Yes...the H9000 is real!

Discussion in 'The Rack Space' started by hendrik7, Dec 19, 2016.


  1. DieHon

    DieHon Member

    Messages:
    482
    Joined:
    Mar 15, 2011
    So now we mortals can buy the H8000?

    If someone attends to this event please ask about VSIG or something related to an oppen architecture.
     
  2. italo de angelis

    italo de angelis Member

    Messages:
    4,303
    Joined:
    Mar 21, 2011

    And if the algorithms size can be the same as an H8000 algorithm or bigger.
    And if the DSPs (engines) can go monolithic for bigger algorithms.

    These 3 aspects will determine its sales, big time!
     
  3. Blix

    Blix Member

    Messages:
    18,247
    Joined:
    Mar 11, 2010
    Location:
    Stabekk, Norway
    That was less costly than I expected, even if it's a pretty solid wad of cash.
     
    chlorinemist likes this.
  4. italo de angelis

    italo de angelis Member

    Messages:
    4,303
    Joined:
    Mar 21, 2011
    Need to add a few hundreds for the break out cables for the 8 analog I/Os and AES/EBU I/Os. Who wants to be stuck with 2 channels only?
     
  5. italo de angelis

    italo de angelis Member

    Messages:
    4,303
    Joined:
    Mar 21, 2011
    The screen above...
    Black Hole from the pedals... NOT the original from the 4000>8000.
    Scary!
     
  6. amphog

    amphog Silver Supporting Member

    Messages:
    3,442
    Joined:
    Nov 28, 2009
    Location:
    NY
    I hope to hear it at AES next week !
     
  7. Gone Fission

    Gone Fission Member

    Messages:
    238
    Joined:
    Feb 24, 2006
    That IS scary. I get the convergence around ARM hardware--why not develop software and algorithms for just one sort of processor?--but the possibility of dumbing down their past, well, that's just dumb. If you have the open architecture and modules reasonably ported, you should be able to do better than H9 approximations of 4000 algorithms. Seems like a bad bean-counter decision to run pedal algorithms on flagship hardware.
     
  8. italo de angelis

    italo de angelis Member

    Messages:
    4,303
    Joined:
    Mar 21, 2011

    It's very likely simpler than that in this case. They might just have chosen one algorithm over the other.
    Maybe both are available, under two different banks... one of them being labeled "H9" or "Space".
    But the whole modules porting isn't the easy part, for sure.
    A LOT of question marks for those who know this platform well and have used it thru the last 2 decades...

    we'll see....
     
  9. italo de angelis

    italo de angelis Member

    Messages:
    4,303
    Joined:
    Mar 21, 2011
    What I am ***REALLY*** worried about is the way the quad core chip power is being used here.
    Thru 3 steps of development in the years one would get different max. sizes of an algorithm running on a DSP.
    So the 4000 had a certain size, the 7000/7500/Orville per DSP had 4xthat size and finally the H8000 had about 5 or 6xthat size. In monolithic mode it gets around 9/10x.
    From what they say it seems that the H9000 runs 16 "H8000 class size algorithms". Which means one is not going to be able to build bigger or more elaborate algorithms in a single engine... not a good choice as it would be a FULL STOP to the whole point of faster DSPs available for good price.
    Pretty much like an octopus now being sixtyfourtopus... not stronger at all... just wider.
    I can perfectly understand the studio workflow, where most recording engineers would be willing to run up to 16 great effects on many tracks at the same time, like plugins... BUT like plugins, whose development is limited by any platform specs to address pre_made slots of no more than XXX size (one reason you don't see complex algorithms in plugs world), you'd be limited to pre_made DSP engines/algorithms/presets sizes.
    In simpler words... I wouldn't be able to develop anything bigger that what I have on my H8000 because the limitations are preset! That really sucks!
    Actually it may even be true that ONE H8000 monolithic algorithm could be larger than any algorithm running on the H9000... which is embarassing.
    So the question is... will these fx engines be able to become monolithic to load bigger sizes algorithms, like 2, 3 or 4 times bigger than an H8000?
    Some kind of dynamic way the engines "get together" making space to host heavy structures for the job.
    It would be not the best choice if this ain't possible, making the H9000 a perfect choice for the recording studio but a poor choice for the creative musician... who doesn't really need the power of 8 H8000 running 16 structures. Believe me... I have gigged with 3 H8000s... and I do a lot of crazy & extreme stuff... I wouldn't know what to do with 8 of them. Take a musician with more "regular" needs... what is he going to do with 16 algorithms?
    Whay is he paying money for power he will never/ever use? A lee powerful H8600 would be a smarter choice for him, as for most musicians.
    The creative guy would end up in a routing madness where he would have to build a mega algorithm splitting in into parts running in separate engines and route them properly for the job... which is not a good way to work.
    If you need a mastering class truly fantastic stereo reverb, with early reflections and diffusion, and pre_verb stereo dual precision filters + post_verb stereo dual precision filters... you may end up not having that kind of power in a single FX engine... requiring to design two or even three separate algorithms to run in series AND parallel (that's often the case in reverberation)... something a bit cumbersome and quite hard to properly manage.
    That really worries me...
    And development of 5.1 matrix reverbs? That may become a stop to certain modules development.
     
    Man with Gas likes this.
  10. Saint Luminus

    Saint Luminus Member

    Messages:
    217
    Joined:
    Aug 18, 2013
    Location:
    Los Angeles
    Ohhh, you think Eventide is going horizontal rather than vertical with the H9000. I hadn't thought of that. Poking around, it seems some guys are asking and thinking this will replace many of their plugins. You're right Italo, useful in a studio doing standard production.

    I would like to think Eventide will take advantage of technology and allow a creative user to access a monster algorithm which would "Eclipse" the H8000. :)
     
  11. italo de angelis

    italo de angelis Member

    Messages:
    4,303
    Joined:
    Mar 21, 2011

    I have a strong feeling it's going horizontal for the studio folks... with software to insert the machine into the DAW workflow.
    It's still a powerful concept... but I can see a guy with an H8000 adding a cheaper one to it... and living in peace.
     
  12. chlorinemist

    chlorinemist Member

    Messages:
    400
    Joined:
    Jul 20, 2016
    The H9000 will unquestionably be able to load all the algorithms from the H8000.

    A friend of mine currently has one in his rack. He has emphatically confirmed Eventide's claims; loading up a chain of up to 16 of the H8000s most CPU heavy (non-monolithic) 8-channel presets (e.g. 8 reverse pitch shifters, 8 precision filters, 8 harpeggiators) is easily achievable with the H9000, i believe even at 96khz.

    This is particularly notable when considering that loading just one 8 channel pitch shifter on my H8000 above 44.1khz causes it to become unstable and crash periodically. Everyone I talk to, including eventide reps, swear that this isn't normal for the H8K, but they also insist that mine doesn't need a repair, so I don't really know what to think about that.
     
  13. italo de angelis

    italo de angelis Member

    Messages:
    4,303
    Joined:
    Mar 21, 2011

    We know that. It will load the H8000 algorithms. It would be stupid if it couldn't. But the real question is... "Is that it?"
    I'm not sure it will load those algorithms running in monolithic mode though.
    All previous units were able to load all older units algorithms and new bigger ones. If this can't load bigger structures... it gets embarassing.
    Now...regarding your unit... mine loads 5419 PitchShift 8ch as monolithic @96 KHz, 5420 ReverseShift 8ch non monolithic @96K, 5424 10v Arpegg Thick (10 voice shifter) monolithic @96K, 5438 MultiShift_8Mod monolithic @96K. All work perfectly fine, without crashing the system.
    There is no reason why one unit would load them and work and another doesn't. DSPs are the same. I have never seen something like this.
    I would try reinstalling the operative system (mine is on V5.5 which is a well tested one) and check.
    If you are doing everything correctly (clock), I can only think of a problem in delay memory, which may affect intensive algorithms using that resource. But you should also have other symptoms, like noise on large delay presets...
     
  14. italo de angelis

    italo de angelis Member

    Messages:
    4,303
    Joined:
    Mar 21, 2011
    An important detail...
    monolithic mode @44.1/48KHz is capable to load a sigfile about 180% bigger than what a single DSP can load.
    Monolithic mode is a way to get huge DSP power to load monster structures or the big ones that would not normally run @96K on a single DSP.
    The statement about being capable to load 16 H8000 class algorithms is very vague and unprecise, because the size of many structures can be much bigger than others.
    96K requires the unit to run at double speed, loosing size capability. 96 KHz cuts DSP resources in half... so it should be used ONLY when really, really necessary... which means almost never.
     
    rsm likes this.
  15. Scrapperz

    Scrapperz Member

    Messages:
    6,264
    Joined:
    Sep 16, 2013
    Location:
    Like Land
    I would wait for the H10,000. You know your gonna want it. :D

    [​IMG]
     
    eclecticsynergy likes this.
  16. italo de angelis

    italo de angelis Member

    Messages:
    4,303
    Joined:
    Mar 21, 2011
    A genius from the dark side?
     
  17. chlorinemist

    chlorinemist Member

    Messages:
    400
    Joined:
    Jul 20, 2016
    Ah well the answer to that I'd say is definitively yes. AFAIK Eventide has not implemented ANY new algos into the H9000. Everything is from pre-existing products. Their focus with this edition is definitely NOT on making bigger better algorithms. This machine is 100% intended to be an expansion in the areas of functionality, power, utility, and user-friendliness.
    The Eventide of today is not the same company inventing new fx left and right that it once was in the glory days, no doubt about it. In a way it's sad, but given they have at their disposal immense treasure trove of still-unmatched legacy software + technology to draw from, and a less sparkling history when it comes to user-friendliness, it makes sense that they instead chose to focus their efforts on eliminating the limitations of the legendary H8000 (2 algos is a huge limitation vs 16) in the development of their new flagship.
    When you consider the current market's obsession with recreating the past, it really doesn't surprise me that Eventide decided not to devote R+D to more advanced algorithms. And when you think about it, building an fx chain using a chain of 16 algorithms, with virtually every effect type at your disposal, with macro controls, is not too dissimilar from buiding your own custom algorithm/preset, no? It at least achieves similar ends to what I'd imagine I'd want to program in vsig
     
  18. valhalladsp

    valhalladsp Member

    Messages:
    344
    Joined:
    Nov 4, 2010
    Location:
    Seattle, WA
    A bit of speculation on the DSP power of the H9000: As far as I know, all of the Quad Core ARMs tend to run at 1 GHz or faster clock speeds. Each core will have its own NEON 4-way floating point SIMD engine. This is roughly equivalent in power to the various flavors of SSE found on Intel processors. At least equivalent to SSE2, and maybe to more advanced versions.

    SIMD makes for INSANELY fast DSP processing. A 4 way SIMD engine can calculate 4 32-bit floating point operations PER CYCLE. By programming various low level operations in SIMD, and/or by calling optimized libraries, a SIMD processor can vastly exceed the performance of a scalar DSP at the same clock speed.

    The higher end ARM processors also have larger amounts of cache memory. This is critical for shorter delay lines, such as the ones used in reverbs. In my experience, the increase in cache size found in the Intel Core processors (i.e. Core 2 Duo and up) resulted in a huge increase of these processors to calculate algorithmic reverbs, compared to the Pentium 4. My guess is that something similar will happen with these new ARMS in the H9000.

    Italo's worries about spreading algorithms across multiple cores are probably justified. Very few plugins in the computer world span more than 1 processor core. However, the increase in clock speed, the large cache memories, and the ability to use SIMD, will probably contribute to many times the performance per core as found on the H8000.
     
    chlorinemist likes this.
  19. italo de angelis

    italo de angelis Member

    Messages:
    4,303
    Joined:
    Mar 21, 2011

    But it doesn't look like that... per Eventide words. "16 H8000 class algortihms running"... and we don't even know if that covers the monolithic ones too, which are almost double size than the non monolithic.
    We'll see...
     
  20. italo de angelis

    italo de angelis Member

    Messages:
    4,303
    Joined:
    Mar 21, 2011

    Wow...
    is that fan boys attitude or you must be the new marketing guy!
    Any and every machine EVER produced by ANY company has always added content to its heritage, not only more power and/or UI changes.
    A PCM80... does a lot more than the PCM70, thanks god... and it must be shown thru the presets.

    So Eventide it's no longer as in its glory days? Your words. Any company wants to be in its glory day now!

    Recreating the past isn't really the main thing in this kind of very high end market. That's more a thing for pedals users and some racks users.
    And your contradiction is even bigger... because if they would ever worry about a market of vintage fans, they'd have to develop a shitloat of new modules and presets.

    Splitting things across different algorithms, fx engines and probably even chip cores isn't a smart thing. You loose sight of control. Phases are no longer properly aligned, and it becomes a mess if you need to make the slightest change. This is stuff I have been doing for ages and know very well what the results are with 2 or 4 algorithms... imagine 16.
    It's nothing like working in Vsigfile, where you make things coherent and logically designed and laid out on the UI.

    I'd say... wait until you know how to work in Vsig, what it is like when you need to make a change to something you have done already... in a SINGLE sigfile... rather than getting into the hell of a change made in a single algorithm that will have consequences on all the others you need to use across many fx engines... I wouldn't want to be in such INFERNO!

    We'll see if they got it right...
     
    rsm likes this.

Share This Page