Page 1 of 2
Posted: Tue Jan 25, 2005 5:27 pm
by Shayne White
In addition to their Linux/Mac developments, they'd better start working on a 64-bit version of Scope so it can work with Windows XP 64-bit and Sonar x64. Otherwise, it'll become absolete on Windows as well as other platforms...
PCI-E and new DSP chips would be nice, too.
Oh, and only DirectX works with 64-bit. Cakewalk says VST needs a complete overhaul to be 64-bit compatible. So CWA had better work on a DX version of XTC, too.
I hope they're busy!
Shayne
Posted: Tue Jan 25, 2005 5:58 pm
by astroman
On 2005-01-25 17:27, Shayne White wrote:
...Otherwise, it'll become absolete on Windows as well as other platforms...
the only thing that
is obsolete is a 64-bit OS for Workstations...
cheers, Tom
Posted: Tue Jan 25, 2005 6:24 pm
by time_chase
wont Win XP64 be compatible with the 32bit stuff? I dont think many developers would be in a hurry to get 64bit going...XP64 aint even out...knowing MS it will be delayed for another 2yrs!! The OS is still a long way off.
The rage in 2005 will be dual core cpus....that's something to look forward to.
Posted: Tue Jan 25, 2005 7:40 pm
by wsippel
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 plugin host without some 'magic' in the host...).
the only thing that is obsolete is a 64-bit OS for Workstations...
Yeah, and 640kB will always be enough...

I like my 64bit OS, and it _has_ it's benefits!
<font size=-1>[ This Message was edited by: wsippel on 2005-01-25 21:15 ]</font>
Posted: Tue Jan 25, 2005 7:53 pm
by astroman
which of them are relevant for non-esotheric purposes ?
Posted: Tue Jan 25, 2005 8:06 pm
by Shayne White
Posted: Tue Jan 25, 2005 9:13 pm
by wsippel
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, but worthwile nevertheless...
Some info on AMD64 vs x86:
http://h200001.www2.hp.com/bc/docs/supp ... 238028.pdf
Posted: Tue Jan 25, 2005 10:21 pm
by darkrezin
I'd like to see some PC motherboards with more than 4 RAM slots

Posted: Tue Jan 25, 2005 10:48 pm
by wsippel
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 2005-01-25 22:50 ]</font>
Posted: Tue Jan 25, 2005 10:50 pm
by kensuguro
don't worry about the slots. By the time you can afford to fill them up, they'll come out with some new standard and make all your RAM obsolete.
Posted: Tue Jan 25, 2005 10:52 pm
by wsippel
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...
Posted: Wed Jan 26, 2005 5:30 am
by astroman
On 2005-01-25 21:13, wsippel wrote:
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, but worthwile nevertheless
I'd consider the first 2 points esotheric (in the workstation, not server, context), I agree on the improved integer processing and the last 2 points are CPU (not OS) related.
In fact that's pretty few

There is not a single cycle of performance increase by such an architecture, but the adaption of the majority of data to 64bit will cause a massive overhead of processing.
Some major suppliers didn't even manage to get a 16bit OS right, and most of current 32bit software is buggy as hell. All to be recompiled with new error introduction
It should be clear that the driving force behind this is 'sales, sales, sales'. It's new, it's bigger, it's better - a welcome justifice for a lot of companies to continue business instead of moving their stuff where it belongs.
You have mentioned the 'Bittware' pages with their TigerSharc boards and tools.
There's an excellent examination about the obstacles for an optimized algorithmic design on a DSP and a (RISC) CPU (TS versus G4).
THOSE are THE problems that have to be solved for performance increase at the workstation level - and those problems are completely OS independant
cheers, Tom
Posted: Wed Jan 26, 2005 1:13 pm
by wsippel
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 worksataion apps like Softimage|XSI, per-process memory requirements are not uncommon to exceed 2GB (this also includes swap and shared memory of course), and since systems with 2GB RAM are not that rare anymore, it's likely we'll see systems with 4GB - 8GB RAM next year, something hard to accomplish using 32bit architectures in an efficient way...
The ability to process 64bit ints in a single cycle also much improves the performance for specific operations - see oggenc and other audio/ video codecs for example - speed improvements of 350% are not uncommon. In fact, processing of longs is about 7-8 times faster on a 64bit OS with current compilers like gcc 3.4, and expected to improve even more with gcc 4.x...
Posted: Wed Jan 26, 2005 3:32 pm
by darkrezin
Hi Willie..
Yes, I was aware of server boards... but I think it's still quite rare to use these kind of boards for DAW systems. I sure wish I had one though!
Posted: Wed Jan 26, 2005 3:56 pm
by wsippel
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.

It has an unbelievable bandwidth, as half the PCI(X) slots are connected to CPU1, the other slots are connected to CPU2. Same goes for PCIE, one connected to each CPU - that makes the K8WE the only board with two PCIE 16x slots on the market...
<font size=-1>[ This Message was edited by: wsippel on 2005-01-26 16:04 ]</font>
Posted: Wed Jan 26, 2005 4:04 pm
by darkrezin
Yeah... cost is definitely a barrier
I look forward to the day that consumer boards get this kind of functionality (I'm not holding my breath....)
Posted: Thu Jan 27, 2005 6:12 am
by rob604
Actually you can get dual PCIe SLI boards for much cheaper than the dual-proc Tyan from companies like Asus, MSI and/or Gigabyte. <a href="
http://usa.asus.com/prog/spec.asp?m=A8N ... s=09">Asus A8N-SLI Dlx</a> would be my choice.
Posted: Thu Jan 27, 2005 3:20 pm
by Basic Pitch
I dont think its going to be an issue any time in the near future, all people will do is what they have done for ages now any way.
Just run a dual boot system one boot to Win64 and one with XP pro on it. Its the same as not with some people refusing to leave Win 2000 or even some who still use 98SE.
Cheers!
Posted: Thu Jan 27, 2005 5:33 pm
by wsippel
rob604,
yes. But you wont find a single CPU board with two full speed PCI-E slots...

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' page:
*SLI mode : x8 , x8
*Default(Single VGA) mode : x16, x1
<font size=-1>[ This Message was edited by: wsippel on 2005-01-27 17:38 ]</font>
Posted: Fri Jan 28, 2005 5:01 am
by rob604
man those numbers make my head buzz.... whatever would you do with all that bandwidth? you are, of course, correct, wsippel - 16x slots. i guess you'd need two cpus just to handle that overhead.
