PlayerUnknown's Battlegrounds Graphics Performance Revisited: 44 GPUs Tested

How did you manage 54fps on Ultra with the 3gb 1060? Mine is full of stuttering due to Vram issues, even in more modest settings.
 
How did you manage 54fps on Ultra with the 3gb 1060? Mine is full of stuttering due to Vram issues, even in more modest settings.
I've learned this from the crypto mining arena, but there are two major manufacturers of VRAM on 1060s. I know one is made by Samsung and I can't remember the other, but they behave very differently for whatever reason. The difference in performance could be attributed to the differences VRAM manufacturers
 
How did you manage 54fps on Ultra with the 3gb 1060? Mine is full of stuttering due to Vram issues, even in more modest settings.

The average can be high(ish) even with stuttering, but maybe the reason is not solely the VRAM. It's possible that the 5GHz i7-8700K with 3200MHz RAM is able to handle possible VRAM overflow so that stuttering is less severe than on your system (which I'm assuming is not quite as fast, since you're using a 3GB GTX 1060).
 
How did you manage 54fps on Ultra with the 3gb 1060? Mine is full of stuttering due to Vram issues, even in more modest settings.

How much system memory do you have? That does mask the issue, so possibly a bit misleading sorry.
 
PC gaming is doomed and this game proves it. We have over 10 year old games that look better than this and yet the performance is abysmal. Its not just PUBG its also a big issue in most Ubisoft games.

Do they seriously expect us to buy a 500$ GPU in 2018 just so we can play something that looks 10 years old in 1080p with 60 fps?

Then there's the bugs, the crashes, the netcode, the cheaters... Give the masses the one thing that they want and the rest can be abysmal.
 
How much system memory do you have? That does mask the issue, so possibly a bit misleading sorry.
16gb @ 3200mhz.

Just putting textures to ultra with everything also on very low stills stutters like crazy. Only with textures on medium the game is actually playable.

Ok, my bad, I haven't tested ultra settings at v.1. I just tried it, your averages seem okay. I got 55 to 70 fps mostly, but the thing is, when I get into a building the fps just nosedives. I mean, it goes as low as 24 for 5-10 seconds. I'm guessing that's when new textures need to be loaded. Also, besides the highish average, the game is unplayable like that. I guess the frametimes are all over the place, since there is serious microstuttering. Especially when I get caught in a crossfire.

So, conclusion is, this card is seriously gimped by the 3gb decision from Nvidia. It packs a bunch, even in Ultra.

PS1. But I tested on the old map, and for some reason it seems to require more Vram than the new one. Of course it also depends on where you land.
 
Last edited:
The average can be high(ish) even with stuttering, but maybe the reason is not solely the VRAM. It's possible that the 5GHz i7-8700K with 3200MHz RAM is able to handle possible VRAM overflow so that stuttering is less severe than on your system (which I'm assuming is not quite as fast, since you're using a 3GB GTX 1060).
It's an r5 1600 @ 4ghz. I know the 8700k is faster, but I've never heard CPU makes a difference when it comes to Vram issues.
 
I've learned this from the crypto mining arena, but there are two major manufacturers of VRAM on 1060s. I know one is made by Samsung and I can't remember the other, but they behave very differently for whatever reason. The difference in performance could be attributed to the differences VRAM manufacturers
The same is true for every video card. There is samsung / hynix / micron. They all run the same speed, the difference is overclocking. The samsung can handle up to 500mhz of OC, hynix and micron can do anything from 200 to 350, depending on your luck. My 1060 has Samsung, thank god for that, it's a beast when mining at +650 memory.
 
16gb @ 3200mhz.

Just putting textures to ultra with everything also on very low stills stutters like crazy. Only with textures on medium the game is actually playable.

Ok, my bad, I haven't tested ultra settings at v.1. I just tried it, your averages seem okay. I got 55 to 70 fps mostly, but the thing is, when I get into a building the fps just nosedives. I mean, it goes as low as 24 for 5-10 seconds. I'm guessing that's when new textures need to be loaded. Also, besides the highish average, the game is unplayable like that. I guess the frametimes are all over the place, since there is serious microstuttering. Especially when I get caught in a crossfire.

So, conclusion is, this card is seriously gimped by the 3gb decision from Nvidia. It packs a bunch, even in Ultra.

PS1. But I tested on the old map, and for some reason it seems to require more Vram than the new one. Of course it also depends on where you land.

Also are you using the latest GeForce driver?
 
Also are you using the latest GeForce driver?
Did you see my edit? Yes I'm using the latest drivers.

I can confirm that your fps metrics are accurate. But the game is still unplayable for me with these settings.

Although as I've said, I test on Erengel, in the area just above Georgopolis with lots of buildings.
 
AMD Radeon performance is still a bit weak it seems. Favours Nvidia.

I would have liked to see a few more setting tested. I don't think many people play this game on ultra on the basis the performance is so shonky you need to give yourself headroom for worst case scenario, unless you have king kong of a gaming PC.

Even then you likely aren't at all ultra because it's not as if you bought a 1080ti for 1080p, and you need at least 60FPS on this game! Only 1080ti looks just about good enough at 1440p and ultra. Madness.

I would say some high/medium setting tests are in order. Try all medium at 4K, or all high at 1080/1440. I would be quite interested in seeing how much the settings affect performance.
 
"Fellow game developers must be scratching their heads wondering how such a half-baked game can be so unbelievably popular."

For some unbelievably stupid reason people keep defending PUBG saying that they don't care that it looks a decade old and performs worse than some of the best looking modern games.
From the server to the client, everything is so badly optimised that it ruins the gaming experience.
 
Can you clarify how exactly it was tested: did you replay the same game recording, or was it a new live match each time?
 
"Fellow game developers must be scratching their heads wondering how such a half-baked game can be so unbelievably popular."

For some unbelievably stupid reason people keep defending PUBG saying that they don't care that it looks a decade old and performs worse than some of the best looking modern games.
From the server to the client, everything is so badly optimised that it ruins the gaming experience.
I'm not defending PUBG here, I want to make that clear, What I'm doing though is explaining an issue the whole industry has.

PUBG is fun to play, entertaining and has sold so well (despite all the issues) because it's new, came up with a good idea for a game mode and isn't the easiest thing in the world, it doesn't have a corridor you walk down, it doesn't have a massive arrow you just follow, you can kill anyone from any angle, it's pretty awesome compared to 99% of what most AAA developers come up with these days.

That's why so many people bought it and defend it, not because it's the most polished game, because everything else is the same boring corridor with some bad guys you fire at or is made so you can barely ever lose or die. And games that are decent (Battlefront 2 is actually a good game) is ruined by Loot Boxes and Pay-to-Win schemes.
 
Having never played PUBG, I'm curious as to what exactly is "new" about this particular game mode. We've already had "large" player maps with large numbers of players (don't even know if they were the first or not, but I know for sure that BF4 has them), as well as "open" maps where you're not just running down corridors or fighting in small rooms (again, not only BF4 but BF3 for sure have maps like that; heck, come to think of it, that describes a lot of the Halo:CE maps from back in 2003). And "every man for himself" deathmatch-style games have also been around for a long time (again, see Halo:CE from 14 years ago, & that's even assuming they were the first ones to have that...but I seem to recall at the very least Quake 2 having that 20+ years ago).

And if the reason is, "oh, this is a battle royale-style game where the safe zones get smaller as the game progresses"...um...isn't that the identical description of H1Z1 (aka H1Z1: King of the Hill), which Brendan Greene helped to develop? Again, reusing a concept & making cosmetic gameplay changes doesn't qualify as "new".
 
Having never played PUBG, I'm curious as to what exactly is "new" about this particular game mode.
Nothing, but they focused on it and no one has seen a game before where this is the only game mode available.
We've already had "large" player maps with large numbers of players (don't even know if they were the first or not, but I know for sure that BF4 has them), as well as "open" maps where you're not just running down corridors or fighting in small rooms (again, not only BF4 but BF3 for sure have maps like that; heck, come to think of it, that describes a lot of the Halo:CE maps from back in 2003).
The two maps in PUBG completely destroy the size of any Battlefield Map and Halo Maps I wouldn't even consider big even back then. And again, all the games you mentioned are old games, Last year, there really wasn't many good games.
And "every man for himself" deathmatch-style games have also been around for a long time (again, see Halo:CE from 14 years ago, & that's even assuming they were the first ones to have that...but I seem to recall at the very least Quake 2 having that 20+ years ago).
Not against 100 other players but the mode you're right, again, nothing decent has come out for a while that made this game mode work.
And if the reason is, "oh, this is a battle royale-style game where the safe zones get smaller as the game progresses"...um...isn't that the identical description of H1Z1 (aka H1Z1: King of the Hill), which Brendan Greene helped to develop? Again, reusing a concept & making cosmetic gameplay changes doesn't qualify as "new".
True, it's just been done better this time round.
 
I won't spend a coin on such a poor game, nobody should. Playing on pc should mean a great experience, specially on the technical side.
 
"Fellow game developers must be scratching their heads wondering how such a half-baked game can be so unbelievably popular."

For some unbelievably stupid reason people keep defending PUBG saying that they don't care that it looks a decade old and performs worse than some of the best looking modern games.
From the server to the client, everything is so badly optimised that it ruins the gaming experience.
I'm not defending PUBG here, I want to make that clear, What I'm doing though is explaining an issue the whole industry has.

PUBG is fun to play, entertaining and has sold so well (despite all the issues) because it's new, came up with a good idea for a game mode and isn't the easiest thing in the world, it doesn't have a corridor you walk down, it doesn't have a massive arrow you just follow, you can kill anyone from any angle, it's pretty awesome compared to 99% of what most AAA developers come up with these days.

That's why so many people bought it and defend it, not because it's the most polished game, because everything else is the same boring corridor with some bad guys you fire at or is made so you can barely ever lose or die. And games that are decent (Battlefront 2 is actually a good game) is ruined by Loot Boxes and Pay-to-Win schemes.
PUBG is also being ruined by loot crates. Players cheating to get crates then selling items for real money.
I wouldn't say most defend this game, most do NOT. While it may be fun at TIMES it isn't fun when being team killed or being killed by a cheater just when you got everything going for you, here comes the aimbotters n no recoil cheats.
People are simply getting tired of the Chinese flooding ALL servers, wont communicate and as we all already know, they cheat.
 
It's an r5 1600 @ 4ghz. I know the 8700k is faster, but I've never heard CPU makes a difference when it comes to Vram issues.

Like I said, I suspected the issue was not solely due to VRAM. When VRAM fills up and stuff needs to redirected to system memory (or worse, the page file) and back, which of course doesn't happen "for free". Now I'm not knowledgeable enough to say how that extra data transfer is handled exactly - as it likely depends on the OS, the game engine, the CPU architecture etc. - but I would assume a system with more computing resources at its disposal can handle the extra load better than a system with less computing resources, which in turn could lead to worse stuttering on the latter system. I could be wrong. In any case, I'm glad an explanation was found for the discrepancy between the benchmark results and what you had observed.
 
PUBG is also being ruined by loot crates. Players cheating to get crates then selling items for real money.
That's the exact same issue Counter Strike has, wouldn't call that game ruining. Unable to unlock the second map until it dropped in a loot crate? That's an issue, people selling items for real money is the same issue most loot crate based system's have.
I wouldn't say most defend this game, most do NOT. While it may be fun at TIMES it isn't fun when being team killed or being killed by a cheater just when you got everything going for you, here comes the aimbotters n no recoil cheats.
If no one defended this game, it wouldn't have nearly got game of the year would it? Cheating can be an issue, granted, I cannot say anything here since I haven't played it in so long. Being killed by team mates though? Why would you play PUBG in a squad of 4 randomers?! You just go at it alone or get some friends online if you wanted to work as a team. Randomers almost never work.
People are simply getting tired of the Chinese flooding ALL servers, wont communicate and as we all already know, they cheat.
Can't comment on this, but if this is currently an issue, it's rectifiable by the developers. They just need to invest in more Chinese servers and they can "cheat" each other to death.
You are right, Fortnite's Battle Royale is indeed really well done :D
Have you actually played Fortnite? Don't get me wrong, definitely polished but it's, for the lack of a better word, Soulless. Something's missing, or isn't right, I don't know if it's the art style, the Physics or what but the Battle Royale Style gameplay in Fortnite isn't anywhere near as intense, not even close. Maybe it's how often you come across med-kits and shields? I don't know.

P.S.
H1Z1 (aka H1Z1: King of the Hill), which Brendan Greene helped to develop? Again, reusing a concept & making cosmetic gameplay changes doesn't qualify as "new".
H1Z1 is on sale here in the UK on Steam, Was going to get it for £5, until I saw the reviews, From the looks of it, the game simply isn't close to PUBG, the devs started taking ideas from PUBG and have been rubbish at updating the game, I'll keep my £5 I think.
 
Last edited:
Back