I just heard something about it from NV :p
Quote:with my recent installation of Windows 7, I've noticed that games under Dolphin have been running slower than ever with Xenoblade Chronicles running at a record low 8 frames per second. As an FYI, on Windows XP, Xenoblade's FPS ranged from 19-25.
Then you must have misconfigured something. Switching your OS isn't going to magically drop your performance by 66%.
Quote:I just heard something about it from NV
I'm going to call BS on that until you can get me a source. As far as I know DefenderX is correct.
(10-30-2012, 03:46 AM)NaturalViolence Wrote: [ -> ]Quote:I just heard something about it from NV
I'm going to call BS on that until you can get me a source. As far as I know DefenderX is correct.
I'll never find it but I'll look :p
Edit 1: lol I found this XD, but as you can see I said the same thing I said here :p
(06-17-2012, 11:30 AM)ExtremeDude2 Wrote: [ -> ]In theory faster RAM should help (more RAM won't really help anything with dolphin [assuming you have enough in the first place 
]) with EFB to RAM, but NV says there is no proof, and I haven't tried so yeah :p
Edit 2: Found this:
(04-29-2012, 08:06 AM)NaturalViolence Wrote: [ -> ]squall leonhart Wrote:not entirely true
the bandwidth of the memory controller and host write/read to pci-e performance do affect dolphin performance when using efb copy to ram, since this function cannot be easily multithreaded it only has single threaded access across the dram channels which is usually limited to the bandwidth provided by a single stick.
so in that case Faster dram per channel provides an immediate improvement to games that require efb to ram
While you should be correct this has never actually been observed to make any significant difference in performance. A number of dolphin users have upgraded their ram to much higher frequencies without seeing even a 1% difference in gamespeed in any game with efb copy to ram. The same goes for pci-e 1.1 vs. 2.0 vs 3.0.
Where is the thread link? I'm pretty sure that if you keep reading you'll find that squall was right. I don't think it's explicitly related to efb copy to ram, just how often dolphin stalls on memory <-> video memory transfers. In some situations there appears to be no speedup while in other situations there appears to be a small speedup.
Makes sense, i changed my RAM speed to 1066MHZ and EFB->RAM was kinda slower
@NV y u no press green arrow?
ALOT of people dont seem to know about the green arrow
Pci-e would become the bottleneck at some point. But until then memory will become the bottleneck while dolphin is stalling on memory <-> vram transfers.
For reference:
PCI-E 1.1 x16: 4GB/s each way (unidirectional)
PCI-E 2.0 x16: 8GB/s each way (unidirectional)
PCI-E 3.0 x16: 16GB/s each way (unidirectional)
DDR2 dual channel 800MHz: Up to 12.8 GB/s (bidirectional)
DDR2/3 dual channel 1066MHz: 17.1 GB/s (bidirectional)
DDR3 dual channel 1333MHz: 21.3 GB/s (bidirectional)
DDR3 dual channel 1600MHz: 25.6 GB/s (bidirectional)
DDR3 triple channel 1066MHz: 25.6 GB/s (bidirectional)
DDR3 dual channel 1866MHz: 30 GB/s (bidirectional)
DDR3 triple channel 1333MHz: 32 GB/s (bidirectional)
As you can see most DDR2 setups are good enough to feed pci-e 2.0 while most DDR3 setups are good enough to feed pci-e 3.0. However there is also caching to consider. Higher memory bandwidth means faster cache refills.
I'll applaud him just for taking the time to actually give such an answer. I got lost halfway through though.