StarCraft II Revisited: How Much Gaming PC Do You Need?

Page 3 - 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
[citation][nom]oxxfatelostxxo[/nom]This doesnt really make much sense to me... i have to be missing something...[/citation]

It would seem logical to deduce that what you're missing is the variable between this machine and the test machine, most likely the benchmark.

You may also be making assumptions as to the frame rate of your machine during large battles. Did you record FRAPS runs? Or do you just guesstimate the 50 fps min figure?

Those are probably the things you're missing. :)
 

cleeve

Illustrious
[citation][nom]mresseguie[/nom]I'm curious now how much improvement may be had by running double graphics cards. Too bad that option wasn't included.[/citation]

Well, there's not really much point to running CrossFire or SLI when the title is so blatantly limited by CPU speed.
 

superflykicks03

Distinguished
Sep 9, 2010
56
0
18,640
Great article! I can confirm the CPU recommendations: In my experience, the CPU is one of the biggest bottlenecks for this game for my rig.

Athlon II X3 @2.8 = lag on ultra settings 1920x1080 no AA on large vs AI battles.

Athlon II X3 @ 3.4 = smooth on ultra settings 1920x1080 - have not had a slowdown in my experience.

Video is a Radeon 5850@ stock, 4GB of DDR3@1600
 

Conner Macleod

Distinguished
Feb 19, 2010
33
0
18,530
[citation][nom]hixbot[/nom]I'm not sure why I was voted down. I do play this game on my Pentium 4 2.8ghz with 3gb ddr 266 ram, and a HD3850 AGP.Minimum requirements are just a P4 2.6ghz with a 9800pro.I'm stating a fact, not an opinion.[/citation]

Probably because you didn't provide what detail setting you have it on, and saying it runs fine is too vague to really mean anything. I voted you up though to balance it out;)

I also play with a P4 but it's a 3.4 GHz, 8800 GT 512 MB, 4 GB SD RAM dual-channel 667 MHz, RAID 0 320 GB 7200 RPM Hard drives running at 1280x1024 with medium detail setting and it's slow. Surprisingly I can run Crysis on high with an average of 30+ fps, but SC2 puts Crysis to shame. I know I need to upgrade but I'm waiting on prices to drop for SSD and for faster video cards to come out, I might upgrade next year.
 

HKH

Distinguished
Dec 15, 2009
63
0
18,630
I use:
i7 930 4ghz, 480GTX, Ultra + 4xAA
gets at least 60+ even in the most intense situations
clocking it at 3.06ghz isn't really a good comparison
since SC2 bursts out more fps when clocked higher
3.06ghz and a 5870 getting 40ish fps isn't good enough
clocking it to 4ghz could probably boost it by alot

And overclocking a 920/930 to 3.8-4ghz is relatively easy with a decent cooler
 
G

Guest

Guest
There has to be something wrong when looking at the compared 460/470 cards. If that 460 carried the same % overclock as the 700MHz GigaByte 470, they would've pretty much equaled.

460 > 470 for StarCraft 2? -.-
 

Killua

Distinguished
Jun 29, 2010
12
0
18,510
I have built 5 computers to run starcraft 2 specifically and these benchmark results has much lower results then what I have obtained. I only needed a Athlon 2 x4 640 with a 5770 to run the game at a constant 60 frames ultra settings no AA 1680x1050 (no overclock on all parts).

The way I benchmark my computers is by playing marine madness (8 players with over 500 units each on tiny map, no doodads/fog/etc) which to my opinion is a much more realistic benchmark then what Tom's hardware has done. Most custom maps and regular maps has large amounts of unit battles on small to large sized maps. Almost NO ONE plays on the biggest sized map available with lots of fog/doodads/etc. Map size with doodads affects computer performance. Based on the pictures you provided, there was not alot of units compared to what custom maps can do, yet in my opinion I would say the doodads were causing your abysmal performance.

You really only need about $750 computer to play this game on max settings, 4x AA on 1920x1080 with an emphasis on CPU (to deal with stuff happening OFF screen because this is where most of the lag will come from).
 

cleeve

Illustrious
[citation][nom]Voids[/nom]There has to be something wrong when looking at the compared 460/470 cards. If that 460 carried the same % overclock as the 700MHz GigaByte 470, they would've pretty much equaled.460 > 470 for StarCraft 2? -.-[/citation]

They would have equaled. The game is CPU-limited, it doesn't matter what kind of card you have as long as it's enough to hit the CPU bottleneck.
 

cleeve

Illustrious
[citation][nom]killua[/nom]yet in my opinion I would say the doodads were causing your abysmal performance.[/citation]

Once the battle was over, the framerate leaps to 100+ fps on faster graphics cards.

The doodads remain on the map, yet performance becomes very fast when there are no AI battles to conduct. Therefore the doodads are not to blame.

If you like, I will ask if I am allowed to share the benchmark. Not sure what kind of issues there would be but I will look into it.
 
G

Guest

Guest
So you're saying that with an i7 at higher speeds (4.0-4.4), a GTX 470 would pull ahead again?
 

cleeve

Illustrious
[citation][nom]Voids2[/nom]So you're saying that with an i7 at higher speeds (4.0-4.4), a GTX 470 would pull ahead again?[/citation]

I'm saying that with the CPU bottlenecked removed, the performance bottleneck would move to the graphics cards and yes, the GTX470 would perform better than the 460.

You can see this when 4xAA is enabled. The Graphics cards become the bottleneck, and at 24560x1600 the GTX470 performs about 25% faster that the GTX460.

Also, in the test setup page, note that all of the graphics cards were clocked down to reference levels for these tests.
 

Killua

Distinguished
Jun 29, 2010
12
0
18,510
I would like to have the custom map to benchmark my computer so I can post results. I could be wrong with how I view SC2 performance but Id like to test it with what tom's hardware did.

I get the frame readings from SC2's internal FPS reading...CTRL + ALT + F. The average frame is just my approximation but I never get lag spikes. Also this review doesn't include the full system specifications. RAM and CPU is a biggest bottleneck and you can get by with a mid range graphics card.
 

Killua

Distinguished
Jun 29, 2010
12
0
18,510
My bad I somehow missed the hardware specs page. I would still like to somehow try your custom map benchmark.
 

cleeve

Illustrious


I've got lots of requests for the benchmark. I'm looking into the possibility of releasing the bench for public use.

I'll update you guys here either way. :)
 

jazz84

Distinguished
Mar 24, 2010
80
0
18,630
Just another frustrating example of how programmers are too darn lazy to take advantage of present-day multiprocessing. When will these people get it through their heads that they're not even coming close to taking advantage of the current hardware? I suppose RTM deadlines will always trounce putting out a top-quality product...

And yes, clearly the game isn't flat-out unplayable for the majority of people out there, but it's just a shame that it could be SO much better. In contrast to my previous statement, in this case it's not as though Blizzard was under any major pressure to release this quickly, so it's all the more frustrating that they simply couldn't be bothered to properly optimize their engine.

 

RazberyBandit

Distinguished
Dec 25, 2008
2,303
0
19,960
This was a great comparison, and nice example of why developers should work to make their games more multi-threaded.

One question: Whatcha gonna do with all those Gigabyte video cards, hmm? :D
 

oxxfatelostxxo

Distinguished
Aug 17, 2007
157
0
18,710
You may also be making assumptions as to the frame rate of your machine during large battles. Did you record FRAPS runs? Or do you just guesstimate the 50 fps min figure?

Done with Fraps, with myself and 2 computer AI running. during large battles, it would drop from about 63fps, to around 54, rarely dipping lower than 50
 

sarwar_r87

Distinguished
Mar 28, 2008
837
0
19,060
something seems off in these benchies

my 5770 on PII x4 @ 3GHz can hit 44fps on average with 27fps as minimum. resolution is 1680x1050

i don get y do these benchies show only 14fps with 5770 at 1680x1050 :S
 

cleeve

Illustrious



When it 'rarely' dipped below 50, how low did it go? It's called the 'minimum' frame rate, not the 'usual low' frame rate...

In any case, you answered your original question. The difference is clearly the benchmark, as there are four AI players with large forces as described in the article. Your setup is not the same. ;)

 

cleeve

Illustrious



The answer is simple:
Your benchmark method isn't as stressful as ours.

Without using the same map result comparisons are futile.
 

Tmanishere

Distinguished
Jun 23, 2009
198
0
18,690
[citation][nom]Conner Macleod[/nom]Probably because you didn't provide what detail setting you have it on, and saying it runs fine is too vague to really mean anything. I voted you up though to balance it outI also play with a P4 but it's a 3.4 GHz, 8800 GT 512 MB, 4 GB SD RAM dual-channel 667 MHz, RAID 0 320 GB 7200 RPM Hard drives running at 1280x1024 with medium detail setting and it's slow. Surprisingly I can run Crysis on high with an average of 30+ fps, but SC2 puts Crysis to shame. I know I need to upgrade but I'm waiting on prices to drop for SSD and for faster video cards to come out, I might upgrade next year.[/citation]


I'm with you on this one. I haven't tried SC2 (I still dabble SC1 every now and then), but it's very helpful to see what people experience with older computers. I myself, have Athlon64 3400+ ATI3850 at home and don't plan on upgrading until early next year. I think that'll be be best time to upgrade with Win7 SP1, Portal 2, Diablo 3, and Valve's "surprise" all hitting stores around that time. Not to mention the SSD getting cheaper by the minute.
 
Status
Not open for further replies.