Diablo III Performance, Benchmarked

Page 5 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Status
Not open for further replies.

cleeve

Illustrious


Reynod, you're the man!

But I restored them. I can take it, I've got a thick skin. The comments are valid, even if they're ridden with typical internet forum unkindness. :)

------------------------------------------------------------------------------

As for you gents who suggested the bench is useless:

I did try to bench the game during large fights. As Chris mentioned, the game has no movie or save function in a battle, so the results were completely useless for benchmarking purposes because they weren't repeatable.

When i did test large battles, I didn't notice a significant frame rate drop when using the overclocked Core i7. In fact, the frame rate drop was higher in Old tristram than in some large battles as this area seems harder to render.

in each benchmark run I reloaded the level, so as I took the bench a lot of baddies were chasing me and on the map. i wasn't running around with nothing there. On top of that, I had a hireling with me who engaged multiple enemies as I ran around.

After a lot of tests, I think this is the best I could do to get repeatable results in a game with no save feature. If you have a better method we can certainly discuss it and we might even try it in a follow up.

Just please keep the tone light and positive. There's no need to put anyone down. :)
 

griever989

Honorable
May 7, 2012
1
0
10,510
Hey guys, I noticed that some of you said it didn't work with multiple monitors, but here's a screenshot I took in the beta: http://i.imgur.com/ab1KR.jpg

I didn't do anything special, I literally just installed the game and jumped right in and it worked the first time.
 

pita

Distinguished
Oct 28, 2009
23
0
18,510
[citation][nom]Cleeve[/nom]If you have a better method we can certainly discuss it and we might even try it in a follow up.[/citation]

It puzzles me that you guys even bothered with beta as a credible source of hardware benchmarking. Not because the graphics engine is unfinished or anything, but in full games u have a whole range of other monsters casting different spells and attacks, as well as the higher level players casting higher level spells with more dramatic effects, not to mention at higher difficulties u might have more monsters around on the screen (not sure about that). The problem is not the methodology you implemented but the fact that Beta barely scratches the surface of the entire game and all those elements may affect the performance further, making the entire data published feeling a bit incomplete.

At first I thought you guys had early access to the game and was excited about seeing some more realistic benchmark numbers. I mean, this is a reputable site, I bet you guys could have afforded to hold off until the full game is released. Or at least make a statement saying these benchmarks were done in beta and things may be a bit different when playing the rest of the game.
 

cleeve

Illustrious


We worked directly with Blizzard on this. They are aware of our test method and had no reservations.

The beta is just as demanding as the final. There's no magical thing that happens the two weeks between the end of beta and launch. There may be levels slightly harder on the frame rate, there may be levels a bit easier on the frame rate, too. But this is a valid representation of what a person will experience when they play the final game, in singleplayer mode at least. I don't think we ever said multi-player mode would get the same results, but it's a given that networks add all sorts of variables that a graphics card and CPU have little to do with.

I'm curious why you would think the final game will produce such a drastic difference in frame rates when it's half a month from the end of beta. Blizzard has been polishing this thing for months, they're not a second-rate developer.


 

cbag

Distinguished
Dec 11, 2011
306
0
18,810
[citation][nom]gnookergi[/nom]I still use a 9600GT, anyone have any idea how it would work with that?[/citation]

I used a 7950gt w/ a x2 5000+ cpu. It was ok but I had to lower the resolution to 1360x768
 

pita

Distinguished
Oct 28, 2009
23
0
18,510
[citation][nom]Cleeve[/nom] I'm curious why you would think the final game will produce such a drastic difference in frame rates when it's half a month from the end of beta. Blizzard has been polishing this thing for months, they're not a second-rate developer.[/citation]

It is not so much about how much blizzard could have done to modify the game from beta to final that concerned me, but the fact that beta only provided limited variety of monsters and level designs as well as the spells cast from the main characters and monsters. What if say inferno difficulty will have increased number of monsters? It made me wonder if those will impact the performance and by how much, in particularly for the mid/low end of the spectrum. Multilayer is a big part of the game, performance hit by having all four players on one screen casting heavy spells will be of interest for everyone as well. In any case, you mentioned you worked closely with Blizzard on this, and so I will take your words for it and leave this topic alone.
 

youssef 2010

Distinguished
Jan 1, 2009
1,263
0
19,360
I never played the original Diablo. I played Diablo II but I was too young to understand it. It's been more than 10 years since the last Diablo release. And while this sequel seems like a good game. I can't help but feel a bit disappointed about the graphics. I was expecting something in the league of Dragon Age at least. Oh well, guess we can't have everything.
 
G

Guest

Guest
You can zoom in with the 'Z' key man... Why do all reviewers leave this out? In the first place, do they even know if it's zoomable?!
 

mchn

Distinguished
Aug 4, 2011
48
0
18,530
[citation][nom]gnookergi[/nom]I still use a 9600GT, anyone have any idea how it would work with that?[/citation]
I ran the open beta with that card's 512MB version at 1920x1080 and lowest detail. It ran OK.
 

Thrin

Honorable
May 5, 2012
39
0
10,560
The main reason why I believe there is a difference between running around in town and being out in the field and encountering events is because of how people complained and experienced slow downs in those specific environments (being the situations whereby there is a low of action going on and multiple other players).

My suspicion is that the performance degradation stems from CPU or maybe even RAM issues as opposed to graphics cards since the majority of posts I saw in regards to this experience had "good enough" GPUs which is also furthered demonstrated by this article.

I think a good portion of how Diablo 3 plays is based off of the CPU and not just graphics. As such, to test for performance one needs to make certain that other variables are isolated and tested in situations where they are actively engaged. In town or nearby there aren't a lot of physics engine collisions / activities / events going on.
 
[citation][nom]Anonymous[/nom]so can anyone recommend a good laptop that can play this no lag for under 800?[/citation]Sign up and start a thread and send me a PM.

The Asus A6-3400M w/ dedicated Radeon 6650M I recently bought for $356 would beast through this game. The point being, you can easily find something for $550 that gets it done, such as some of HP's A6's w/ dedicated 7470M's (PLEASE, pay the $25 to upgrade to a 7690M).

Note the the Dells have DDR3 where the HP's have GDDR5 for most laptop dedicated graphics.
 

Scotty99

Distinguished
Aug 21, 2010
434
0
18,810
How is the 6770 ahead of the 6850 at some points? Also, how is a 6770 beating a gtx 460.

IM gonna wait for a few more benchmarks before i build my cousin a rig.
 

silverblue

Distinguished
Jul 22, 2009
1,199
4
19,285
[citation][nom]scotty99[/nom]How is the 6770 ahead of the 6850 at some points? Also, how is a 6770 beating a gtx 460.IM gonna wait for a few more benchmarks before i build my cousin a rig.[/citation]
Probably the faster core clock offsetting the reduced execution hardware (850MHz vs. 775MHz). D3 doesn't look particularly heavy on such things.
 

MegaTheJohny

Honorable
May 8, 2012
40
0
10,530
you have to test FPS rate with 4 players in party, with max details and spam some spells. Than you will see FPS drop till 20. (I have I7 2.7 Ghz, 6GB RAM, ATI Raedon HD 5900 series.)
 
G

Guest

Guest
One of the things missing from this article is graphics during concurrent game play. From the looks, they were all taken with a single player and a limited skill set due to character level. From personal experience, with three to four people in the game spamming a number of abilities (even more so in a boss fight or event), I saw some serious dips in fps. Running Q9450 @ 3.67GHz w/ 6950 xFire was enough for me, but I dipped into the 40 range frequently.
 

spleenbegone

Distinguished
Feb 15, 2012
37
0
18,530
Things must have changed a bit since the open beta, my crossfired 6850's were averaging around 185-190 fps at 1920x1080 according to FRAPS. They have a slight overclock on them, but nothing that would make up a 40fps difference.
 

superfula

Distinguished
Dec 30, 2011
11
0
18,510
[citation][nom]rantoc[/nom]Early beta wasn't optimized at all, bet the latest beta version is as close as you get from the final but without a doubt blizzard will optimize it further in patches[/citation]

No part of the beta was optimized. Blizzard has said as much.
 
Status
Not open for further replies.