Question graphics card optimum efficacy

roieco

Reputable
Dec 4, 2020
128
1
4,595
hi all,

my hw system is as in my signature.
i was wondering if there is a way to know (test) if my graphics card operates to the maximum of its ability.
i'd prefer a synthetic test, as in one that isnt impacted by the rest of the system.

thanks!
 
will it be able to test the gpu independently from the rest of the system?
Technically this is impossible, because the GPU is a device that can't do anything without input from the system. Which means the general system performance can impact the performance of the GPU.

Anyway, there are plenty of apps that put a synthetic or repeatable load on the GPU. These include:
  • 3D Mark using the Fire Storm Extreme, Time Spy, or Port Royal tests.
  • Unigine Superposition
  • FurMark, MSI Kombuster, or other stress test apps
 

roieco

Reputable
Dec 4, 2020
128
1
4,595
here is my result on Unigine Superposition with 1080p Extreme:

1DHwlOF.png


is this result expected with my card and system?
 

roieco

Reputable
Dec 4, 2020
128
1
4,595
Last edited:

roieco

Reputable
Dec 4, 2020
128
1
4,595
Make sure you close Afterburner and all screen overlays before running.

INFO: Checking GPU configuration
INFO: Starting GPU Benchmarks
INFO: Launching NGPU detection
INFO: NGPU Name: NVIDIA GeForce RTX 3080
INFO: NGPU Driver: 517.48
INFO: NGPU Bios: 94.02.42.80.dc
INFO: NGPU Total RAM MB: 10240
INFO: NGPU Max C MHz: 2100
INFO: NGPU Max M MHz: 4750
INFO: Launching AGPU detection
INFO: Completed GPUStats successfully
INFO: GPU score on Plane = 295.5 FPS
INFO: GPU score on Cube = 254.8 FPS
INFO: GPU score on Stones = 322.6 FPS
INFO: GPU score on Swarm = 174.7 FPS
INFO: GPU score on Galaxy = 239.7 FPS
ERROR: Sphere GPU benchmark failed, aborting further tests on this GPU
INFO: Completed GPU benchmarks

i disabled nvidia overlay and uninstalled afterburner.
 

Aeacus

Titan
Ambassador
is this result expected with my card and system?

It's on the lower side but there are systems that have done worse.

Unigine leaderboards,
link: https://benchmark.unigine.com/leaderboards/superposition/1.x/1080p-extreme/single-gpu
In the Search bar, type in 3080 and pick Nvidia RTX 3080 from results. 4 pages would be displayed and yours is at the bottom of 3rd page, meaning doing worse than most, of who have submitted their results.

--

Is your GPU brand new, or do you buy it used? :unsure:

ERROR: Sphere GPU benchmark failed, aborting further tests on this GPU

Possible fix: https://evanscomputers.com/index.ph...fails-sphere-graphics-test-possible-solution/
Try it and if it works, share your Userbenchmark run with us.
 
  • Like
Reactions: roieco
Honestly at this point I'm wondering if there's any real point in worrying about if the video card is being used to its "maximum potential" or not. The thing is, the video card could have lots of "maximum potential" that requires a lot to get to that point. For example, this is one of the best scores for an RTX 3080 with Time Spy: https://www.3dmark.com/spy/28148958

At first you'll think "wow! I want that!" but then if you notice something, you'll see the benchmark reported an average GPU temperature of -42C. This implies they used either liquid nitrogen or a phase change cooler to overclock the snot out of the video card. Though of course, they probably also had a card that was built for it and yours may not actually get that far, but throwing those factors out, this is the "maximum potential" that the RTX 3080 could achieve.

So let's worry about something that's more practical: is it getting the performance you want out of games? Set a reasonable set of requirements, like resolution, graphics quality preset, and frame rate. If the card is hitting all three, then there's nothing to worry about. Also, make sure your requirements are modest, otherwise you'll be chasing after them and burning your wallet doing so.
 

roieco

Reputable
Dec 4, 2020
128
1
4,595
Honestly at this point I'm wondering if there's any real point in worrying about if the video card is being used to its "maximum potential" or not. The thing is, the video card could have lots of "maximum potential" that requires a lot to get to that point. For example, this is one of the best scores for an RTX 3080 with Time Spy: https://www.3dmark.com/spy/28148958

At first you'll think "wow! I want that!" but then if you notice something, you'll see the benchmark reported an average GPU temperature of -42C. This implies they used either liquid nitrogen or a phase change cooler to overclock the snot out of the video card. Though of course, they probably also had a card that was built for it and yours may not actually get that far, but throwing those factors out, this is the "maximum potential" that the RTX 3080 could achieve.

So let's worry about something that's more practical: is it getting the performance you want out of games? Set a reasonable set of requirements, like resolution, graphics quality preset, and frame rate. If the card is hitting all three, then there's nothing to worry about. Also, make sure your requirements are modest, otherwise you'll be chasing after them and burning your wallet doing so.

thank you for your reply.
i played a bit with the results and compared online. it seems that if i compare only to systems with the same cpu, same gpu and same gpu core clock and gpu memory clock as mine:
  • in time spy, my system is 40'th out of 322
  • in fire strike, my system is 3rd out of 52
so unless you think otherwise, it would seem to me that the system behaves as expected.

the reason i began thinking of this was that people around were throwing numbers such as 300 fps in top, new games with max settings and no dlss. when i tried those games on my system i didn't receive the same numbers which led me to suspect that something isn't right.

as to your question of my expectaation from the system - i play mostly rpgs and strategy (no shooters, no sport) - things like dragon age, skyrim, cyberpunk etc. so i dont need crazy high fps. good (~80) fps on max quality in 4k would be wonderful.
 

roieco

Reputable
Dec 4, 2020
128
1
4,595
It's on the lower side but there are systems that have done worse.

Unigine leaderboards,
link: https://benchmark.unigine.com/leaderboards/superposition/1.x/1080p-extreme/single-gpu
In the Search bar, type in 3080 and pick Nvidia RTX 3080 from results. 4 pages would be displayed and yours is at the bottom of 3rd page, meaning doing worse than most, of who have submitted their results.

--

Is your GPU brand new, or do you buy it used? :unsure:



Possible fix: https://evanscomputers.com/index.ph...fails-sphere-graphics-test-possible-solution/
Try it and if it works, share your Userbenchmark run with us.

thanks! that worked.

my Userbenchmark results are:
UserBenchmarks: Game 218%, Desk 105%, Work 223%
CPU: Intel Core i7-11700K - 100.6%
GPU: Nvidia RTX 3080 - 221.5%
SSD: Samsung 980 Pro NVMe PCIe M.2 2TB - 287.9%
SSD: Samsung 870 EVO 4TB - 601.3%
RAM: G Skill Intl F4-4000C17-16GTZRB 2x16GB - 132.4%
MBD: Asus ROG STRIX Z590-E GAMING WIFI

and for the graphics card:
loWQr7O.png


are those good?

also i see something strange for the boot ssd.
on top it says:

9G4aODB.png


but down at the ssd section it says:

bS2Prhy.png


that result is also very slow comparing to magician's results:

sks0qST.png


any idea why the ssd works so slow on Userbenchmark's test?
 
Unless you have really expensive ram and have tweeked everything to the max already, your system is running as expected.
You can not ever expect to be near the top of the charts unless you LN2/Phase change/ your processor/memory and GPU.
You are about where you should be.
So enjoy your system or join the overclockers club where you spend hours upon hours slowly tweeking one setting at a time until you find its MAXIMUM .
Then the next setting .
Then the next. ........................................
OH <Mod Edit> its 3 oclock and I have to get up in a few hours.
IT begins!
 
Last edited by a moderator:

Aeacus

Titan
Ambassador
and for the graphics card:
loWQr7O.png


are those good?

Looks good to me, performing better than average.

Overall, i'd say your GPU performs average, in terms of all RTX 3080 out there. And being average isn't bad.

but down at the ssd section it says:

bS2Prhy.png


that result is also very slow comparing to magician's results:

sks0qST.png


any idea why the ssd works so slow on Userbenchmark's test?

Here's one reason;
Proprietary drivers (e.g. Intel RST or Samsung NVME) often offer better performance than windows stock drivers.
Source: https://www.userbenchmark.com/Faq/What-is-the-UBM-PC-status/94

--

Overall, i'd say that your PC works fine. Hunting for the perfection (or best efficiency) will waste a lot of time and effort, with little (if any) performance gains.

Now, if you'd have serious performance issues (which you don't), then it would be good to figure out what is going on. Else-ways, don't stress about it.
 

roieco

Reputable
Dec 4, 2020
128
1
4,595
Looks good to me, performing better than average.

Overall, i'd say your GPU performs average, in terms of all RTX 3080 out there. And being average isn't bad.



Here's one reason;

Source: https://www.userbenchmark.com/Faq/What-is-the-UBM-PC-status/94

--

Overall, i'd say that your PC works fine. Hunting for the perfection (or best efficiency) will waste a lot of time and effort, with little (if any) performance gains.

Now, if you'd have serious performance issues (which you don't), then it would be good to figure out what is going on. Else-ways, don't stress about it.

980 pro is not supported by the samsung nvme drivers. should i install intel rst? i was told that with rst i will lose DirectStorage.

i ran an ssd test by 3dmark and here are the results:

ovIGDti.png


when comparing online it says that i'm on the bottom 39% for overall score and bottom 37% for bandwidth :(
 

roieco

Reputable
Dec 4, 2020
128
1
4,595
980 pro is not supported by the samsung nvme drivers. should i install intel rst? i was told that with rst i will lose DirectStorage.

i ran an ssd test by 3dmark and here are the results:

ovIGDti.png


when comparing online it says that i'm on the bottom 39% for overall score and bottom 37% for bandwidth :(

correction. bottom 1% (!) when comparing with same drive type and size.
 

Aeacus

Titan
Ambassador
980 pro is not supported by the samsung nvme drivers

True, based on what is said on Samsung website. However, what your 980 Pro has, is firmware, that you can download and update, if needed.

correction. bottom 1% (!) when comparing with same drive type and size.

Your MoBo has 4x M.2 slots. In which slot is your 980 Pro installed?
  • M.2_1
  • M.2_2
  • M.2_3
  • M.2_4
?

--

Download CrystalDiskInfo,
link: https://crystalmark.info/en/software/crystaldiskinfo/

Run it and look what it says. (Can share screen here as well.) What i'm most interested about, is Transfer Mode.
 

roieco

Reputable
Dec 4, 2020
128
1
4,595
True, based on what is said on Samsung website. However, what your 980 Pro has, is firmware, that you can download and update, if needed.



Your MoBo has 4x M.2 slots. In which slot is your 980 Pro installed?
  • M.2_1
  • M.2_2
  • M.2_3
  • M.2_4
?

--

Download CrystalDiskInfo,
link: https://crystalmark.info/en/software/crystaldiskinfo/

Run it and look what it says. (Can share screen here as well.) What i'm most interested about, is Transfer Mode.

thank you for the reply, @Aeacus .

my ssd is connected at M.2_1 .

here's a screenshot from CrystalDiskInfo:

qa7u13V.png