OTOY Forums

Hi all,

just having a quote put together for a new rig.

Does the speed of the RAM have any influence on Octane rendering, as far as loading the scene etc.

The quote I'm looking at has DDR3 at 1600.
Is this ok, or should I go for 2100?

cheers,

Steve

As far as I am aware 2100 is just a marketing thing. It does not have any sort of benefits in any software. Last time I checked not even windows was optimized beyond 1600. I might be wrong although.

00ghz is right, mostly.
The frequency of the memory is not a reliable method to ensure the ram actually works faster.

I was overclocking as a hobby for 10 years and found that when ram throughput (read speed, write speed, etc) was higher systems always performed snappier. This was true for opteron, core2 and all i7 builds I've done.
And, about 7 years ago low latency memory went out of favour for high frequency higher (relative) latency memory. Generally they were getting more predictable, memory throughput results using high freq. higher latency; also, when overclocking the cpu, higher frequency memory allows higher FSB & therefore cpu speeds.

What does this mean for your original questions; if you buy high frequency RAM you are likely get faster RAM(assuming you don't buy cheap stuff), but you'll also be paying the 'high frequency' surcharge due to the popularity of overclocking kits.
Unless you plan on overclocking, I recommend buying the best kit you can get at 1600mhz (should be 9-9-9-24 or something) otherwise you'll be paying for something you don't need.

As to weather it actually speeds octane voxelizing (or app performance) I doubt it will be significant; the slight increase in ram speed even if you did overclock would shave milliseconds from the 'data juggling' between cpu calculations. Overall load times would decrease, and you'd probabily 'feel' that when loading proxies or scenes into the working viewport, but you're not going to have any comparably smaller benchmark figures to show off.

In systems like ours RAM is relatively cheap (especially if you're using an LGA2011 multi GPU system) so don't scrimp on RAM; for the sake of better overall system performance, but don't spend money on overclock ability which you won't use. Just go with 1600mhz with cl9

True and true, to the best of my knowledge. However, correct me anyone if I am wrong, but I am fairly certain that there are huge gains by having a mobo with quad channel memory, versus dual channel. If you can build your workstation starting with quad channel mobo that has 4 double-slot PCI-E, you are going to be in pretty good shape. And if said mobo does not require ECC ram, better yet. That shit is ridiculously expensive - and probably not all that important, provided that GPU rendering will be the main task of the beast.

CAS Latency is the key!
Search for RAM with low CL value.

Awesome, cheers for the info.

I'm not planning to over clock, so I'll stick with the 1600.

cheers,

Steve

prodviz wrote: I'm not planning to over clock, so I'll stick with the 1600.

it's better to stay with cheeper, but good 1600 than faster & way more expensive (or lower quality fast). You'll not find any difference for a lot of apps. I've seen a video where Guys used a lot of fast ram & converted it into RAMdisk for for some assets - It was great, but for daily use, You better stick with lower speeds & more reliable SKUs as unstable RAM influencs a lot of crashes.