(03-03-2017, 02:42 AM)DarkHacker Wrote: [ -> ]No one expected it to be better than Kabylake on dolphin. I'm a realist. I simply expect that AMD could be a good option to run modern emulators (like Pcsx2 and Dolphin).
I imagine that for this purpose, Zen 4c\4t and 4c\8t are the best options, for surely will have highest clocks.
Supposedly coming out in Q2 this year.
(03-09-2017, 04:36 AM)likeike12 Wrote: [ -> ]Supposedly coming out in Q2 this year.
That's for the 6-core variant(s) (and
maybe a single 4core/8thread variant).
Lower variants aren't until the second half of 2017, which is the same release window for Raven Ridge (Zen-based APUs).
I'm trying REALLY REALLY hard to catch up on playing stuff and testing it out. I sadly work a pretty crappy job having nothing to do with tech so between that and other responsabilties I don't have much time.
But here is
Last Story footage. Also, shameless plug, please subscribe if you feel compelled to. I am trying to grow my channel and would really love to reach 100 subs. Also have a giveaway video for when I do......END shameless plug lol.
I'll provide more footage of games soon though.
BY THE WAY. Castle scene I was suggested to test out is like 3:34 or 3:35 in the video.
It's come to light that Ryzen doesn't like it when some program's threads are split between two CCXs (something that can happen
very easily with Windows' core-jumping-happy scheduler), especially in latency-sensitive workloads. In this way, it can be said that 8-core Ryzen performs kind of like a 2-socket CPU system running two 4c/8t CPUs.
Therefore, it may be prudent to retest the Dolphin benchmark(s) with an entire CCX is disabled.
https://www.pcper.com/reviews/Processors/AMD-Ryzen-and-Windows-10-Scheduler-No-Silver-Bullet Wrote:Pings within the same physical core come out to 26 ns, and pings to adjacent physical cores are in the 42 ns range (lower than Intel, which is good).
Hopping across CCX's costs another 100 ns of latency, bringing the total to 142 ns for those cases.
![[Image: attachment.php?aid=15769]](https://forums.dolphin-emu.org/attachment.php?aid=15769)
The Dolphin benchmark uses a 3 and a half year old Dolphin version. '4.0-652'
and we are currently on 5.0-2877, So as a benchmark, it's obsolete. So don't trust it.
(03-15-2017, 02:04 PM)Gir Wrote: [ -> ]The Dolphin benchmark uses a 3 and a half year old Dolphin version. '4.0-652'
and we are currently on 5.0-2877, So as a benchmark, it's obsolete. So don't trust it.
It's a synthetic benchmark, the point is to have results that allow to compare different cpus. It doesn't matter that the cpu emulation is much faster now, and you get better results.
I'd be interested in results with several cores disabled too, maybe even up to only enabling one core. But then you'd have to do the same on an intel cpu, to see how that changes the results there.
(03-15-2017, 05:09 PM)mimimi Wrote: [ -> ]It's a synthetic benchmark, the point is to have results that allow to compare different cpus. It doesn't matter that the cpu emulation is much faster now, and you get better results.
Their point was moot anyway because they completely missed the newer luabench-based Dolphin benchmark which is being run on Dolphin 5.0.
(03-15-2017, 05:09 PM)mimimi Wrote: [ -> ]I'd be interested in results with several cores disabled too, maybe even up to only enabling one core. But then you'd have to do the same on an intel cpu, to see how that changes the results there.
The only thing I can see being useful from running a single core is testing AMD's implementation of SMT vs Intel's.
By comparison, disabling an entire Ryzen CCX should let us simulate how single-CCX Ryzen and Raven Ridge APUs would perform (though the L3 cache on those products would be smaller).