USB Driver Stack: improvements for the USB Audio driver

The legacy usb driver stack version 10.x branch is constantly being update by Lars Erdmann and Wim Brul. In 2017 the USB Audio driver got a lot of fixes and new functionality. One part of the patches is for the stream handling that fixes problems after a stream resume in case another stream was played. Starting with driver version 10.217 the usbaudio driver comes with an additional wps ioproc module called “usbwave” which does the on fly sample rate conversion, this is required because the usb audio hardware often supports a fixed sample rate only.

The ioproc need to be installed separately from the USB Audio driver:

usbaudio wave ioproc installation

In case it is correctly installed it should be listed in the WPS Multimedia Codes page:

usbaudio wave ioproc mmcodecs

Audio applications which are using the pmmm interface will work in this setup but audio application/mixer that may talk directly to the audio driver, e.g. uniaud or use the ioclt90 interface will not work.

Team OS/4: OS/4 kernel support for AMD Ryzen CPU

The OS/4 Kernel Team recently updated the OS/4 kernel to support the AMD Ryzen CPU, which has the famous VME bug found by Michal from the os2museum. AMD fixed the issue with a microcode update found in the AGESA 1.0.0.6 BIOS blob.

Downloads:

• updated oemhlp kernel driver which contains a patch for the AMD Ryzen VME bug
• kernel support for the AVX CPU instructions found in recent AMD and Intel CPUs
• new pmdd.sys/pmddk.sys which talks to the new KEE interface but it is also compatible with the IBM Aurora CP SMP Kernel

OS/2 on AMD’s new “Richland” A8 6500T 45W processor

AMD now offers a 45W CPU from the Richland family which was released in the 3/4q 2013. There was again a problem with ACPI like on the Haswell machine, but this time not a crash right from the bootup but a boot problem in stage 2.

OS/2 Desktop running on AMD’s new “Richland” A8 6500T processor

OS/2 on the 4th gen. Core i (Haswell)

The new Haswell CPU from Intel is now available and therefore we could run some tests with this new CPU. The tested model of the CPU was a Core i5-4570. OS/2 can work on this kind of CPU but at the moment it looks like there is a SMP problem with the ACPI PSD driver that prevents the use of all CPU cores, thus with only one core enabled everything functions correctly and the system boots just fine.

OS/2 desktop running on an Intel Haswell Core i5-4570 CPU

First contact: OS/2 on Intels new Ivybridge CPU

Intel released the 3rd generation of the Core family code name IvyBridge. On the 7th of may we had luck to get the hands on a new machine with such a CPU, it was shipped with a Core i7-3770. A desktop screenshot running some benchmarks is below.

Most of the things like ahci and network were working fine, but uniaud failed to initialize and the usb 3.0 controllers were still waiting for the right driver. Sysbench was working ok for all but the CPU bench – it simply froze the system. As it never happened before somebody with a similar CPU can maybe test it and report back. The benchmark result table will be available later, hopefully the next new CPU test will be AMDs already released Mobile Trinity CPU  or the desktop version that will be available in august 2012.

Update July 2012: the CPU bench problem is maybe not related to the IvyBridge CPU itself. There have been some talks about an usb3 driver in the os2world forum. The Trinity Desktop CPU is already available in OEM systems but for end users most likely later than august 2012.