Vienna Symphonic Library Forum
Forum Statistics

194,315 users have contributed to 42,914 threads and 257,953 posts.

In the past 24 hours, we have 0 new thread(s), 13 new post(s) and 81 new user(s).

  • Unabashed Rumor Mongering

    last edited
    last edited
    I found this online today. Let’s go off the deep end and pretend that 30-50% of it could be true.

    @Another User said:

    Macs are about to get faster in 2007 -- and not just with better processors, graphics or memory. Fast data storage is computing's greatest bottleneck....and Apple is about to break the bottle wide open.
    Many new Macs in the year ahead will be getting multiple internal hard drives stock; in an effort to attack computing's narrowest bottlenecks -- primarily storage and memory -- Apple will start shipping Macbook Pro and Mac Pro models with at least two hard drives stock and more stock models will gain high RPM, big-cache drives with hardware RAID support offering users a boost in performance (RAID 0 "Striping") or data reliability (RAID 1 "mirroring"). Eventually, even the entry-level Macbooks, iMac and Macbook Thin will have at least some dual-HDD models.
    And all of the above will gradually adopt Intel's powerful new high-end Flash memory caching technology to put frequently accessed data from any storage device(s) into a Flash cache that performs at thousands of times the speed of any hard drive. This will help offset some of the power drainage that comes from dual hard drives in laptop applications....and make more of a noticeable difference in real-world performance than practically any other kind of hardware upgrade in the past ten years. Intel has other technology that Apple is working hard to be an early adopter of.....attacking computing's other key bottlenecks. Use of ultra-high-performance RAM caches at key points between the memory and CPU, between the memory and key peripheral busses, between the memory/CPU and the GPU....et cetera
    Notably, the next few generations of Core processors will gain fast increasing on-chip Level 2 cache sizes and improve the "crossbar" connections between multicore chips so that they can spread data between them with optimum speed and as close to zero latency as possible.
    What we're really excited about is the advent of entry-level Macbooks and iMacs with dual high-performance hard drives. A Macbook with two 7200RPM, 16MB cache 2.5-inch Serial ATA hard drives in a hardware RAID 0 array would be a quantum leap forward in portable performance and the iMac would be even sexier with two 15,000RPM 32MB cache 3.5-inch SATA drives to say the least!


    Now I’m not posting this to get anyone more excited about the new Macs than necessary. But I’m hoping that VSL will look far enough forward in their new software instrument development to take advantage of what’s coming down the road. Will Mir and more complex sample handling be expected from our favorite sample library suppler? I realize that VSL lives in the real world and is busily readying the new Brass, Woodwinds and Choirs so we can play with them and they can stay in business. However, part of the company has to look as far down the road as possible.

    Compounding the mystery of all this is when the new 8-processor Mac Pros come out and they have to use OS 10.4.9 for some time to come. I imagine that 10.4.9 will be a very important release because many users will have to ‘make do’ on a lot of computers for a long time. I say this because I speculate that OS 10.5 will be the 64-bit system optimized for the new Intel chipped CPU’s, leaving some aspects of 10.5 unrealized on older Macs. I would think that Apple would have enough ‘forward reaching’ implementation in 10.4.9 to make it in some ways compatible with features that are planned for 10.5.

    May you live in exciting times.

  • well, well ... i don't want to break down any excitement ... and be assured VSL is looking far down the road ... and even at and behind the posterwalls on the roadside.

    - raid in general: 2 drives in a raid can be mirrored (safe, same speed as a single disk) or striped (almost twice as fast but insecure - loose one disk and loose all)
    - raid in detail: what to raid - the system is already fast enough, - the samples are recommended not to be on the system drive. so you'd need at least 3 drives, 4 if you want to use raid level 5 and have your system seperate.
    - hybrid drives: IMO just a new hype, you could use any fast flash device for access to eg. a pagefile or startup data (keep in mind number of write-cycles (corr. from read-cycles) are limited with flash)
    - cache: not neccessarily an improvement (does only speed up subsequent aka. sustained reads), sample access is random read of tiny blocks.
    - crossbar technology: SGI invented this somewhen last century, even my VW320 has this since 1997
    - CPU power: we have enough of it so far (except you want to run lots of altiverbs on the same machine of course, then it's never enough)

    - sATA disks with higher density (perpendicular recording) and/or higher speed (15.000 rpm, always loud) _do_ increase performance (look at the new seagate models, eg. the 2,5" SAS)
    - solid state disks: could become affordable with some of the new flash-disks coming up (sandisk, PQI, adtron) and reduce latency significantly
    - IBM's millipede would be a real quantum leap, let's take us by surprise if they really appear this year ...

    what looks great for the average grafic designer or home-video cutter is unfortunately often not enough for the advanced audiophile using samples.
    let's see what the upcoming year will really bring to us and how usable it will be in the sampling world ...
    christian

    edited typos and corrected write- from read-cycles

    and remember: only a CRAY can run an endless loop in just three seconds.
  • Thanks cm,

    What do think actually would be important to you to break the barriers of limitation?

  • Not to "Hi Jack" the thread but:

    Hi Jack!

    Where was this spec posted? I'm just curious.

    I hope that all of what was said will be seen sooner than later, of course. For me it's always in two stages-- the wait and see stage with hardware and then the wait and see stage with the software to take advantage of the hardware updates.

  • It was posted on Macosrumors.com, which is like saying it was posted on the wall in a public toilet. [:)]

    They have a very low batting average, and I'm always annoyed at myself for having wasted the time to read that site. "WOW!!!! [insert current year] is shaping up to be the most fast and furious one for Apple ever!!! We have some unbelievable dish coming up this weekend - stay tuned! [they never do] Our reliable sources inform us that there's going to be a new model coming out!!!! Steve Jobs did it again! What a great man!"

    And yet I read a couple of times a week... [:O]ops:

  • Please note that I carefully presaged my commentary indicating it was not to be taken literally or in its totality. I fully expected Nick to chime in saying it was not true and the source was unreliable as he has in the past regarding this source. Of course that's true to a certain extent and Nick is correct about that. Hence my title, 'Unabashed Rumor Mongering'. How much it may be true or not true can only be borne out by experience. My stated concern in posting it was regarding VSL's long range vision and to a degree cm answered that.

    Apple is very circumspect in its pre-release information to the public. Let's just wait a few weeks and see what the true offering turns out to be.

  • barriers ... let's keep aside the harddisk thing, it's the same like with batteries in notebooks, always too slow, always too little capacity ...

    the next bottleneck we will face is the bus from memory to processor - 4 cores on a die is great, but what about the I/O if we run applications which do not just process data already held in the cache? carefully designed chipsets are needed urgently ... and affordable soundcards with PCIe

    kind of friday evening discussion [;)] christian

    and remember: only a CRAY can run an endless loop in just three seconds.