It’s just the average of the lowest fps dips. Ideally you want it as close to the average fps drawn
GABE_EDD
Let’s say you’re playing a game and averaging 100FPS, but it’s a little “stuttery” at times, and 1% of the time you’re actually at 40FPS because one or a few frames take longer than normal to render, your 1% low would be 40FPS. The more buttery-smooth your frame rate is, the closer your 1% lows are to your average FPS.
Atrocious1337
Your lowest FPS dips 1% of the time (there are 0.1% lows too).
But their functional purpose is to measure your frame rate consistency and micro stutters. Like if you are trying to play at a steady 60 FPS, your average FPS should be 60, and your 1% lows should also be 60 or at least extremely close (e.g 59fps, 58fps).
El_Androi
Imagine you’re playing at an average 60 fps. If your 1% lows are close to it, it means most of those frames are evenly spread across every second and it feels smooth. However, imagine you had 120 frames in the first half of the second, and zero for the second half of it. It would still be 60 fps in average, but It’d feel way more janky. That’s roughly what low 1% lows are but of course it’s way more irregular in reality.
Eziolambo
Each second has separate frames like, 60,59,52,60… and so on. It will only pick lowest of them all and take its average.
poinguan
The bigger the fps number, the better it is. That’s all you need to know.
aberroco
It’s the worst 1% of your framerate for each frame. So, if it’s just a bit lower than your average, the game is butter smooth. If it’s in single digits – the game or your setup or both have stuttering issues and microfreezes.
Upd.: for extra clarification, it’s usually in triplet numbers: low 1%, average and max. The max shows the best your system could show in perfect conditions. Average is just average, pretty self-explanatory, and 1% as I just explained. The closer they are – the better the game is optimized, usually. It means that the load on your GPU is even and there’s no freezes or stuttering. The greater the difference between them the more troubling it might be. Except if benchmark was done badly and it shows you a clear sky where the max just jumps to hundreds because there’s literally nothing to render. But that’s not very reliable, as 1% low might be just at the very beginning, when resources are still loading dynamically, and it’s not really that bad or annoying, as the rest of the benchmark is smooth. Good benchmarks should be done to account all that, so, no scenes that differ too much in complexity, and long enough to minimize the effect of initial loading, if it happens.
JerbearCuddles
All I know is the closer my 1% lows is to my average FPS the better. Having jarring frame drops is not fun to look at.
ARatOnPC
If you have low 1% average you will experience stutters and freezing.
NotJustBibbit
Imagine having 100fps. 1% lows is the average of the lowest 1% of the frames. So if your lowest frame was 60 then 1% lows would be 60. You want it as close to the normal average as possible to prevent stuttering
bvkimi
The simplest way I heard it being explained by some Youtuber was that, whatever that 1% Low figure is, 99% of the time, your FPS is above that.
Example: If you play a game for 10,000 seconds, you record the FPS for each second that passes, and then pick out the lowest 100 samples (1% of 10,000). That is your 1% Low.
*(If this thinking is wrong, anyone feel free to correct. I don’t mind.)*
Salty_Theory2742
The human eye can not see over 60 FPS Lows.
/s
kingawsume
It’s deviation from the average. Average of the 1%/0.1% of the most slowly generated frames. You want it to be as close to the average of 100% of the frames as possible.
MediocreRooster4190
Imagine you have 60 frames in a second and half are spaced 33.2ms apart and half are 8.3ms apart. The time between each frame is not stable this can cause bad frame pacing or frametimes.
1% lows is like that but when you drop frames suddenly and drastically, like 100fps to 55fps. It is just a metric that shows where your rates drop and how deep the difference is between average and minimum essentially.
Basically the idea is to show numbers that can tell you how smooth or jumpy the experience can be.
nmathew
Fuck I hate what happened to Techreport. DON’T go there, the good old articles had the useful images scrubbed and it’s advertising click bait now.
Editor was asked in a TV interview what it’s like to actually play games on a card, and he realized he couldn’t answer the question. So he actually took a step back and tried that.. only to find that average fps wasn’t revealing what he experienced.
He’s saw bad “micro stutters” in multi GPU configurations (common at the time and worse with AMD”. And stutter in general on all cards. (Spoiler: Nvidia and AMD did a ton of driver work to fix these issues.)
This lead to investigating “fps inside the second.” This came to with revolutionary benchmarks looking at the time delay to deliver the next frame. Basically, GPUs don’t deliver a static frame rate… It varies.
Keep in mind this predates dynamic Freesync and G-sync, so missing a frame at 60 Hz was a 16.7 ms stutter.
Short story long, lots of effort was put into the experience of the player during poor frame rates. It’s generally understood that people notice sudden hitches and slowdowns more than fast vs faster
The 1% low is probably what you really want to compare. It’s the delay (or effective ‘fps’) of the 1% worst frame delivery times. Take the benchmark, and what do the worst 1% of frames look like.
Sadly, no one’s is doing as well at Techreport was doing a decade ago. They had graphs letting you look at the frames with delays over the requirement to deliver steady 240, 120, 60, 30, etc Hz.
Atheist-Gods
1% lows is the 1% worst fps you get. So 99% of the time you will have at least that much fps. It’s a metric that tries to put a number on how bad your “stutters” are.
16 Comments
It’s just the average of the lowest fps dips. Ideally you want it as close to the average fps drawn
Let’s say you’re playing a game and averaging 100FPS, but it’s a little “stuttery” at times, and 1% of the time you’re actually at 40FPS because one or a few frames take longer than normal to render, your 1% low would be 40FPS. The more buttery-smooth your frame rate is, the closer your 1% lows are to your average FPS.
Your lowest FPS dips 1% of the time (there are 0.1% lows too).
But their functional purpose is to measure your frame rate consistency and micro stutters. Like if you are trying to play at a steady 60 FPS, your average FPS should be 60, and your 1% lows should also be 60 or at least extremely close (e.g 59fps, 58fps).
Imagine you’re playing at an average 60 fps. If your 1% lows are close to it, it means most of those frames are evenly spread across every second and it feels smooth. However, imagine you had 120 frames in the first half of the second, and zero for the second half of it. It would still be 60 fps in average, but It’d feel way more janky. That’s roughly what low 1% lows are but of course it’s way more irregular in reality.
Each second has separate frames like, 60,59,52,60… and so on. It will only pick lowest of them all and take its average.
The bigger the fps number, the better it is. That’s all you need to know.
It’s the worst 1% of your framerate for each frame. So, if it’s just a bit lower than your average, the game is butter smooth. If it’s in single digits – the game or your setup or both have stuttering issues and microfreezes.
Upd.: for extra clarification, it’s usually in triplet numbers: low 1%, average and max. The max shows the best your system could show in perfect conditions. Average is just average, pretty self-explanatory, and 1% as I just explained. The closer they are – the better the game is optimized, usually. It means that the load on your GPU is even and there’s no freezes or stuttering. The greater the difference between them the more troubling it might be. Except if benchmark was done badly and it shows you a clear sky where the max just jumps to hundreds because there’s literally nothing to render. But that’s not very reliable, as 1% low might be just at the very beginning, when resources are still loading dynamically, and it’s not really that bad or annoying, as the rest of the benchmark is smooth. Good benchmarks should be done to account all that, so, no scenes that differ too much in complexity, and long enough to minimize the effect of initial loading, if it happens.
All I know is the closer my 1% lows is to my average FPS the better. Having jarring frame drops is not fun to look at.
If you have low 1% average you will experience stutters and freezing.
Imagine having 100fps. 1% lows is the average of the lowest 1% of the frames. So if your lowest frame was 60 then 1% lows would be 60. You want it as close to the normal average as possible to prevent stuttering
The simplest way I heard it being explained by some Youtuber was that, whatever that 1% Low figure is, 99% of the time, your FPS is above that.
Example: If you play a game for 10,000 seconds, you record the FPS for each second that passes, and then pick out the lowest 100 samples (1% of 10,000). That is your 1% Low.
*(If this thinking is wrong, anyone feel free to correct. I don’t mind.)*
The human eye can not see over 60 FPS Lows.
/s
It’s deviation from the average. Average of the 1%/0.1% of the most slowly generated frames. You want it to be as close to the average of 100% of the frames as possible.
Imagine you have 60 frames in a second and half are spaced 33.2ms apart and half are 8.3ms apart. The time between each frame is not stable this can cause bad frame pacing or frametimes.
1% lows is like that but when you drop frames suddenly and drastically, like 100fps to 55fps. It is just a metric that shows where your rates drop and how deep the difference is between average and minimum essentially.
Basically the idea is to show numbers that can tell you how smooth or jumpy the experience can be.
Fuck I hate what happened to Techreport. DON’T go there, the good old articles had the useful images scrubbed and it’s advertising click bait now.
Editor was asked in a TV interview what it’s like to actually play games on a card, and he realized he couldn’t answer the question. So he actually took a step back and tried that.. only to find that average fps wasn’t revealing what he experienced.
He’s saw bad “micro stutters” in multi GPU configurations (common at the time and worse with AMD”. And stutter in general on all cards. (Spoiler: Nvidia and AMD did a ton of driver work to fix these issues.)
This lead to investigating “fps inside the second.” This came to with revolutionary benchmarks looking at the time delay to deliver the next frame. Basically, GPUs don’t deliver a static frame rate… It varies.
Keep in mind this predates dynamic Freesync and G-sync, so missing a frame at 60 Hz was a 16.7 ms stutter.
Short story long, lots of effort was put into the experience of the player during poor frame rates. It’s generally understood that people notice sudden hitches and slowdowns more than fast vs faster
The 1% low is probably what you really want to compare. It’s the delay (or effective ‘fps’) of the 1% worst frame delivery times. Take the benchmark, and what do the worst 1% of frames look like.
Sadly, no one’s is doing as well at Techreport was doing a decade ago. They had graphs letting you look at the frames with delays over the requirement to deliver steady 240, 120, 60, 30, etc Hz.
1% lows is the 1% worst fps you get. So 99% of the time you will have at least that much fps. It’s a metric that tries to put a number on how bad your “stutters” are.