• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Diablo 3 - GPU and CPU Performance

I know we're keeping a thread alive that doesn't belong, but...

I don't mind the look. D2 looked as it did because everything was super low resolution. Using modern rendering it wouldn't have that same look.
 
"With no in-game benchmark tool available, we had to devise one of our own. The best way to get reliable and repeatable performance was to run around the edge of Old Tristram, from the portal and back again, and record the frame rate with Fraps."

All this tests proves is that in Diablo 3 you can run in a circle, by yourself, on nearly any hardware. Even then, at 1920x1080 with highest detail, the 192-bit GTX 460 will drop below 60 fps. At 2560x1600 you would need a 7870 or GTX 580 to stay above 60 fps.

While running in a circle. Not fighting. By yourself.

Surely the game is relatively easy to run, but this test is hugely misleading. Once you add in four players, spells casting, the physics being put into action, with tons of mobs on the screen, you will be glad you spent some extra money on your hardware.
 
I ran a bunch of test with the beta. An Intel HD3000 with everything turned to minimum at 1080P (Including the decrease particle effects button) runs ok most of the time. Drops frames in battles. At 768P is actually runs fine, no issues at all (everything on low), however the beta did not have any really big battles as it was only on normal difficulty in the first act.

Them doing a test of running in a circle by themselves is a joke though. Wait till hell difficulty with 4 players, its going to take a lot more hardware to runt hat smoothly.
 
Even then, at 1920x1080 with highest detail, the 192-bit GTX 460 will drop below 60 fps.

This is a relatively slow moving hack and slash game, not a FPS or a racing game. Most people don't need 60 fps minimum in this style of game. You can spend $500-1000 on GPUs if you want that, but I would imagine for this type of game 60 fps min is not necessary for a smooth Diablo 3 experience for the majority of people.

At 2560x1600 you would need a 7870 or GTX 580 to stay above 60 fps.

It's highly unlikely that gamers with 2560x1600 monitors aren't already using HD7970/GTX680 or similar. Only 0.17% of Steam PC gamers have a 2560x1600 monitor. It's a safe bet most gamers using such a monitor are running the latest and greatest or something like GTX570 SLI/HD6950 CF, etc.

Surely the game is relatively easy to run, but this test is hugely misleading. Once you add in four players, spells casting, the physics being put into action, with tons of mobs on the screen, you will be glad you spent some extra money on your hardware.

What physics? There are hardly any physics effects in this game. Also, Blizzard wouldn't make a game that needed a $500 GPU for the full gaming experience. I mean sure if some gamer out there wants 120 fps minimum framerate with 3D vision for their 120hz Monitor, by all means they can spend $2000+ on Quad-SLI GTX690s. That's not what Blizzard games are about and its reflective in the graphics which are fairly basic. Even looking at Low to High details just adds shadows and a bit more detail on the ground.

HD6870 should be plenty fast for 1080P:

diablo%203%201920x1080.png


Either way before people go out and start wasting $ on hardware, I would wait for more benchmarks and feedback from gamers based on the final game, not on the beta....
 
Last edited:
I'd agree that in static scenarios it is fast. 2 points:

1) This is blizzard, they like having a wide audience. I don't have an issue with that as long as the gameplay is good.

2) This is also a heavy multiplayer game. When the action is heavy in MP with lots of spells flying around, it can bog down lower setups - WoW is similar, if you are in a 25 man raid with lots of chaos going on even high end systems can slow down a bit. Most of that stuff is CPU limited (multiplayer blizzard titles)
 
Just means I'm going to turn on SSAA.

It also means those Llanos and IB chips can play this game a pretty good FPS.
 
This is a relatively slow moving hack and slash game, not a FPS or a racing game. Most people don't need 60 fps minimum in this style of game. You can spend $500-1000 on GPUs if you want that, but I would imagine for this type of game 60 fps min is not necessary for a smooth Diablo 3 experience for the majority of people. ....

Agreed. What concerns me though, is if you like to play with V-sync to avoid tearing and you buy a card based on these charts, you are likely to be looking at 30fps, or less, in combat. I'm unsure that 30fps is good enough.


It's highly unlikely that gamers with 2560x1600 monitors aren't already using HD7970/GTX680 or similar. Only 0.17% of Steam PC gamers have a 2560x1600 monitor. It's a safe bet most gamers using such a monitor are running the latest and greatest or something like GTX570 SLI/HD6950 CF, etc.

You got me there. I play at 2560x1600 and just upgraded to a GTX 680 from CF 6950. I have seen so many comments that this game will run maxed out on nearly any hardware, that I wanted to expose the fallacy of those comments. Barring optimizations, a high-end GTX 580 can barely do so outside of combat according to this test.


What physics? There are hardly any physics effects in this game. Also, Blizzard wouldn't make a game that needed a $500 GPU for the full gaming experience. I mean sure if some gamer out there wants 120 fps minimum framerate with 3D vision for their 120hz Monitor, by all means they can spend $2000+ on Quad-SLI GTX690s. That's not what Blizzard games are about and its reflective in the graphics which are fairly basic. Even looking at Low to High details just adds shadows and a bit more detail on the ground.

Yes the physics effects are minimal, but they are there. The way this test was structured, they put the lightest load possible on the CPU. The conclusion reached in this test was that a 2.0ghz Phenom 2 would be fine; that is unlikely to be the case once multiplayer and actually playing the game is factored in.


Either way before people go out and start wasting $ on hardware, I would wait for more benchmarks and feedback from gamers based on the final game, not on the beta....

Agreed. I hope that the tests on the retail client are actually representative of gameplay, not a quick lap around Old Tristram.
 
Last edited:
What physics? There are hardly any physics effects in this game. Also, Blizzard wouldn't make a game that needed a $500 GPU for the full gaming experience. I mean sure if some gamer out there wants 120 fps minimum framerate with 3D vision for their 120hz Monitor, by all means they can spend $2000+ on Quad-SLI GTX690s. That's not what Blizzard games are about and its reflective in the graphics which are fairly basic. Even looking at Low to High details just adds shadows and a bit more detail on the ground.

There is not a lot of physics, but there is some. The single largest thing that causes FPS drop (Outside of loading new areas) if the particle effects. IE: Spells and such. You get all those lighting and particle effects going and things DO slow down.
 
I'm mostly concerned about it not taking advantage of more than 2 cores...Starcraft 2 also runs fine at the beginning, but the more stuff that happens later on, the more you can see the Phenom IIs and Core 2 and other old/weak CPUs getting behind
 
I'm mostly concerned about it not taking advantage of more than 2 cores...Starcraft 2 also runs fine at the beginning, but the more stuff that happens later on, the more you can see the Phenom IIs and Core 2 and other old/weak CPUs getting behind

I never saw it use more than 2 cores on any of my machines.
 
"With no in-game benchmark tool available, we had to devise one of our own. The best way to get reliable and repeatable performance was to run around the edge of Old Tristram, from the portal and back again, and record the frame rate with Fraps."

All this tests proves is that in Diablo 3 you can run in a circle, by yourself, on nearly any hardware. Even then, at 1920x1080 with highest detail, the 192-bit GTX 460 will drop below 60 fps. At 2560x1600 you would need a 7870 or GTX 580 to stay above 60 fps.

While running in a circle. Not fighting. By yourself.

Surely the game is relatively easy to run, but this test is hugely misleading. Once you add in four players, spells casting, the physics being put into action, with tons of mobs on the screen, you will be glad you spent some extra money on your hardware.


Yep, you are right. 4 players with max level spells spamming constantly and hordes of enemies on screen also firing off tons of particle effects and suddenly 60fps becomes 30fps.

There's other options to enable too if you got spare performance like ambient occlusion, high end AA, etc.
 
This game has a lot of stuttering on a variety of systems no matter how high end. Hopefully it got fixed in the final. Also got my new Ivy to hopefully smooth it out lol.
 
You cant judge from a beta anyways. New games always have multiple patches later on that improve everything so, expect it to be the same with this game. I am sure it will take a few months and many patches to iron out and improve performance.
 
"With no in-game benchmark tool available, we had to devise one of our own. The best way to get reliable and repeatable performance was to run around the edge of Old Tristram, from the portal and back again, and record the frame rate with Fraps."

All this tests proves is that in Diablo 3 you can run in a circle, by yourself, on nearly any hardware. Even then, at 1920x1080 with highest detail, the 192-bit GTX 460 will drop below 60 fps. At 2560x1600 you would need a 7870 or GTX 580 to stay above 60 fps.

While running in a circle. Not fighting. By yourself.

Surely the game is relatively easy to run, but this test is hugely misleading. Once you add in four players, spells casting, the physics being put into action, with tons of mobs on the screen, you will be glad you spent some extra money on your hardware.

This. Running around solo in beta my old 5850 at 1080p max everything barely hit 50% gpu load and it was 60fps vsync. Got into a game with a few people and went dungeon crawling with mobs exploding left and right and spells going off, it got up to 99% load at some points.

Also note that SC2 beta was relatively easy on the gpu at max settings, in retail, max (retail included ultra settings) was much more demanding.
 
Benchmarks don't really capture the hell can break lose on screen with 4 characters spamming their abilities and a screen full of monsters in act 3 heading to siegebreaker with all the video settings to max.

No way your mom's HP can handle that 🙂.

That said, yeah, the game itself was/is a disappointment. The initial playthroughs and tinkering with the classes is fun enough, but the endgame is a joke.

Ah well, here's hoping we get a pleasant surprise from TL2, GW2, Grim Dawn, WoW: MoP, etc.
 
I'm kinda pissed it looks so cartoony - D2 looked dark and Gothic. I have to play it first before I can say for sure - one week from today.

D3 is a great game, however seeing characters running around dressed like harry potter definitely kills the whole " you're in hell " vibe.
 
gtx 470, I7 860, 8GB ram. 7200 rpm hard drive, all settings maxed, and at 1080p.

I have no issues. I've been running around in Hell with groups in the Sin Heart area with and without V-sync on. I see no difference.
 
D3 is a great game, however seeing characters running around dressed like harry potter definitely kills the whole " you're in hell " vibe.

I have to disagree with that. Walking around in a place where everyone dresses and acts like they are in Harry Potter sounds EXACTALLY like hell to me...
 
Back
Top