PlanetZ forums - scopeusers.com

Sam,

I know that what you said is pretty much what STI had in mind with CELL. I simply think that's not feasible. How is the communication between the nodes supposed to work? Large cluster systems usually depend on Myrinet or Infiniband. Everything else lacks the needed latency and bandwidth, and ...

ooo000ooo,

yes, you'd need an EZlite board. Scope cards can't use unencrypted modules, and you can't encrypt them without the Scope SDK. I don't know, maybe you could also use a ZP board or something like that, but Scope cards won't give you raw access to the DSP's...

And even if CSound is open ...

Tom,

you're right that clockspeed-talk is nonsense. It's easy to see by comparing AMD and Intel CPU's (especially P4), or Intel's P4 and P-M, or even comparing the PPC970/ G5 and AMD's amd64 line.

When it comes to pure number-crunching, amd64 CPU's, still based on a heavily improved x86 core ...

The CELL is no match for a PC. IBM quote from an article, published by heise.de:

"CELL is up to ten times faster, compared to CPU's currently used in entertainment and gaming systems."

Marketing speech. Keep in mind, IBM had a prototype CELL running at 4.6GHz (but didn't mention the cooling ...

Strange, I really like the new logo - the "@" was so... 1992! It was one of the ugliest logos I've ever seen. But I still don't like the interface design of SFP, looks a little to "toyish" for me. :smile:

But other than that, there's not really much that changed from 3.1c to 4.0...

To say it again, I'm not exactly sure about the CSound thing. I remember that there was a port advertised and offered by AD, called XTCSound or Extenended CSound. And there was this rumor several years ago (1999/ 2000), stating that Scope was a XTCSound variant...

There were other companies back ...

ooo000ooo,

no, that won't help creating plugins. As the OS handles the decryption of the plugins, you won't be able to encrypt and load them. And, of course, CW wouldn't open-source the DSP files, as they are platform independant anyway.

Other than that, anyone _could_ write software for Sharc ...

There's an easy solution for this problem: don't trust releases you got from anyone except CW... :smile:

Maybe we could even provide a nice, distribution independent binary package for Linux, using a nifty GUI installer (Loki, InstallAnywhere, Autopackage...). And if you want to compile SFP on ...

DementiaT,

I think there are some things to clear up a little more in-depth?

See, first of all, quite a few companies are indeed releasing the sources for their drivers, or at least support the development of open source drivers by releasing hardware specs and answering questions (eg HP, Intel ...

Pentium MMX 233 subnotebook, 96MB RAM - still works like a charm (Fujitsu Biblo B110)! It's used as a license manager these days...

You'll get the highest possible PCI bandwidth with NForce pro based Opteron boards - far beyond anything offered elsewhere...

<font size=-1>[ This Message was edited by: wsippel on 2005-02-03 22:20 ]</font>

rob604,

yes. But you wont find a single CPU board with two full speed PCI-E slots... :smile:
The total PCI(X/E) bandwidth of the K8WE should reach at least 16GB/s, plus 8GB/s for CPU<->CPU communications, and additional 16GB/s memory throughput (2x3 1000MHz DDR 16/16 HT links).

EDIT: To quote ASUS ...

Well, the K8W series are workstation boards - who would need an AGP slot in a server?

Especially the K8WE should make an extremely powerful DAW, but it's quite expensive I'm afraid. :sad: It has an unbelievable bandwidth, as half the PCI(X) slots are connected to CPU1, the other slots are connected ...

Of course, the additional registers are not really 64bit specific. But there is no other way to introduce them on x86 (and have them used on a large scale). And the double register size is 64bit specific, of course...

The improved memory addressing is not only suited for servers, btw. Using ...

AMD plans not to move to DDR2 'tll late 2006 - that doesn't mean they'll not use DDR533 or something like that this year, of course...

dArKr3zIn,

8 slots (Workstation boards. Up to 16GB):
http://tyan.com/products/html/thunderk8we.html
http://tyan.com/products/html/thunderk8w.html

16 slots (Granted, server board. Up to 32GB):
http://tyan.com/products/html/thunderk8qspro.html


<font size=-1>[ This Message was edited by: wsippel on ...

Increased per-process memory space, increased phyiscal memory size (without address translation), increased virtual memory pool size, 64bit integer opps in a single CPU cycle (faster/ higher precision), double the integer registers + double the register size, 8 additional SSE(2) registers - not much ...

time_chase,

no. You may use 32bit software on Win x64, but _not_ 32bit drivers. And, if I can draw a conclusion from using 64bit Linux, it's not always possible to 'connect' 32bit and 64bit applications (exchange informations - ASIO drivers come to mind, or 32bit VST plugins may not work on a 64bit ...

Tom,

just tried to get some more info, and you're right. Not binary compatible, and $199 each at 10.000 units - quite expensive for a DSP!

Well, no TigerSharcs for now...

valis,

IIRC, the TigerSharc is indeed pin- and binary compatible to the Sharc, but much faster and quite cheap (not as cheap as the Sharc, which costs about $ 10 each, but cheaper than the Motorola G4, which is only slightly faster at floating point ops - but then again, you could use more than one ...