EPTThis? http://www.techspot.com/review/305-starcraft2-performance/page13.html It's dated but it still shows that there's a major difference in clock speed.
Is i5 2400 enough for Sc2? - Page 4
| Forum Index > Tech Support |
|
skyR
Canada13817 Posts
This? http://www.techspot.com/review/305-starcraft2-performance/page13.html It's dated but it still shows that there's a major difference in clock speed. | ||
|
Shikyo
Finland33997 Posts
For instance this bench is only for 1024x768 and still gets low FPS, indicating its CPU-limited, which it is. When we go up to 1080p it takes much more from the GPU but the CPU requirements stay about the same. That's why for instance in Metro 2033 in 1440p it's almost completely GPU-limited until you get to a tri-SLi 580 setup or something like that. Notice also that those aren't with max settings, furthering the CPU-limitedness. They have to make the settings CPU-related so that they can compare the CPUs. Now would you happen to have CPU-benches for say, 1080p Crysis with Enthusiast settings? I'll bet it's completely GPU-capped. On September 25 2011 08:46 skyR wrote: For Starcraft II? No, it's still CPU limited even at 1080p. I don't see how this is strange since mine at high 1080p drops to ~30... This? http://www.techspot.com/review/305-starcraft2-performance/page13.html It's dated but it still shows that there's a major difference in clock speed. I wasn't talking about SC2, I said the majority. SC2 is one of the rare games that are CPU-capped, didn't I say that?... (And yes that bench as well is with an overkill GPU) EDIT: Though yeah if this is solely for SC2 you definitely want to OC the 2500k, the difference is going to be 10-20 frames depending on the settings, though you can just leave the advanced physics off. I guess it's about what you're willing to pay for it. | ||
|
skyR
Canada13817 Posts
No one (or very few) bench CPUs at 1080p since ya... well most games will be GPU limited before CPU. | ||
|
Shikyo
Finland33997 Posts
EDIT: Hold the phone... If the computer indeed is strictly for Starcraft 2, why are you getting a 560 Ti twinfrozr??? Get something like the 100$ GTX 460 from NCIX. You'll be CPU-limited anyway and you save 150$. | ||
|
Heinstar
Australia61 Posts
On September 25 2011 08:56 Shikyo wrote: By the way thank you for the discussion, always learn a lot, hope I wasn't too obnoxious~ EDIT: Hold the phone... If the computer indeed is strictly for Starcraft 2, why are you getting a 560 Ti twinfrozr??? Get something like the 100$ GTX 460 from NCIX. You'll be CPU-limited anyway and you save 150$. 1.) Because 560 Ti seems to be the good value GPU at the moment just like HD 6950. Although not as important, I'm also going to try other games like Skyrim, Dragon Age II, GW 2, Tera, WoW and Diablo III. 2.) I have HD 5770 which is currently used by my work PC. I could give this GPU to my gaming PC but I don't think this GPU is powerful enough to run above games at 1920x1080 is it? 3.) Even if it is, I'd need to buy another weaker GPU for my work PC if I was to do that. I'd need to buy decent GPU that can handle dual-monitor and 3d modeling software. On September 25 2011 08:19 Shikyo wrote: That ram = no just no. Also you probably should get 8gb instead. 1. 2-3 in most games 2. It tends to be GPU-limited so should be the same in most games, not SC2 though. 3. Depends on what you want to achieve. Paying 149$ for a filco when AU has leopolds for 109 is pretty silly imo. 4. It shouldn't go that low with either setup but you really don't even need advanced physics or whatever CPU-intensive settings there are. 1.) Yeah but I heard that Leopold has worse key caps and feels mushier to press compared to Filco so I'm going with Filco even if it's $40 more. To me it's worth the extra $40 since it looks better and has better quality. On September 25 2011 08:22 skyR wrote: This depends on how far you overclock your 2500k to. It is definitely not a 2-3 fps difference as Shikyo said. And I have no clue where you pulled the number 18 from. And you're wasting money on 1600MHz cas7 memory: http://www.xbitlabs.com/articles/memory/display/sandy-bridge-ddr3.html http://www.anandtech.com/show/4503/sandy-bridge-memory-scaling-choosing-the-best-ddr3 http://techreport.com/articles.x/20377 If you read my above post, you can see I estimated it: + Show Spoiler + On September 22 2011 23:54 Wabbit wrote: Generally speaking, 30 FPS is a widely accepted "minimum" for which most games are playable. I've experienced 20ish on my old laptop in big battles and didn't like it. Aside from making the gameplay feel choppy, lower FPS introduces additional delays. 60 FPS = 1 frame every 16.6ms, 30 FPS = 1 frame every 33.2 (assuming linear frame times, an unlikely oversimplification but sufficient for understanding). While it's not likely you can notice an extra 10-20ms delay on top of all the other delays (especially from having to play via BNet, I think it's like 200+ ms no matter what; there was a thread recently, somebody analyzed this), I personally don't mind making improvements wherever possible, small as they may be. This site shows visual difference between 15fps vs 30fps vs 60fps: http://www.boallen.com/fps-compare.html It looks like 30fps is acceptable but 60fps is still slightly better and you would never want to go 15fps. As for whether 2500K is worth it or not, when I look at this benchmark for Sc2: http://www.xbitlabs.com/articles/cpu/display/core-i7-2600k-990x_9.html#sect0 1. It shows that 2500K at stock speed(3.3GHz) can get 51.4fps in 1680x1050 resolution. 2. It gets additional 18fps when it's OC to maximum of 4.7GHz in 1680x1050 resolution. 3. 2400(3.1Ghz) will be slightly worse than 2500K at stock speed(3.3GHz) so it would get like 50.7fps in 1680x1050 resolution. So by paying extra $117 for OC 2500K, I'll probably get additional 20fps in Sc2. On September 22 2011 00:12 Wabbit wrote: @Heinstar: As you can see in this link that skyR also posted, SC2 performance scales pretty linearly with clock speed (at least for these Sandy Bridge CPU's). This is an oversimplification, but if you do a little math, it scales almost linearly: + Show Spoiler + As a personal experience: + Show Spoiler + And If you're getting 40fps in 200v200 1v1 at 3.6GHz then 2400(3.1GHz) will probably get ~35fps in 200v200 1v1. Which is still visually acceptable. So if I get a 2400, in the worst case scenario I'll still get a acceptable game play in 1v1 200/200 at 30~35fps. But if I get OC 2500K, even in the worst case scenario I'll always get the smoothest game play in 1v1 200/200 at +55fps. Is it worth it? I'm still not sure. 1.) So how much is really the gain in FPS? is it more than 18FPS? 2.) So is the test showing that tight timing and 1600MHz will barely increase FPS? If so then should I save money and just get the cheapest 2x2GB RAM for $29 instead? | ||
|
Heinstar
Australia61 Posts
1.) If it turns out I prefer 1 PC for both gaming and studying instead of having separate Gaming Rig and study PC, then I can just upgrade my study PC. Worst case scenario I'll only lose $59 on Antec three-hundered. 2.) Sc2 is a very CPU intensive game and 2500K is the best one for it right now. There is hardly any FPS difference between 2500K and 2600K in Sc2. 3.) Z68 Mobo will last a long time since DDR4 won't arrive until 2013 - 15 4.) If it turns out I want to use the components for my study PC then the Z68 Mobo will allow me to SLI to get triple monitor. So I just need to know which RAM I should go for? Can you choose a good priced one that actually gives performance difference and is worth the money?: http://www.pccasegear.com/index.php?main_page=index&cPath=186_538_913&vk_sort=4 2x2GB vs 2x4GB? 1333 vs 1600 vs 1800 vs 2000? 9-9-9-24 vs 8-8-8-24 vs 7-7-7-24? | ||
|
Phaded
Australia579 Posts
| ||
|
jaj22
United Kingdom1376 Posts
On September 26 2011 12:29 Heinstar wrote: So I just need to know which RAM I should go for? Best and only SC2 RAM benchmark: http://www.teamliquid.net/blogs/viewblog.php?topic_id=245087 That's a triple-channel system, so RAM speed should be even more important for a 2500K unless Intel screwed up somewhere. | ||
|
JingleHell
United States11308 Posts
| ||
|
mahnini
United States6862 Posts
| ||
|
CatNzHat
United States1599 Posts
| ||
|
nalgene
Canada2153 Posts
also... take apart your entire case before adding anything to it ( if it's your first time building one as you'll see ) and rout the cables to the back if possible ( holes should all be made for most modern cases, at least some of the coolermaster ones do ? ) if not, just get a cutter/care should be taken and make some holes ( this would improve airflow and allows you to hide cables away until you need them to pop out at specific spots to connect to the motherboard ) also helps if the psu is modular ( modifiable/able to add/remove cable ) although not necessary | ||
|
Heinstar
Australia61 Posts
On September 26 2011 14:24 nalgene wrote: You can get to 5.2 with a loop / 4.7 is more doable / and 4.5 on heatsinks/fans on both the i7-2600k and i5-2500k ( not all chips are made the same and some can't hit those numbers regardless of what you do ) I heard that water coolers are no better than air cooling. In fact, air cooling will be cheaper to do especially in the long run because you don't have to replace the liquid. Do you think it's worth spending money on better Case and better CPU cooler than CM Hyper 212+ to overclock the 2500K to 4.7GHz? Or leave it at 4.5GHz since there's no guarantee I'll get to 4.7GHz ? I saw JingleHell's RAM test. So getting faster MHz is more important than lower timing right? So I should probably go for CL9 1600MHz 2x2GB/2x4GB RAM then. I don't think I can get CL6 from PCCG anyway. | ||
|
nalgene
Canada2153 Posts
On September 26 2011 16:04 Heinstar wrote: I heard that water coolers are no better than air cooling. In fact, air cooling will be cheaper to do especially in the long run because you don't have to replace the liquid. Do you think it's worth spending money on better Case and better CPU cooler than CM Hyper 212+ to overclock the 2500K to 4.7GHz? Or leave it at 4.5GHz since there's no guarantee I'll get to 4.7GHz ? I saw JingleHell's RAM test. So getting faster MHz is more important than lower timing right? So I should probably go for CL9 1600MHz 2x2GB/2x4GB RAM then. I don't think I can get CL6 from PCCG anyway. Uh.... it's not necessary, but those are already done for you...? You can still drill the holes for the antec 300... ( most modern ones come with the holes already made though, that's the only difference, but there's nothing stopping you from just drilling your own holes for cable management ) //unrelated water is always better than a heatsink/fan(s) if cost isn't added in ( but the latter is obviously cheaper ) | ||
Womwomwom
5930 Posts
Watercooling is completely pointless these days unless its a hobby - CPU coolers like the Thermalright Archon, Silver Arrow, and Noctua D14 can reach 5Ghz overclocks if you really want to and they do it for cheaper while producing far less noise. Of course, spending so much on cooling is completely stupid since overclocking that extra .3 Ghz gets you so little performance benefit. If you want to think about overclocking in a pragmatic way, that 0.3ghz is not going to make a game that isn't unplayable, playable. The only decent reason for spending more than $40 on a CPU heatsink is to decrease noise. Anyway...get the cheapest RAM you can find because all RAM is basically the same in performance/quality. Reread JingleHell's RAM tests, the conclusion is that neither gives any real form of performance benefit. Sure, its likely not a perfectly done test but its enough to prove that 1600mhz vs 1333mhz gives you like 1% extra performance at best. Of course use some common sense here...if a 1600mhz kit is like $3 more than a 1333mhz kit, you might as well spend extra on the 1600mhz kit. Also, ditch the Antec 300. For $60, its absolutely not worth the price. It has a bottom mounted power supply but it has no cable management features; it has a lot of airflow but has absolutely no dust filters (unless you stretch some pantyhose over the intakes); it does bareback hard drive mounting so its going to sound like shit if you have any mechanical hard drives in there. The only good thing it still has compared to the competition is 0.8mm steel construction. | ||
|
JingleHell
United States11308 Posts
Basically, the more effect a component has on gaming ability, the more important it is to put part of your budget into it. Meaning, if you aren't buying the best of everything, RAM is the last thing to spend extra on for performance gains. | ||
|
nalgene
Canada2153 Posts
this one says 59.99 and 49.99 after rebate ( but he's in australia ??? ) has all the features he'd most likely use ( holes are already made for routing cables to the back and then out to motherboard/other components ) the one in the OP is 58.0 //1200/5-5-5-15 can beat 1600/cas7 ( these aren't really noticeable either ) | ||
|
SoKHo
Korea (South)1081 Posts
| ||
|
PonyWA
19 Posts
| ||
|
JingleHell
United States11308 Posts
On September 27 2011 02:03 nalgene wrote: http://www.newegg.com/Product/Product.aspx?Item=N82E16811119233 this one says 59.99 and 49.99 after rebate ( but he's in australia ??? ) has all the features he'd most likely use ( holes are already made for routing cables to the back and then out to motherboard/other components ) the one in the OP is 58.0 //1200/5-5-5-15 can beat 1600/cas7 ( these aren't really noticeable either ) Uhm, timings and clock on RAM are pretty variable for different pieces of software, different CPUs, different clocks, and different IMC frequencies. You can't just generalize like that. And seeing as my tests pretty much show that in SC2, timings make so little difference it can be chalked up to human error in testing, and clock makes a good bit more, at least with recent generation Intel, I'm just not going to believe you until you at least explain how you came to that conclusion. | ||
| ||
