SiSoft Sandra - Bandwidth Efficiency - How do I get it Higher?

TheBigFatClown

Posts: 1,110   +495
I have questions about the "Bandwidth Efficiency" values in SiSoftware Sandra memory benchmarks test.

I'm getting approximately 73% integer and float bandwidth efficiency when I run the aggregated "Memory Bandwidth" benchmark in SiSoftware Sandra.

Is this good for my set of memory sticks? Will the memory bandwidth efficiency go up with faster memory speeds or is this more a characteristic of the CPU and memory controller?

My system right now is:
AMD 2200G, ASRock AB350M Pro 4, 16GB (2x8GB) 3000MHz Patriot Vipers @ 2934MHz
 
The memory sticks I have are actually not listed on that page you posted a link to. It's the "Memory QVL (Raven Ridge)" page where mine can be seen. It's basically what you posted but with an RR at the end of the URL. Although, sneaky ASRock lists my 16GB kit model number on the RR link, it puts up a second number in parentheses which I have yet to verify on my chips because I don't know how. (11BF2)

My Vipers have been "working" in my new system from day 1. But that's all they have been doing is working. @ 2133MHz, which sucks. That's what my system boots up at if I leave the memory options in the BIOS set to AUTO. I put the timings in my BIOS that Patriot recommends on their website and I did get up and running @ 2934MHz for a couple of days before Windows 10 crashed. So, worst case scenario I run these stable @ 2133MHz indefinitely. Worst case scenario is not gonna work for me. I've been doing a lot of research to find out what other things I need to do to get the maximum performance from these vipers.

I should have just bought a Playstation 4 Pro. It would have been a hell of a lot easier!
But then again, all those games are overpriced and I've always been a PC gamer so, it is what it is for now!

I've been reading on ASRock forums that a new BIOS is coming out soon that will supposedly make things better. *fingers crossed*
 
Last edited:
I've been thinking a bit more on this. The ratio of 2934/3000 is approximately 97% so I don't think that's how the bandwidth efficiency percentages are calculated, at all. That's what I really want to know.

I'm thinking it's comparing my usable bandwidth to my "theoretical maximum" bandwidth. Aren't memory timings inversely proportional to usable bandwidth? In other words, if your memory timings go up, I.e., more waiting, then your usable bandwidth, goes down? I'm not happy at all with 3/4 of what I should be getting.
 
Back