FX-6300 and 7850 performance
|
FX-6300 and 7850 performance
|
|
05-06-2013, 04:10 AM
Thanks for the replies. I won't base what performance I could get on this alone, but this video shows SMG running on a PC with lower specs than I mentioned just fine: http://www.youtube.com/watch?v=VHDgl5kubxI
If I can get the extra money, I will upgrade to the 3570k, but the chances of me getting the money don't seem very high.  (05-06-2013, 04:10 AM)Sato Wrote: Thanks for the replies. I won't base what performance I could get on this alone, but this video shows SMG running on a PC with lower specs than I mentioned just fine: http://www.youtube.com/watch?v=VHDgl5kubxI That video is not a very good point of reference. He is using a very old revision which runs faster but is less accurate and has more bugs. HLE audio is being used which is why the music cuts out quickly, also is using skip EFB from CPU which whilst speeds up the game you cannot complete certain sections of the game without this enabled (e.g pull stars). 05-06-2013, 10:05 AM
@lor is totally correct
SMG requires i5 3570k @ 4.2GHz , no less Laptop: (Show Spoiler)
Mini PC :: (Show Spoiler)
05-06-2013, 10:53 AM
(This post was last modified: 05-06-2013, 11:04 AM by NaturalViolence.)
 pauldacheez Wrote:Yes, yes, it uses multiple threads, but "single-threaded performance" matters to Dolphin because if the most demanding thread can't run fast enough, the other threads have to slow down. This "lightly-threaded" thing makes sense now that you've explained it, though. (Still applies better to the app's description, though, since nobody really benchmarks two-thread performance on quad-cores.) I started using "lightly threaded" instead of "single threaded" awhile back because like Thorhian I find it to be a better description of the type of workload the application is doing. I don't really care if benchmarks don't usually talk about it because it's still a more accurate term. There are plenty of applications that do depend on single-threaded performance (as in a dual core cpu will provide no speedup) and plenty of applications that will gain some additional performance from additional cores but in a very limited way that are incorrectly labeled "single-threaded" in benchmarks. I have thus decided to call the latter category "lightly threaded" to differentiate it from the former, and dolphin definitely falls into this category. One of the reasons why this bothered me enough for me to start using that term was the recent release of bulldozer/piledriver which had people all across the internet saying it has "good multi-threaded performance and bad single-threaded performance". You see it everywhere but it's not true at all. The vast majority of multi-threaded apps can only use 2-3 cores tops and don't have anywhere close to a linear performance increase from the extra cores. Thus they perform just as bad as single-threaded apps do compared to sandy bridge based cpus. You can see how differentiating between the two becomes important in this scenario. I also like the term that nvidia has coined for certain types of apps, "laughably parallel (LP)". Describing applications whose performance can scale perfectly or near perfectly with core count up to some enormous number of cores. These tend to be applications that simulate some type of real world model (3D graphics, physics, chemical interactions, etc.) or are related to video/image processing in some way. These are generally the types of applications where piledriver is competitive with sandy/ivy bridge. But even with this type of workload they still trade blows depending on the exact application instead of one being clearly better than the other. Not to mention SB/IB do it with lower power consumption and an on die IGP running. Anything that doesn't fall into this category (the vast majority of current applications) will without question run better on a SB/IB cpu. Which is why AMD had to price the FX series cpus so much lower to remain competitive. I haven't started using laughably parallel yet because nobody would have any idea wtf I was talking about if I did so I just use "heavily multi-threaded", which in my opinion is a bad term. Of course this entire model breaks down when you have to compare dual core cpus against quad core cpus since "lightly threaded" is not descriptive enough.
"Normally if given a choice between doing something and nothing, I’d choose to do nothing. But I would do something if it helps someone else do nothing. I’d work all night if it meant nothing got done."
-Ron Swanson "I shall be a good politician, even if it kills me. Or if it kills anyone else for that matter. " -Mark Antony (05-06-2013, 06:58 AM)lor Wrote:(05-06-2013, 04:10 AM)Sato Wrote: Thanks for the replies. I won't base what performance I could get on this alone, but this video shows SMG running on a PC with lower specs than I mentioned just fine: http://www.youtube.com/watch?v=VHDgl5kubxI Darn... Oh well, thanks for the help. (05-06-2013, 06:58 AM)lor Wrote: That video is not a very good point of reference. He is using a very old revision which runs faster but is less accurate and has more bugs. HLE audio is being used which is why the music cuts out quickly, also is using skip EFB from CPU which whilst speeds up the game you cannot complete certain sections of the game without this enabled (e.g pull stars). And what will that particular version do to Xenoblade? Do you happen to know? Anyone? FX 6300 user here with a GeForce GTX 660. If there's a previous version that will work better with Xenoblade with my hardware, please advise. I would overclock but I am using stock cooling. I tend to notice performance is good with the game but then the audio starts to break up and performance gets all out of whack. I may not have it configured properly. (06-24-2013, 02:04 PM)niccolo83 Wrote: If there's a previous version that will work better with Xenoblade with my hardware, please advise. I would overclock but I am using stock cooling. If you overclock your CPU , you should use latest Dolphin version If not , check post #126 in this thread for settings and Dolphin version Note that : Use PAL version (25FPS is fullspeed) without 30FPS patch and disable "PAL60" option in Dolphin settings will help you reach fullspeed easier than using NTSC 30 or PAL with 30FPS patch So if you had NTSC-US/JAP version of the game , you should trade with someone for PAL version Laptop: (Show Spoiler)
Mini PC :: (Show Spoiler)
(06-24-2013, 02:12 PM)admin89 Wrote:(06-24-2013, 02:04 PM)niccolo83 Wrote: If there's a previous version that will work better with Xenoblade with my hardware, please advise. I would overclock but I am using stock cooling. As I said I cannot OC it past 3.5GHz much (maybe 3.7) because I am using stock cooling and then it gets >60C which shortens the FX CPU lifespan. At stock, the CPU does have 4.1GHz turbo but it probably does nothing for Dolphin. There has to be more optimizations coming for Dolphin that makes these games run faster. I will check the thread, thank you Quote:it gets >60CFull load with Prime95 ? Don't know about FX series TJ max but for Intel : 60C full load is good temp 70C full load is normal 80C full load is high but this doesn't affect CPU lifespan much >85C is overheating Most Intel CPUs have 100C TJ max (still operate at 90-99C) , some CPUs have 105C TJ max Laptop: (Show Spoiler)
Mini PC :: (Show Spoiler)
It is disputed... some people say 62W, others 71W...
The 6300 is a 95W processor however. 6350 is 125W and actually does not OC as well. btw how much would I have to OC it? edit: I believe 62.5C is the max safe temp for this CPU. Tjmax info is incorrect. http://forums.amd.com/game/messageview.cfm?catid=446&threadid=160921 Yes, note this thread. An AMD Tech Director is quoted as saying 61C . |
|
« Next Oldest | Next Newest »
|
