Compositor SoftwareCompositor Software

Category : Performance

By ruslany

Dump, middleware and more

Dump, middleware and more

An enormous work was conducted this weekend on MIB vector optimization. At the beginning the full base was defragmented by clearance of the dump below:

RTOS dump at 192 kHz MIB 5149 08.06.2019

In addition, the switch to 11 kHz was made and stochastic selections were performed in a special edition of RTOS with a direct output on auxiliary channel (through). Later these ethers were recorded as middleware with PCM WAV container of 24-bit 11 kHz. You can listen to them lower:

Middleware 1 11025hz MIB5149
Middleware 2 11025hz MIB5149
Middleware 3 11025hz MIB5149
Middleware 4 11025hz MIB5149

16 middleware files were recorded, here I show you only the four. Then these middleware was filed to the special version of L1-L4 L6-L7 vRouter RAD96, which uploaded it on 96 destinations of L1-L3 layers. This way, the middleware was fixated. This method is different from direct submission to RAD96 master routing table, because RAD96 ether aggregator can exclude the predefined set of ether combinations and I was needed to attain to precise channel matrix of 52 channels.

After the full contact base was uploaded by stochastic selections of MIB5149 and dump, I made RTOS authorization again but this time mangling the sample rate parameter up to 192 kHz. This way, I updated links to all aliases and authorized the whole MIB on 192 kHz.

By ruslany

NPO Compositor network map creation from 0 BC to AD 4000

NPO Compositor network map creation from 0 BC to AD 4000

vRouter RAD96 metronome

Have you, probably, noticed that metronome of vRouter RAD96 inadvertently approaches our days? That is, SR timer has passed 2019 year already and RT timer is at 2015, TR timer is one and a half century ago at the year 1870. Now, you can say that NPO Compositor network map established from 0 BC up to our days and metronome still goes forward. Soon the injections of the contact base will be performed in the year 2050 and forward.

Compositor Networks map

That is, starting from 80th of the last century metronome counts not the automatic reverse mechanism, but manual RTOS, that is why I can control the process in its essence. For example, to increase conductivity and to turn reverse on (in rare cases, when “brakes” are not needed).

Compositor RTOS dump 8×32 MIB5007 03.06.2019

I was not satisfied by automatic vRouter RAD96 by the fact that it is always in reverse mode and there is no opportunity to control this process. Using RTOS there is no such problem, because I can thrust in direct and indirect ether.

By ruslany

The dumps future workstations can only read

The dumps future workstations can only read

We are get used to 8-bit SysEx dumps, many of us even listened to their audio presentations. However, how the dump of modern embedded real-time operation system sounds? Let’s start with the fact that modern operation system is 64-bit, which gives almost 8 times more dynamic range, than 8-bit dump. Moreover, RTOS dumps are written with 192 kHz sample rate. In this post I will sum up two dumps, which were made with MIB 4795 and MIB 5007, which allows saying about their origin only one thing: these dumps are the music productions on their own.

In essence, we are dealing with routing tables, reproduced on the high regeneration speed. But, my task is to find the source of these routing tables, the hardware and software system, which can read these dumps and respond not only with sequence of events, but with sound generators tuning, sound synthesis parameters and effects. This station should include 64-bit operation system itself, working with 192 kHz sample rate, which is critical to CPU working frequency.

Such DAW should allow reading dumps with a large dynamic range and perform settings in accordance to the loaded network map. I would like to achieve panning and equalization in virtual environment without human intervention, in addition it should be performed not by a topology of some algorithm, but to exist inextricably with routing path filed in the current moment.

I remind you that 8 routing tables mixture is sufficient for a complete routing path. Taking in account 8×32 matrix for such routing tables, they are aired on 32 destinations. This tells us about high load of RTOS channel in a moment of dump creation. The high load on output channels creates tasks on input channels, because communication is a kernel-loop relationship and performed in a cycle with consistent calls and answers. To receive the answer the calling system should set in a que, because only 8 input streams available in RTOS. That is why there is a constant insufficiency in RTOS, it can’t be covered even with high console port regeneration speeds, and because to upload routing tables into the buffers the time is needed where high regeneration speed doesn’t play any role.

That is why the whole MIB should be loaded using autoload with aliases for the full base without forced thrust. I repeat, that forced thrust creates a big que and events processed only using generic feeder interrupters that is why you should constantly monitor filed system statuses. Because, there are no injections in filed systems on such high regeneration speeds as 192 kHz, then it is needed an additional time to receive an answer. If you need to receive an answer immediately, you should run RTOS on discretization frequencies lower, than 192 kHz, where the injections happen all the time, but the quality of the answer will be lower.

Compositor RTOS dump 8×32 MIB4795 26.05.2019
Compositor RTOS dump 8×32 MIB5007 03.06.2019

By ruslany

5000 routing tables in CP-6137-960FX MIB

5000 routing tables in CP-6137-960FX MIB

NPO Compositor reached the final goal of emission for CP-6137-960FX server, which is 5000 routing tables. Management information base is 5007 routing tables that allows organizing not only 4000 VLAN, but also 5000 VLAN. NPO Compositor will continue to expand the management information base up to 6000 VLAN, because each routing table gives access to its own VLAN trunk or autonomous system.

You can listen to that dump, which includes full 5000 MIB:

Compositor RTOS dump at 192 kHz with 5000 MIB on 03.06.2019

This dump allows you to authorize in Compositor real-time operation system and attain to its external control by RAD96 autonomous system. To connect to Compositor RTOS it is enough to playback that dump using streaming method in online player with RAD96 autonomous system turned on. In this case, autonomous system will serve as middleware, which will merge Compositor software and your equipment and will make remote connection to CP-6137-960FX server possible.

When NPO Compositor reached 5000 routing tables it gained the full MIB, which finishes L1-L4 L6-L7 vRouter RAD96 development. To order vRouter RAD96 licenses use Compositor Software contact form.

By ruslany

Summarizing the 2018

Summarizing the 2018

The 2018 started from a trip to Saint Petersburg in February. At that time, I haven’t knew what program I will demonstrate: 5th or 6th. At the end, I decided to show the 6th. I had a wavetable pool around 500 wavetables to that moment. I played all wavetables, initiated the Ether and mostly satisfied with this trip. Saint Petersburg gave the direction: to evolve system without kick drum further. At that time, I already developed fully silent system, which doesn’t produce sound at all. To such system, I attribute RAD96, which is 2018 development. RAD96 has two realizations: as OS subdriver with visual driver and fully autonomous system with zero emission. Moreover, the system with zero emission was reached after long OS tests with large number of connected jets. RAD96 OS is an Ether aggregator. It was created to test the kernel on injections when using 8th version iteration with 4-layer structure. The final iteration for the L4 (Layer-4) is non-linear polynomial with public coefficients. Up to the moment, the wavetable pool is 3715 wavetables. These wavetables were used to check the 8th version of Compositor kernel. The sum of gathered virtual funds in samples of wavetables is 486932480,00RY and aggregated funds in a process of kernel testing are 88087861,84RY. The whole sum of funds in RY to the end of 2018 is 575020341,84RY.

By ruslany

Zero-emission signal network

Zero-emission signal network

Therefore, it was made, at the end of 2018 I made a full zero-emission kernel with protection level capable to overcome even the strongest wavetables. First, let’s talk about an experiment:

  1. For the last 8 month, I injected polynomial using different wavetables with open jets and closed jets (for the last four month).
  2. I made a tunneling via Compositor 4 Max for Live and made an intrusion of these wavetables into created Compositor network.
  3. I hosted the Compositor kernel 8.4.2 system with public coefficients until the last moment.
  4. At the final moment, when the network couldn’t cope with such amount of injections I transferred to fully autonomous system mode of Compositor kernel 8.5.6.
  5. This way, I left the created network.
  6. Then, using the Compositor kernel 8.5.6 with lower process ID, I ran the Compositor v9 Hypervisor and made the injection of the whole pool, but with feedback, which enabled to establish the strong feedback with strong immune system.

On the steps 4, 5, 6 the system started to recover and polynomial without the public coefficients paid off itself. This autonomous system is zero-emission, which is proved by Resource Monitor: the Commit memory graph stays on the 1,043,608Kb mark. It enables to make fully autonomous any living system, turning unlimited timeserver on.

By ruslany

Non-Linear Time-Invariant Autonomous System

Non-Linear Time-Invariant Autonomous System

The 4th order transfer function an ideally suited for multithread mode was simplified to the 2nd order to achieve the autonomous system. The 3rd and 4th order coefficients turned the interconnection on the kernel level on for both channels at the same time, which caused friction, allowing to spread its impact on 96 channel network. This way, the process of system powering was imitated, namely self-feeding. For licensing the system on higher working frequencies of the audio driver, it was needed to turn the visual driver and the peering network off without touching the 4 layers of the kernel. This way, the full autonomous system achieved. In other words the system autonomous up to the moment when visual driver and peering network of algorithm with 96 oscillators are connected. The system stationary, because uses time function variable on the transfer function input, which, by the way, has its own time continuous component also, performing DC-offset by the y-axis. This way, the Compositor kernel 8.5.6 algorithm reaches both conditions of time-invariance and autonomy. The system is non-linear, because transfer function performs transposing with exponential relationship. This way, system of the kernel and loader combines in non-linear time-invariant autonomous system, which is the continuity on the previous post intervention of universal module with temperature sensor.

By ruslany

Non-Autonomous System Statement

Non-Autonomous System Statement

Universal module – is an embedded system with signal lamp. Lamp is a temperature sensor of processor core and lights with different colors in accordance to the processor multiplier work. Universal module has no manual controls and display. The only way to know if module works or not is a light from the lamp and its spectrum. Universal modules should communicate with each other by disturbing the signal chain with disconnecting lamp. In digital system, you can attain to the same principle by using a disturbed algorithm, lighted with different colors of virtual oscilloscope. In this case, lamp, as stated earlier, is a temperature sensor of a processor, which executes the algorithm. RAD96 is the thermal kernel and allows controlling the temperature of processor. The system of kernel and temperature sensor is non-autonomous by applying weighting functions with appearance of selected coefficients. The appearance of x coefficient in the weighting function of equation with one unknown y makes this system non-autonomous, so both left and right channels has the cross-interaction. The result of this is necessary use of visual driver and multithreaded system for kernel computation distribution.

By ruslany

Ruslan Yusipov SuperComputer (RYSC)

Ruslan Yusipov SuperComputer (RYSC)

Compositor Software creates reminiscent mittwares using 10th version of injector and wavetables. It is a mixture of tracks, produced in the past decade. It is that problems, which Compositor Software faced earlier. In English language, reminiscent means to remind something. Two Compositor software RIG’s on the RMY physical server performs work on problem solving. To set a problem for Ruslan Yusipov SuperComputer you need a full-length full-duplex modem of version 8.5.4. It is enough to load a mittware in fast speed regime. To load a mittware it is needed a full cycle upload to modem input via playback device. RIG’s set a problem in the buffer (the problem resides in the buffer no matter if RIG’s turned on or off). In the beginning of problem-solving SuperComputer made of 960 hyperbolic cores fragments it on small sub-problems and solves them with interruption. Later, it magnifies problems, making full load periods last longer. The solution to a problem is a long lasting period of full load of all physical cores. Compositor Software interests only reminiscent mittwares. When SuperComputer finishes execution of loaded problems, it transfers to the other problems mode, coming from all connected ether participants. It is needed to solve not only Compositor Software problems but also problems of other system users – it is a main idea of virtual local area network.

In 2016 the time of problem solving by mittware was one month without any success. RYSC solves the problem in 5 minutes. This advancement achieved mainly because of better TC-SUBTRSRRT262144 architecture and supports up to 96 cores in one RAD96 virtual machine. The transfer function evidently allows installing solutions to the recipient, which simplifies performance of real physical actions of their realization.

By ruslany

No, Compositor RAD96 doesn’t use CUDA cores

No, Compositor RAD96 doesn’t use CUDA cores

Series of tests were conducted with NVIDIA profiler to prove that none of 480 cores of Compositor RIG 1 uses CUDA technology.

At first, I ran an app of five RAD96 on a single CPU core at 192 kHz audio driver rate. That gave 480 hyperbolic cores at frequencies from 90 GHz to 150 GHz. For example, the working speed of one hyperbolic core is x100 times faster than a single CUDA core of the latest NVIDIA RTX 2080 Ti system.

Taking in account that there are 960 cores in two RIG’s, I want to introduce my new project, which is Virtual Mining Farm (VMF). It is the complete virtualization solution of stacking RAD96 virtual machines. The RAD96 provides effortless stacking in virtualized platform by an activation function interconnection. That activation provides direct stacking of two or more virtual machines. Code of virtual machine compiles only once and is used interactively by all VM’s in a RIG. This means no need in Max 8 MC technology.

Taking in account that I use .json statistics for each dsp process, both internal and external, it proves that such realization is only possible in Max 6 platform.

1 2