When I bought the new system, I went a bit cheap on the memory. It's DDR3 1600 11-11-11-28. My Windows Experience for memory is 5.9, which makes me since everything else is 7+.
I'm wondering if Handbrake would be happier with memory that has faster timing, or if it might only be marginally faster?
Bring up task manager and watch the physical memory under performance as you encode a Bluray file. If you don't see it going over 50% use, "more RAM" isn't going to be a big help. As Rollin mentions, though, faster RAM may help, and if it's a bigger space... That's a bonus for running other things.
But faster doesn't necessarily mean the CPU can use the faster speed. If the motherboard was designed for 1600, that may be all the bandwidth that will be used, unless you overclock.
Faster timings just won't make much difference. It's bandwidth that helps HandBrake, and only in certain circumstances. I.e when using NLMeans with HD/4K content.
As long as you have free Memory, it's not worth it.
Woodstock wrote:Bring up task manager and watch the physical memory under performance as you encode a Bluray file. If you don't see it going over 50% use, "more RAM" isn't going to be a big help. As Rollin mentions, though, faster RAM may help, and if it's a bigger space... That's a bonus for running other things.
But faster doesn't necessarily mean the CPU can use the faster speed. If the motherboard was designed for 1600, that may be all the bandwidth that will be used, unless you overclock.
Thanks!
s55 wrote:Faster timings just won't make much difference. It's bandwidth that helps HandBrake, and only in certain circumstances. I.e when using NLMeans with HD/4K content.
As long as you have free Memory, it's not worth it.
Thanks s55!
I have a memory gadget on the desktop and with 4 GB, I never seem to max out my free memory.
Yeah, I always assumed windows has so many services/processes running that you can never get exact number. I wonder if Linux would be better as I imagine its easier to kill unnecessary tasks.
Rodeo wrote: ↑Fri Feb 24, 2017 10:26 am
The difference is way too small to be statistically significant, unless you maybe ran each encode 100 times and average the resulting encoding speeds.
I was thinking about that too. I wasn't interested in doing that many encodes though! I wanted to get that memory in. I guess I should have done it at least three times each, but I wasn't interested in that either. I guess I could have just clicked on add to the queue a couple of times...
Oddly I did a bit more mouse clicking on stuff when I was doing the first test to get my baseline so I figured that would have been the slowest.
rollin_eng wrote: ↑Fri Feb 24, 2017 10:38 am
Yeah, I always assumed windows has so many services/processes running that you can never get exact number. I wonder if Linux would be better as I imagine its easier to kill unnecessary tasks.
I had quite the nightmare trying to get Ubuntu installed and up to date when I bought this system a few years ago. I guess the LTS version was just about due to be updated and that was the only one I could get that would install easily, but then updating it 5 times make for too much
I doubt you'll find much difference. Encoding a single video isn't that memory bandwidth intensive.
Outside of memory benchmarks and (large) file compression, the difference between fast and slow memory, double / triple channels are hard to measure.
Once, I had a colleague who had a choice between 12 GB triple channel or 32 GB double channel memory. In the end he opted for the first cos memory benchmark showed it was 10-15% faster. In real-world usage, the difference is maybe 0.5-1%.
But he really needed more memory to run his VMs, so I think it was a poor trade-off.