that is why I bought an SP 404MK2 for longer samples and resample feature plus the FX are super amazing. Love my Octatrack but the Roland is more immediate.
If you scroll all the way back up to the OP and read it you’ll note that the question is; why are new samplers being produced with only 64MB? There’s not a clear answer to that but it seems like it mostly comes down to economics and/or arbitrary limitation based on percieved user expectations and requirements. No one is really complaining particularly.
Though it would be nice to be able to have longer record times in the cheap samplers.
2 Likes
the initial question was why, and there’s no definite answer
it can’t be a niche use case as clearly the number is growing. in similar vein you could’ve argued that stereo samples is niche — who needs those for one shots?
anyway, the OP question still stands until people with development or music marketing experience chime in I suppose
but the logic ‘less is more’ or ‘I dont need that so you’re a niche’ is hard to understand…
4 Likes
So the question is, there are many devices with more than a gig of RAM, why aren’t there more?
Just do pads in the DT. You can record them into the DT (not import) and then widen them with reverb and delay. Dataline released a Juno samplepack and he did this very thing in the project. It sounds great. Plus when your tune is played in mono it won’t sound like complete ass.
There is always a workaround.
Pretty sure the TE machine uses a small inexpensive CPU that it probably tough to upgrade the RAM on. In other words, this is one way they could offer this thing at $299 and use the CPU/board combo that they wanted to.
So it comes back to workarounds.
1 Like
No not really, you haven’t understood the question.
Do you not think it’s odd that in 2023 when flash storage and RAM is as incredibly cheap as it is compared to what it was 10/20 years ago, that we’re still getting only 64MB on new samplers?
That’s all really. This feels like another version of the whole stereo/mono debate from a while ago which was also quite silly tbh.
3 Likes
I don’t find it odd because you’ve already answered that question — because the people who need more space have other products that fit their needs.
Idk im just trying to be helpful. Unless I’m missing a bunch of complaints about memory in entry level samplers (which I could be) I don’t really understand the issue here.
Shoutout to the M8 <3
2 Likes
For me, 64mb is fine. I’m more disappointed by track count. 16 used to be standard. Then 8. Now 4 seems to be a trend. FFS 
I get that limitations can breed creativity, but if I had 16 tracks, I could still only use 4 of them. I’d rather impose that limitation myself, rather than the gear.
1 Like
It is because adding ram to hardware is not cut and dry. You cannot just wack on some PC style memory on a small microcontroller. You’re usually limited by what is available at quantity for SDRAM chips and pin count / hardware support for the given microprocessor / microcontroller you’re using. Everything flys out the window when you’re using a large microprocessor / embedded linux solution like a raspberry pi etc, which is usually when we see large amounts of onboard memory, but then it suffers in other ways - longer boot times, bulky touchscreen UIs, maybe latency and at that point you may as well use a phone, ipad, or laptop. 
17 Likes
It is interesting. Dunno why. I understand the whole thing with “restrain makes creative” but I think its most important to restrain one self. I have tried minimalistic devices over the years and then end up buying additional gear for what they are lacking for my workflow. So one might say I am going in circles sometimes 
However I really admire some people who are creating some killer sparsely arranged yet still engaging sounding stuff. So much to learn
1 Like
I did a quick search on this, and for microcontroller devices that uses SRAM or DRAM 64MB is basically what is available. Usually a DDR memory controller isn’t available which limits the options. Plus a DDR design is much more complicated from an EDS point (many traces with requirements on length matching etc).
See this related post:
If I remember correctly the Daisy is using the STM32H7.
7 Likes
Looks like an interesting find. Thanks!
Will read it later
Samplers are built with either more general MCU’s or specialized DSP chips. They’re not CPU and they’re work differently. In both cases there is limited amount of memory supported by chip.
For example, AFAIK DT uses Coldfire MCF5441. Datasheet says it supports up to 2GB of DDR.
So, memory limits are often just limitations of specialized audio chips. All-purpose ARM processors can handle more memory, but they are worse for audio applications.
2 Likes
Yeah, OT and M8 are my main samplers these days. Streaming audio from storage makes this a non-issue. The way I use (both of) them would definitely exhaust the memory, but I think I only use them that way because I know streaming from storage is possible. I use other samplers without this functionality differently.
I would totally be happy to see multi-GB of memory on samplers, but won’t be expecting it anytime soon.
2 Likes
Akai Z4, Z8 and MPC4000 could be expanded up to 512MB of RAM.
Kind of overkill IMHO, but it’s also interesting to note how the tendency has reversed and most HW samplers nowadays have less RAM than 20 years ago.
I guess that for longer recordings/orchestral samples/etc. most people now use their computers. It makes sense, I guess.
2 Likes
Frankly I’m amazed most of the manufacturers can source ram chips as small as this. They must be competing with washing machine and microwave oems for them. We no longer live in 1998 and it must take some serious purchasing effort to find 16Mb chips and not just chuck in a 1Gb chip and call it a day.
Clearly a design choice and price is a secondary factor.
The reason for this is there are basically two types of digital music hardware:
- Stuff that is basically an embedded system
- Stuff that is basically a PC
So things like the MPC, Machine+ and Push 3 are basically just low end PCs / Raspberry Pi so adding more RAM costs about the same as on a regular PC.
These devices will usually be large, heavy, expensive, get hot and have short battery life, and typically have higher latency, but will be powerful and easy to program
Other devices are embedded systems / microcontrollers and these typically don’t have a lot of RAM because they don’t need to for 99% of the applications they are used for with the exception of samplers and the hardware sampler market probably isn’t big enough to make a special board for.
Creating an embedded system is very expensive so most people just buy a standard development kit like this one: Projects using the Teensy USB development board which is used by Polyend and Dirtywave and lots of others.
If you are Roland you are big enough to make bespoke embedded systems for your devices (Probably all of your devices are using the same platform which a lot of expensive R&D went into). As a result you can tailor them a bit more and that’s why you tend to get Roland devices with loads of polyphony or sample time that still run on batteries.
12 Likes
What you write makes a lot of sense
Interested to know where DT and M:S (1Gb) fit in here ?