KPA 2.0 - Suggestion for KPA 1.0 upgrade

  • When I assemble my PC the components I use always are in regard to future upgrades. If possible, I purchase a motherboard that will not only be compatible with my CPU, but the motherboard should also be able to interface with CPUs that are more powerful and (at the time) much more expensive. The RAM, expansion and peripheral slots of the motherboard are all considered with an eye on future upgrades.


    Why not design the KPA 2.0 with a vision of upgrading the KPA 1.0 hardware? The KPA 1.0 was a significant investment for many of us and though (in my opinion) it will continue to be relevant for at least a decade to come, having an eye on upgrading the DSP and memory capabilities would not be a waste of resource. I think many of us would invest significant resource to upgrade our KPA 1.0 to a KPA 1.5.


    Am I completely off base on this? Is the hardware of the KPA 1.0 so integrated that any upgrade would require gutting the whole thing?


    Thanks,

    J.W.

  • the components I use always are in regard to future upgrades

    Let's say you bought an Intel 1156 socket board back in 2010 ... there's no way you can upgrade the CPU today to any of the modern CPUs. In other words, the PC world is very "short term" compared to the DSP world.

    Keep in mind that the Kemper Profiler architecture has been developed about 10 years ago and still keeps going strong. I would even say that the "magic" isn't so much in the continuous increase of horsepower but in the (very) smart way to use it.


    If there ever comes a KPA 2.0, the KPA 1.0 will still sound the same and will continue to be very well usable. I don't worry about this topic at all. :)

  • I was mainly wondering if the KPA 1.0 was develop with an eye on modular upgrades. It would be rather odd to me if the KPA I purchased a few months ago (and was built approximately a year or so ago) would have the exact same DSP chips that they installed in the 2012-13 model. If this is indeed the case, then that makes my musings a moot point.


    You are indeed correct that software efficiency buys you much more than computing horsepower, but 6 years in computational technology is a very long time in regard to hardware.


    You will note that I mentioned that the KPA 1.0 would be relevant for a long time to come. That's because of Kemper's commitment to continuous improvement of their software. I was wondering if perhaps they had designed the Kemper hardware with thoughtfulness for hardware upgrades.

  • It's the same DSP chip since it hit the market. A future hardware upgrade would be a KPA 2, not a chip upgrade to the KPA 1. No company would ever ask you to open their device and swap modules inside. And no company would consider e.g. 50000 devices sent in for upgrade requests ... unless they totally lost their minds.

    The thoughtfulness you mention has gone into the initial design. All plans and ideas Kemper had for the KPA 1 can be and have been done without upgrades. Enjoy it, play it, don't worry. :)

  • The problem with upgrading hardware is avoiding bottlenecks. More power in one place doesn’t mean it isn’t held up and wasted farther down.


    Modular works.... but only if the various pieces play well together.

    “Without music, life would be a mistake.” - Friedrich Nietzsche

  • It is indeed a moot point. Perhaps modern computational technology has me spoiled. If they are still using the same DSP chips they installed in the first KPAs off the assembly line then that certainly offends my sensibilities with regard to hardware technology.


    However, the value of the KPA isn't found solely (or actually mostly) in its hardware, but within the software. When purchasing a KPA you are subscribing to the Kemper software service. They seem to have made great progress in squeezing all they can from their algorithms.

  • If they are still using the same DSP chips they installed in the first KPAs off the assembly line then that certainly offends my sensibilities with regard to hardware technology.

    This should help quell your misgivings on the matter, bluz. Something Christoph wrote in June last year:


    Christoph Kemper's Take on the DSP


    The Profiler is not aged. The DSP we use is not the fastest by megahertz, but it has AFAIK the highest code efficiency per megahertz for DSP jobs, of all processors out there. The Eventide H9 and others runs on the same processor family.


    But adding new effects is not a matter of horsepower. We could theoretically add another 500 new effect types. It is all about memory to keep all this code in the hardware. Since we are not running all possible effects at the same time, but only those that you have dialled into your rig, calculation power is not really an issue.


  • Thanks for bringing that comment to my attention, Monkey_Man. My whole question or comment stemmed from the little bit of knowledge I have about early parallel computing when the real concerns were indeed communication and storage. I know things have come a long way since then.

  • I'm a dimwit when it comes to computer technology, etc...


    Surely a huge percentage of the coding in the algorithm of a modeller like the Helix or Axe-FX is wasting memory space....like CK mentions, "It is all about memory to keep all this code in the hardware"


    They model an individual amp, so the coding has to include every aspect of that individual amp.

    But, every amp has it's sweet spots, I've never met a guitarist who didn't introduce me to his tube amp by immediately setting it to the sweet spots he's discovered over the years.

    So all the other non sweet spots settings of the amp that the Helix or Axe-FX coding covers is a waste of memory space and engineering time?


    Yes I know there are those who want the Gain and Tone controls to function like they do on the amp, but Kemper has proved that in the end that's not as important as the final tone we hear and feel...and CK achieves that with less waste of memory space.


    So CK's coding/algorithm is more of a motherboard/CPU upgrade than an actual motherboard/CPU upgrade. :)

  • I don't think hardware upgrade-ability is a viable long-term goal. Nobody knows which innovations the future will bring other than that we know the world is moving forward. Software OHOH is a different animal with a constant stream of fixes and enhancements, some of which gradually add to system load. My wish is is that manufacturers use a bit more powerful components when they design their gear. While the DSP and converters on the KPA still deliver excellent sound the system in general gets slower with every release with certain operations now starting to challenge my patience. There were more powerful components available back when the KPA was designed, at not so very much higher prices. $10-15 more in components back then would have made a huge difference today. We should expect better performing components in premium-priced high-end gear. The challenge here is to keep the beancounters at bay during the design-phase.

  • I think what CK means is that when they shipped the profiler initially and all along right up to today, they shipped it with enough on-board memory to hold eight effects and the amp profile at any one time. They would need to increase the physical memory of the unit to fit in more but because the unit is hardwired for eight effects , they can build a massive catalog of effects but you can only have eight of them at any one time in memory and the trick is to optimize the effects so they do. The coding that has been done for the profiler is incredibly unique in that it was coded in house for a specific DSP over nearly 20 years. When you have that level of detail and optimisation, you can push the hardware much further and with more features than if you used off the shelf APIs with lesser capable DSP. The only solution under that ethos is to add more and faster and hotter DSPs to compensate for the lack of programming skills. The profiler's DSP is still air cooled. It has no fan, no heatsink, no nothing to add to the cost and maintenance of the profiler and that's a testament to CK's ethos.

  • Except no 2 people precisely agree on what those "sweet spots" are. Some light high gain, mid gain, low gain. Some like high-mid focus, others some low-mid squishiness. Some like scooped, others mid-pushed. Some like bright, others like dark. I would contend that modeling an entire amp allows you to have the full pallet to work from in order to find your sweet spot, not another persons, or an engineers idea of what the sweet spot may be. It's no less "wasted" than it is on the amp it's modeling.


    I would only add that part of what can be frustrating about the Kemper currently is that you're basically locked into a profiled sound. Yes, there is a front EQ and deep editing parameters, but they respond like a studio EQ rather than the amp profiled. Having the ability to profile EQ and gain stacks would be a great way around this, and was an idea Christoph spoke of long ago.

  • The only solution under that ethos is to add more and faster and hotter DSPs to compensate for the lack of programming skills. The profiler's DSP is still air cooled. It has no fan, no heatsink, no nothing to add to the cost and maintenance of the profiler and that's a testament to CK's ethos.

    In fact, if it had a fan I'd not have bought one 'cause my racks sit alongside me in the room and I'd have taken the authenticity hit and gone for a Helix rack. That's a huge sacrifice IMHO, but it shows how important it is to me to not have a fan whirring away near my head when I'm trying to make considered judgements about a sound, track or mix.


    I've said many times that this is one of the many reasons I love the KPA, and it's a very-important one IMHO. Not only is it noise-free, but imagine all the dust and grime that would've otherwise entered our beloved beasties over the years and the problems this might've caused. Additionally, everyone knows that the primary enemies of electronic circuitry are heat and moisture.