Release v3.7

_includes/imageheading.html

@@ -1,4 +1,11 @@
 {{ include.separator | default: "<hr />" }}
 <figure class="image">
-  <img src="{{ include.url }}" alt="{{ include.heading }}" class="heading" />
+  <img src="{{ include.url }}" class="heading"
+{% if include.heading %}
+alt="{{ include.heading }}"
+{% endif %}
+{% if include.tooltip or include.heading %}
+title="{{ include.tooltip | default: include.heading }}"
+{% endif %}
+/>
 </figure>

_posts/2022-07-30-86box-v3-7.md

@@ -0,0 +1,116 @@
+---
+layout: post
+title: "86Box v3.7"
+description: "Release for July 2022."
+authors: ["richardg867"]
+image: "/assets/images/v3.7/hero.png"
+---
+
+{% include imageheading.html url="/assets/images/v3.7/hero.png" separator=" " tooltip="Legend has it that only those who suck at Doom could ever get the broken build 4000 to work." %}
+
+This is the July 2022 update to 86Box, bringing initial suport for dual monitors, performance improvements and bugfixes. As always, you can [download 86Box v3.7 from GitHub](https://github.com/86Box/86Box/releases/tag/v3.7).
+
+<hr />
+
+## Dual monitors: color and monochrome
+
+86Box v3.7 brings initial support for pairing a **CGA, EGA or VGA** color video card with a secondary **MDA or Hercules** monochrome card. This kind of dual-monitor arrangement is possible because CGA/EGA/VGA and MDA/Hercules use different memory and I/O addresses which usually don't conflict with each other. Some applications are capable of using both cards for display output:
+
+* **DOS** can switch between cards with the `mode co80` (color) and `mode mono` (monochrome) commands;
+* [**FastDoom**](https://github.com/viti95/FastDoom), pictured above, can display the automap on the monochrome card with the `-hercmap` parameter (you still have to press Tab to enable the map);
+* **Lotus 1-2-3** with the `both` display driver can render a graph on the color card while the spreadsheet is being edited on the monochrome card;
+* Some **debuggers** can use the monochrome card for their user interface while the application being debugged runs on the color card, however, 86Box cannot run many debuggers for performance reasons.
+
+As for newer emulated setups running operating systems such as Windows 98, we have plans to extend this feature to **PCI VGA cards** in the future, since many of those have hardware and software support for multi-monitor operation using multiple cards.
+
+<hr />
+
+## Improving performance by slowing hard disks (no, really)
+
+Another new feature brought by this release is a **hard disk speed emulator** by new contributor [**elyosh**](https://github.com/elyosh). You can pick a **speed profile** when adding or editing an IDE or ESDI hard disk on the **Settings** window, which factors in the rotation speed, physical layout and cache size of an average hard disk from around 1989, 1992, 1994, 1996, 1997, 1998 or 2000; alternatively, the **RAM Disk** profile maintains the previous instant-access (as fast as the host disk can read/write) behavior. The emulator is currently only available for **IDE and ESDI hard disks**, but we're hoping to expand it to other disk types in the future.
+
+But wait, there's more! It turns out throttling disk speeds has a positive side effect of **improving emulation performance** (as measured in %), since it effectively spreads the emulated CPU's I/O workload over time; in other words, the CPU has to work less often than it would with an instant-access disk, reducing or even eliminating the **percentage dips** that occur due to interrupt bursts during I/O.
+
+With the significant improvement to emulation performance consistency afforded by disk throttling, we've opted to set **[1997] 5400 RPM** as the default IDE and ESDI speed profiles when migrating an emulated machine from a previous version of 86Box to v3.7. As a result of this change, **you will see a difference** in your setup's operating system speed and responsiveness if it uses IDE or ESDI hard disks, especially if their images are stored in an SSD, but that's a worthy sacrifice in the name of consistent performance, especially while running games. **You can always restore the old behavior without downgrading 86Box** by switching all IDE/ESDI hard disks to the aforementioned **RAM Disk** speed profile.
+
+<hr />
+
+## Changelog
+
+### Emulator
+
+* Added **Discord integration** support for Apple Silicon Macs
+* Fixed black bar on the **OpenGL 3.0 Core renderer**
+* Fixed screenshots on the **Direct3D 9 renderer**
+* Fixed crash when switching to full screen mode the **Direct3D 9 renderer**
+* Fixed screen area being incorrectly resized in **full screen mode**
+* Attempted fix for random FluidSynth
+* Changed audio back-end to **OpenAL** again, please let us know if you have sound issues!
+* Improved some **Linux app metadata** odds and ends
+
+### User interface
+
+* Added global preference to fix the starting directory for **file prompts on macOS**
+* Fixed **Settings section list** being too narrow on some host machines
+* Fixed incorrect button order on the **hard reset confirmation** message
+* Fixed incorrect **icons** on some warning and error messages
+* Fixed incorrect warning message shown when **Ghostscript** is missing
+* Improved **status bar activity indicator** performance
+* Updated translation: **Portuguese (Brazil)**
+
+### Machines
+
+* Added **Soltek SL-56A5** Socket 7 motherboard
+* Added S3 86C805 internal video controller to the **DECpc LPV**
+* Added ability to load an optional language ROM (`roms/machines/ibmps1_2121/F80000.BIN`) on the **IBM PS/1 model 2121**
+* Fixed **Amstrad MegaPC** internal video controller and CMOS password issues
+* Fixed **MR BIOS 386SX clone** and **Acer V10** allowing selection of a non-existent internal video controller
+
+### Hardware
+
+* **Core**
+  * Added **fast PIT** mode for improved performance when emulating 486 and newer CPUs
+  * Fixed miscellaneous issues on machines with the **ALi M1543(C)** southbridge
+  * Fixed Windows NT 3.1 installation issue on **IBM 486** CPUs
+  * Fixed **RTC** timer interrupt emulation inaccuracy
+  * Fixed **PIC** poll mode emulation inaccuracy
+  * Improved **PIT**, **RTC** and **ACPI** timer emulation performance
+* **Display**
+  * Added support for a [**secondary MDA or Hercules**](#dual-monitors-color-and-monochrome) video card
+  * Added **IBM XGA-2** add-on graphics accelerator for MCA (original) and ISA (clone)
+  * Added **Reply Video Adapter** (Cirrus Logic GD5426) MCA video card
+  * Added **BOCA Research 4610** (Cirrus Logic GD5428) ISA video card
+  * Added **Orchid Kelvin EZ** (Cirrus Logic GD5430 and GD5434) VLB video cards
+  * Fixed hard reset crash on the **Trident TGUI 9400CXi**
+  * Fixed corrupted graphics with **S3** cards on BeOS
+  * Fixed 16-bit 1280x1024 and 1600x1200 modes on **ELSA Winner 2000 Pro/X** S3 cards
+  * Fixed Windows 2.10 display issues on the **IBM 8514/A**
+  * Fixed overscan display issue on the **IBM 8514/A**
+  * Fixed 40x25 CGA text mode on the **IBM PGC**
+  * Improved **VGA CRTC** emulation performance
+* **Input**
+  * Improved **joystick** emulation performance when none is selected
+* **Sound**
+  * Added [**ymfm**](https://github.com/aaronsgiles/ymfm) as a faster alternative for OPL2/3 emulation (can be enabled in the **Sound** settings; the more accurate NukedOPL is still default)
+  * Added **Sound Blaster 16 Reply** MCA sound card
+  * Added IRQ and DMA configuration to the **AdLib Gold**
+  * Added IRQ configuration to the MCA **standalone MPU-401**
+  * Added SB MIDI input configuration to the **Sound Blaster Pro MCV**
+  * Fixed **MT-32 LCD message** staying on the status bar after a reset
+  * Fixed **PnP Sound Blaster** cards incorrectly retaining PnP configuration after a reset
+  * Upgraded MT-32 emulation to **Munt** version 2.7.0
+* **Network**
+  * Added **Western Digital WD8013EP/A** MCA network controller
+  * Improved **AMD PCnet** emulation performance
+* **Storage**
+  * Added [**hard disk speed emulator**](#improving-performance-by-slowing-hard-disks-no-really) for IDE and ESDI disks
+  * Added **Rancho RT1000B-MC** MCA SCSI controller
+  * Fixed **tertiary/quaternary IDE** and **cassette** options resetting themselves
+  * Fixed a cue/bin emulation inaccuracy affecting some **copy-protected CDs**
+  * Fixed **IBM PS/2 SCSI Adapter** detecting multiple copies of the same disk
+  * Fixed **Longshine LCS-6821N** SCSI ASPI driver for DOS
+  * Fixed status bar activity indicator on the **IBM ESDI** MCA disk controller
+  * Fixed emulator crash with a `Bad attention request` message with the **IBM ESDI** MCA disk controller
+  * Fixed emulator crash with a `Write_MULTIPLE` message when exiting DOS mode on Windows 3.1 with an IDE hard disk
+  * Fixed emulator crash with a `Hard disk image` message when a host I/O error occurs on a **VHD image**
+