The DSC Desktop Super Computer!?!? :o

Saw this mentioned on Gearslutz…

http://www.hpcmusic.com/

DSC using Haswel-EP delivers 210x more audio-dedicated throughput than ‘the industry standard’ Pro tools HDX3, 100x more than the 12core Apple Mac Pro and 315x more throughput than 4x UAD2 OCTO. This is the worst-case scenario.

Oh?

Isn’t that the usual crackpot?
Claiming all kind of things not providing any shred of evidence.

I don’t know who you’re or what talking about, I just happened to see this on another forum and thought it was interesting. Are you claiming this is a hoax of some kind? It at least appears that this has been in a prototype phase, and is something ready for production, rather than something that is currently available for sale.

I was also assuming that the components that this is made of, is current available technology. The only difference is how it’s configured…? or it has a specially designed motherboard to make use of multiple CPU’s and RAM modules. The ‘claim’ of this thing is supposed to have a support of both ‘up to’ 288 core’s - 56 threads & 1TB RAM with 1.2 TB memory bandwidth & 16 TB flash based storage.

Their ‘claim’ is that this pushes what would be available by the typical rate of improvement seen. Of course, if one is keen on all computer/component manufacturers, software operating system & DAW developers to keep leading people on a string, tube feeding ever so minor upgrades/improvements for the sole purpose of their own money-flow, by continual milking consumers on a never ending basis…as has been happening, and will continue.

But with the above rate feeding the public only minor developments, it’s easy pickings & continue revenue. Add to this, planned obsolescence and Ca-ching! | |: $$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$ : ||

If nothing else, this is a pretty cool concept design :ugeek:

It’s just that 288 core’s - 56 threads & 1TB RAM would not make a huge difference, it does not scale that way, when we are talking near real time computing. And the site looked like something I had seen a year ago, when someone posted on Gearslutz that he was there to answer the questions of the audio of tomorrow. Some of the DAW builders asked some questions that never got answered, just the usual list of cores and MIPS and a few pictures that showed some computers, but not a audio device in sight. Maybe it is to get some funding, don’t know.

Could be that it’s all about getting attention to gain interest and attract funding of a ‘concept’. As far as “scaling” I’ve heard of this, but I’m not very clear on the term. All I know with my own experiences as a hobby PC builder & user, is with more CPU and RAM, it means more can be done…whether it’s in one single powerful system, or combined resources of multiple systems.

Not knowing how it can even be ‘possible’ to have a single system (at this point an time at least) to utilize this much CPU & RAM…not just the hardware, but also software support, I was wondering if the concept here was more or less, just one big ‘networked’ system or a stacked ‘Cluster’ built into a single, configured & ready to use box. By looking at the enclosure, it appears it could be just that :bulb:

There’s nothing wrong with pushing a ‘concept’ either…after all, people who are conceptual artists, aren’t necessarily engineers themselves. You do have to have an initial ‘idea’ before it can become a reality…much like the things in the original Star Trek series, which later most have become a reality. Though in the case of this DSC Desktop Super Computer, they seem to be trying to pass off as a reality, rather than a concept.

https://www.facebook.com/hpcandmusic


Until there is a shipping model, it’s vaporware. Interesting vaporware, but vaporware nonetheless.

@Qbass: it’s more than that, though. Unfortunately, you can’t slap together more CPUs and memory and expect it to work faster. The biggest problem is that all of the RAM is still shared amongst the CPU cores and, with a single back end bus (on the motherboard, not your Cubase project :wink: ), coordinating access to memory becomes problematic at best, impossible at worst (at least not without specialized hardware). And then you get into issues with real time interrupts on one CPU while another CPU is currently accessing memory or other system resources, etc.

Would love to se Steinberg take a shot at vst-link again, using lan as an easy build in feature to use more computers. Maybe a ASIO Guard 3 that could offload plugins to other computers on the lan.

Right swamp as you say, interesting vaporware…although on that link there sure seems to be a lot of hoopla, so we shall see what this apparent ‘concept’ brings in the future.

Well Larry, I’m still not even sure what this think is. It either has to be an intended specially designed hardware to make the ‘good stuff’ happen without the CPU/Ram limitations, or it has to be an intended purpose built run of the mill super computer networked cluster, with the limitations that exist on that. The 1st will need a LOT of cooperation from hardware & software companies. the 2nd, still a lot but at least uses existing technology.

With computers of the future, there had been a conversation some time ago here about a concept of designing it so that it would be able to install and utilize as much CPU & ram you wish (modular-like)…either larger capacity components (CPU & Ram) or multiples of them could be ‘added’ to your system. This way, at least everything would be designed once, rather than designing lesser capability & more capability hardware. A pipe dream yes.

Peak, I can only assume VST System Link (VSL) is still available within the current Cubase versions, but just not talked about much anymore, and hasn’t been developed any further. What I’m unclear about is if audio over Ethernet is analog or digital (?) or can be either (?) but since s/pdif digital audio can actually be ran through ordinary RCA audio cables, I’d think digital audio could be ran through Ethernet too.
Since VSL was specifically designed to be used through digital I/O’s and cables for it’s data as well as audio…
I wonder if it’s possible to run VSL through an existing application that already allows you to run your audio through your Ethernet (assuming that it’s digital audio capable) and Cubase/VSL seeing the Ethernet as just another I/O available? If not, I don’t see why Steinberg couldn’t make this possible within the VSL application as further development.
Unless… if you’re just talking about Steinberg, turning VSL into Vienna Ensemble Pro application :question:

If you look at the YouTube video’s it’s about RDMA for clustering and the usage of graphics processors in a single system. This looks like something that is existing and available.

RDMA supports zero-copy networking by enabling the network adapter to transfer data directly to or from application memory, eliminating the need to copy data between application memory and the data buffers in the operating system. Such transfers require no work to be done by CPUs, caches, or context switches, and transfers continue in parallel with other system operations. When an application performs an RDMA Read or Write request, the application data is delivered directly to the network, reducing latency and enabling fast message transfer.

https://www.youtube.com/watch?v=HYlWxPeL9-k

Maybe send them an info request ?
The contact field is at the bottom of that webpage.

Probably they will present that desktop at NAMM ?
That is very soon.

kind regards,
R.

http://www.hpcmusic.com/#!partners/c1h6a

http://www.hpcadvisorycouncil.com/council_members.php

and this is also interesting:

www.youtube.com/user/hpcadvisorycouncil

But strange that they have so few views…