EVGA

Hot!What does, "equal or greater performance" really mean?

Author
bavor
New Member
  • Total Posts : 32
  • Reward points : 0
  • Joined: 2008/02/20 21:24:43
  • Status: offline
  • Ribbons : 0
2019/11/05 12:36:59 (permalink)
My EVGA GTX 1080 ti FTW3 Hybrid had problems recently.  I contacted support and received a RMA.  I assume that no replacement GTX 1080 ti FTW3 Hybrids were available because I received a RTX 2080 XC Hybrid in it place.  What I've noticed so far, is that at both stock speeds and at its maximum stable memory and GPU overclock, the 2080 Hybrid is giving me lower scores in some 4K resolution graphics benchmarks and lower FPS in games at 4K resolution.  Also, possibly due to the lower memory capacity and lower memory bandwidth, encoding video in adobe premiere pro CC/adobe media encoder with CUDA acceleration is taking longer compared to the same video projects when I had a properly functioning GTX 1080 ti FTW3 Hybrid.
 
I noticed the RTX 2080 gives equal or better performance in 1080p gaming and benchmarks, but doesn't consistently give equal or better performance in 4K gaming and 4K game benchmarks.  For example, with my RTX 2080, the two FireStrike Extreme benchmark scores I've been able to obtain was 12,878 with a graphics score of 14,307 and 12,920 with a graphics score of 14,282.  My GTX 1080 ti scored 13,701 with a graphics score of 15,500.  My RTX 2080 is able to have a higher stable overclock on both the GPU and memory than my GTX 1080 ti was capable of.  Yet the RTX 2080's benchmark scores at 4K are lower at both stock clock speed and overclocked.  The the results of in the game benchmarks that I tried at 4K at lower with the RTX 2080.  It doesn't appear that there is anything wrong with my RTX 2080, so is the GTX 1080 ti just that much better when overclocked?
 
RTX 2080 XC Hybrid


 
EVGA GTX 1080 ti FTW3 Hybrid

I tried with three different drivers and couldn't get better results for the RTX 2080.
 
I also looked up the TFLOP performance of the cards.  I know TFLOPS doesn't equal FPS and is just a rough estimate of the GPU performance in certain situations.  This is the info I found from Nvidia for the founders edition/reference design cards:
GTX 1080 ti 11.3 TFLOPS
RTX 2080 Super 11.1 TFLOPS
RTX 2080 TFLOPS 10.1 TFLOPS
Based on that, the GTX 1080 ti looks "faster" than the RTX 2080.
 
What benchmarks or games does EVGA use to determine equal performance?  Is it just that the games I have do worse with RTX cards at 4K resolution for some reason?  Are the cards only compared at 1080p?  Did I get some magical golden silicon lottery jackpot GTX 1080 ti GPU on my FTW3 Hybrid?
 
I had another question about the warranty.  I purchased the extended warranty and the advanced RMA for my GTX 1080 ti.  The warranty from the 1080 ti carried over to the RTX 2080, but the advanced RMA didn't.  Why is that?
#1

19 Replies Related Threads

    Sajin
    EVGA Forum Moderator
    • Total Posts : 39112
    • Reward points : 0
    • Joined: 2010/06/07 21:11:51
    • Location: Texas, USA.
    • Status: offline
    • Ribbons : 195
    Re: What does, "equal or greater performance" really mean? 2019/11/05 13:37:41 (permalink)
    Sounds like you need to tell evga you want a 1080 ti instead of the 2080. As far as the advanced rma goes read the terms... https://www.evga.com/support/ear/

    Want to save 5 to 10% on your next EVGA purchase? Just click on the associates banner to save, or enter the associates code at checkout on your next purchase. If you choose to use my code I want to personally say "Thank You" for using it.
     

     
    #2
    Cool GTX
    EVGA Forum Moderator
    • Total Posts : 18701
    • Reward points : 0
    • Joined: 2010/12/12 14:22:25
    • Location: Folding for the Greater Good
    • Status: offline
    • Ribbons : 114
    Re: What does, "equal or greater performance" really mean? 2019/11/05 13:50:26 (permalink)
    bavor
     
    I had another question about the warranty.  I purchased the extended warranty and the advanced RMA for my GTX 1080 ti.  The warranty from the 1080 ti carried over to the RTX 2080, but the advanced RMA didn't.  Why is that?




     
    1 use & Done.  However - You have an Option to purchase a New A-RMA within 30 days for your RMA card's registration

    RTX Project EVGA X99 FTWK-5930K,EVGA2080Ti EVGA-1200P2  Nibbler EVGA X99 3-GPU Water Loop 1600P2  AIO Folding  1080Ti FTW3 Hybrid, 1200P
    Thank You for Your Support


    #3
    EVGATech_DanielM
    EVGA Tech Support
    • Total Posts : 421
    • Reward points : 0
    • Joined: 2017/07/07 19:01:01
    • Location: Brea, CA
    • Status: offline
    • Ribbons : 1
    Re: What does, "equal or greater performance" really mean? 2019/11/05 21:10:31 (permalink)
    Hello
     
    it does not transfer and you have the option to buy a new one. As it is a paid service it simply does not transfer once used. In regards to the performance, can you provide us the links to these tests rather than screenshots? 

     
     
     
    -Daniel_EVGA
    #4
    bavor
    New Member
    • Total Posts : 32
    • Reward points : 0
    • Joined: 2008/02/20 21:24:43
    • Status: offline
    • Ribbons : 0
    Re: What does, "equal or greater performance" really mean? 2019/11/06 06:51:15 (permalink)
    EVGATech_DanielM
    In regards to the performance, can you provide us the links to these tests rather than screenshots? 



    RTX 2080 12,878 overall score:
    https://www.3dmark.com/fs/20807891
     
    RTX 2080 12,920 overall score:
    https://www.3dmark.com/fs/20805440
     
    GTX 1080 ti 13,701 overall score:
    https://www.3dmark.com/fs/20010235
     
    These were the highest stable settings I could achieve in Precision X1 with the RTX 2080:

    Those settings gave me my highest scores in the Firestrike benchmark.
     
     
     
     
    #5
    EVGATech_DanielM
    EVGA Tech Support
    • Total Posts : 421
    • Reward points : 0
    • Joined: 2017/07/07 19:01:01
    • Location: Brea, CA
    • Status: offline
    • Ribbons : 1
    Re: What does, "equal or greater performance" really mean? 2019/11/06 10:24:51 (permalink)
    So you are overclocking both cards? You are getting a 2063 on the 1080Ti which is really good, I can only get 2000 on mine. Looks like you RMA'd a unicorn. Unfortunately, we do not take overclocking into consideration when we base performance in RMA as we only guarantee the clocks we advertise. Also, the 3DMark definitions are different on both tests and sometimes you need to take that into consideration. With the combined test between both cards there is a 1 FPS difference in your system so while you are judging the GPU score your combined overall test (which is in game, which matters) has not changed much at all. You will probably see better performance gains with an updated CPU as that appears to be holding you back a bit, not too much to notice but a better CPU would help you all around. Keep in mind the memory is much quicker in the 2080. But once again, you are comparing 2 different generations of card that handle operations very differently. 
     
    https://www.3dmark.com/compare/fs/20805440/fs/20010235
     
    As you are using Firestrike Extreme, we do not base performance on these tests as this is not what the average user is using. We base most of our tests on 1080p as you questioned. Are you noticing any difference in game? 

     
     
     
    -Daniel_EVGA
    #6
    EVGATech_DanielM
    EVGA Tech Support
    • Total Posts : 421
    • Reward points : 0
    • Joined: 2017/07/07 19:01:01
    • Location: Brea, CA
    • Status: offline
    • Ribbons : 1
    Re: What does, "equal or greater performance" really mean? 2019/11/06 10:53:16 (permalink)
    So for boredom before work I went ahead and ran the same bench as you with your Hybrid 1080 Ti vs mine (air) with a higher mem OC and no GPU OC 
     
    https://www.3dmark.com/compare/fs/20010235/fs/20878622#
     
    As you can see we have very similar systems but my overall score beats yours but your actual GPU tests beat mine slightly. This is typically what we mean in terms of performance. Based on that test as well your OS is a bit outdated compared to mine but you ran that test about 3 months ago. Not all cards clock the same but if you are experiencing in game performance issues we can look in to it. Keep in mind I normally game at 1080 144hz.  You are more than welcome to send me a PM and we can chat it up 

     
     
     
    -Daniel_EVGA
    #7
    bavor
    New Member
    • Total Posts : 32
    • Reward points : 0
    • Joined: 2008/02/20 21:24:43
    • Status: offline
    • Ribbons : 0
    Re: What does, "equal or greater performance" really mean? 2019/11/06 12:31:17 (permalink)
    My concern is the RTX 2080 scores at the highest possible stable overclock are only around as good as my GTX 1080 ti scores were at the stock factory default settings with a lower CPU speed.  So it seems that I received a card with worse performance than the one it replaced.
     
    Here are the scores stock vs stock.
    https://www.3dmark.com/compare/fs/20878961/fs/13568207
    The combined and both graphics scores are lower on the RTX 2080.  The GTX 1080 ti results are with the CPU at a much lower speed.
     
    This is the best overclocked RTX 2080 results vs the stock GTX 1080 ti results.
    https://www.3dmark.com/compare/fs/20805440/fs/13568207
    As you can see the RTX 2080 on the highest possible overclock is only a little faster than my GTX 1080 ti on the default non overclocked settings.  The GTX 1080 ti results are with the CPU at a much lower speed.
     
    When I compare my stock RTX 2080 vs stock GTX 1080 ti Firestrike Ultra settings, I noticed that the combined and graphics scores are 10-15% higher on the old GTX 1080 ti.
    https://www.3dmark.com/compare/fs/20879005/fs/13568192
    Aagin, the GTX 1080 ti results are with the CPU at a much lower speed.
     
    I can run the tests again with the RTX 2080 with my CPU at 4.0 GHz to get a more apples to apples comparison with the RTX 2080, but I'm sure that won't improve the results.  The RTX 2080 results are with my CPU at 4.6 or 4.8 GHz while the GTX 1080 ti results are at 4.0 GHz.  That's why the Physics scores are higher and that adds to improving the overall score.
     
    Most of my gaming is either at 3440x1440 or 4K depending on the game.  There were a few instances in games at 3440x1440 and 4K where the RTX 2080 didn't seem as smooth as the GTX 1080 ti.  
     
     
     
     
    #8
    EVGATech_DanielM
    EVGA Tech Support
    • Total Posts : 421
    • Reward points : 0
    • Joined: 2017/07/07 19:01:01
    • Location: Brea, CA
    • Status: offline
    • Ribbons : 1
    Re: What does, "equal or greater performance" really mean? 2019/11/06 13:02:11 (permalink)
    Comparing a test from 2017 when you first acquired the card to one you are testing today isn't very fair. As you already sent your card back it's hard to say your card will score from where it was 2 years ago on the same system with a different version of Windows 10 at stock settings. Variables change. The scoring metric changes over time. How about this, we have a test bench in the office and I will bench my card against a 2080 on a clean OS and I will post results, won' be able to until Sunday as it isn't too busy. You may very well have received an under performing card and of course we can look into that if that's the case.

     
     
     
    -Daniel_EVGA
    #9
    bavor
    New Member
    • Total Posts : 32
    • Reward points : 0
    • Joined: 2008/02/20 21:24:43
    • Status: offline
    • Ribbons : 0
    Re: What does, "equal or greater performance" really mean? 2019/11/06 13:55:37 (permalink)
    I agree that its hard to get an apples to apples comparison because I don't benchmark my hardware monthly.  So, the only results I have to go by are older.  When I saw that I was receiving a RTX 2080 as a replacement for my GTX 1080 ti, I assumed there would be no decrease in overall performance between that any my GTX 1080 ti, except in the rare situations when there was a limit in VRAM or VRAM bandwidth. 
     
    It could be that I just was really lucky with the silicon lottery with my GTX 1080 ti and it performed well above average.  When I looked at the Firestrike Extreme results with a RTX 2080 and i7-6700K, my non overclocked GTX 1080 ti results would have placed in the top 140 out of the nearly 900 tests of the RTX 2080 with the i7-6700K.
    https://www.3dmark.com/search#/?mode=advanced&url=/proxycon/ajax/search2/cpugpu/fs/X/2005/1208/500000?minScore=0&cpuName=Intel%20Core%20i7-6700K%20Processor&gpuName=NVIDIA%20GeForce%20RTX%202080&gpuCount=1
     
    Also my results from July with the GTX 1080 ti would have placed it 2nd of the list of i7-6700K and RTX 2080 systems with was obtained in March.
    https://www.3dmark.com/compare/fs/18854954/fs/20010235
     
    I'd be interested in seeing those results you obtain with a GTX 1080 ti and RTX 2080.  I couldn't even use my GTX 1080 ti to do comparison benchmarks because it wouldn't run above the base clock speed and no troubleshooting here and through support fixed the issue.  That's why I obtained a RMA.
    #10
    EVGATech_DanielM
    EVGA Tech Support
    • Total Posts : 421
    • Reward points : 0
    • Joined: 2017/07/07 19:01:01
    • Location: Brea, CA
    • Status: offline
    • Ribbons : 1
    Re: What does, "equal or greater performance" really mean? 2019/11/06 14:08:39 (permalink)
    Ah so it even had a unstable factory OC for the 1080Ti. Got it, that makes more sense. What PSU are you using? I will be interested to see the comparison myself. Ill update Sunday. Have a good rest of your day. 




     
     
     
    -Daniel_EVGA
    #11
    bavor
    New Member
    • Total Posts : 32
    • Reward points : 0
    • Joined: 2008/02/20 21:24:43
    • Status: offline
    • Ribbons : 0
    Re: What does, "equal or greater performance" really mean? 2019/11/06 14:50:04 (permalink)
    I am using a Seasonic SS-760XP2 760 watt 80+ Platinum power supply.  My GTX 1080 ti had the same issue of not going above base clock speed when it was tested in another PC with an EVGA 850 watt 80+ Gold power supply.  
     
    I also remembered that the benchmarks my GTX 1080 ti wouldn't run at 2101 MHz GPU speed were unigine heaven and unigine superposition.  Otherwise I would have allowed it to run at a higher frequency for everything.
    #12
    EVGATech_DanielM
    EVGA Tech Support
    • Total Posts : 421
    • Reward points : 0
    • Joined: 2017/07/07 19:01:01
    • Location: Brea, CA
    • Status: offline
    • Ribbons : 1
    Re: What does, "equal or greater performance" really mean? 2019/11/10 17:15:46 (permalink)
    Hello! I was able to bring in my 1080Ti FTW3 DT and benched it against the 2080 XC 2183-KR 
     
    https://www.3dmark.com/compare/fs/20915547/fs/20915472  
     
    This test was on stock clocks, fresh OS install, and fans cranked at 75% for each card. As you can see, the performance of the 1080Ti FTW3 is 1.0% faster and this is only using the 1080p free edition and is how we would judge a direct replacement. While the definition of equal performance is a slippery slope, we feel these cards are equal performing and there would be no way a customer could identify the difference between the two in game. We do not have the Extreme or Ultra tests here at the office but I do have it at home and I can try to get an accurate testing between the 2. As you can also see instead of your 6700K we used a 6600K to try and mirror your set up but not identical and even the CPU scores increased on the 2080 test line. 

     
     
     
    -Daniel_EVGA
    #13
    bavor
    New Member
    • Total Posts : 32
    • Reward points : 0
    • Joined: 2008/02/20 21:24:43
    • Status: offline
    • Ribbons : 0
    Re: What does, "equal or greater performance" really mean? 2019/12/03 05:38:58 (permalink)
    Before you posted a link to your 1080 ti results that showed a clock speed of 1987 MHz when compared ot mine:
    https://www.3dmark.com/compare/fs/20010235/fs/20878622#
     
    The comparison you posted showed the 1080 ti at a much lower clock speed of 1886 MHz with the 2080 at 1995 MHz.  Also the memory clock was at 1377 MHz vs the previous results showing 1477 MHz.  
    https://www.3dmark.com/compare/fs/20915547/fs/20915472
    I understand they both were at stock clocks.  I understand the policy, but who buys a FTW3 Hybrid card to run it at stock clocks?  
     
    I still feel that my replacement GPU is slower than the one I had previously, but there is nothing I can do about it because of the policy.  It seems that the policy is out of touch with the end user who buys the higher tiered products with better cooling for the intention of overclocking.  The few game benchmarks that I have from the 1080 ti are still higher FPS at the same settings than the 2080 at its maximum stable overclock.  However the newest 1080 ti results that I have are from spring and summer of 2019, so they are all from at least a few months before I received the 2080.  Its just hard to believe that scores in benchmarks and FPS in games can change 10% or more in a few months.  I have tried two Corsaid fans meant for watercoolign radiators in a push pull configuration on the radiator of my 2080 to make it run cooler and that still can't give me the same results as my 1080 ti did.  I put the 2080 in a new Ryzen 3800X system where the CPU water cooled and over clocked and still can't obtain the same graphics benchmark scores I received with my 1080 ti.  I'll do a fresh Windows 10 install on my old 6700K system to see if I can get the 2080 to give any better results and to compare the drivers from last summer vs now to see if there is any difference.
     
    Basically, I feel that the policy is the same as if Intel replaced an i7-3770K with an i7-3700 under warranty because they both give the same benchmark results when run at stock speeds.  Yes I know old CPUs in the example, but that was back when the K series CPUs and the non K CPUs had the same base and boost clocks.
     
     
    #14
    RainStryke
    The Advocate
    • Total Posts : 15335
    • Reward points : 0
    • Joined: 2007/07/19 19:26:55
    • Location: Kansas
    • Status: online
    • Ribbons : 48
    Re: What does, "equal or greater performance" really mean? 2019/12/03 06:12:13 (permalink)
    You can request for them to give you a GTX 1080Ti in replacement. That's what most have done from what I have seen. Gauging various benchmarks i've done recently, the GTX 1080Ti is still a little better than a RTX 2080. I had to push my RTX 2080 really hard to get it to be about 5% faster than a GTX 1080Ti on the Tomb Raider benchmark.
     
    On a side note, the memory on that RTX 2080 is good for a 1000MHz+ to overclock.

    Main PC|CPU: Intel i7 8086K|CPU Cooler:EK-Velocity|Motherboard:Gigabyte Aorus Z390 Master|GPU:NVIDIA RTX 2080 Ti with EKWB Vector|Monitor:Dell S2716DG|RAM:32GB G.Skill Trident Z @ 4266MHz CL17|PSU:Seasonic X-1250|SSD:Samsung EVO 970 1TB|Case:Lian Li PC-011
    #15
    bavor
    New Member
    • Total Posts : 32
    • Reward points : 0
    • Joined: 2008/02/20 21:24:43
    • Status: offline
    • Ribbons : 0
    Re: What does, "equal or greater performance" really mean? 2019/12/03 10:10:21 (permalink)
    RainStryke
    On a side note, the memory on that RTX 2080 is good for a 1000MHz+ to overclock.



    The memory of this RTX 2080 hasn't been stable much above a 700 MHz overclock.  I've tried going much higher and it never works.  
     
    Also, the latest benchmarks I've seen show that the RTX 2080 with the newer drivers should be faster than the GTX 1080 ti.  It doesn't seem that there is anythign wrong with this RTX 2080, but I'm nowhere near begin faster(or even the same) in game FPS or benchmarks than my GTX 1080 ti.
    post edited by bavor - 2019/12/03 10:17:55
    #16
    EVGATech_DanielM
    EVGA Tech Support
    • Total Posts : 421
    • Reward points : 0
    • Joined: 2017/07/07 19:01:01
    • Location: Brea, CA
    • Status: offline
    • Ribbons : 1
    Re: What does, "equal or greater performance" really mean? 2019/12/03 20:56:52 (permalink)
    bavor
    Before you posted a link to your 1080 ti results that showed a clock speed of 1987 MHz when compared ot mine:
    https://www.3dmark.com/compare/fs/20010235/fs/20878622#
     
    The comparison you posted showed the 1080 ti at a much lower clock speed of 1886 MHz with the 2080 at 1995 MHz.  Also the memory clock was at 1377 MHz vs the previous results showing 1477 MHz.  
    https://www.3dmark.com/compare/fs/20915547/fs/20915472
    I understand they both were at stock clocks.  I understand the policy, but who buys a FTW3 Hybrid card to run it at stock clocks?  
     
    I still feel that my replacement GPU is slower than the one I had previously, but there is nothing I can do about it because of the policy.  It seems that the policy is out of touch with the end user who buys the higher tiered products with better cooling for the intention of overclocking.  The few game benchmarks that I have from the 1080 ti are still higher FPS at the same settings than the 2080 at its maximum stable overclock.  However the newest 1080 ti results that I have are from spring and summer of 2019, so they are all from at least a few months before I received the 2080.  Its just hard to believe that scores in benchmarks and FPS in games can change 10% or more in a few months.  I have tried two Corsaid fans meant for watercoolign radiators in a push pull configuration on the radiator of my 2080 to make it run cooler and that still can't give me the same results as my 1080 ti did.  I put the 2080 in a new Ryzen 3800X system where the CPU water cooled and over clocked and still can't obtain the same graphics benchmark scores I received with my 1080 ti.  I'll do a fresh Windows 10 install on my old 6700K system to see if I can get the 2080 to give any better results and to compare the drivers from last summer vs now to see if there is any difference.
     
    Basically, I feel that the policy is the same as if Intel replaced an i7-3770K with an i7-3700 under warranty because they both give the same benchmark results when run at stock speeds.  Yes I know old CPUs in the example, but that was back when the K series CPUs and the non K CPUs had the same base and boost clocks.
     
     


    I completely understand where you are coming from but we cannot take any type of overclocking into consideration when replacing a product based on performance. The tests I posted are from the driver from that date on a clean OS. This is exactly how we gauge the replacement options. The cards I tested were factory spec and still boosted above the advertised clocks. I do apologize you are frustrated. This would not be the same as swapping a K for a non K chip as it simply has features that are locked away from the consumer with a 3770 and that is not remotely close to what we do. Not all cards clock the same but if you sent in a silicon lottery GPU which it sounds like you did, we would not go through every card to see if we can test each individual one to match the exact performance. The performance difference you are describing are a few FPS at the most. The original 1987 speed was on my system and the GPU simply boosted to that speed and no overclock was set and your previous 1080Ti in GPU still beat mine. It just sounds like you sent in an extremely clockable 1080Ti. 

     
     
     
    -Daniel_EVGA
    #17
    Hoggle
    EVGA Forum Moderator
    • Total Posts : 5156
    • Reward points : 0
    • Joined: 2003/10/13 22:10:45
    • Location: Eugene, OR
    • Status: offline
    • Ribbons : 3
    Re: What does, "equal or greater performance" really mean? 2019/12/03 21:24:45 (permalink)
    The performance should be pretty close to equal and it's also worth considering the ray tracing feature which is starting to become important I feel but admit it's something that is different to each gamer. I personally would rather play Cyberpunk 2077 on the 2080 then a 1080Ti.

    Use an Associates Code & SAVE 5% - 10% on your purchase. Just click on the associates banner to save, or enter the associates code at checkout on your next purchase. If you choose to use my code I want to personally say "Thank You" for using it. 
     
     
    #18
    EVGATech_DanielM
    EVGA Tech Support
    • Total Posts : 421
    • Reward points : 0
    • Joined: 2017/07/07 19:01:01
    • Location: Brea, CA
    • Status: offline
    • Ribbons : 1
    Re: What does, "equal or greater performance" really mean? 2019/12/03 21:34:59 (permalink)
    Hoggle
    The performance should be pretty close to equal and it's also worth considering the ray tracing feature which is starting to become important I feel but admit it's something that is different to each gamer. I personally would rather play Cyberpunk 2077 on the 2080 then a 1080Ti.


    A thing to note is we do not take RTX features into account just as overclocking in regards to performance replacement options. 

     
     
     
    -Daniel_EVGA
    #19
    browns72
    New Member
    • Total Posts : 9
    • Reward points : 0
    • Joined: 2018/07/21 18:59:39
    • Status: offline
    • Ribbons : 0
    Re: What does, "equal or greater performance" really mean? 2019/12/04 03:39:13 (permalink)
    I'm far from an expert, but I also have a 1080 ti FTW 3 Hybrid. I got lucky being able to do a stable overclock to 2025 MHz and found it placed pretty high in Superposition (1080p). I believe when it comes to 4k that memory plays a huge part in testing (11 gb vs 8). If you want a 1080ti back, you should get on the horn with evga. I remember when I bought the hybrid cooler I got sent an older model that shorted the card and was unable to display. I RMA'd the cooler and got the same issue so they sent me a prebuilt hybrid in exchange for the cooler and card. The customer service in my experience does really well in this sort of thing. Hope this helps!
     
     
    #20
    Jump to:
  • Back to Mobile