which video card can handle Starcraft II reasonably fine?

I'm thinking to buy a new notebook, but I don't want a gaming notebook. which video card (ati/nVidia) can handle Starcraft II with medium quality?

if only gaming notebooks have the power to run SCII in medium quality so I prefer an integrate one and not playing SCII...

PS: how bad is ati 6630 or nvs4200m runing SCII? which is better?

 

Starcraft 2 at medium settings with a resolution of 1366x768.

HD 6630M ~ 50 FPS
AMD Radeon HD 6630M - Notebookcheck.net Tech

NVS 4200M ~ 30FPS
NVIDIA NVS 4200M - Notebookcheck.net Tech

 

We still need a SC2 benchmark that we can standardize.

Just a custom map with a preset AI action & camera that does a time demo.

With how incredibly random SC2 is in normal play its absolutely impossible to determine how good/bad the fps ratings are from another user and even less from a site in general that you cant ask how they determined the numbers.

I could roll @ 60fps easy with my desktop and saw it drop to 2FPS when I had a mothership cloak my mass army once :/ it can drop to 10fps with no special effects in a mass unit game like evolves.

So ideally the time demo would sweep over a nice high unit count base for all 3 races and include all the variables like a mothership cloaking some stuff, zerg creep, particle effects, etc.

 

2 vs 2 with 200 army supply each can be hard for most cards that can handle it on high otherwise. Just get best gpu that you can afford — 95% of modern gpus are able to run SC2, the only difference is settings.

Also ye its hard to just list all the gpus that can run SC2. can you be more specific what options you are looking at?

 

That type of bench would result in artificially low framerates compared to in-game rates in a standard game. Most games have relatively low unit counts and no motherships. You might want a set of time-demos that can be run individually.

 

2v2s easily get to late game like 50% of the time.
As for "standard", well most "new" players while playing "campaign" mode will try to mass 150-200 food army before attacking *just to be safe*, so thats rather heavy load.

Also its not like you will be like "oh its getting to late game, gotto reduce my setting else in the key moment of the battle i may be looking at slide show". So you need to factor the extremes, its an online game — your framerate must be good ALL the time or you loose.
I am playing on medium switching to medium-low when i do 2v2, even though my system can easily get 50fps on high most of the time.

 

It does not mean it will be artificially low, it just depends on how many units you put on there.

A average army size of 200ish seems about right (100 vs 100 actually quite easy I have seen 200v200v200v200 on more than one occasion). No matter if its hard or not it wont matter what matters if having a standardized place for comparison.

Think of the Crysis benchmark, on Very High people were in the 7-10fps, that was fine, just as long as you know thats what to expect and when you bench a different gpu have a place for comparison.

It would be best to make it "hard" on purpose so you know it stressed the gpu properly and wont become cpu limited. This is why you have to use 3dmark Vantage and cant use 3dmark 06 for benchmarks these days because 3dmark06 is cpu bound on any moderately strong computer giving falsified gpu scores.

You can of course make a bench based on cpu (tons of AI commands) and another for GPU (tons of eyecandy & effects) to break it down better.

In current testing is using 2 identical laptops with the same settings in game, same hardware, same drivers, same everything; One person could post 30fps and the next 50fps average just based on what they were looking at and doing while benching. Benchmarks are useless completely useless without standardization and a point of comparison.

I could tell you my computer scores 5000 on 3DMark 2012, you would not know if that is good or bad if you didnt have other scores to compare it with and know what kind of score to expect.

Having random numbers from a non-standardized testing procedure is just as useless.

 

artificially low compared to expected normal in-game frame rates in standard 1v1 ladder matches*

SC2 is relatively diverse and I agree there is really no way to handle the fact that your normal 1v1 competition game tends to end at about 200-300 total supply (sometimes less, and can certainly go to 400) - whereas larger games with multiple players can easily hit 800 supply or more, and custom games can go arbitrarily high and bring down ANY computer.

And, so you are right, you can design the benchmark as a stress test and use it for hardware comparison.

I would personally also like to see a different type of benchmark, that gives you a direct idea of expected performance without requiring comparison in the standard 1v1 game. ie- run this benchmark and get a direct idea of how well sc2 will run on your system.

The rationale behind my philosophy is that plenty of benchmarks like 3dmark already exist, we don't really need GPU or CPU comparison-style stress tests. It would be great to be able to run something to get a sense of how well sc2 will run on a particular set of hardware directly. just my 2 cents.

I had already built in a compromise, it would be cool to have a bench that runs through different scenes and gives you separate information for each. Different scenes could give you the ability to try to represent 1v1 games and larger format games. Of course, I'm not sure how to set up a timedemo in SC2 aside from doing a replay, so it's sort of a moot point. I guess we could build a custom game, but I'm not sure if there is a framerate api available, I sort of doubt it.

 

so... would 6630m enough for a decent quality 2v2 game?

 

I think so. I tried it on a HD5650M and it runs really smooth in 720p in High.
notebookcheck:
HD6630M= 47.8 FPS in High Details.

 

2v2 can hit 800 supply, so it depends on how you play.

the reality of the situation is that you start to need a ridiculous computer to handle ridiculous amounts of units. a 6630 would probably be OK in 1v1, and most 2v2 games, unless you are getting into situations where everyone is capped out on supply at 800 units. Probably low/medium settings. The game looks fine as long as shading is set to medium.

 

Youre already talking about CPU performance. Supreme commander cant run with 8KI´s and a 50km x 50km map on a i7 Hexacore @ 3.33 Ghz
after 15 Minutes fluently. That has nothing todo with GPU Performance anymore.

 

I misread the question. 6630m should be fine.

 

6630 should give rather comfortable 1v1 on strong medium settings
2v2 — might need to turn it a bit towards lower end but still i dont see a huge problem if you got a good deal with this card.

 

You'll also probably need a good CPU for starcraft 2 at those numbers of units...probably a 3.0Ghz+ Dual core or a Quad core

Mr. Mysterious