StarCraft II: Wings of Liberty GPU & CPU Performance

Great article, this is what brought me to Techspot! However, it is getting dated pretty quickly and I was hoping we could get an update on this for a couple of reasons

1) Catalyst 10.9 claims to have fixed CrossFireX scaling; ("ATI CrossFireX™ now functions properly, and in-game Anti-Aliasing can now be enabled in “StarCraft II”")

2) The new addition of the Radeon HD 6800 series and how it matches up against the GTX 460 (I myself moved from 5770CF to GTX460 solely because of Starcraft II)

3) I recall Blizzard saying the single player campaign uses higher quality models, and therefore taxes the graphics card more than multiplayer. Although this can change as the amount of units/players increase in multiplayer. In that regard, some custom games are impossible to play on Ultra, no matter how good of (commercial) rig you own.

Thanks Techspot!
 
when this game is running at optimal speed its beautiful and any one that wants to get the best out of there machine and this game should look at doing the upgrade it will be worth it!
 
I don't know how he's getting such an increase in performance when overclocking an i5 750 to 3.7Ghz. I overclocked mine to 3.645 and haven't seen any increase on a GTX 460 1GB. Does anyone know how he's doing it?
 
Its called a GPU bottleneck, they tested with a GTX 480 and that is how they did it.
 
But I thought anything beyond a GTX 460 is unnecessary in this game.
 
hi im about to buy a nvidia 9800 and i have a amd phemon x4 underclocked to 1.8ghz now i read that it has a huge front load bus 3200 so if i remove the underclock and over clock a lil :) do u think it would be ok to use? im on a tight budget for now so if not im boned lol. also to note 4 gigs of ram and as i said gonna slap a nvidia 9800 in it as the comp currently has an onboard graphics card. PLZ get back to me asap becasue if its not gonna work then i dont wanna buy the card lol ... PC model is HP a6700y :) thanks again lookin foward to response
 
Great article and much appreciated! Is it possible to replicate this same test for your humble readers? I have a Gateway FX6840-03E that I am looking to upgrade the graphics card and overclock the processor, so I'd like to see if I can replicate the results/gains. As with any test, I'm sure your mileage may vary, but this looks like a great baseline.
 
Starcraft 2 is more about CPU than GPU.

I have a 5 cpu lan one is a single core A64, three are dual cores and one is a I7 920 OC to 3.5ghz

Video cards are
=============
ATI 5770
Nvidia GTX 460 Fermi
ATI 5870
ATI 4670

They all play the game well on a basic map like a 1on1 except the single core. Sc2 need two cores, that computer plays it but have to really tweak it and no 4 on 4s or it just lags.

But I can take the fastest video card the Ati 5870 and put it in one of the dual cores and put the 4670 or 5770 in the I7 and the I7 with the lessor video card will play the game better.

Its not how fast the game goes, its HOW SLOW IT GETS.

I play a 3 on 3 mod called Desert Strike. Anyone that plays it and makes it to round 81 knows that all the onscreen action will bring the strongest computer to its knees.

All the dual cores will get a slide show not matter how powerful the graphics chip.
Dual Cores = 1fps (yes 1 frame per second or less.)

I7 920 = 8-14 fps even with a 5770 or Nvidia 460

I took the graphics settings down and it really did not make any difference on the I7 computer.

So take this review with a grain of salt. Starcraft II will play decently with many graphics cards you really need a I7 to get the most out of it in mods with thousands of troops.

Its probably one of the few games where I would say keep your graphics card and upgrade your cpu.

This review focused on the wrong thing.

You can play the game at 20fps and its ok, but when you hit 1FPS you are basically locked and have to scroll the screen away from the action so you do not freeze.

A side note:

People like to ALT-Tab during the game and it will lag you or give a time out. Also when you alt-tab back into the game it will take like 10 seconds.

Go into options and set the screen to "Windowed"
If you do this you can alt-tab back and forth an it will be super fast. You can also use the print screen function which seems to be messed up in true full screen mode.

Bliz said that windowed mode might be slower. I did not notice that but it does sort of mess up the mouse position sometimes.

A torture test for your computer is Desert Strike play it a few times to level 81 overtime and hold your mouse over the "menu" drop down to see the FPS. If you get over 2 your doing good.

Most dual cores running at 3ghz or more will play it fine with just about any modern video card.

i7 is the only system of the 5 that is actually playable ie. You can see the action and scroll around. I look at the other cpus and the screen look frozen until enough of the enemies die off to raise the frame rate.
 
This review focused on the wrong thing.

You can play the game at 20fps and its ok, but when you hit 1FPS you are basically locked and have to scroll the screen away from the action so you do not freeze.

Yeah and you failed to read page 13 :S
 
Great article! Just clearing up one thing though. Bizarrely, blizzard decided to show *more* terrain in 16:9 vs 16:10 or 4:3. See the following thread for more, and screenshots: http://www.teamliquid.net/forum/viewmessage.php?topic_id=112931

I'm basically speccing a system around this game, that's how much I like it, and I'll definitely be going for 16:9 resolution. Very helpful article tho, I think I'll be getting an i7 27" iMac of all things -- with a 5750 card (not actually a 5750 as advertised, but that's a different topic). I'm hoping for around 30 fps, running win 7.
 
Thanks for a great review, credible information on hardware requirements for the newer Blizzard offerings hasn't been all that easy to come by. Think the minimum requirements for this game on the Blizzard site are a bit understated.
 
Why does this benchmark seem soooo wrong?

I know that the 4890 > 4870 > GTX 260 for fact. These are stock values, and the 4890 is a great overclocker. OC'ed or NOT, the 4890 should be getting better FPS results, but in the benchmark the 260 GTX is noticeably higher. Why is this the case??
 
Why does this benchmark seem soooo wrong?

I know that the 4890 > 4870 > GTX 260 for fact. These are stock values, and the 4890 is a great overclocker. OC'ed or NOT, the 4890 should be getting better FPS results, but in the benchmark the 260 GTX is noticeably higher. Why is this the case??

Its as well known fact that Nvidia had better support for SC2 in the early days.

Also where are you getting your info from? The Radeon HD 4890 is far from superior to the GTX 260. Of course it depends on the game...

GTX 260 much faster than HD 4970 is Back Ops.
https://www.techspot.com/review/336-cod-black-ops-performance/page5.html

GTX 260 much faster than HD 4970 is MOH.
https://www.techspot.com/review/324-medal-of-honor-performance/page5.html

GTX 260 much faster than HD 4970 is Mafia II.
https://www.techspot.com/review/312-mafia2-performance/page5.html
 
Hi, I'm building a new PC and was wondering if the new I7 2600 3.4 Ghz with an ATI HD 5870 and a Soundblaster titanium soundcard would be a good choice.

Thanks.
 
To add on to the above I want to run Ultra on 1900x1200 with about 60 fps.
 
blizzard removed the lan setting from sc2 becouse it ware held inhouse tournaments in Korea and other countries without advertising for blizzard and they got pissed so its gone...sad but true
 
Actually I do WAY better than what's advertised here for a AMD Phenom II 965 BE @ stock 3.4 ghz (mine is step C2) combined with a EVGA GTX570 SC, around 75-120 FPS, on extreme preset in full HD 1920x1080 on a 27" monitor (close enough to a 1920-1200 monitor).
All runing at default clocks (I do not overclock)
I'm sorry to put it that way, but it is obviously another fake test that seems to give the advantage to Intel. We are talking video games here, not video encoding.
 
To answer the guy who ask if a sandy bridge processor would do well with a HD5870 and a titanium sound card would do good:

I would say yes, but if you want more performance, I would buy a higher end videocard like mine, a GTX570 (GTX560ti would be good too). Still, a HD5870 still rocks with today's games.
 
Actually I do WAY better than what's advertised here for a AMD Phenom II 965 BE @ stock 3.4 ghz (mine is step C2) combined with a EVGA GTX570 SC, around 75-120 FPS, on extreme preset in full HD 1920x1080 on a 27" monitor (close enough to a 1920-1200 monitor).
All runing at default clocks (I do not overclock)
I'm sorry to put it that way, but it is obviously another fake test that seems to give the advantage to Intel. We are talking video games here, not video encoding.

Yes it is obviously a fake test designed to favor Intel, you got us again.

First of all the testing in this article was done using a much earlier patch, there have been about a dozen since then. It is possible AMD processors perform better now, I am not sure I would have to re-test.

Finally I am not sure that you are not testing under the same conditions as we did, this would have a massive impact on the CPU performance since we used four AI players.
 
Back