Page 8 of 9 FirstFirst ...
6
7
8
9
LastLast
  1. #141
    Quote Originally Posted by n0cturnal View Post
    You are still missing the point that the 8350 would have to overclock to around 7GHz to outscale the 3570k by that much.
    Which would take ln2 or dice to do it. Then you could take the 3570k to 6ghz or so. At that point who gives a shit?

  2. #142
    Brewmaster Majesticii's Avatar
    Join Date
    Feb 2010
    Location
    Netherlands
    Posts
    1,413
    Quote Originally Posted by glo View Post
    What else would I be referring to?
    There's no math in there. Honestly, the amount of difficulty you're having with this is a little depressing.
    I'll draw you a excel plot.

  3. #143
    Quote Originally Posted by Majesticii View Post
    There's no math in there.
    Yeah, scaling numbers and their percentages against a static constant has nothing to do with math. My bad.
    i7-4770k - GTX 780 Ti - 16GB DDR3 Ripjaws - (2) HyperX 120s / Vertex 3 120
    ASRock Extreme3 - Sennheiser Momentums - Xonar DG - EVGA Supernova 650G - Corsair H80i

    build pics

  4. #144
    I am Murloc! Cyanotical's Avatar
    Join Date
    Feb 2011
    Location
    Colorado
    Posts
    5,255
    Quote Originally Posted by Majesticii View Post
    Cyanotical can maybe help us by proving the 3960x is enough to max 99% a 7970 on FC3. Pretty sure he even maxes out the quad SLI on that game.
    if you want to send me the 7970, sure

    i7-4790K | Z97 Class. | 8GB DDR3-2133 | GTX-690 Quad SLI | RAIDR | 512GB Samsung 830 | AX1200 | RV05
    Dell U2711 | Ducky Shine3 YoS | Steelseries Sensei | Xonar Essence One | KRK RP8G2s

  5. #145
    Brewmaster Majesticii's Avatar
    Join Date
    Feb 2010
    Location
    Netherlands
    Posts
    1,413
    Quote Originally Posted by glo View Post
    Yeah, scaling numbers and their percentages against a static constant has nothing to do with math. My bad.
    Hardly any math then, and if there was, it was beside the point.

    Quote Originally Posted by Cyanotical View Post
    if you want to send me the 7970, sure
    proving a 690 gets 99% will also be sufficient.

  6. #146
    Not exactly relevant at this point, but I thought I'd format Majesticii's numbers into a more readable format. Not sure if I can actually be arsed to extend this to the other benchmarks, admittedly.



    Note: Numbers chosen were lowest of samples for minimum, median for average and highest for maximum.

  7. #147
    Brewmaster Majesticii's Avatar
    Join Date
    Feb 2010
    Location
    Netherlands
    Posts
    1,413
    Fine, another attempt. Assume 70-75fps is the absolute max avg the 7970ghz can do, which i'm pretty convinced it is.
    So 75fps represents 99% load, and 0 fps means 0% load.



    The AMD starts at a much lower GPU load, represented by 44fps, in regard to the intel, represented by 65fps.
    Hence increasing the frequence starting up from 2.5ghz, the AMD still has a performance gain at a seaminly higher scaling, whereas the Intel almost saturates the 7970 @ 3-3.5ghz and hardly scales. However, this means it has a much higher efficieny per-frequency, and thus leaving more headroom for the future. Because it takes 4.5ghz on the AMD just to max out a single 7970, when you can do the same on a stock 3570K.
    Any newer, or faster card, and the AMD will start to bottleneck.
    Last edited by Majesticii; 2013-01-27 at 02:04 AM.

  8. #148
    Quote Originally Posted by Majesticii View Post
    Assume 75fps is the absolute max avg the 7970ghz can do, which i'm pretty convinced it is.
    Are you missing the point of your entire theory hinging on that? Your graph means nothing when it's based on an apparent FPS wall that you made up.
    i7-4770k - GTX 780 Ti - 16GB DDR3 Ripjaws - (2) HyperX 120s / Vertex 3 120
    ASRock Extreme3 - Sennheiser Momentums - Xonar DG - EVGA Supernova 650G - Corsair H80i

    build pics

  9. #149
    Quote Originally Posted by glo View Post
    I'm not being baseless, there's an entire linear progression from 2.5Ghz all the way up to 4.5Ghz. I'm suggesting that factoring in every other benchmark available for the processor not showing any sign of a scaling wall beyond 4.5Ghz -should- translate into gaming gains.

    You're sitting there saying just because they didn't test that high that any informed guess is immediately invalid.
    There's a reason extrapolation is not allowed in 'real science'. Unless you can prove that the FX-8350 behaves linearly beyond 4.5 Ghz, you cannot extrapolate.

    Pointing to the fact that the FX-8350 is linear from 2.5-4.5 Ghz is not real proof that the FX-8350 will behave linearly beyond 4.5 Ghz for this case.

    Like I've said before, using other benchmarks doesn't help since the 'wall', as Majesticii put it, moves around depending on game, settings and GPU used.
    Last edited by yurano; 2013-01-27 at 02:10 AM.

  10. #150
    Brewmaster Majesticii's Avatar
    Join Date
    Feb 2010
    Location
    Netherlands
    Posts
    1,413
    Quote Originally Posted by glo View Post
    Are you missing the point of your entire theory hinging on that? Your graph means nothing when it's based on an apparent FPS wall that you made up.
    Oh for ..., are you being deliberatly thick here. I said i didn't make it up, because i can say in relation to my own experience, that it is.
    With my own i5-760, i get 99% load on a overclocked 670. Which relates to a 680 stock (in the test) which is on par with the 7970 on that test.
    Meaning you don't get extra fps after that threshold.

    YES, they are assumptions. But they're valid assumptions, not just simply made up.
    Last edited by Majesticii; 2013-01-27 at 02:10 AM.

  11. #151
    Quote Originally Posted by yurano View Post
    There's a reason extrapolation is not allowed in 'real science'. Unless you can prove that the FX-8350 behaves linearly beyond 4.5 Ghz, you cannot extrapolate.

    Pointing to the fact that the FX-8350 is linear from 2.5-4.5 Ghz is not real proof that the FX-8350 will behave linearly beyond 4.5 Ghz in this situation.
    Which is why I said it "should". I have previous testing history from 2.5-4.5Ghz and an entire internet's worth of 8350 benchmarks to back up my "should". You have nothing by scientific definitions you learned in school saying that my "should" is invalid.
    i7-4770k - GTX 780 Ti - 16GB DDR3 Ripjaws - (2) HyperX 120s / Vertex 3 120
    ASRock Extreme3 - Sennheiser Momentums - Xonar DG - EVGA Supernova 650G - Corsair H80i

    build pics

  12. #152
    Continuing the trend of illustrated benchmark, this time in Metro 2033...



    Still same methodology, which showed its flaw when one of the runs had a stupid high max fps lol.

  13. #153
    Brewmaster Majesticii's Avatar
    Join Date
    Feb 2010
    Location
    Netherlands
    Posts
    1,413
    Quote Originally Posted by Drunkenvalley View Post
    Continuing the trend of illustrated benchmark, this time in Metro 2033...

    Still same methodology, which showed its flaw when one of the runs had a stupid high max fps lol.
    Wierd, i didn't alter anything. Just copy-paste.
    And thanks, i will edit them into the post later when youre done.

  14. #154
    Quote Originally Posted by glo View Post
    Which is why I said it "should". I have previous testing history from 2.5-4.5Ghz and an entire internet's worth of 8350 benchmarks to back up my "should". You have nothing by scientific definitions you learned in school saying that my "should" is invalid.
    'Should' is not enough to be a concrete conclusion. If you want to prove that the FX-8350 scales linearly to 5.0 Ghz, you must replicate TechSpot's test and include the 5.0 data point, and maybe a 5.5 data point as well.
    Last edited by yurano; 2013-01-27 at 02:17 AM.

  15. #155
    The Lightbringer Tekkommo's Avatar
    Join Date
    Nov 2011
    Location
    Scotland
    Posts
    3,734
    Because people are cheap.

  16. #156
    Brewmaster Majesticii's Avatar
    Join Date
    Feb 2010
    Location
    Netherlands
    Posts
    1,413
    Quote Originally Posted by Tekkommo View Post
    Because people are cheap.
    I guess were done here.

  17. #157
    Quote Originally Posted by Majesticii View Post
    Oh for ..., are you being deliberatly thick here. I said i didn't make it up, because i can say in relation to my own experience, that it is.
    With my own i5-760, i get 99% load on a overclocked 670. Which relates to a 680 stock (in the test) which is on par with the 7970 on that test.
    Meaning you don't get extra fps after that threshold.
    So you're claiming that because your GPU is at 99% that a processor upgrade would have no impact. Are you even listening to yourself? o_O
    i7-4770k - GTX 780 Ti - 16GB DDR3 Ripjaws - (2) HyperX 120s / Vertex 3 120
    ASRock Extreme3 - Sennheiser Momentums - Xonar DG - EVGA Supernova 650G - Corsair H80i

    build pics

  18. #158
    Brewmaster Majesticii's Avatar
    Join Date
    Feb 2010
    Location
    Netherlands
    Posts
    1,413
    Quote Originally Posted by glo View Post
    So you're claiming that because your GPU is at 99% that a processor upgrade would have no impact. Are you even listening to yourself? o_O
    Yes because unless you have a graphics card that goes over 99%, there is no framerate to be gained.
    Are you even still comprehending this, I sincerely doubt it at this point.

    They only scenario where a CPU upgrade would affect my performance in FC3, is if i couldn't reach 99%, or had big GPU load drops. Which i don't. It stays on 99%.
    Last edited by Majesticii; 2013-01-27 at 02:22 AM.

  19. #159
    Yeaaaaaaaah, I think I'll have to figure out a better way to represent the runs you had lol. Some of the outcomes are a bit strange. Maybe the median of the three values next time...


  20. #160
    Brewmaster Majesticii's Avatar
    Join Date
    Feb 2010
    Location
    Netherlands
    Posts
    1,413
    Quote Originally Posted by Drunkenvalley View Post
    Yeaaaaaaaah, I think I'll have to figure out a better way to represent the runs you had lol. Some of the outcomes are a bit strange. Maybe the median of the three values next time...

    My guessing some values are just incorrect and should be disregarded. Therefor you run the benchmark 3 times, to spot anomalies.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •