Well ofc not, I don't have the hardware to do that. I did say I'm assuming.
I was doing affinity testing today, and the game showed lack of performance for each thread I reduced. It's an assumption based on testing I did today.
Maybe it wont, but that's why I've been looking at 2~4 core performance in this thread.
Round and round we go...........I know it was an assumption. I said that, and you followed it up with a post that alluded to you doing testing and it not being an assumption (how I read it). But it had to be an assumption because you could not test with the number of cores!
Round and round we go...........I know it was an assumption. I said that, and you followed it up with a post that alluded to you doing testing and it not being an assumption (how I read it). Anyway, move on.
Erm yeah, testing on my own current hardware...
Kinda at least have to assume that I'm asking about certain hardware that i might purchase that I might not exactly own the hardware to test beyond what I have. Savvy?
Even though Wildstar distrubutes its load over a lot of cores, there is always one core which will be close to 100% that houses the main thread of the game.
This is actually incorrect, the game doesn't behave like this at all now. Perhaps it did over a year ago in the beta client, but it doesn't act like this at all now.
The CPU usage hovers around 19~20% across all cores with one thread hovering aroud 50~60%, and then the rest 10~20ish %, in the city of thayd.
I'm literally walking around in game right now testing this.
For some reason the game seems to put its most usage on the 4th thread and the least usage on the 5th and 6th(Numbering Reference: CPU 0~7 [8 threads])
The info you linked is outdated so here's some current data/test.
Ok, so one core is showing significantly more use than another. That still lends credence to the underlying message (it runs the main thread off one core) of that link, no? If it was truly multicore, wouldn't the loads be much more similar across all threads instead of similar on most except for one showing around double the use of the others?
I mean, we can argue semantics all day long, but looking at the forest through the trees, I still see his(my links) point in your up to date testing (one core is stressed a lot more than the others).
The CPU usage hovers around 19~20% across all cores with one thread hovering aroud 50~60%, and then the rest 10~20ish %, in the city of thayd.
I'm literally walking around in game right now testing this.
Ok, so one core is showing significantly more use than another. That still lends credence to the underlying message (it runs the main thread off one core) of that link, no? If it was truly multicore, wouldn't the loads be much more similar across all threads instead of similar on most except for one showing around double the use of the others?
I mean, we can argue semantics all day long, but looking at the forest through the trees, I still see his(my links) point in your up to date testing (one core is stressed a lot more than the others).
Most stressful location I could find atm, bumped the CPU load to 29%
With a bit more testing, I found that when I limit the game to just 5 threads, my turbo reaches up to 4.15ghz, instead of staying at 4.05ghz.
WildStar seems to slightly drop performance if I limit it to 4 cores, and drops significantly more if I limit it to 3 cores.
So from the looks of it, 5 cores is producing the best results on my CPU, so anymore than 5 threads seems to be useless for the game.
OFC this means that the assumption that the i7 5820K would increase any noticable performance with 12 threads is wrong.
However both the i7 5820K and 4790K are haswell archetechture. So now begs the question, how much of an improvement is Devils Canyon and/or Haswell-E over Ivybridge (fcPGA 988 85Watt vs LGA 1150 X Watts)?
How much of an improvement in performance is haswell vs ivy?
Edit: this thread should probably be moved to the intel subform now.
5% improvement+ the extra power with a good overclock.
Since the 3740QM is fcPGA988, and the Performance difference between an fcPGA988 Ivybridge(QM), and an LGA 1155 Ivybridge(3770K) comes out to about 7~9%, clock for clock(4ghz), either CPU choice is going to be a significant performance step up.
Sounds like a good deal to me.
Thanks for the info everyone.
I just did some research on the 4790's vs. 5820's. As far as gaming goes, it looks like the 4790 is superior ATM, I only found one game where the 5820 seemed to be superior. If you're using a lot of threads and processes the 5820 is better. Of course, if you're really using that many cores, maybe you should consider 8 cores instead of 6.
Ironically, I've only see one gaming benchmark that absolutely blasted all the cores on my 5820: the Call of Pripyat benchmark.
Have you given a thought to the second hand market? You could certainly grab an i5 IvyBridge (3570k) platform for $200 or so...
Same as Mandrake here: I've used both platforms until very recently (last Intel was a 4790K, sold last week), and there is a noticeable difference on CPU intensive titles, at the Intel advantage.
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.