 

Nah Allen, I get screen everytime but with many artifacts (of course I can see the image/benchmark test). Also voltage is set 1.65v~ from evbot which is 1.73 - 1.75v load @ -135C.

[phil]

All I want to say is that if someone looks carefully at ALL scores,there is no way to say which score is bugged and which is not.

[FireKillerGR]

Based on what Phil posted.

 

On my 2852 score (which should be considered as Legit from all of you I guess) I got 14.26 FPS on GT1 @2090.3 MHz.

On Coldest's score GT 1 (which is mostly Core depended for the last 2-3 years as far as I know and also the easiest to get "bugged" as far as Demac said) he got 14.72 FPS @1920.2 MHz - +0.5 FPS with -170 MHz on core and -100 MHz on mem.

On Demac's 3057 Score GT1 = 16.10 FPS at 2142.6 FPS (+2 FPS with just +50 MHz) and on his 3122 Points score GT = 16.78 FPS @2168.7 (+2.5 FPS with +80 MHz).

 

At stock the FPS / (default) MHz ratio should be > than FPS / OCed MHz ratio cause everyone knows that if you double the frequency you wont get double the GT Score (I got 0.0068 FPS per MHz on my legit OCed score while I had 0.0083~ ratio at default). These scores have more than 0.0075 FPS per MHz (after years of running FS I dont know that you could have such impact on the GT score on the same freq and card) and some even have 0.00773 if my calculations are correct. So we are talking about almost the perfect efficiency (perfect scaling from stock clocks to increased clocks)!

[Demac]

Oh lets take for expample then Doc`s score 2890 graphics score with 1700 core then and please redo your whole thinking again man... you know because you tested that a normal 2100 mhz run should be around 3030 points overall test.If you do not want to admit it then it is your problem man.

I will not stay here and try to become even more petty trying to prove the obvious.It's a real bummer though seeing you react like that really...

Good luck to you and all the other guys for the rest of the comp...I'm gone do my best in the remaining stages.