how does passmark do their percentile rankings? And is mine accurate?

[StunnaFan34]

So I ran a passmark test with my 2080 super and got a score of 21,414. Thats good but what makes me question how useful passmark is because it put me in the 98th percentile for 3D graphics with my 2080 super NON overclocked. Doesnt this seem a little high?

[Kilrah]

No. There's only one card that's more powerful and it's expensive, so people benchmaring a 2080Ti or even an OCd 2080S will be a drop in the ocean compared to all those who have mid end cards.

[LIGISTX]

3 minutes ago, StunnaFan34 said:

So I ran a passmark test with my 2080 super and got a score of 21,414. Thats good but what makes me question how useful passmark is because it put me in the 98th percentile for 3D graphics with my 2080 super NON overclocked. Doesnt this seem a little high?

Think of all the video cards in the world right now in computers....... in reality, having a 2080 super is likely top .01%.... 

 

Now, scale that back down to just "gamers and content creators", your still in the highest range of cards that you can buy. The 2080ti is faster, and quadro and tesla cards do have faster options, but those are even more expensive and not suited for gaming. If anything, 98% is low. If you look at the steam survey, I bet A LOT of people still run 970's...

[StunnaFan34]

3 minutes ago, Kilrah said:

No. There's only one card that's more powerful and it's expensive, so people benchmaring a 2080Ti or even an OCd 2080S will be a drop in the ocean compared to all those who have mid end cards.

ah ok. so what youre saying is i can trust it because the vast majority of people are testing with cards worse than mine?

[StunnaFan34]

1 minute ago, LIGISTX said:

Think of all the video cards in the world right now in computers....... in reality, having a 2080 super is likely top .01%.... 

 

Now, scale that back down to just "gamers and content creators", your still in the highest range of cards that you can buy. The 2080ti is faster, and quadro and tesla cards do have faster options, but those are even more expensive and not suited for gaming. If anything, 98% is low. If you look at the steam survey, I bet A LOT of people still run 970's...

how would 98 percent be low?

[Kilrah]

1 minute ago, StunnaFan34 said:

ah ok. so what youre saying is i can trust it because the vast majority of people are testing with cards worse than mine?

It simply means your GPU's performance is in the top 2% of people who have run a benchmark with their tool.

[StunnaFan34]

1 minute ago, Kilrah said:

It simply means your GPU's performance is in the top 2% of people who have run a benchmark with their tool.

ok thank you. still seems unbelievably high but im sure its accurate

[Chris Pratt]

Just throwing it out there. Not sure if this is the case, but running SLI, you can sometimes nerf benchmarks like these, even if your individual cards are in a lower class. However, that's why synthetic benchmarks have very limited utility, because in real world performance, a single 2080 super is going to still run circles around those setups, just because SLI is dead and very few games support it.

[LIGISTX]

3 hours ago, StunnaFan34 said:

how would 98 percent be low?

Because most people in the world playing PC games can't afford the second most expensive gaming GPU on the planet.... lol

 

How many billions of computers are in the world? How many millions of those do people play games on? How many do you think have a 2080 or higher..? 

 

Steam hardware survey would agree with 3dmarks numbers.... I think your neglecting to remember most people buy cards in the ~150-250 dollar range.... as seen in the data. The top card USED to be the 970, but even then, still more 970's out in the wild than 2080's. There are more people gaming on integrated Intel graphics than 2080's... I admit, a good % of those integrated graphics results are from PC's that are not really playing games beyond side scrollers etc, but the point remains, MOST people in the world playing video games don't have 2080 tier cards. This is why 98 percentile is low. I would expect 2080 to be top half a percent or higher, which would be 99.5 percentile. Looks like according to steam its more like 98% which matches passmarks number. 

[LogicalDrm]

17 hours ago, StunnaFan34 said:

ok thank you. still seems unbelievably high but im sure its accurate

Ok, you don't seem to understand what that means.

 

Passmark is paid benchmarking software, and not very popular. Its somewhat good tool as they do test very wide range of cards, CPUs etc, so making comparisons between stuff is easy. However, since they do benchmark very wide range of units, and the amount of overall tested units is low compared to 3Dmark and some other more popular benchmarks, having top-end GPU, even non-OC'd or with factory OC, means that compared to WHOLE pool of tested units, you will end up in top end of the bunch.

 

Thats what percentile means. Just like being millionaire in US would place you at top 1-5% of wealthy people. Doesn't mean you are actually that rich. You would be just rich compared to all rest of the population.

[StunnaFan34]

On 8/17/2020 at 4:32 AM, LIGISTX said:

Spoiler

 

Because most people in the world playing PC games can't afford the second most expensive gaming GPU on the planet.... lol

 

How many billions of computers are in the world? How many millions of those do people play games on? How many do you think have a 2080 or higher..? 

 

Steam hardware survey would agree with 3dmarks numbers.... I think your neglecting to remember most people buy cards in the ~150-250 dollar range.... as seen in the data. The top card USED to be the 970, but even then, still more 970's out in the wild than 2080's. There are more people gaming on integrated Intel graphics than 2080's... I admit, a good % of those integrated graphics results are from PC's that are not really playing games beyond side scrollers etc, but the point remains, MOST people in the world playing video games don't have 2080 tier cards. This is why 98 percentile is low. I would expect 2080 to be top half a percent or higher, which would be 99.5 percentile. Looks like according to steam its more like 98% which matches passmarks number. 

 

 

sorry im just seeing this now. I understand thank you

 

On 8/17/2020 at 6:24 PM, LogicalDrm said:

Spoiler

 

Ok, you don't seem to understand what that means.

 

Passmark is paid benchmarking software, and not very popular. Its somewhat good tool as they do test very wide range of cards, CPUs etc, so making comparisons between stuff is easy. However, since they do benchmark very wide range of units, and the amount of overall tested units is low compared to 3Dmark and some other more popular benchmarks, having top-end GPU, even non-OC'd or with factory OC, means that compared to WHOLE pool of tested units, you will end up in top end of the bunch.

 

Thats what percentile means. Just like being millionaire in US would place you at top 1-5% of wealthy people. Doesn't mean you are actually that rich. You would be just rich compared to all rest of the population.

 

 

ohhh ok sorry i get it. sorry im responding late. 

 

6 hours ago, StunnaFan34 said:

sorry im just seeing this now. I understand thank you

so is there something wrong with my card? on tests like timespy i get a good score

 

On 8/17/2020 at 4:32 AM, LIGISTX said:

Spoiler

 

Because most people in the world playing PC games can't afford the second most expensive gaming GPU on the planet.... lol

 

How many billions of computers are in the world? How many millions of those do people play games on? How many do you think have a 2080 or higher..? 

 

Steam hardware survey would agree with 3dmarks numbers.... I think your neglecting to remember most people buy cards in the ~150-250 dollar range.... as seen in the data. The top card USED to be the 970, but even then, still more 970's out in the wild than 2080's. There are more people gaming on integrated Intel graphics than 2080's... I admit, a good % of those integrated graphics results are from PC's that are not really playing games beyond side scrollers etc, but the point remains, MOST people in the world playing video games don't have 2080 tier cards. This is why 98 percentile is low. I would expect 2080 to be top half a percent or higher, which would be 99.5 percentile. Looks like according to steam its more like 98% which matches passmarks number. 

 

 

sorry i accidentally responded to myself. So should my card be 99% or should it be 98%? 

[LIGISTX]

2 hours ago, StunnaFan34 said:

sorry i accidentally responded to myself. So should my card be 99% or should it be 98%? 

The percentile is extremely meaningless; if you were off by multiple %, like low low to mid 90's then something would be wrong. I would have guessed a 2080 super would be higher than 98, hell higher than 99, but it really doesn't matter nor does it mean much. Your in the correct range so everything is fine :).

[StunnaFan34]

14 hours ago, LIGISTX said:

The percentile is extremely meaningless; if you were off by multiple %, like low low to mid 90's then something would be wrong. I would have guessed a 2080 super would be higher than 98, hell higher than 99, but it really doesn't matter nor does it mean much. Your in the correct range so everything is fine :).

ok thank you