EVGA

Hot!Test your driver variance always

Page: 123 > Showing page 1 of 3
Author
GloR1ouS_
SSC Member
  • Total Posts : 767
  • Reward points : 0
  • Joined: 2008/07/06 02:14:09
  • Status: offline
  • Ribbons : 2
Flagged as Spam (2)
2021/10/11 14:06:28 (permalink)
https://www.youtube.com/watch?v=gbclWij-Yqk
 
Basically a 3070 was outperforming my 3080 TI on the 3070's "golden driver" which was not a good driver for the 3080 TI. It's why its so key to actually use a program as good as CapFrameX. Honestly this program has helped me so much in optimizing my games its not even funny.
 
As a side note, W11 is great but its a real pain. Its an actual task to completely remove gamebar and "gamebar presence writer". I'd say W10 21H1 is just overall a superior gaming experience.

My Setup
-ASUS PG259QN, XG248Q
-Lian Li 011 Dynamic
-MSI Z490 Meg Ace
-10900K 5.2GHZ

-Artic Freezer II 360MM
-16GB Viper 4400MHZ
-3080 TI Founders

-EVGA 750 P2

#1

66 Replies Related Threads

    ty_ger07
    Insert Custom Title Here
    • Total Posts : 18753
    • Reward points : 0
    • Joined: 2008/04/10 23:48:15
    • Location: traveler
    • Status: offline
    • Ribbons : 215
    Re: Test your driver variance always 2021/10/11 14:38:30 (permalink)
    No thank you. Gaming doesn't need to be a chore. I wouldn't have the time for all that. If it works, it's good enough for me.
    #2
    GloR1ouS_
    SSC Member
    • Total Posts : 767
    • Reward points : 0
    • Joined: 2008/07/06 02:14:09
    • Status: offline
    • Ribbons : 2
    Re: Test your driver variance always 2021/10/11 15:26:06 (permalink)
    ty_ger07
    No thank you. Gaming doesn't need to be a chore. I wouldn't have the time for all that. If it works, it's good enough for me.



    It's not a chore, its called optimization. Why would you buy say a 3090 and allow for 3070 performance? How would you even know if you don't put in the time? may as well stick with pre-builds. Also, 18k + posts? seems you have more time than me guy.
    post edited by GloR1ouS_ - 2021/10/11 15:32:04

    My Setup
    -ASUS PG259QN, XG248Q
    -Lian Li 011 Dynamic
    -MSI Z490 Meg Ace
    -10900K 5.2GHZ

    -Artic Freezer II 360MM
    -16GB Viper 4400MHZ
    -3080 TI Founders

    -EVGA 750 P2

    #3
    ty_ger07
    Insert Custom Title Here
    • Total Posts : 18753
    • Reward points : 0
    • Joined: 2008/04/10 23:48:15
    • Location: traveler
    • Status: offline
    • Ribbons : 215
    Re: Test your driver variance always 2021/10/11 16:39:50 (permalink)
    I would never buy a 3090.
    #4
    CraptacularOne
    Omnipotent Enthusiast
    • Total Posts : 13632
    • Reward points : 0
    • Joined: 2006/06/12 17:20:44
    • Location: Florida
    • Status: offline
    • Ribbons : 189
    Re: Test your driver variance always 2021/10/11 17:12:48 (permalink)
    GloR1ouS_
    ty_ger07
    No thank you. Gaming doesn't need to be a chore. I wouldn't have the time for all that. If it works, it's good enough for me.



    It's not a chore, its called optimization. Why would you buy say a 3090 and allow for 3070 performance? How would you even know if you don't put in the time? may as well stick with pre-builds. Also, 18k + posts? seems you have more time than me guy.


    So you passive aggressive attack him because he doesn’t agree with your methods? What a great way to voice your opinion. 

    I agree with him though. With a family other responsibilities I don’t have time to play the ”driver optimization game” either. When I get time to myself and a good game I’m just gonna turn it on and go. 

    Intel i7 11700K ..................................Ryzen 9 5900X H2O
    PNY XLR8 RTX 3090 ............................Sapphire RX 6900 XT Toxic 
    Viotek SUW49C 32:9 Ultra Wide............Oculus Rift S
    MSI MPG Z590 Gaming Edge WiFi..........ASUS TUF Gaming X570 WiFi
    32GB TForce Xtreem 3600Mhz..............32GB Dominator Platinum 3600Mhz
    EVGA G3 SuperNova 1000w..................EVGA G3 SuperNova 850w
    #5
    GloR1ouS_
    SSC Member
    • Total Posts : 767
    • Reward points : 0
    • Joined: 2008/07/06 02:14:09
    • Status: offline
    • Ribbons : 2
    Flagged as Spam (4)
    Re: Test your driver variance always 2021/10/11 21:41:24 (permalink)
    CraptacularOne
    GloR1ouS_
    ty_ger07
    No thank you. Gaming doesn't need to be a chore. I wouldn't have the time for all that. If it works, it's good enough for me.



    It's not a chore, its called optimization. Why would you buy say a 3090 and allow for 3070 performance? How would you even know if you don't put in the time? may as well stick with pre-builds. Also, 18k + posts? seems you have more time than me guy.


    So you passive aggressive attack him because he doesn’t agree with your methods? What a great way to voice your opinion. 

    I agree with him though. With a family other responsibilities I don’t have time to play the ”driver optimization game” either. When I get time to myself and a good game I’m just gonna turn it on and go. 




    I wouldn't call it passive aggressive tbh. It honestly makes me think of people who buy a 144hz monitor and leave it at 60hz and cannot mentally see or know that there's actually a difference. If you play on high end hardware 360hz ect you should be able too FEEL when something is wrong. Either way people that say "im too busy" are the most annoying people on the planet. I probably work more than them anyways.

    My Setup
    -ASUS PG259QN, XG248Q
    -Lian Li 011 Dynamic
    -MSI Z490 Meg Ace
    -10900K 5.2GHZ

    -Artic Freezer II 360MM
    -16GB Viper 4400MHZ
    -3080 TI Founders

    -EVGA 750 P2

    #6
    Celeras
    SSC Member
    • Total Posts : 865
    • Reward points : 0
    • Joined: 2010/01/07 19:19:40
    • Status: offline
    • Ribbons : 3
    Flagged as Spam (5)
    Re: Test your driver variance always 2021/10/11 21:59:29 (permalink)
    I think he is lost. This is an enthusiast forum. Somehow he racked up 20,000 posts thinking it was Facebook 


    If I helped you in some way, or you just think I'm awesome.. please use my associate code! It gives you a discount
    #7
    bavor
    SSC Member
    • Total Posts : 694
    • Reward points : 0
    • Joined: 2008/02/20 21:24:43
    • Status: offline
    • Ribbons : 1
    Re: Test your driver variance always 2021/10/12 12:17:26 (permalink)
    GloR1ouS_
    https://www.youtube.com/watch?v=gbclWij-Yqk
     
    Basically a 3070 was outperforming my 3080 TI on the 3070's "golden driver" which was not a good driver for the 3080 TI. It's why its so key to actually use a program as good as CapFrameX. Honestly this program has helped me so much in optimizing my games its not even funny.



    I've never seen more than a 1-2% difference in different driver versions with the same card when testing game and synthetic benchmark results except for the few time when Nvidia released new drivers optimized for a game.  I've never had significantly worse results in a game or synthetic benchmark with a newer driver version. 
     
    Perhaps there were issues with the testing methodology in that video?

    Ryzen R9 5950X, EVGA RTX 3090 Kingpin SLI/NVLink, 64 GB DDR4 4000 MHz B-Die, 1TB PCIE 4.0 NVMe SSD, 4TB PICE 3.0 NVMe SSD, 2TB 2.5" SSD, 12TB 7200 RPM HDD, 1500 Watt BeQuiet PSU, Lian-Li O11 Dynamic XL case.
    #8
    zippytek
    Superclocked Member
    • Total Posts : 228
    • Reward points : 0
    • Joined: 2020/10/06 17:25:35
    • Status: offline
    • Ribbons : 0
    Re: Test your driver variance always 2021/10/12 12:26:21 (permalink)
    GloR1ouS_Either way people that say "im too busy" are the most annoying people on the planet. I probably work more than them anyways.

     
    you sure about that, clark
    #9
    GloR1ouS_
    SSC Member
    • Total Posts : 767
    • Reward points : 0
    • Joined: 2008/07/06 02:14:09
    • Status: offline
    • Ribbons : 2
    Flagged as Spam (2)
    Re: Test your driver variance always 2021/10/12 13:02:21 (permalink)
    bavor
    GloR1ouS_
    https://www.youtube.com/watch?v=gbclWij-Yqk
     
    Basically a 3070 was outperforming my 3080 TI on the 3070's "golden driver" which was not a good driver for the 3080 TI. It's why its so key to actually use a program as good as CapFrameX. Honestly this program has helped me so much in optimizing my games its not even funny.



    I've never seen more than a 1-2% difference in different driver versions with the same card when testing game and synthetic benchmark results except for the few time when Nvidia released new drivers optimized for a game.  I've never had significantly worse results in a game or synthetic benchmark with a newer driver version. 
     
    Perhaps there were issues with the testing methodology in that video?




    You live in a shell. CapframeX and Ldat do not lie and completely counter what you say.

    My Setup
    -ASUS PG259QN, XG248Q
    -Lian Li 011 Dynamic
    -MSI Z490 Meg Ace
    -10900K 5.2GHZ

    -Artic Freezer II 360MM
    -16GB Viper 4400MHZ
    -3080 TI Founders

    -EVGA 750 P2

    #10
    bavor
    SSC Member
    • Total Posts : 694
    • Reward points : 0
    • Joined: 2008/02/20 21:24:43
    • Status: offline
    • Ribbons : 1
    Re: Test your driver variance always 2021/10/12 17:01:34 (permalink)
    GloR1ouS_
    bavor
    GloR1ouS_
    https://www.youtube.com/watch?v=gbclWij-Yqk
     
    Basically a 3070 was outperforming my 3080 TI on the 3070's "golden driver" which was not a good driver for the 3080 TI. It's why its so key to actually use a program as good as CapFrameX. Honestly this program has helped me so much in optimizing my games its not even funny.



    I've never seen more than a 1-2% difference in different driver versions with the same card when testing game and synthetic benchmark results except for the few time when Nvidia released new drivers optimized for a game.  I've never had significantly worse results in a game or synthetic benchmark with a newer driver version. 
     
    Perhaps there were issues with the testing methodology in that video?



    You live in a shell. CapframeX and Ldat do not lie and completely counter what you say.



    Do you have proof of the reliability and accuracy of the testing methods?  I see a lot of tech reviews that are done in a poor and unscientific manner.
     
    I obviously live in a shell and don't know what I'm talking about when it comes to testing different driver versions since I have between 25 and 30 1st place scores on HWBOT and multiple top 5 and top 10 scores with both RTX 2000 and RTX 3000 series cards.
     

    Ryzen R9 5950X, EVGA RTX 3090 Kingpin SLI/NVLink, 64 GB DDR4 4000 MHz B-Die, 1TB PCIE 4.0 NVMe SSD, 4TB PICE 3.0 NVMe SSD, 2TB 2.5" SSD, 12TB 7200 RPM HDD, 1500 Watt BeQuiet PSU, Lian-Li O11 Dynamic XL case.
    #11
    CraptacularOne
    Omnipotent Enthusiast
    • Total Posts : 13632
    • Reward points : 0
    • Joined: 2006/06/12 17:20:44
    • Location: Florida
    • Status: offline
    • Ribbons : 189
    Re: Test your driver variance always 2021/10/12 18:55:22 (permalink)
    bavor
    GloR1ouS_
    bavor
    GloR1ouS_
    https://www.youtube.com/watch?v=gbclWij-Yqk
     
    Basically a 3070 was outperforming my 3080 TI on the 3070's "golden driver" which was not a good driver for the 3080 TI. It's why its so key to actually use a program as good as CapFrameX. Honestly this program has helped me so much in optimizing my games its not even funny.



    I've never seen more than a 1-2% difference in different driver versions with the same card when testing game and synthetic benchmark results except for the few time when Nvidia released new drivers optimized for a game.  I've never had significantly worse results in a game or synthetic benchmark with a newer driver version. 
     
    Perhaps there were issues with the testing methodology in that video?



    You live in a shell. CapframeX and Ldat do not lie and completely counter what you say.



    Do you have proof of the reliability and accuracy of the testing methods?  I see a lot of tech reviews that are done in a poor and unscientific manner.
     
    I obviously live in a shell and don't know what I'm talking about when it comes to testing different driver versions since I have between 25 and 30 1st place scores on HWBOT and multiple top 5 and top 10 scores with both RTX 2000 and RTX 3000 series cards.
     


    Don't even bother engaging him, he's combative and condescending in all his replies to anyone who opposes his "advice". He tried to tell me about "high end" hardware when both of my systems are higher end than his as if he's illuminating me somehow  
     
    He's just trying to plug his youtube video for views. 

    Intel i7 11700K ..................................Ryzen 9 5900X H2O
    PNY XLR8 RTX 3090 ............................Sapphire RX 6900 XT Toxic 
    Viotek SUW49C 32:9 Ultra Wide............Oculus Rift S
    MSI MPG Z590 Gaming Edge WiFi..........ASUS TUF Gaming X570 WiFi
    32GB TForce Xtreem 3600Mhz..............32GB Dominator Platinum 3600Mhz
    EVGA G3 SuperNova 1000w..................EVGA G3 SuperNova 850w
    #12
    Nozler
    FTW Member
    • Total Posts : 1991
    • Reward points : 0
    • Joined: 2009/10/28 10:51:49
    • Location: Ottertail county,Minnesota,USA
    • Status: online
    • Ribbons : 1
    Re: Test your driver variance always 2021/10/12 20:10:31 (permalink)
    😆 Did the dimm timing change? Maybe I'm not getting it right? Op care to explain? Did miss I something probably so.   These ol meat hooks r terrible with those videos
    post edited by Nozler - 2021/10/13 19:55:27

    heatware
     
    #13
    rottentreats
    Superclocked Member
    • Total Posts : 188
    • Reward points : 0
    • Joined: 2011/10/07 23:54:35
    • Status: offline
    • Ribbons : 6
    Re: Test your driver variance always 2021/10/12 20:27:46 (permalink)
    OP how about giving us a TL;DR so we don't have to watch a video... maybe the driver version that benefited you most? I personally just always run the latest and never have issues and my computer performs very well.  Not sure taking the time to optimize would change my experience or my ability to frag everyone with precision.
     
    post edited by rottentreats - 2021/10/13 13:05:02

    GPU: EVGA GeForce RTX 3080 Ti FTW3 ULTRA HYBRID

    Spoiler
    this should be on a trunk of a car
    close

    #14
    rzelek506
    Superclocked Member
    • Total Posts : 157
    • Reward points : 0
    • Joined: 2021/05/20 19:44:12
    • Status: online
    • Ribbons : 1
    Re: Test your driver variance always 2021/10/12 21:13:53 (permalink)
    Yeah, OP, I would also rather see a TL;DR or maybe a short write up with the apples-to-apples comparison and graphs for this driver testing; that would help me understand your conclusion better.
    And results from a more GPU intensive game than Fortnite would be best, otherwise the GPU driver may not be the isolated bottleneck (I think you mentioned in your video that increasing the memory frequency improved the FPS performance).
     
    My understanding of GPU drivers has always been that the latest stable driver should always be the best performing and the most reliable.
    And if the latest driver is not the best, then there is a bug with that driver and a new driver should be made to fix that bug while having either the same or better performance/reliability.
     
     
    post edited by rzelek506 - 2021/10/12 21:31:46
    #15
    GloR1ouS_
    SSC Member
    • Total Posts : 767
    • Reward points : 0
    • Joined: 2008/07/06 02:14:09
    • Status: offline
    • Ribbons : 2
    Flagged as Spam (2)
    Re: Test your driver variance always 2021/10/13 01:53:48 (permalink)
    CraptacularOne
    bavor
    GloR1ouS_
    bavor
    GloR1ouS_
    https://www.youtube.com/watch?v=gbclWij-Yqk
     
    Basically a 3070 was outperforming my 3080 TI on the 3070's "golden driver" which was not a good driver for the 3080 TI. It's why its so key to actually use a program as good as CapFrameX. Honestly this program has helped me so much in optimizing my games its not even funny.



    I've never seen more than a 1-2% difference in different driver versions with the same card when testing game and synthetic benchmark results except for the few time when Nvidia released new drivers optimized for a game.  I've never had significantly worse results in a game or synthetic benchmark with a newer driver version. 
     
    Perhaps there were issues with the testing methodology in that video?



    You live in a shell. CapframeX and Ldat do not lie and completely counter what you say.



    Do you have proof of the reliability and accuracy of the testing methods?  I see a lot of tech reviews that are done in a poor and unscientific manner.
     
    I obviously live in a shell and don't know what I'm talking about when it comes to testing different driver versions since I have between 25 and 30 1st place scores on HWBOT and multiple top 5 and top 10 scores with both RTX 2000 and RTX 3000 series cards.
     


    Don't even bother engaging him, he's combative and condescending in all his replies to anyone who opposes his "advice". He tried to tell me about "high end" hardware when both of my systems are higher end than his as if he's illuminating me somehow  
     
    He's just trying to plug his youtube video for views. 


    Your lack of knowledge is just funny to me. You think 1-2% variance is the case. You are the type of guy who buys a car but knows nothing about how it works or whats under the hood. You buy the best hardware on the market and assume it's perfect out of the box.

    As for people who are wondering what the methodology is, its very simple. CapFrameX uses rivatuner statistics and captures your 0.1% low and 1% low and average fps. These actually matter on the grand scale of things and input lag at its finest. Youd be amazed how 1 driver difference can tank these numbers.

    Watch battle(non)sense or fr33thy for their scientific analogy with Ldat and system latency. As for those who don't have Ldat, capframex will paint you an idea of whether or not your system is performing.

    Why do people buy an 8k pulling rate mouse? Lower latency. A system with higher latency can literally be the difference between getting a kill in a game or not. Taking a wall in fortnite or not, losing tile control ect. These things matter and people on an enthusiast forum should be more aware of the differences out there.

    Did you guys know for the last few nvidia drivers there was an issue with the driver constantly writing? You can google it.

    Also just because you can benchmark high doesnt mean a damn thing. My 3080ti founders benched 21.5K in timespy, I must be a wizard for hitting a number that has nothing to do with system latency or how a driver performs with a game. 3dmark ect all mean nothing in reality to game performance. They give you an idea if your hardware is stable or on par with factory spec. CapFrameX will give you a way better idea.
    post edited by GloR1ouS_ - 2021/10/13 02:01:37

    My Setup
    -ASUS PG259QN, XG248Q
    -Lian Li 011 Dynamic
    -MSI Z490 Meg Ace
    -10900K 5.2GHZ

    -Artic Freezer II 360MM
    -16GB Viper 4400MHZ
    -3080 TI Founders

    -EVGA 750 P2

    #16
    GloR1ouS_
    SSC Member
    • Total Posts : 767
    • Reward points : 0
    • Joined: 2008/07/06 02:14:09
    • Status: offline
    • Ribbons : 2
    Re: Test your driver variance always 2021/10/13 02:11:43 (permalink)
    On a side note, its crazy to me that people spend thousands on hardware but don't pay attention to optimization. I can understand people think Nvidia would release great drivers but they don't.

    A 6900xt with a 5950x gets 500 fps + in cod mw. A 3090 gets like 300. If you use nvidia filters, it tanks your fps. If you use AMD sharpening you lose zero fps. This just an example of optimization on a driver end.

    As for people wondering about why I mention gamebar, its layer of desktop compositer that runs overtop of your game causing input lag and on w11 its definitely noticeable.

    https://youtu.be/KJpraVLbgBg

    Apples to apples w11/w10 latency test.

    My Setup
    -ASUS PG259QN, XG248Q
    -Lian Li 011 Dynamic
    -MSI Z490 Meg Ace
    -10900K 5.2GHZ

    -Artic Freezer II 360MM
    -16GB Viper 4400MHZ
    -3080 TI Founders

    -EVGA 750 P2

    #17
    rjohnson11
    EVGA Forum Moderator
    • Total Posts : 86638
    • Reward points : 0
    • Joined: 2004/10/05 12:44:35
    • Location: Belgium
    • Status: offline
    • Ribbons : 75
    Re: Test your driver variance always 2021/10/13 02:28:07 (permalink)
    So far this thread has been fairly clean, but please take care when responding. 

    AMD Ryzen 9 5950X,  Corsair Mp600 Pro M.2 2TB, 128GB DDR4  Crosshair VIII Hero, Asus RX 6900 XT Associate Code: H5U80QBH6BH0AXF

    #18
    redteamgo
    Superclocked Member
    • Total Posts : 244
    • Reward points : 0
    • Joined: 2013/11/16 13:20:08
    • Status: online
    • Ribbons : 0
    Re: Test your driver variance always 2021/10/13 07:17:53 (permalink)
    What’s the difference between nvidia experience and capframex? Experience applies recommended settings. Capframex does the same but has some bar charts?
    #19
    ObscureEmpyre
    SSC Member
    • Total Posts : 546
    • Reward points : 0
    • Joined: 2012/01/15 14:40:05
    • Status: online
    • Ribbons : 3
    Re: Test your driver variance always 2021/10/13 07:55:21 (permalink)
    I’ve noticed variances in FPS across benchmark and several game titles using different drivers. Bottom line: if it’s not a drastic decrease, I don’t care that much. 5 FPS less isn’t a deal-breaker for me, but I do wonder why that is the case with some driver releases. Some decrease and some increase.

     
     
     
     
     

    #20
    GloR1ouS_
    SSC Member
    • Total Posts : 767
    • Reward points : 0
    • Joined: 2008/07/06 02:14:09
    • Status: offline
    • Ribbons : 2
    Re: Test your driver variance always 2021/10/13 08:15:17 (permalink)
    ObscureEmpyre
    I’ve noticed variances in FPS across benchmark and several game titles using different drivers. Bottom line: if it’s not a drastic decrease, I don’t care that much. 5 FPS less isn’t a deal-breaker for me, but I do wonder why that is the case with some driver releases. Some decrease and some increase.


    Ive seen variances of 100+ fps on the average. 471.61 averaged around 160 on the 0.1% where as 472.12 netted me 210 on the 0.1% my average fps jumped over 50. The latest driver that just came out with the new cache option delivered a noticeably smoother fortnite performance at 600+ fps. More fps = lowest latency.

    My Setup
    -ASUS PG259QN, XG248Q
    -Lian Li 011 Dynamic
    -MSI Z490 Meg Ace
    -10900K 5.2GHZ

    -Artic Freezer II 360MM
    -16GB Viper 4400MHZ
    -3080 TI Founders

    -EVGA 750 P2

    #21
    ObscureEmpyre
    SSC Member
    • Total Posts : 546
    • Reward points : 0
    • Joined: 2012/01/15 14:40:05
    • Status: online
    • Ribbons : 3
    Re: Test your driver variance always 2021/10/13 08:58:49 (permalink)
    GloR1ouS_
    ObscureEmpyre
    I’ve noticed variances in FPS across benchmark and several game titles using different drivers. Bottom line: if it’s not a drastic decrease, I don’t care that much. 5 FPS less isn’t a deal-breaker for me, but I do wonder why that is the case with some driver releases. Some decrease and some increase.


    Ive seen variances of 100+ fps on the average. 471.61 averaged around 160 on the 0.1% where as 472.12 netted me 210 on the 0.1% my average fps jumped over 50. The latest driver that just came out with the new cache option delivered a noticeably smoother fortnite performance at 600+ fps. More fps = lowest latency.

    That’s crazy! I’ve never seen FPS swings that drastic. 5-10 at most in either direction. The only thing that’s ever had a huge effect for me is a GPU upgrade or a CPU/RAM upgrade to a lesser extent.

     
     
     
     
     

    #22
    CraptacularOne
    Omnipotent Enthusiast
    • Total Posts : 13632
    • Reward points : 0
    • Joined: 2006/06/12 17:20:44
    • Location: Florida
    • Status: offline
    • Ribbons : 189
    Re: Test your driver variance always 2021/10/13 09:01:40 (permalink)
    GloR1ouS_
    CraptacularOne
    bavor
    GloR1ouS_
    bavor
    GloR1ouS_
    https://www.youtube.com/watch?v=gbclWij-Yqk
     
    Basically a 3070 was outperforming my 3080 TI on the 3070's "golden driver" which was not a good driver for the 3080 TI. It's why its so key to actually use a program as good as CapFrameX. Honestly this program has helped me so much in optimizing my games its not even funny.



    I've never seen more than a 1-2% difference in different driver versions with the same card when testing game and synthetic benchmark results except for the few time when Nvidia released new drivers optimized for a game.  I've never had significantly worse results in a game or synthetic benchmark with a newer driver version. 
     
    Perhaps there were issues with the testing methodology in that video?



    You live in a shell. CapframeX and Ldat do not lie and completely counter what you say.



    Do you have proof of the reliability and accuracy of the testing methods?  I see a lot of tech reviews that are done in a poor and unscientific manner.
     
    I obviously live in a shell and don't know what I'm talking about when it comes to testing different driver versions since I have between 25 and 30 1st place scores on HWBOT and multiple top 5 and top 10 scores with both RTX 2000 and RTX 3000 series cards.
     


    Don't even bother engaging him, he's combative and condescending in all his replies to anyone who opposes his "advice". He tried to tell me about "high end" hardware when both of my systems are higher end than his as if he's illuminating me somehow  
     
    He's just trying to plug his youtube video for views. 


    Your lack of knowledge is just funny to me. You think 1-2% variance is the case. You are the type of guy who buys a car but knows nothing about how it works or whats under the hood. You buy the best hardware on the market and assume it's perfect out of the box.

    As for people who are wondering what the methodology is, its very simple. CapFrameX uses rivatuner statistics and captures your 0.1% low and 1% low and average fps. These actually matter on the grand scale of things and input lag at its finest. Youd be amazed how 1 driver difference can tank these numbers.

    Watch battle(non)sense or fr33thy for their scientific analogy with Ldat and system latency. As for those who don't have Ldat, capframex will paint you an idea of whether or not your system is performing.

    Why do people buy an 8k pulling rate mouse? Lower latency. A system with higher latency can literally be the difference between getting a kill in a game or not. Taking a wall in fortnite or not, losing tile control ect. These things matter and people on an enthusiast forum should be more aware of the differences out there.

    Did you guys know for the last few nvidia drivers there was an issue with the driver constantly writing? You can google it.

    Also just because you can benchmark high doesnt mean a damn thing. My 3080ti founders benched 21.5K in timespy, I must be a wizard for hitting a number that has nothing to do with system latency or how a driver performs with a game. 3dmark ect all mean nothing in reality to game performance. They give you an idea if your hardware is stable or on par with factory spec. CapFrameX will give you a way better idea.



    First of all I wasn't the one stating 1-2% variance next time instead of slobbering your Advanced GG all over the keyboard in your ignorant rush to insult someone at least make sure you're directing it at the appropriate party. I've forgotten more about testing, benchmarking and overclocking than you'll probably ever know kiddo, you have no idea who you're talking to. Take your inept BS somewhere else junior. 
     
    post edited by CraptacularOne - 2021/10/13 09:05:43

    Intel i7 11700K ..................................Ryzen 9 5900X H2O
    PNY XLR8 RTX 3090 ............................Sapphire RX 6900 XT Toxic 
    Viotek SUW49C 32:9 Ultra Wide............Oculus Rift S
    MSI MPG Z590 Gaming Edge WiFi..........ASUS TUF Gaming X570 WiFi
    32GB TForce Xtreem 3600Mhz..............32GB Dominator Platinum 3600Mhz
    EVGA G3 SuperNova 1000w..................EVGA G3 SuperNova 850w
    #23
    redteamgo
    Superclocked Member
    • Total Posts : 244
    • Reward points : 0
    • Joined: 2013/11/16 13:20:08
    • Status: online
    • Ribbons : 0
    Re: Test your driver variance always 2021/10/13 09:26:43 (permalink)
    glorious, 
     
    so what is the tldr/method/formula you used to get +50 average fps in fortnite? if you want to explain it, i'll test it here as well.  I am on 472.12 with fortnite settings at Nvidia experience "optimal" at 2160p
    #24
    ty_ger07
    Insert Custom Title Here
    • Total Posts : 18753
    • Reward points : 0
    • Joined: 2008/04/10 23:48:15
    • Location: traveler
    • Status: offline
    • Ribbons : 215
    Re: Test your driver variance always 2021/10/13 10:05:34 (permalink)
    Smells of too good to be true. A 31% improvement between driver versions sounds like a headline. I wonder what else was at play behind the scenes.
    post edited by ty_ger07 - 2021/10/13 10:07:16
    #25
    GloR1ouS_
    SSC Member
    • Total Posts : 767
    • Reward points : 0
    • Joined: 2008/07/06 02:14:09
    • Status: offline
    • Ribbons : 2
    Re: Test your driver variance always 2021/10/13 10:51:14 (permalink)
    One of the things I have read from Nvidia was talking about drivers. How advanced they are for how many different configurations of hardware exist. Sometimes its bad engine coding playing nicer with a different driver.

    On the latest driver that just came out, they gave us an option to select the amount of shader cache nvidia can store. Reading dev notes from epic games talking about unreal engine, they did a driver that specifically stated driver cache improvements. I'm no software engineer but I can understand why a certain driver plays nice with 20 games and not nice with 5. Theres so many bad drivers out there. It took AMD 6 months to release a driver that allowed a 6900xt to get more than 60fps in the tangled shore in destiny 2. A 3080 got 200+ in that spot.

    Another thing for me personally, I play on 360hz and when you play on it long enough you start to feel abnormalities in a game. The second I installed 472.12 on my 3070 I knew I was having issues in fortnite. Same goes with 471.61 on the 3080 ti. What good is 360fps if you have a driver causing your 0.1% low to constantly give you a micro stutter.

    All my testing was done on a fresh install and a well optimized windows 10/11.

    My Setup
    -ASUS PG259QN, XG248Q
    -Lian Li 011 Dynamic
    -MSI Z490 Meg Ace
    -10900K 5.2GHZ

    -Artic Freezer II 360MM
    -16GB Viper 4400MHZ
    -3080 TI Founders

    -EVGA 750 P2

    #26
    bavor
    SSC Member
    • Total Posts : 694
    • Reward points : 0
    • Joined: 2008/02/20 21:24:43
    • Status: offline
    • Ribbons : 1
    Re: Test your driver variance always 2021/10/13 13:45:59 (permalink)
    GloR1ouS_
    Your lack of knowledge is just funny to me. You think 1-2% variance is the case. You are the type of guy who buys a car but knows nothing about how it works or whats under the hood. You buy the best hardware on the market and assume it's perfect out of the box.

     
    I love how you know nothing about a person or their knowledge and experience, but go right into personal insults. That's a great way to argue your case. Really mature!
     
    GloR1ouS_
    As for people who are wondering what the methodology is, its very simple. CapFrameX uses rivatuner statistics and captures your 0.1% low and 1% low and average fps. These actually matter on the grand scale of things and input lag at its finest. Youd be amazed how 1 driver difference can tank these numbers.
    Watch battle(non)sense or fr33thy for their scientific analogy with Ldat and system latency. As for those who don't have Ldat, capframex will paint you an idea of whether or not your system is performing.

     
    You really haven't explained your methodology in detail enough for it to be considered reliable, accurate, or repeatable.
    I've taken the FPS data logs from multiple games and synthetic benchmarks for different Nvidia driver versions and haven't seen the drastic difference that you claim there are on 0.1% lows, 1% lows, Average FPS, 99th percentile FPS, etc... after recording the data multiple times for each game and benchmark.
     
    GloR1ouS_
    Why do people buy an 8k pulling rate mouse? Lower latency. A system with higher latency can literally be the difference between getting a kill in a game or not. Taking a wall in fortnite or not, losing tile control ect. These things matter and people on an enthusiast forum should be more aware of the differences out there.
    Did you guys know for the last few nvidia drivers there was an issue with the driver constantly writing? You can google it.

     
    I know about it and a driver update at the end of August fixed the issue for me. It hasn't occurred again since then and ony of the several systems I have here. Different systems may have different results.
     
    GloR1ouS_
    Also just because you can benchmark high doesnt mean a damn thing.

     
    Obviously you assume that I haven't tested different Nvidia driver version multiple times to see if any give consistent and repeatable higher scores. Do you think I actually set all those highest scores and top 5 and top 10 scores by just firing up a system with the latest driver and running the benchmark on overclocked hardware with no prep ot testing at all?
    You make too many assumptions of others that have more experience in testing hardware, software, and drivers than you.
     
    GloR1ouS_
    My 3080ti founders benched 21.5K in timespy, I must be a wizard for hitting a number that has nothing to do with system latency or how a driver performs with a game. 3dmark ect all mean nothing in reality to game performance. They give you an idea if your hardware is stable or on par with factory spec. CapFrameX will give you a way better idea.

     
    It seems that you know nothing about benchmarks and how scores are calculated. Any frame stutter or drops in FPS are noticeable in the score and the data recorded for the score. That would show up in different drivers if the issue was widespread. I've data logged during game and synthetic benchmark runs to see if there was a difference in performance in drivers.
    People involved in competitive benchmarking will even test different windows color schemes or color theses to see if there is a difference in performance. In some versions of windows there is an actual difference in the color theme used regarding game and benchmark performance.
    Saying your timespy score with no additional information about the rest of the system, cooling used, etc... doesn't mean anything.
    After testing multiple driver versions on multiple hardware configurations over the past few years, I've noticed that its rare to see any signifigant difference in the performance of Nvidia drivers that is outside of normal tun to run variance in both games and benchmarks unless you are using drivers that were released before the game or benchmark.
    There were a few exceptions, but it was usually limited to issues that Nvidia noted and fixed in later releases. However they were rare and didn't always have an effect every hardware configuration.
     
    GloR1ouS_
    On a side note, its crazy to me that people spend thousands on hardware but don't pay attention to optimization. I can understand people think Nvidia would release great drivers but they don't.

     
    The massive performance differences in drivers versions you are claiming are rare and uncommon.
     
    GloR1ouS_
    A 6900xt with a 5950x gets 500 fps + in cod mw. A 3090 gets like 300. If you use nvidia filters, it tanks your fps. If you use AMD sharpening you lose zero fps. This just an example of optimization on a driver end.

     
    You are changing the topic to divert attention from the main topic I see.
     
    Moving on....
     
    One of the main issues with your claims is that you don't have detailed enough explanations of your methods, procedures, system monitoring, hardware used, etc... Instead you just insult people who disagree with you despite their experience level.
     
    Was that a typo in the video where you had several 4.71.61 driver version results along with a 4.71.71 driver result and said they were tests 1 through 4? You noted an inconsistent result in the 3080 ti results at one point. You saw a significant improvement in the 3080 ti results with a faster RAM speed. Maybe your testing should use a different RAM speed if it had a significant effect in performance or caused run to run variation. That could eliminate another external factor. At one point it looked like you were only collecting 20 seconds of data. Using only 20 seconds of data can yield a ton of inconsistencies in results. Were the FPS captures started and stopped at the exact same spot every time? In your video there were significant differences in your two YouTube example tests results that can be explained just be starting and stopping at different points. The differences were over 10 FPS on average and over 20FPS in the lows. The video gives the impression that your methods at best are inconsistent and sloppy, not repeatable, and not reliable.
     
    There are so many other factors that can make system performance vary that your data can't be considered reliable and accurate. You don't give enough data for other people to repeat your testing in the same conditions. Stating that you use a fresh install of windows and that its "optimized" in your own opinion isn't good enough for people to take your claims seriously. Any claims of drastic performance differences between driver versions needs to be able to be repeated by others using the exact same or very similar conditions on multiple systems.
     
    What BIOS versions? What chipset driver versions? What exact operation system versions were used? What did you do to "optimize" the operating systems? Was it a fresh OS install between each driver version or did you install the drivers then install new ones after rebooting? Did you use a different method between driver changes? What options were selected in the driver install? Did you repeat your results on multiple systems with different hardware configurations or were the result confined to one system? How long did you collect data for each result?  What was the ambient room temperature during testing? Did the ambient room temperature change during testing? What were the hardware temperatures of the GPU, VRAM, CPU, SSD, etc.... during testing? What was the GPU speed with the different driver versions? When there were performance differences noted, what were the changes in power consumption and clock speed of the components? What did you do to eliminate, minimize or control all external factors in the results and just isolate the drivers?
     
    Its great that you are trying to find differences in system performance, but your methods are lacking in many areas. So they can't be taken seriously.  Insulting people that disagree with you based on their personal experience with thorough, repeatable, and reliable testing methods just makes you look immature.  It also makes people question your results even more because you are not open to other's expensive experience on the topic.

    Ryzen R9 5950X, EVGA RTX 3090 Kingpin SLI/NVLink, 64 GB DDR4 4000 MHz B-Die, 1TB PCIE 4.0 NVMe SSD, 4TB PICE 3.0 NVMe SSD, 2TB 2.5" SSD, 12TB 7200 RPM HDD, 1500 Watt BeQuiet PSU, Lian-Li O11 Dynamic XL case.
    #27
    Sajin
    EVGA Forum Moderator
    • Total Posts : 45539
    • Reward points : 0
    • Joined: 2010/06/07 21:11:51
    • Location: Texas, USA.
    • Status: offline
    • Ribbons : 198
    Re: Test your driver variance always 2021/10/13 15:19:24 (permalink)
    Never seen this big of a swing myself. Not saying it couldn’t happen though.

    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.
     

     
    #28
    GloR1ouS_
    SSC Member
    • Total Posts : 767
    • Reward points : 0
    • Joined: 2008/07/06 02:14:09
    • Status: offline
    • Ribbons : 2
    Re: Test your driver variance always 2021/10/13 16:23:09 (permalink)
    I honestly don't know how you say capframex is unreliable. You could have the exact same hardware as 20 people and all have variations on a specific game engine. Capframex is an amazing software for testing ones own system. For anyone confused by how bad driver variation is I incourage you to download the software and run tests in games on your system. High density ram could tank your fps to an unsuspecting person. This has been proven by multiple tech channels. You could bench incredibly in a synthetic bench but get way less performance in a game without even knowing it. The knowledge is out there, it's whether people care to actually go in depth and care about system latency and maximizing their expensive hardware.

    Gamers nexus did an amazing video on 0.1% lows and 1% lows and how they effect your actual average fps. You could be getting 600 fps (lowers latency) but if your 0.1% was say 100 fps, you would get noticeable stutters destroying any latency advantage you obtained by buying "the best of the best" I myself am a competitive gamer and want every edge I can get vs these absolute cracked out players that exist in every game these days from gaming growing over the years.

    Im not sitting here like some kind of mad scientist writing numbers down on a spreadsheet. I simply try a new driver and see how it performs. Its not rocket science to capframex and see my 0.1% lows tanked or 1% tanked. I could feel a stutter in the game and wanted to see what was going on. 150 fps on a 0.1% vs 210-220 is just common sense. I'm not reading into it too hard, I was simply putting it out there for people that this actually exists. Do, don't, it doesnt matter to me.
    post edited by GloR1ouS_ - 2021/10/13 16:34:51

    My Setup
    -ASUS PG259QN, XG248Q
    -Lian Li 011 Dynamic
    -MSI Z490 Meg Ace
    -10900K 5.2GHZ

    -Artic Freezer II 360MM
    -16GB Viper 4400MHZ
    -3080 TI Founders

    -EVGA 750 P2

    #29
    Guydodge1021
    Superclocked Member
    • Total Posts : 147
    • Reward points : 0
    • Joined: 2017/05/28 10:14:18
    • Status: offline
    • Ribbons : 0
    Re: Test your driver variance always 2021/10/13 17:27:48 (permalink)
    GloR1ouS_
    On a side note, its crazy to me that people spend thousands on hardware but don't pay attention to optimization. I can understand people think Nvidia would release great drivers but they don't.

    A 6900xt with a 5950x gets 500 fps + in cod mw. A 3090 gets like 300. If you use nvidia filters, it tanks your fps. If you use AMD sharpening you lose zero fps. This just an example of optimization on a driver end.

    As for people wondering about why I mention gamebar, its layer of desktop compositer that runs overtop of your game causing input lag and on w11 its definitely noticeable.

    https://youtu.be/KJpraVLbgBg

    Apples to apples w11/w10 latency test.

    so let me get this straight your a competitive gamer you have a 4k monitor and a 1080 monitor and your choice of vid card was
    a 3080 ti founders.and you want to get the most out of your rig for competitive gaming..intresting choices

    ASUS TUF Gaming GT501 Mid-Tower Computer Case W/smoked tempered glass
    Gigabyte Z390 AORUS MASTER
     Intel Core i9-9900K Desktop Processor 8 Cores up to 5.0 GHz Turbo unlocked
    HyperX Predator Black 32GB kit 3333MHz DDR4 CL16 DIMM XMP memory
     EVGA RTX 3090 FTLOSS 3 ULTRA 24GB
    1-- Samsung 970 EVO Plus SSD 250GB - M.2 NVMe Interface Internal Solid State (O.S only)
    #30
    Page: 123 > Showing page 1 of 3
    Jump to:
  • Back to Mobile