Jump to content
Search In
  • More options...
Find results that contain...
Find results in...


  • Content Count

  • Joined

  • Last visited

Posts posted by rockmassif

  1. It's not surprising that Maxwell struggles with Valley/Heaven, and I doubt any driver updates will improve this. Maxwell gains most of its strength from better color compression and faster clocks, and Unigine benchmarks have a lot of movement and color fluctuation - partly due to day/night cycling and extreme tessellation, not to mention weird camera panning and movement. 3DMark Firestrike on the other hand is a poster child for proving color compression techniques: there's not a plethora of different colors, not as much bizarre camera movement (there is some lol), and in one scene only a small percentage of the pixels on screen are changing. 


    I'm no expert on this stuff and the thing I'm gonna say will sound really stupid but 980 really feels like a dumbed down version of 780Ti with 4GB VRAM. Doesn't really feel powerful and unique.

  2. The i5 he has now is fine for gaming, just benchmarks reward the higher end Intel CPU's alot more then they reward AMD/Intel i5 CPU's


    We're talking about benchmarking tho, not gaming  :)

    3D Mark benefits highly from both high clock speeds, higher end processors. That's what I said in my earlier post. I'm getting close to 10K physics score with an i5. Thus allowing me to score 1K more than him with an R9 280x.





    Oh don't get me wrong the 5960X is state of the art... but the 5930K supports already the 40 native PCIe lanes. This and the OC potential won me over to the 6 core CPU. If I'd get my hands on a good 5960X (at least 4.4GHz on all cores) sample I wouldn't hesitate to sell my CPU right away.


    I'm kinda obsessed with the number of cores, I think they should always be a power of 2 :P So there is no way I'm getting a 6-core CPU. And 5960X is literally just too expensive for me. :D

  3. yeah 3dmark really rewards high-end processors. give it a few months and I might be trolled into buying a 5820k


    Plus higher clock speeds.


    If that was a 4690K running around 4.6 you would still be fine imo.

  4. The specs are like this;

    • Fan Size: 140mm
    • Bearing Type: HDB (Hydro Dynamic Bearing)
    • RPM: 600 -1300 RPM
    • CFM: 89.55 CFM
    • Max static air pressure: 2.76 mm H2O
    • Noise Level: 22.2 dB(A)


    And looking at these I thought it would be a good radiator fan so I decided to change my stock H110 fans with these but I'm pretty disappointed. Don't know if I did something wrong(pretty unlikely that I did) but these fans are pretty loud even at 1000RPM compared to stock fans. And the cooling performance is not that great I think, getting 40-42C while typing these and watching stream(4670K@1.28v).


    Is there anyone tried to use these as radiator fans? What was your experience?

  5. Furmark is definitely a no-no. It puts immense amount of pressure to the card that no real world application will even come close. I never tested any of my cards with Furmark. It's better to just let Valley run 30 min loop without artifacts and other things. That's how I test stability and temps.

  6. AMD Advantages: Provides 3-4GB VRAM(which is something I really need) for their high end cards, cheaper, AMD fanboys

    NVIDIA Advantages: Got more partners in the industry, better driver support, much better utility compared to shitty CCC, NVIDIA fanboys, Overclocks much better


    For me the story is just like this.


    I ain't gonna like, if 780Ti had 4GB VRAM, I wouldn't even consider with going 290X.

  7. Why don't you just Overclock via bios or IETU? I have the same mobo I honestly didn't even install that utility because it's complete shit. 


    Plus you should not run Prime95 when the voltage is adaptive, make it fixed first. And getting 55Celsius on Prime95 with only 212 Evo doesn't seem realistic.

  8. I think if both work at the same clock speeds, 780Ti should beat 980 in some games and especially on heaven/valley or 3D Mark(if it's not biased towards new architectures somehow).


    I remember Kingpin making 780Ti work at 1933MHz Core and the 3D Mark score was around 8400.(he didn't have 5960X on that test setup)

    Now he has 980 working at 2137MHz(204Mhz huge difference imo), and the score is around 9600.


    Of course it's safe to assume that 780Ti basically don't have the efficiency to go above 2100 like 980, but then again that's just for pure testing. A decent 780Ti(Kingpin-Matrix) card can get easily 1400-1500 Core like 980 on air.


    http://www.3dmark.com/fs/3155351 5960X@5.1GHz, 16GB 2133MHz Memory, 980@1868,2053 Score: 8391

    http://www.3dmark.com/fs/3160094 5960X@5.3GHz, 16GB 2133MHz Memory, 780Ti@1756, 2100 Score: 8733


    Closes test setups and clock speeds I could find.

  9. funny this topic comes up. I just got into a long argument with an idiot on youtube about how there really is no difference once they are overclocked, he wanted to say it's a big difference and that the 980 blows the 780ti away and even the 970 is faster in most games. lol The 970 isn't even faster than a 780 in basically all games when the cards are overclocked. Sounded like someone who moved from a 780/780ti to a 980 and is desperately trying to convince themselves it was a good idea.


    Currently that might be true but with MFAA that story might change a little.


    If I had 780Ti I wouldn't actually bother with upgrading at all but for new specs it's safe to go with 980. You can't rule out the fact that it has more vram, low power consumption and new features like MFAA.


    But I have to add, none of the current 980 models interest me at all. If we see 980 Kingpin or 980 Matrix Platinum only then I would consider buying one.

  10. Not really sure if you're asking for Dual Titan Z or Dual Titan.


    3x980 will beat Titan Z of course but has less VRAM per GPU. If you're fine with 4 then that's OK.

    It's really not advisable to pay that much for Titan Z at this stage unless you have a very specific reason to buy, I have no idea why the prices didn't drop yet.