Who here are using 6 to 8 core CPU's & 64 GB's ram?

That’s only partially true. Multi-channel direct access memory is still far faster than even dedicated SSDs. So, who the hell needs that? How about people who do orchestration? I’ve had over 30 gig of samples loaded on a Giga system. If you load a large number of articulations over a large number of instruments, it would be very easy to get to 64Gig. But again, improvements in streaming and SSD capabilities are definitely closing the gap on those use cases. But, just because YOU don’t need something doesn’t mean someone else doesn’t.

And who in hell needs 8 cores? Your question sounds like a matter of quantity vs. quality.

Again, maybe someone is using a large number of VSTi. I have a 980x OCed to 4.5Ghz. That’s a 12 core setup. I can crush my CPU with just a handful of DIVA instances.

And, by the way, there are more uses for computers than music. Modeling can chew up and spit out even the most robust CPU’s. The most powerful commercially viable workstations are still feeble compared to the special purpose systems. Multi-core is just the current flavor of performance increase.

It isn’t that it doesn’t work right. It is more about the fact that in the PC world people think that a $1 stick of memory should function identically to a $100 stick of memory, and that a $10 MB should work the same as a $250 MB. In the case of 32 and 64 gig of RAM, to get that to be stable, you need high end parts across the board. MB, power supply, matched memory etc… In a few years, commodity parts will probably be able to handle it. But, when I was building in 2010, I could have gone with 24Gib triple channel. But, folks were having problems getting 6 sticks of RAM to cooperate. ESPECIALLY if you wanted to OC. That’s been over 2 years ago, so some of that may have been addressed in commodity parts. But, I doubt it.

The best way to think about it is … If it is at the edge of the current technology, the less chance commodity parts will be stable supporting it. You will need higher end, well matched parts. Also, that cutting edge system may rock for one application and totally blow up trying to do something else. Cutting edge is always fiddly.

Yeah, upgrading is always a “what I need it for” question. Those that constantly bitch about the upgrade train forget that the general consumer is often not the driver. There are professional concerns that must be addressed. We’d still be pounding on logs with sticks if everyone just went with the “good enough for me” line of reasoning.

Wow, I want your system! Makes me wonder, though, if an SSD drive would improve my machine (for samples, right?). I have a space for it. When I got my machine, year and a half ago, I couldn’t make the ASIO meter go red. Then I got NI Studio Drummer… now I can!

Yes, SSD for the samples. SSDs have different strengths. For sample playback you are looking for Read speed and total bandwidth. You can get an excellent 256Gig SSD for $200ish. It makes a HUGE difference for sample playback.

I changed to an all SSD system and do really like it. Projects start up up much quicker because VSTis like Omnisphere, Superior2 and Kontakt etc. load samples a lot faster. Windows itself starts up quicker and the drives don’t make any noise nor generate much heat (nice for us overclockers) at all.

Well worth it if can afford it, and it really does’nt cost a fortune anymore. The three drives in my system cost me around AU$600.00. The money is not wasted either, as when upgrading your setup, just whack the same drives into your new machine.

Mauri.

faster drives allow streaming to feed samples without having to load them completely into memory. That’s why less memory is needed now.

All Romplers like drum machines, Kontakt, Halion etc… Wave table synths don’t benefit. This is simply about sample players that load large libraries.

It doesn’t really matter if it is a Rompler or a Sampler. It has to do with playing back large sample sets. BFD, Steven Slate Drums, Superior Drummer etc… all use large sets of samples for their drum kits. I have ~200Gig of just drum samples. IDE and SATA HDs are still not all that great at streaming these large libraries. Lets say you have 4Gig … By the time you load your DAW and all the other programs, you may have 1 Gig or less to play with. Heck on 32bit systems, this was a lot harder, but 64bit systems have really simplified this as well. Now, it is very easy to have a drum rompler take up that gig of RAM if you loaded in the entire kit. Leaving you nothing for other parts of the project. Streaming just has multiple ways of managing how much of a kit to load into memory so that you don’t have to wait for the sound when you hit a snare … but a snare might have 6 articulations with as 10 or more velocity layers. What happens when the loud left hand rim click is not pre-loaded and it gets hit? HD bandwidth and read seek time are the key factors in how well the rompler can respond.

It’s even worse for a violin section …

First Violin can have 8 to 20 solo articulations (pizza, lagato, trill, WT, HT, gliss etc…)
Then do that again for a combi - (4 or more playing those same articulations)

So just for First Violin you can have upwards of 30 basic articulations spread across the note range of a violin. Toss in the velocity layers for each note and that can be thousands of samples. That single instrument can be 2 or 3 hundred MB or more compressed. JUST for first violin. Now, multiply that by each instrument in an orchestra! So, if you do sample based orchestration, large amounts of RAM or kick assed streaming are a must.

Even with some of the new Hybrid (large mem cache or SSD front end) SATA II 6GB regular HDDs, the random read and throughput of just regular consumer SSDs is not even close. Throughput can be double to triple, read times can be 50 times faster. Since those are the two most prominent performance indicators for streaming, it is a no brainer that SSDs are killer for streaming.