Audiocontrol Hyperion Processor

[mglobe]

If you want to dismiss the fact that MTBF is the defecto standard for everyone in the world to determine whether electronics and built to a high standard or a poor standard that’s your choice. But if you want to expand beyond the information I’ve provided, actual data would be beneficial to support your case

Sir, the only thing I have questioned is your apparent assumption on the architecture of the Altitude CI.

 

[Dj7675]

@Highfidelity Coming in quite hot it seems… On the one hand claiming Amir owes you all of the success he has built here and on the other hand challenging him to ban you? And all in quite a flurry of posts. Just seems odd. But… welcome I guess

 

[Highfidelity]

Just a few too many tequila’s in the Panama sun. Passed out hard at 11pm. Reviewing the fallout now. I was quite behaved I’d say.

 

[Highfidelity]

Under the lid of the CI. Looks like a fan cooled Intel CPU to me. 3 fans total. Moving parts to wear out and clog with dust. DPR-16 has 0 fans.

 

[Highfidelity]

Lazy design if you ask me. Stock Intel fan. Stamped steel case like you’d find in a $50 Chinese 2U PC case. Even $800-1000 PC builds using an equivalent motherboard and CPU would have a passive fan-less cooling solution and a heavier aluminum plate construction with thick heatsink side panels. Other than the consumer grade general purpose motherboard they only have like 2 PCB’s. The HDMI switch card and the main I/O board. In which both run cool for sure and wouldn’t require fan cooling if in a proper design. It’s like they’re shouting from the mountain tops that their clients are too stupid to know any better and $26000 is what they’ll pay for 32 channels because we say so. Completely underestimating the world in 2026. Things are gonna change, and this BS will end. Mark my words.

 

[Highfidelity]

Pumped to see HDMI CEC compatibility with the APR. likely the case for the DPR as well. 1 more advantage over the Trinnov. Who wants to fumble with multiple remote controls to turn volume up and down in 2026? I love just using my Apple TV Siri Remote to do everything. Not possible with Trinnov, but possible with Hyperion.

 

[Highfidelity]

Audiocontrol will have a big booth next week at ISE 2026. Official launch of the Hyperion Axis-10. And hopefully they’ll unveil the MSRP of the DPR-16

 

[welwynnick]

Under the lid of the CI. Looks like a fan cooled Intel CPU to me.

I think that's exactly what it is. Trinnov say the Altitude 16 uses an Intel Core i312100 Gen 12 CPU.
I presume the ALCI uses something two or three generations newer, with more cores, slower clock, more processing power and lower power consumption.

When the Nova came out, I thought the new efficient architecture with a low power ARM processor on a single PCB was the start of an exciting way forwards for Trinnov.
Of course the Nova only processed up to 6 channels, and didn't handle HDMI, video, Dolby or Atmos etc. But it seemed like a start.

So when they announced the ALCI, I presumed this used the same architecture. But no, it's still PC based. Ah well.

However, all Trinnov processors run their DSP at 96 or 192 kHz, depending on model, channel count and software.
If the price to be paid for native rate processing is a $200 CPU and fan cooling, I think that's a small price.

However this is a Hyperion thread. I like the idea of the DPR-16, and will quite likely buy one, but I'm not expecting any more than 48kHz processing

 

[Highfidelity]

It doesn’t matter if Trinnov runs the DSP at 96 or 192 if they downsample it to 48kHz before sending out via Dante. The DPR-16 just does it before the DSP processing to make it more efficient, reliable and cool running. Without fans to wear out and clog with dust. And I save $12000 for 16 channels. I’ll allocate any heavy lifting post-processing DSP tasks to my military grade fan-less rig that can run 24/7 for 30 years without a hitch. I’m using this rig regardless for my active crossovers. Now I just get to raise the CPU max load from 5% to 8-9%.

 

[welwynnick]

It doesn’t matter if Trinnov runs the DSP at 96 or 192 if they down-sample it to 48kHz before sending out via Dante.

Who cares? They can output 96 or 192 over AES3, or just send it to the DACs.
Trinnov themselves say Dante is there for installation convenience, and not sound quality.

I’ll allocate any heavy lifting post-processing DSP tasks to my military grade fan-less rig that can run 24/7 for 30 years without a hitch. I’m using this rig regardless for my active crossovers. Now I just get to raise the CPU max load from 5% to 8-9%.

I will probably do the same, and I have something suitable in mind.

 

[mglobe]

I think that's exactly what it is. Trinnov say the Altitude 16 uses an Intel Core i312100 Gen 12 CPU.
I presume the ALCI uses something two or three generations newer, with more cores, slower clock, more processing power and lower power consumption.

When the Nova came out, I thought the new efficient architecture with a low power ARM processor on a single PCB was the start of an exciting way forwards for Trinnov.
Of course the Nova only processed up to 6 channels, and didn't handle HDMI, video, Dolby or Atmos etc. But it seemed like a start.

So when they announced the ALCI, I presumed this used the same architecture. But no, it's still PC based. Ah well.

However, all Trinnov processors run their DSP at 96 or 192 kHz, depending on model, channel count and software.
If the price to be paid for native rate processing is a $200 CPU and fan cooling, I think that's a small price.

However this is a Hyperion thread. I like the idea of the DPR-16, and will quite likely buy one, but I'm not expecting any more than 48kHz processing

So I communicated with the folks at Trinnov, and they said that "the Altitude CI is a combination between the x86 and ARM architecture"

I suspect the way forward will ultimately be ARM, but that they have some work to do to get there. Note that this last comment is speculation on my part. The info from Trinnov is fact.

 

[3ll3d00d]

So I communicated with the folks at Trinnov, and they said that "the Altitude CI is a combination between the x86 and ARM architecture"

I suspect the way forward will ultimately be ARM, but that they have some work to do to get there. Note that this last comment is speculation on my part. The info from Trinnov is fact.

The TAC2 is the arm chip, it's described a basically an audio interface (signal routing etc). The processing is on the (x64) CPU

 

[mglobe]

The TAC2 is the arm chip, it's described a basically an audio interface (signal routing etc). The processing is on the (x64) CPU

Thanks for that clarity.

 

[Highfidelity]

Who cares? They can output 96 or 192 over AES3, or just send it to the DACs.
Trinnov themselves say Dante is there for installation convenience, and not sound quality.

I will probably do the same, and I have something suitable in mind.

With a proper implementation using Dante instead of AES offers no sound quality compromises. Dante keeps the signal bitperfect from the AVP output to the DAC. If you want to use AES out you’ll deal with other compromises that can result in degraded performance overall in a final system implementation. Such as long speaker cables from rack mounted amps to the speakers. Dante allows me to place the amps 1m away from the speaker drivers for all speakers. And my speaker cable runs are 1m max from the amp outputs to each driver. And I can do this across 100’s of channels in sync in massive 10000sqft+ homes. With the DPR-16 as the single source. Try doing that with AES.

 

[Highfidelity]

So I communicated with the folks at Trinnov, and they said that "the Altitude CI is a combination between the x86 and ARM architecture"

I suspect the way forward will ultimately be ARM, but that they have some work to do to get there. Note that this last comment is speculation on my part. The info from Trinnov is fact.

And all the Storm processors also use an Arm chip in the RPI they use to run the user interface. Every AVP uses an Arm chip somewhere in 2026.

 

[welwynnick]

Dante keeps the signal bit-perfect from the AVP output to the DAC.

Yes, but not inside the AVP. That probably matters with something like the Nuprime H16-AIP, but less so with the Hyperion DPR-16 or Trinnov ALCI.

Dante allows me to place the amps 1m away from the speaker drivers for all speakers. And my speaker cable runs are 1m max from the amp outputs to each driver.

That would be a reason for using Dante. You can still do it with AES, but the cabling is much more difficult.
I'm glad that the Hyperion and the Trinnov give us the option to use either.

 

[Highfidelity]

Yes, but not inside the AVP. That probably matters with something like the Nuprime H16-AIP, but less so with the Hyperion DPR-16 or Trinnov ALCI.

That would be a reason for using Dante. You can still do it with AES, but the cabling is much more difficult.
I'm glad that the Hyperion and the Trinnov give us the option to use either.

With the DPR-16 no matter if you use Dante or AES out you’ll get the exact same digital output. Bitperfect match between both. With the Trinnov you skip the resample to 48Khz on either 2 channel stereo or old school 5.1 or 7.1 Dolby True HD and DTS that is originally in 24/96 or 24/192 (very limited selection) from the AES outputs. Everything that’s actually encoded in Atmos that’s available today is already in 24/48 only. So it passes through both the Trinnov and Hyperion bitperfect 24/48 in and out from the Dante outputs.

So to recap the only advantage you gain with AES out on the Trinnov is with 2 channel stereo and 5.1-7.1 multichannel if 24/96 or 24/192. Most people don’t care about that. Most people will listen to music on an Apple TV that outputs a fixed 24/48 with both stereo and Atmos. Or similar streaming box. It’s only the fringe few who have 5.1 to 7.1 multichannel on blueray or computer rips. If listening to 2 channel stereo no need to pipe it through the processor anyways.

 

[welwynnick]

So it passes through both the Trinnov and Hyperion bit-perfect 24/48 in and out from the Dante outputs.

The player isn't synchronised to the Dante network, so that can't be true, it's got to be re-clocked and hence re-sampled so it isn't bit-perfect any more.
Of course it may not matter much, and the difference may be inaudible, that's fine, but it isn't bit-perfect.
We've been over this a few times, it seems.
Please go an buy a DPR-16 and be happy. I will probably join you.

 

[Highfidelity]

The player isn't synchronised to the Dante network, so that can't be true, it's got to be re-clocked and hence re-sampled so it isn't bit-perfect any more.
Of course it may not matter much, and the difference may be inaudible, that's fine, but it isn't bit-perfect.
We've been over this a few times, it seems.
Please go a buy a DPR-16 and be happy. I will probably join you.

Everything gets re-clocked with both processors regardless of what outputs or sample rates you’re using. This is how all the clocks sync up with the HDMI. The re-clocking is also necessary to clean the jitter from the HDMI. All stereo DAC’s also re-clock all incoming audio. Re-clocking doesn’t mess with the data. It only syncs the clocks. The data from input to output only loses bitperfectness from the Atmos processing and DSP in post processing. Not from re-clocking. And all sample rates from both outputs can’t avoid that.

 

[welwynnick]

Everything gets re-clocked with both processors regardless of what outputs or sample rates you’re using. This is how all the clocks sync up with the HDMI. The re-clocking is also necessary to clean the jitter from the HDMI. All stereo DAC’s also re-clock all incoming audio. Re-clocking doesn’t mess with the data. It only syncs the clocks. The data from input to output only loses bitperfectness from the Atmos processing and DSP in post processing. Not from re-clocking. And all sample rates from both outputs can’t avoid that.

Yes I understand what clocking is, I've been banging on about it for decades.
Re-clocking can be done synchronously or asynchronously.
When it's synchronous, the clock is low-pass filtered to remove the jitter, typically with a PLL or two, but the data stays the same (though it may go through a FIFO).
When it's asynchronous, new timing and data are synthesised by an ASRC using a local clock. The new data is different, but that doesn't make it bad.
Some DACs have an ASRC and some don't. If it's done internally, then I think we have some assurance that it's done well because it would be reflected in the chip specs.
In the case of Dante, the digital output always goes through an ASRC, because the Dante network and the source aren't synchronised.
I'm sure I read that Trinnov stated they didn't re-sample incoming data, EXCEPT with the Dante output on the ALCI where is was re-sampled.