1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

Other Geekbench scores

Discussion in 'Hardware' started by Bloody_Pete, 14 Dec 2011.

  1. GeorgeStorm

    GeorgeStorm Aggressive PC Builder

    Joined:
    16 Dec 2008
    Posts:
    6,597
    Likes Received:
    383
    Nothing stopping you if you wanted to :)
    I got back into overclocking last year and got a skylake setup for it, my main pc is an ivy (maybe) i5 locked, with 1 stick of ddr3 hah #priorities
     
  2. true_gamer

    true_gamer Well-Known Member

    Joined:
    4 May 2009
    Posts:
    6,622
    Likes Received:
    1,049
    It's one of those benchmarks that will take advantage of all threads, clock speeds, memory speeds, and timings. So someone could come along with a similar setup with higher clocks and beat your benchmark.

    Back in the day the 3960X was the best chip, and I managed to get hold of a golden sample that clocked to 5.4GHz on all 6 cores with HT under water. Which was pretty impressive. :)

    True, though I wouldn't know where to start as I've been a bit out of touch with the PC market, so I'm only just trying to get to grips with Intel's and AMD's new offerings, which are so confusing... As when it comes to benching in 3Dmarks, is more core/threads better than clock speed when you get to the realms of 8 cores plus?...
     
  3. GeorgeStorm

    GeorgeStorm Aggressive PC Builder

    Joined:
    16 Dec 2008
    Posts:
    6,597
    Likes Received:
    383
    Let me know if you want to get into overclocking more, there's the r/oc discord that I've recently joined which isn't bad. Don't have to use new stuff, something like www.hwbot.org lets you compete against others using the same hw regardless of age.

    The newer 3dmarks like cores due to the physics component of the test, older/legacy 3dmarks (01/3/5/6) don't need anything more than 4 threads and like mhz, other than 06 which scales to 6ish.
     
  4. Vault-Tec

    Vault-Tec Green Plastic Watering Can

    Joined:
    30 Aug 2015
    Posts:
    11,769
    Likes Received:
    1,838
    It is a combination of both that improves scores. Scaling is literally linear, the same as Cinebench*. The Physics score will be the score to watch for the CPU, and of course you will get a better combined score having a better CPU too.

    * I spent two days doing tons of research on both and if you disable two cores then the loss in performance is predictable and linear. IE - let's say you had a 10 core CPU that scores 10,000 in a perfectly linear benchmark when you disable 2 cores that score will drop to 8000. It is also linear on the clock speed, too, so if you halved the clock speed you would halve the score, too.

    Geekbench does not do that nor work like that and that is why I wouldn't even waste my time on it (because I have in the past and the results make no sense). Just don't use 3Dmark 11 as it only uses six cores properly. Firestrike and Timespy don't, they will take whatever you have.
     
  5. true_gamer

    true_gamer Well-Known Member

    Joined:
    4 May 2009
    Posts:
    6,622
    Likes Received:
    1,049
    I'm sure I did some benchmarks on Hwbot several yrs ago. So I might have to look into it again especially when I upgrade in the near future. :)

    Yeah, I gathered that, but looking at the hall of fame, it looks like everyone is using that 7980XE, which is more of a workstation CPU than for general use like gaming, benchmarking, and some video editing.
    I haven't ran 3Dmarks 06 for a longtime, loved watching that back in the day of owning a ATI HD2900XT. (Time for a bench.) :D

    I completely understand that process of how that all works. - But in real world, to test a system as a whole, rather than individual components, which is how we use our computers, then Cinebench is fine for just a CPU benchmark on it's own. Geekbench scales with better faster memory, so you are getting results from not just the CPU performance, but from the memory controller, and RAM speed. In which Ryzen scales better with faster memory. So as a whole, people can see how well the systems can scale when comparing others results.

    But as we've mentioned, we can setup a few threads with different benchmarks to have scores to compare.

    My main goal for when I upgrade my Board, CPU and RAM, is to get the best for 4K, but most benchmarks I see only show CPU performance at 1080p which then becomes the CPU as the bottleneck. 4k scaling - AMD and Intel seem to be on par from what I've seen, but for me to chose either team, I want to make sure from all the results we gather, which is the best for 4k gaming, video editing, and bragging rights. :D
     
  6. Vault-Tec

    Vault-Tec Green Plastic Watering Can

    Joined:
    30 Aug 2015
    Posts:
    11,769
    Likes Received:
    1,838
    4k gaming you could shove a potato in the socket. CPU is almost meaningless. Few years back I had a 3970x running 5ghz with a pair of Titan Black. Board fets died, CPU was throttling to 1.1ghz I lost about 4 FPS.

    The CPU I am running now (Broadwell E) clocks to 2.9ghz on over 4 cores. I get a solid 70 FPS out of Fallout 4 and any dips are down to the GPU not coping fully with large outside areas (in the city for example).

    It will be a very long time before we see a GPU powerful enough to make the onus lean back onto the CPU at 4k.
     
    true_gamer likes this.
  7. GeorgeStorm

    GeorgeStorm Aggressive PC Builder

    Joined:
    16 Dec 2008
    Posts:
    6,597
    Likes Received:
    383
    Go for it :)

    Well yeah if you want the top scores you need the biggest/most expensive hardware most of the time :p

    7980xe has enough cores to do well in the physics, but can also be clocked high enough to give the gpu enough power for the game tests.
     
    true_gamer likes this.

Share This Page