MSI GTX 970 mini review with rFactor 2 performance

Discussion in 'General Discussion' started by DrR1pper, Sep 25, 2014.

  1. DrR1pper

    DrR1pper Registered

    Joined:
    Apr 29, 2012
    Messages:
    3,294
    Likes Received:
    35
    Corrrected performance graph with the GTX 970 in PCI-e 3.0 x16 mode should yield the following rf2 results:

    [​IMG]

    [​IMG]

    [​IMG]
     
  2. Spinelli

    Spinelli Banned

    Joined:
    Jan 28, 2012
    Messages:
    5,290
    Likes Received:
    31
    I cant tell the differences between the 3 graphs. They all have the same cards, at the same clocks with the same games, just different %, how does that work?
     
  3. Prodigy

    Prodigy Registered

    Joined:
    Nov 2, 2012
    Messages:
    949
    Likes Received:
    64
    The one card with 100% is the base line and other with lower or higher percetange are showing how much they are performing better or worse comparing to the base card.

    The FPS counter would maybe be more interesting and accurate to tell how big the difference is. Now we see 100% but we don't know from how much. Only if we read through thread to see in posts.
     
    Last edited by a moderator: Oct 1, 2014
  4. Spinelli

    Spinelli Banned

    Joined:
    Jan 28, 2012
    Messages:
    5,290
    Likes Received:
    31
    Ya but what's with the 3 graphs? Can't we see the relative differences/percentages from one GPU to the next just by looking at the first chart? Are the second and third charts the same as the first but just using a different GPU for the baseline 100% mark?...
     
  5. DrR1pper

    DrR1pper Registered

    Joined:
    Apr 29, 2012
    Messages:
    3,294
    Likes Received:
    35
    correct.

    Yeah, that was the initial intent but then i saw how different the fps were to each game/benchmark that it made it too variable and useless for comparison of each card and clock speeds as a whole. Also a pain to try and normalise the graphs with non equal fps values for each so i can do the same 100% for each game and benchmark in a specific card and clock.
     
  6. DrR1pper

    DrR1pper Registered

    Joined:
    Apr 29, 2012
    Messages:
    3,294
    Likes Received:
    35
    Well yeah but this way you can pick one you want as the baseline to compare the other 3 to and it's easier and quicker (or at least i find) to see values of relative performance differences (> or < than 100%) for a particular card.

    I'm going to repeat these graphs anyway when i get my new system which should be friday. The other game values seemed to match what others were getting on the internet but it will be interesting to double check over anyway, especially since it's now only Metro that don't share the same performance gains vs a gtx 770 as seen in rf2 and the rest.
     
  7. Spinelli

    Spinelli Banned

    Joined:
    Jan 28, 2012
    Messages:
    5,290
    Likes Received:
    31
    Got it DrR, wasn't sure about the differences between the 3 graphs, but you've clarified that for me. Thanks :)
     
  8. Descoat

    Descoat Registered

    Joined:
    Apr 17, 2013
    Messages:
    86
    Likes Received:
    0
    Does this new GPU with new drivers improves the anti aliasing filter?

    By the way, Does any have bought a 980 gtx and try with triple screen? Just to compare with this 970 gtx.

    Thanks for your review
     
  9. DrR1pper

    DrR1pper Registered

    Joined:
    Apr 29, 2012
    Messages:
    3,294
    Likes Received:
    35
    I did actually test this for myself with 0xMSAA vs 4xMSAA in the nvidia control panel with the rf2 live benchmark and found between my 770 and 970, there was not difference in the percentage impact between the two cards. However, i will to test this with a pci-e 3.0 motherboard to tell if this is really true or not. I can test/compare after tomorrow when the new parts arrive.
     
  10. Descoat

    Descoat Registered

    Joined:
    Apr 17, 2013
    Messages:
    86
    Likes Received:
    0
    Thanks bro! Specially I want to know if the quality of the anti aliasing is better in games such as GSC Extreme or rfactor2. I´ve got an AMD now and multisampling offers a good quality.
     
  11. DrR1pper

    DrR1pper Registered

    Joined:
    Apr 29, 2012
    Messages:
    3,294
    Likes Received:
    35
    Gotch you but i actually use 2xMSAA with 2xSGSSAA (Spare Grid Super Sampling Anti Aliasing) when in playing rf2 myself. The visual results are pretty amazing for the performance and i could even run with the much visually better 4xMSAA with 4xSGSSAA but it was a little low in fps for my liking at times so i stuck with very smooth 2x. With the 970 however, i suspect 4x is going to be no problem whatso (considering based on the data the 970 overlocked will perform 50% better than my overclocked 770.

    I'll report back with a battery of tests over the weekend.
     
    Last edited by a moderator: Oct 2, 2014
  12. Spinelli

    Spinelli Banned

    Joined:
    Jan 28, 2012
    Messages:
    5,290
    Likes Received:
    31
    DrR, haven't neard from you in this thread or the other 2 threads (live benchmark, and pci-e threads). Are you still working on all the tests?...
     
  13. SCampbell

    SCampbell Registered

    Joined:
    Feb 8, 2012
    Messages:
    64
    Likes Received:
    1
    That's pretty crazy now with the new 970 you can get those frames. I now feel like the situation with FPS is becoming less of an issue with rF2 because it seems now that the graphics cards have now eclipsed what the game engine is capable of delivering. 2 years later and the FPS is what I recall the FPS was for rF1 when I first bought it. I figured that the technology needed to catch up a bit with the software.

    For reference, I am still running a GTX 680 and its been very good, about 70-100FPS normally all the time with settings pretty high up (full detail, but lower on the anti-aliasing and a few other settings). I can only run 1680x1050 because my monitor is old, but still looks great.
     
  14. stenio

    stenio Registered

    Joined:
    Nov 13, 2014
    Messages:
    2
    Likes Received:
    0
    No i dont see a problem, what you see is what the actual performance is compared to other cards truthfully when both are pushed. and what misled you and multitude of others is the online reviews that only has the New maxwells overclocked in graphs comparison. i have 3 brothers all gamers and we got just about every card, we have kept score of every card that has come thru and ur scores is correct. i have 970 sli and did many single tests. my bro got 780's sli and i use to have 780's aswell, and couple years ago i had 770. the reviews only overclock the card under review and Maxwells are extremely high clocked right outt of the box for example reference 780 is clocked 30 % lower vs reference 970. but when both cards are pushed the 780 beats 970 in every benchmark/game. my bros 780 is a average chip and at 1251 mhz core it beats my 1600 mhz 970 in everything. maxwells are very cool and efficient! but they have very little raw horse power. 970 got 21,632 cudas.. thats almost as much as your 770 ... 780 got 27,648. I get these cudas from the number of cores (x) the number of clusters (Sms). but because they need little wattage, thus produce less heat and so invidia said well hey! they run so cool we ought to clock them 30% higher than 780's and still be safe from RMA's because at 30 % higher clock the 970 runs cooler than 780 still. but make no mistake even a 780 on air will beat 970 when both are pushed. now if u add a little water cooling 780 destroys the 970. keep in mind 780's win even with .038 less voltage and 30% lower core clock and Much less Vram clock, and this is because 780 got significant more amount of cores. 780 and 780 ti is still a overclockers choice.

    now show you whhat u see in reference clock comparison.. 780 - 27.8 % more cores than 970. but 970 clocked 30% higher than 780 because of safe temps and higher voltage allowed by those temps. even tho if both were same clock the 780 has much more raw power its retrained in reference clocks a ton by invidia.
    970 30 % higher freq minus -27.8 % less performance = 3 % fps still in reference clocks than 780. but if you push both it loses and if u have good cooling is no comparison.
     
  15. stenio

    stenio Registered

    Joined:
    Nov 13, 2014
    Messages:
    2
    Likes Received:
    0
    No its not, and it sits right behind the 780 and in front of 680. when the cards are overclocked and pushed it loses in every benchmark score/ game to the 780. when both are pushed. https://www.youtube.com/watch?v=MK6AYFcSxCs now every 970 is overclocked around 30 % higher than 780's at stock clocks but they 27.8 % less cores. because maxwell runs so efficiently and cool nvidia said oh heck ! lets give them high clocks right off the start!. but if gona overclock them and push them both beyond what nvidia recommended the 780 wins man. i have 3 brothers and we got all the cards. the reviews only overclock the card that its reviewing and to begin with 970 is already clocked much higher so its horrribly mis leading. thats why u got this crazy hype out thereright now. but if people keep their old card around and compare like this man did here.. they will see its nothing like they were led to believe.
    also keep in midn if u want to speak just performance and forget about temps. 780 destroys 970, if both cards are watercooled there isnt even a comparison. 970 even is allowed 0.038 more voltage than 780 by invidia because of its temps. some water cooling on both cards and on same voltage has has 970 getting smashed! and on air + stock voltage has 780 wining wtill by avg of 8 - 12 %
     
  16. DrR1pper

    DrR1pper Registered

    Joined:
    Apr 29, 2012
    Messages:
    3,294
    Likes Received:
    35
    Stenio, whilst very interesting there is but one possible problem that i can see with this video (https://www.youtube.com/watch?v=MK6AYFcSxCs) review comparing the different cars at the max overclock.

    It does not tell us the overclock they could achieve with their specific cards they received and they do not report back what their overclocks were for each card. This can really skew the results to make the 780 appear better overall (equally it could correct), but without knowing the overclock values they were able to achieve and run with, i can't trust this comparison completely.

    For example, the GTX 970 they have used is an Asus GTX 970 Strix which based on the reviews i can find online have managed to overclock the card to around no higher than 1450-1500mhz. Where as we know that the better 970 cards such as the gigabyte gtx 970 G1 can hit 1550-1600mhz. That's an additional 7% overclock/performance.

    Unfortunately, we don't even know which brand of gtx 780 was used in their test as well as how much they overclocked it by so it's really not fair to make any real conclusions from the comparison at all.

    But, if we assumed for a moment that we knew it was the best gtx 780 clocked to the max (which we can't and do not know hence this is just hypothetical), then accounting for the asus gtx 970 strix lower overclock ability vs the best gtx 970 brands, this would be the corrected results (again only hypothetical and if the assumption made were true and also making the assumption that the percentage increase in overclock matches perfectly with the percentage performance game in performance):

    Crysis 3 - Very High, 1080p, MSAA8x

    GTX 780 (assumed highest overclock possible) - 46.1 (100%)
    GTX 970 Strix - 40.0 (87%)
    GTX 970 (estimated gigabyte g1) - 42.8 (93%)

    Crysis 3 - High, 4k, AA off

    GTX 780 (assumed highest overclock possible) - 36.5 (100%)
    GTX 970 Strix - 31.4 (86%)
    GTX 970 (estimated gigabyte g1) - 33.6 (92%)

    Bio Shock Inf. - Ultra, 1080p, Alt. Post. proc

    GTX 780 (assumed highest overclock possible) - 144.9 (100%)
    GTX 970 Strix - 139.5 (96%)
    GTX 970 (estimated gigabyte g1) - 149.3 (103%)

    Bio Shock Inf. - Ultra, 4k, Alt. Post. proc

    GTX 780 (assumed highest overclock possible) - 46.7 (100%)
    GTX 970 Strix - 41.6 (89%)
    GTX 970 (estimated gigabyte g1) - 44.5 (95%)

    Far Cry 3 - Ultra, 1080p, MSAAx8

    GTX 780 (assumed highest overclock possible) - 54.6 (100%)
    GTX 970 Strix - 48.6 (89%)
    GTX 970 (estimated gigabyte g1) - 52.0 (95%)

    Far Cry 3 - Ultra, 4k, MSAAx2

    GTX 780 (assumed highest overclock possible) - 28.4 (100%)
    GTX 970 Strix - 25.9 (91%)
    GTX 970 (estimated gigabyte g1) - 27.7 (98%)

    Watch_dogs - Ultra, 1080p, MSAAx4

    GTX 780 (assumed highest overclock possible) - 60.6 (100%)
    GTX 970 Strix - 57.5 (95%)
    GTX 970 (estimated gigabyte g1) - 61.5 (101%)

    Watch_dogs - High, 4k, AA off

    GTX 780 (assumed highest overclock possible) - 33.6 (100%)
    GTX 970 Strix - 31.8 (95%)
    GTX 970 (estimated gigabyte g1) - 34.0 (101%)

    So based on this, even with corrected results and considering the best case scenario for the gtx 970 to appear better, you appear to be correct in stating that the gtx 780 with max overclock possible in games on average beats the gtx 970 (but only by a small margin but still, wins a win so kudos to you for raising this point and thank you).

    The results in the video make the GTX 980 really shine out as a serious winner. I just checked prices and it's cheaper than the GTX 780 Ti's and performance a good bit better. In the UK, i find the GTX 970's are still a fair bit cheaper than the GTX 780's (especially when comparing the best overclocking cards from both).
     
  17. Jim Par

    Jim Par Registered

    Joined:
    Jan 9, 2012
    Messages:
    11
    Likes Received:
    1
    Hello guys, I have a question regarding the GTX970's performance on triples 23". I currently use a GTX660 MSI 2GB. I use low-med graphics as I dont care much about them, I just want good and stable performance over the triples with multiview enabled.
    That's where I want to focus with buying a GTX970 4GB, will it make it better for me on triples and multiview enabled (multiview is important as it gives a good hit on fps) and how much better, regarding fps with 15-20 vehicles on track and at least 12 of them visible. Thank you for your time, Jim.
     
  18. DrR1pper

    DrR1pper Registered

    Joined:
    Apr 29, 2012
    Messages:
    3,294
    Likes Received:
    35
    "triples 23"" doesn't tell us what resolution they're at. :)
     
    Last edited by a moderator: Dec 29, 2014
  19. Jim Par

    Jim Par Registered

    Joined:
    Jan 9, 2012
    Messages:
    11
    Likes Received:
    1
    lol you are right. The monitors are at 1920x1080. Currently I'm using 3840x720 in rF2 for even better performance. It can go up to 5820x1080 but I don't need that atm cause I lose fps.
     
    Last edited by a moderator: Dec 28, 2014
  20. DrR1pper

    DrR1pper Registered

    Joined:
    Apr 29, 2012
    Messages:
    3,294
    Likes Received:
    35
    I'm just speculating here but should be a significant improvement over your 660 performance, especially with such a large res.
     

Share This Page