Seeking suggestions to run GLideN64 with mupen64plus on RK3328 Mali-450

[wn2000]

Background:
I'm trying to run mupen64plus on a RK3328 Mali-450 aarch64 platform.
Tried several video plugins but always have the issue that video is only rendered at a lower left corner of the screen, similar to the issue reported here: RetroPie/RetroPie-Setup#3118
My monitor's resolution is 1280x720, so I can use --resolution 1280x720 to force it fill the screen. But games run a lot slower than --resolution 320x240. What I wanted to achieve is to render at an internal resolution of 320x240 but scale it to the 1280x720 resolution for display.
I was told that GLideN64 supports that with the UseNativeResolutionFactor = 1 option so I'm very interested to try it out.

Issues
I built it with cmake -DEGL=On -DMUPENPLUSAPI=On ../src but got a segfault. Did some tweaking based on an old issue and disabled the check of EGL extensions (EGLImage was not used in case of GLES2 anyway) to finally pass the segfault.
But then the screen is all blank. The music plays in the background and sounds smooth.
Then I found I have to set EnableLegacyBlending = True for the graphics to show. But once this is set, it becomes really slow (~10 FPS).

So I want to see if anybody has successfully used this on Mali-450 with good performance.

[dankcushions]

continuing our discussion from mupen64plus/mupen64plus-core#813, please provide your entire mupen64plus.cfg and your launch log (with --verbose may be useful)

FYI, the GLideN64 part of my CFG (for a pi4) is

[Video-GLideN64]

# Settings version. Don't touch it.
configVersion = 29
# Bilinear filtering mode (0=N64 3point, 1=standard)
bilinearMode = 1
# Enable frame and|or depth buffer emulation.
EnableFBEmulation = True
# Frame buffer size is the factor of N64 native resolution.
UseNativeResolutionFactor = 1
# Do not use shaders to emulate N64 blending modes. Works faster on slow GPU. Can cause glitches.
EnableLegacyBlending = True
# Enable threaded video backend
ThreadedVideo = True
# Swap frame buffers (0=On VI update call, 1=On VI origin change, 2=On buffer update)
BufferSwapMode = 2
# Enable hybrid integer scaling filter, this can be slow with low-end GPUs
EnableHybridFilter = False
# Enable/Disable MultiSampling (0=off, 2,4,8,16=quality)
MultiSampling = 0
# Enable/Disable Fast Approximate Anti-Aliasing FXAA
FXAA = False
# Screen aspect ratio (0=stretch, 1=force 4:3, 2=force 16:9, 3=adjust)
AspectRatio = 1
# Remove halos around filtered textures.
enableHalosRemoval = False
# Max level of Anisotropic Filtering, 0 for off
MaxAnisotropy = 0
# Enable dithering pattern on output image.
EnableDitheringPattern = False
# Enable hi-res noise dithering.
EnableHiresNoiseDithering = False
# Dither with color quantization.
DitheringQuantization = True
# Dithering mode for image in RDRAM. (0=disable, 1=bayer, 2=magic square, 3=blue noise)
RDRAMImageDitheringMode = 3
# Enable LOD emulation.
EnableLOD = True
# Enable hardware per-pixel lighting.
EnableHWLighting = False
# Use persistent storage for compiled shaders.
EnableShadersStorage = True
# Enable writing of fragment depth. Some mobile GPUs do not support it, thus it made optional. Leave enabled.
EnableFragmentDepthWrite = True
# Use GLideN64 per-game settings.
EnableCustomSettings = True
# Make texrect coordinates continuous to avoid black lines between them. (0=Off, 1=Auto, 2=Force)
CorrectTexrectCoords = 0
# Render 2D texrects in native resolution to fix misalignment between parts of 2D image. (0=Off, 1=Optimized, 2=Unoptimized)
EnableNativeResTexrects = 0
# Render backgrounds mode (HLE only). (0=One piece (fast), 1=Stripped (precise))
BackgroundsMode = 1
# Copy auxiliary buffers to RDRAM
EnableCopyAuxiliaryToRDRAM = False
# Enable N64 depth compare instead of OpenGL standard one. Experimental. (0=Off, 1=Fast, 2=Compatible)
EnableN64DepthCompare = 0
# Force depth buffer clear. Hack. Needed for Eikou no Saint Andrews.
ForceDepthBufferClear = False
# Disable buffers read/write with FBInfo. Use for games, which do not work with FBInfo.
DisableFBInfo = True
# Read color buffer by 4kb chunks (strict follow to FBRead specification)
FBInfoReadColorChunk = False
# Read depth buffer by 4kb chunks (strict follow to FBRead specification)
FBInfoReadDepthChunk = True
# Enable color buffer copy to RDRAM (0=do not copy, 1=copy in sync mode, 2=Double Buffer, 3=Triple Buffer)
EnableCopyColorToRDRAM = 0
# Enable depth buffer copy to RDRAM  (0=do not copy, 1=copy from video memory, 2=use software render)
EnableCopyDepthToRDRAM = 2
# Enable color buffer copy from RDRAM.
EnableCopyColorFromRDRAM = False
# Enable resulted image crop by Overscan.
EnableOverscan = True
# PAL mode. Left bound of Overscan
OverscanPalLeft = 0
# PAL mode. Right bound of Overscan
OverscanPalRight = 0
# PAL mode. Top bound of Overscan
OverscanPalTop = 0
# PAL mode. Bottom bound of Overscan
OverscanPalBottom = 0
# NTSC mode. Left bound of Overscan
OverscanNtscLeft = 0
# NTSC mode. Right bound of Overscan
OverscanNtscRight = 0
# NTSC mode. Top bound of Overscan
OverscanNtscTop = 0
# NTSC mode. Bottom bound of Overscan
OverscanNtscBottom = 0
# Texture filter (0=none, 1=Smooth filtering 1, 2=Smooth filtering 2, 3=Smooth filtering 3, 4=Smooth filtering 4, 5=Sharp filtering 1, 6=Sharp filtering 2)
txFilterMode = 0
# Texture Enhancement (0=none, 1=store as is, 2=X2, 3=X2SAI, 4=HQ2X, 5=HQ2XS, 6=LQ2X, 7=LQ2XS, 8=HQ4X, 9=2xBRZ, 10=3xBRZ, 11=4xBRZ, 12=5xBRZ), 13=6xBRZ
txEnhancementMode = 0
# Deposterize texture before enhancement.
txDeposterize = False
# Don't filter background textures.
txFilterIgnoreBG = False
# Size of memory cache for enhanced textures in megabytes.
txCacheSize = 100
# Use high-resolution texture packs if available.
txHiresEnable = True
# Allow to use alpha channel of high-res texture fully.
txHiresFullAlphaChannel = True
# Use alternative method of paletted textures CRC calculation.
txHresAltCRC = False
# Enable dump of loaded N64 textures.
txDump = False
# Zip textures cache.
txCacheCompression = True
# Force use 16bit texture formats for HD textures.
txForce16bpp = False
# Save texture cache to hard disk.
txSaveCache = True
# Use file storage instead of memory cache for enhanced textures.
txEnhancedTextureFileStorage = False
# Use file storage instead of memory cache for HD textures.
txHiresTextureFileStorage = False
# Path to folder with hi-res texture packs.
txPath = "/home/pi/.local/share/mupen64plus/hires_texture"
# Path to folder where plugin saves texture cache files.
txCachePath = "/home/pi/.cache/mupen64plus/cache"
# Path to folder where plugin saves dumped textures.
txDumpPath = "/home/pi/.cache/mupen64plus/texture_dump"
# File name of True Type Font for text messages.
fontName = "FreeSans

that said, it's not that unusual for SBCs to tank in certain areas/games in GLideN64 - examples: #2329. However something like mario kart 64 or mario 64 should be mostly full speed in game.

[wn2000]

Thanks @dankcushions .
Here is my mupen64plus.cfg:

# Mupen64Plus Configuration File
# This file is automatically read and written by the Mupen64Plus Core library

[64DD]

# Filename of the 64DD IPL ROM
IPL-ROM = ""
# Filename of the disk to load into Disk Drive
Disk = ""


[Audio-SDL]

# Mupen64Plus SDL Audio Plugin config parameter version number
Version = 1.000000
# Frequency which is used if rom doesn't want to change it
DEFAULT_FREQUENCY = 33600
# Swaps left and right channels
SWAP_CHANNELS = False
# Size of primary buffer in output samples. This is where audio is loaded after it's extracted from n64's memory.
PRIMARY_BUFFER_SIZE = 16384
# Fullness level target for Primary audio buffer, in equivalent output samples. This value must be larger than the SECONDARY_BUFFER_SIZE. Decreasing this value will reduce audio latency but requires a faster PC to avoid choppiness. Increasing this will increase audio latency but reduce the chance of drop-outs.
PRIMARY_BUFFER_TARGET = 8192
# Size of secondary buffer in output samples. This is SDL's hardware buffer. The SDL documentation states that this should be a power of two between 512 and 8192.
SECONDARY_BUFFER_SIZE = 1024
# Audio resampling algorithm. src-sinc-best-quality, src-sinc-medium-quality, src-sinc-fastest, src-zero-order-hold, src-linear, speex-fixed-{10-0}, trivial
RESAMPLE = "trivial"
# Volume control type: 1 = SDL (only affects Mupen64Plus output)  2 = OSS mixer (adjusts master PC volume)
VOLUME_CONTROL_TYPE = 1
# Percentage change each time the volume is increased or decreased
VOLUME_ADJUST = 5
# Default volume when a game is started.  Only used if VOLUME_CONTROL_TYPE is 1
VOLUME_DEFAULT = 80
# AUDIO_SYNC = True
AUDIO_SYNC = False


[Core]

# Mupen64Plus Core config parameter set version number.  Please don't change this version number.
Version = 1.010000
# Draw on-screen display if True, otherwise don't draw OSD
OnScreenDisplay = False
# Use Pure Interpreter if 0, Cached Interpreter if 1, or Dynamic Recompiler if 2 or more
R4300Emulator = 2
# Disable compiled jump commands in dynamic recompiler (should be set to False)
NoCompiledJump = False
# Disable 4MB expansion RAM pack. May be necessary for some games
DisableExtraMem = False
# Increment the save state slot after each save operation
AutoStateSlotIncrement = False
# Activate the R4300 debugger when ROM execution begins, if core was built with Debugger support
EnableDebugger = False
# Save state slot (0-9) to use when saving/loading the emulator state
CurrentStateSlot = 0
# Path to directory where screenshots are saved. If this is blank, the default value of ${UserDataPath}/screenshot will be used
ScreenshotPath = "/tmp"
# Path to directory where emulator save states (snapshots) are saved. If this is blank, the default value of ${UserDataPath}/save will be used
SaveStatePath = ""
# Path to directory where SRAM/EEPROM data (in-game saves) are stored. If this is blank, the default value of ${UserDataPath}/save will be used
SaveSRAMPath = ""
# Path to a directory to search when looking for shared data files
SharedDataPath = "/tmp/mupen64plus/share/mupen64plus"
# Force number of cycles per emulated instruction
CountPerOp = 0
# Randomize PI/SI Interrupt Timing
RandomizeInterrupt = True
# Duration of SI DMA (-1: use per game settings)
SiDmaDuration = -1
# Gameboy Camera Video Capture backend
GbCameraVideoCaptureBackend1 = ""
# Disk Save Format (0: Full Disk Copy (*.ndr/*.d6r), 1: RAM Area Only (*.ram))
SaveDiskFormat = 1


[CoreEvents]

# Mupen64Plus CoreEvents config parameter set version number.  Please don't change this version number.
Version = 1.000000
# SDL keysym for stopping the emulator
Kbd Mapping Stop = 27
# SDL keysym for switching between fullscreen/windowed modes
Kbd Mapping Fullscreen = 0
# SDL keysym for saving the emulator state
Kbd Mapping Save State = 286
# SDL keysym for loading the emulator state
Kbd Mapping Load State = 288
# SDL keysym for advancing the save state slot
Kbd Mapping Increment Slot = 0
# SDL keysym for resetting the emulator
Kbd Mapping Reset = 290
# SDL keysym for slowing down the emulator
Kbd Mapping Speed Down = 291
# SDL keysym for speeding up the emulator
Kbd Mapping Speed Up = 292
# SDL keysym for taking a screenshot
Kbd Mapping Screenshot = 293
# SDL keysym for pausing the emulator
Kbd Mapping Pause = 112
# SDL keysym for muting/unmuting the sound
Kbd Mapping Mute = 109
# SDL keysym for increasing the volume
Kbd Mapping Increase Volume = 93
# SDL keysym for decreasing the volume
Kbd Mapping Decrease Volume = 91
# SDL keysym for temporarily going really fast
Kbd Mapping Fast Forward = 102
# SDL keysym for advancing by one frame when paused
Kbd Mapping Frame Advance = 47
# SDL keysym for pressing the game shark button
Kbd Mapping Gameshark = 103
# Joystick event string for stopping the emulator
Joy Mapping Stop = ""
# Joystick event string for switching between fullscreen/windowed modes
Joy Mapping Fullscreen = ""
# Joystick event string for saving the emulator state
Joy Mapping Save State = ""
# Joystick event string for loading the emulator state
Joy Mapping Load State = ""
# Joystick event string for advancing the save state slot
Joy Mapping Increment Slot = ""
# Joystick event string for resetting the emulator
Joy Mapping Reset = ""
# Joystick event string for slowing down the emulator
Joy Mapping Speed Down = ""
# Joystick event string for speeding up the emulator
Joy Mapping Speed Up = ""
# Joystick event string for taking a screenshot
Joy Mapping Screenshot = ""
# Joystick event string for pausing the emulator
Joy Mapping Pause = ""
# Joystick event string for muting/unmuting the sound
Joy Mapping Mute = ""
# Joystick event string for increasing the volume
Joy Mapping Increase Volume = ""
# Joystick event string for decreasing the volume
Joy Mapping Decrease Volume = ""
# Joystick event string for fast-forward
Joy Mapping Fast Forward = ""
# Joystick event string for advancing by one frame when paused
Joy Mapping Frame Advance = ""
# Joystick event string for pressing the game shark button
Joy Mapping Gameshark = ""


[Input-SDL-Control1]

# Mupen64Plus SDL Input Plugin config parameter version number.  Please don't change this version number.
version = 2.000000
# Controller configuration mode: 0=Fully Manual, 1=Auto with named SDL Device, 2=Fully automatic
mode = 2
# Specifies which joystick is bound to this controller: -1=No joystick, 0 or more= SDL Joystick number
device = -1
# SDL joystick name (or Keyboard)
name = "Keyboard"
# Specifies whether this controller is 'plugged in' to the simulated N64
plugged = True
# Specifies which type of expansion pak is in the controller: 1=None, 2=Mem pak, 4=Transfer pak, 5=Rumble pak
plugin = 2
# If True, then mouse buttons may be used with this controller
mouse = False
# Scaling factor for mouse movements.  For X, Y axes.
MouseSensitivity = "2.00,2.00"
# The minimum absolute value of the SDL analog joystick axis to move the N64 controller axis value from 0.  For X, Y axes.
AnalogDeadzone = "4096,4096"
# An absolute value of the SDL joystick axis >= AnalogPeak will saturate the N64 controller axis value (at 80).  For X, Y axes. For each axis, this must be greater than the corresponding AnalogDeadzone value
AnalogPeak = "32768,32768"
# Digital button configuration mappings
DPad R = "key(100)"
DPad L = "key(97)"
DPad D = "key(115)"
DPad U = "key(119)"
Start = "key(13)"
Z Trig = "key(122)"
B Button = "key(306)"
A Button = "key(304)"
C Button R = "key(108)"
C Button L = "key(106)"
C Button D = "key(107)"
C Button U = "key(105)"
R Trig = "key(99)"
L Trig = "key(120)"
Mempak switch = "key(44)"
Rumblepak switch = "key(46)"
# Analog axis configuration mappings
X Axis = "key(276,275)"
Y Axis = "key(273,274)"


[Input-SDL-Control2]

# Mupen64Plus SDL Input Plugin config parameter version number.  Please don't change this version number.
version = 2.000000
# Controller configuration mode: 0=Fully Manual, 1=Auto with named SDL Device, 2=Fully automatic
mode = 2
# Specifies which joystick is bound to this controller: -1=No joystick, 0 or more= SDL Joystick number
device = -1
# SDL joystick name (or Keyboard)
name = ""
# Specifies whether this controller is 'plugged in' to the simulated N64
plugged = True
# Specifies which type of expansion pak is in the controller: 1=None, 2=Mem pak, 4=Transfer pak, 5=Rumble pak
plugin = 1
# If True, then mouse buttons may be used with this controller
mouse = False
# Scaling factor for mouse movements.  For X, Y axes.
MouseSensitivity = "2.00,2.00"
# The minimum absolute value of the SDL analog joystick axis to move the N64 controller axis value from 0.  For X, Y axes.
AnalogDeadzone = "4096,4096"
# An absolute value of the SDL joystick axis >= AnalogPeak will saturate the N64 controller axis value (at 80).  For X, Y axes. For each axis, this must be greater than the corresponding AnalogDeadzone value
AnalogPeak = "32768,32768"
# Digital button configuration mappings
DPad R = "hat(0 Right)"
DPad L = "hat(0 Left)"
DPad D = "hat(0 Down)"
DPad U = "hat(0 Up)"
Start = "button(9)"
Z Trig = ""
B Button = "button(1)"
A Button = "button(2)"
C Button R = "button(7)"
C Button L = "button(0)"
C Button D = "button(5)"
C Button U = "axis(3)"
R Trig = "button(12)"
L Trig = "button(6)"
Mempak switch = ""
Rumblepak switch = ""
# Analog axis configuration mappings
X Axis = "hat(0 Left Right)"
Y Axis = "hat(0 Up Down)"


[Input-SDL-Control3]

# Mupen64Plus SDL Input Plugin config parameter version number.  Please don't change this version number.
version = 2.000000
# Controller configuration mode: 0=Fully Manual, 1=Auto with named SDL Device, 2=Fully automatic
mode = 2
# Specifies which joystick is bound to this controller: -1=No joystick, 0 or more= SDL Joystick number
device = -1
# SDL joystick name (or Keyboard)
name = ""
# Specifies whether this controller is 'plugged in' to the simulated N64
plugged = False
# Specifies which type of expansion pak is in the controller: 1=None, 2=Mem pak, 4=Transfer pak, 5=Rumble pak
plugin = 2
# If True, then mouse buttons may be used with this controller
mouse = False
# Scaling factor for mouse movements.  For X, Y axes.
MouseSensitivity = "2.00,2.00"
# The minimum absolute value of the SDL analog joystick axis to move the N64 controller axis value from 0.  For X, Y axes.
AnalogDeadzone = "4096,4096"
# An absolute value of the SDL joystick axis >= AnalogPeak will saturate the N64 controller axis value (at 80).  For X, Y axes. For each axis, this must be greater than the corresponding AnalogDeadzone value
AnalogPeak = "32768,32768"
# Digital button configuration mappings
DPad R = ""
DPad L = ""
DPad D = ""
DPad U = ""
Start = ""
Z Trig = ""
B Button = ""
A Button = ""
C Button R = ""
C Button L = ""
C Button D = ""
C Button U = ""
R Trig = ""
L Trig = ""
Mempak switch = ""
Rumblepak switch = ""
# Analog axis configuration mappings
X Axis = ""
Y Axis = ""


[Input-SDL-Control4]

# Mupen64Plus SDL Input Plugin config parameter version number.  Please don't change this version number.
version = 2.000000
# Controller configuration mode: 0=Fully Manual, 1=Auto with named SDL Device, 2=Fully automatic
mode = 2
# Specifies which joystick is bound to this controller: -1=No joystick, 0 or more= SDL Joystick number
device = -1
# SDL joystick name (or Keyboard)
name = ""
# Specifies whether this controller is 'plugged in' to the simulated N64
plugged = False
# Specifies which type of expansion pak is in the controller: 1=None, 2=Mem pak, 4=Transfer pak, 5=Rumble pak
plugin = 2
# If True, then mouse buttons may be used with this controller
mouse = False
# Scaling factor for mouse movements.  For X, Y axes.
MouseSensitivity = "2.00,2.00"
# The minimum absolute value of the SDL analog joystick axis to move the N64 controller axis value from 0.  For X, Y axes.
AnalogDeadzone = "4096,4096"
# An absolute value of the SDL joystick axis >= AnalogPeak will saturate the N64 controller axis value (at 80).  For X, Y axes. For each axis, this must be greater than the corresponding AnalogDeadzone value
AnalogPeak = "32768,32768"
# Digital button configuration mappings
DPad R = ""
DPad L = ""
DPad D = ""
DPad U = ""
Start = ""
Z Trig = ""
B Button = ""
A Button = ""
C Button R = ""
C Button L = ""
C Button D = ""
C Button U = ""
R Trig = ""
L Trig = ""
Mempak switch = ""
Rumblepak switch = ""
# Analog axis configuration mappings
X Axis = ""
Y Axis = ""


[Rsp-HLE]

# Mupen64Plus RSP HLE Plugin config parameter version number
Version = 1.000000
# Path to a RSP plugin which will be used when encountering an unknown ucode.You can disable this by letting an empty string.
RspFallback = ""
# Send display lists to the graphics plugin
DisplayListToGraphicsPlugin = True
# Send audio lists to the audio plugin
AudioListToAudioPlugin = False


[Transferpak]

# Filename of the GB ROM to load into transferpak 1
GB-rom-1 = ""
# Filename of the GB RAM to load into transferpak 1
GB-ram-1 = ""
# Filename of the GB ROM to load into transferpak 2
GB-rom-2 = ""
# Filename of the GB RAM to load into transferpak 2
GB-ram-2 = ""
# Filename of the GB ROM to load into transferpak 3
GB-rom-3 = ""
# Filename of the GB RAM to load into transferpak 3
GB-ram-3 = ""
# Filename of the GB ROM to load into transferpak 4
GB-rom-4 = ""
# Filename of the GB RAM to load into transferpak 4
GB-ram-4 = ""


[UI-Console]

# Mupen64Plus UI-Console config parameter set version number.  Please don't change this version number.
Version = 1.000000
# Directory in which to search for plugins
PluginDir = "/tmp/mupen64plus/lib/mupen64plus"
# VideoPlugin = "mupen64plus-video-n64.so"
VideoPlugin = "mupen64plus-video-GLideN64.so"
# Filename of audio plugin
AudioPlugin = "mupen64plus-audio-sdl.so"
# Filename of input plugin
InputPlugin = "mupen64plus-input-sdl.so"
# Filename of RSP plugin
RspPlugin = "mupen64plus-rsp-hle.so"


[Video-General]

# Use fullscreen mode if True, or windowed mode if False
Fullscreen = True
# If true, activate the SDL_GL_SWAP_CONTROL attribute
VerticalSync = False
# Rotate screen contents: 0=0 degree, 1=90 degree, 2 = 180 degree, 3=270 degree
Rotate = 0
# Width of output window or fullscreen width
ScreenWidth = 1280
# Height of output window or fullscreen height
ScreenHeight = 720


[Video-Glide64mk2]

# Enable full-scene anti-aliasing by setting this to a value greater than 1
wrpAntiAliasing = 0
# Card ID
card_id = 0
# If true, use polygon offset values specified below
force_polygon_offset = False
# Specifies a scale factor that is used to create a variable depth offset for each polygon
polygon_offset_factor = 0.000000
# Is multiplied by an implementation-specific value to create a constant depth offset
polygon_offset_units = 0.000000
# Vertical sync
vsync = False
# TODO:ssformat
ssformat = False
# Display performance stats (add together desired flags): 1=FPS counter, 2=VI/s counter, 4=% speed, 8=FPS transparent
show_fps = 1
# Clock enabled
clock = False
# Clock is 24-hour
clock_24_hr = True
# Wrapper resolution
wrpResolution = 0
# Wrapper VRAM
wrpVRAM = 0
# Wrapper FBO
wrpFBO = True
# Wrapper Anisotropic Filtering
wrpAnisotropic = True
# Texture Enhancement: Smooth/Sharpen Filters
ghq_fltr = 0
# Texture Compression: 0 for S3TC, 1 for FXT1
ghq_cmpr = 0
# Texture Enhancement: More filters
ghq_enht = 0
# Hi-res texture pack format (0 for none, 1 for Rice)
ghq_hirs = 0
# Compress texture cache with S3TC or FXT1
ghq_enht_cmpr = False
# Tile textures (saves memory but could cause issues)
ghq_enht_tile = 0
# Force 16bpp textures (saves ram but lower quality)
ghq_enht_f16bpp = False
# Compress texture cache
ghq_enht_gz = True
# Don't enhance textures for backgrounds
ghq_enht_nobg = False
# Enable S3TC and FXT1 compression
ghq_hirs_cmpr = False
# Tile hi-res textures (saves memory but could cause issues)
ghq_hirs_tile = False
# Force 16bpp hi-res textures (saves ram but lower quality)
ghq_hirs_f16bpp = False
# Compress hi-res texture cache
ghq_hirs_gz = True
# Alternative CRC calculation -- emulates Rice bug
ghq_hirs_altcrc = True
# Save tex cache to disk
ghq_cache_save = True
# Texture Cache Size (MB)
ghq_cache_size = 128
# Use full alpha channel -- could cause issues for some tex packs
ghq_hirs_let_texartists_fly = False
# Dump textures
ghq_hirs_dump = False
# Alternate texture size method: -1=Game default, 0=disable. 1=enable
alt_tex_size = -1
# Use first SETTILESIZE only: -1=Game default, 0=disable. 1=enable
use_sts1_only = -1
# Use spheric mapping only: -1=Game default, 0=disable. 1=enable
force_calc_sphere = -1
# Force positive viewport: -1=Game default, 0=disable. 1=enable
correct_viewport = -1
# Force texrect size to integral value: -1=Game default, 0=disable. 1=enable
increase_texrect_edge = -1
# Reduce fillrect size by 1: -1=Game default, 0=disable. 1=enable
decrease_fillrect_edge = -1
# Enable perspective texture correction emulation: -1=Game default, 0=disable. 1=enable
texture_correction = -1
# Set special scale for PAL games: -1=Game default, 0=disable. 1=enable
pal230 = -1
# 3DFX Dithered alpha emulation mode: -1=Game default, >=0=dithered alpha emulation mode
stipple_mode = -1
# 3DFX Dithered alpha pattern: -1=Game default, >=0=pattern used for dithered alpha emulation
stipple_pattern = -1
# Check microcode each frame: -1=Game default, 0=disable. 1=enable
force_microcheck = -1
# Force 0xb5 command to be quad, not line 3D: -1=Game default, 0=disable. 1=enable
force_quad3d = -1
# Enable near z clipping: -1=Game default, 0=disable. 1=enable
clip_zmin = -1
# Enable far plane clipping: -1=Game default, 0=disable. 1=enable
clip_zmax = -1
# Use fast CRC algorithm: -1=Game default, 0=disable. 1=enable
fast_crc = -1
# Adjust screen aspect for wide screen mode: -1=Game default, 0=disable. 1=enable
adjust_aspect = 1
# Force strict check in Depth buffer test: -1=Game default, 0=disable. 1=enable
zmode_compare_less = -1
# Apply alpha dither regardless of alpha_dither_mode: -1=Game default, 0=disable. 1=enable
old_style_adither = -1
# Scale vertex z value before writing to depth buffer: -1=Game default, 0=disable. 1=enable
n64_z_scale = -1
# Fast texrect rendering with hwfbe: -1=Game default, 0=disable. 1=enable
optimize_texrect = -1
# Do not copy auxiliary frame buffers: -1=Game default, 0=disable. 1=enable
ignore_aux_copy = -1
# Clear auxiliary texture frame buffers: -1=Game default, 0=disable. 1=enable
hires_buf_clear = -1
# Read alpha from framebuffer: -1=Game default, 0=disable. 1=enable
fb_read_alpha = -1
# Handle unchanged fb: -1=Game default, 0=disable. 1=enable
useless_is_useless = -1
# Set frambuffer CRC mode: -1=Game default, 0=disable CRC, 1=fast CRC, 2=safe CRC
fb_crc_mode = -1
# Filtering mode: -1=Game default, 0=automatic, 1=force bilinear, 2=force point sampled
filtering = -1
# Fog: -1=Game default, 0=disable. 1=enable
fog = -1
# Buffer clear on every frame: -1=Game default, 0=disable. 1=enable
buff_clear = -1
# Buffer swapping method: -1=Game default, 0=swap buffers when vertical interrupt has occurred, 1=swap buffers when set of conditions is satisfied. Prevents flicker on some games, 2=mix of first two methods
swapmode = -1
# Aspect ratio: -1=Game default, 0=Force 4:3, 1=Force 16:9, 2=Stretch, 3=Original
aspect = -1
# LOD calculation: -1=Game default, 0=disable. 1=fast, 2=precise
lodmode = -1
# Smart framebuffer: -1=Game default, 0=disable. 1=enable
fb_smart = -1
# Hardware frame buffer emulation: -1=Game default, 0=disable. 1=enable
fb_hires = -1
# Read framebuffer every frame (may be slow use only for effects that need it e.g. Banjo Kazooie, DK64 transitions): -1=Game default, 0=disable. 1=enable
fb_read_always = -1
# Render N64 frame buffer as texture: -1=Game default, 0=disable, 1=mode1, 2=mode2
read_back_to_screen = -1
# Show images written directly by CPU: -1=Game default, 0=disable. 1=enable
detect_cpu_write = -1
# Get frame buffer info: -1=Game default, 0=disable. 1=enable
fb_get_info = -1
# Enable software depth render: -1=Game default, 0=disable. 1=enable
fb_render = -1


[Video-GLideN64]

# Settings version. Don't touch it.
configVersion = 29
# Enable threaded video backend
ThreadedVideo = True
# Enable/Disable MultiSampling (0=off, 2,4,8,16=quality)
MultiSampling = 0
# Enable/Disable Fast Approximate Anti-Aliasing FXAA
FXAA = False
# Screen aspect ratio (0=stretch, 1=force 4:3, 2=force 16:9, 3=adjust)
AspectRatio = 1
# Swap frame buffers (0=On VI update call, 1=On VI origin change, 2=On buffer update)
BufferSwapMode = 0
# Frame buffer size is the factor of N64 native resolution.
UseNativeResolutionFactor = 1
# Bilinear filtering mode (0=N64 3point, 1=standard)
bilinearMode = True
# Remove halos around filtered textures.
enableHalosRemoval = False
# Max level of Anisotropic Filtering, 0 for off
MaxAnisotropy = 0
# Enable dithering pattern on output image.
EnableDitheringPattern = False
# Enable hi-res noise dithering.
EnableHiresNoiseDithering = False
# Dither with color quantization.
DitheringQuantization = False
# Dithering mode for image in RDRAM. (0=disable, 1=bayer, 2=magic square, 3=blue noise)
RDRAMImageDitheringMode = 0
# Enable LOD emulation.
EnableLOD = True
# Enable hardware per-pixel lighting.
EnableHWLighting = False
# Use persistent storage for compiled shaders.
EnableShadersStorage = True
# Do not use shaders to emulate N64 blending modes. Works faster on slow GPU. Can cause glitches.
EnableLegacyBlending = True
# Enable hybrid integer scaling filter, this can be slow with low-end GPUs
EnableHybridFilter = False
# Enable writing of fragment depth. Some mobile GPUs do not support it, thus it made optional. Leave enabled.
EnableFragmentDepthWrite = True
# Use GLideN64 per-game settings.
EnableCustomSettings = True
# Make texrect coordinates continuous to avoid black lines between them. (0=Off, 1=Auto, 2=Force)
CorrectTexrectCoords = 0
# Render 2D texrects in native resolution to fix misalignment between parts of 2D image. (0=Off, 1=Optimized, 2=Unoptimized)
EnableNativeResTexrects = 1
# Render backgrounds mode (HLE only). (0=One piece (fast), 1=Stripped (precise))
BackgroundsMode = 0
# Enable frame and|or depth buffer emulation.
EnableFBEmulation = False
# Copy auxiliary buffers to RDRAM
EnableCopyAuxiliaryToRDRAM = False
# Enable N64 depth compare instead of OpenGL standard one. Experimental. (0=Off, 1=Fast, 2=Compatible)
EnableN64DepthCompare = 0
# Force depth buffer clear. Hack. Needed for Eikou no Saint Andrews.
ForceDepthBufferClear = False
# Disable buffers read/write with FBInfo. Use for games, which do not work with FBInfo.
DisableFBInfo = True
# Read color buffer by 4kb chunks (strict follow to FBRead specification)
FBInfoReadColorChunk = False
# Read depth buffer by 4kb chunks (strict follow to FBRead specification)
FBInfoReadDepthChunk = True
# Enable color buffer copy to RDRAM (0=do not copy, 1=copy in sync mode, 2=Double Buffer, 3=Triple Buffer)
EnableCopyColorToRDRAM = 0
# Enable depth buffer copy to RDRAM  (0=do not copy, 1=copy from video memory, 2=use software render)
EnableCopyDepthToRDRAM = 0
# Enable color buffer copy from RDRAM.
EnableCopyColorFromRDRAM = False
# Enable resulted image crop by Overscan.
EnableOverscan = False
# PAL mode. Left bound of Overscan
OverscanPalLeft = 0
# PAL mode. Right bound of Overscan
OverscanPalRight = 0
# PAL mode. Top bound of Overscan
OverscanPalTop = 0
# PAL mode. Bottom bound of Overscan
OverscanPalBottom = 0
# NTSC mode. Left bound of Overscan
OverscanNtscLeft = 0
# NTSC mode. Right bound of Overscan
OverscanNtscRight = 0
# NTSC mode. Top bound of Overscan
OverscanNtscTop = 0
# NTSC mode. Bottom bound of Overscan
OverscanNtscBottom = 0
# Texture filter (0=none, 1=Smooth filtering 1, 2=Smooth filtering 2, 3=Smooth filtering 3, 4=Smooth filtering 4, 5=Sharp filtering 1, 6=Sharp filtering 2)
txFilterMode = 0
# Texture Enhancement (0=none, 1=store as is, 2=X2, 3=X2SAI, 4=HQ2X, 5=HQ2XS, 6=LQ2X, 7=LQ2XS, 8=HQ4X, 9=2xBRZ, 10=3xBRZ, 11=4xBRZ, 12=5xBRZ), 13=6xBRZ
txEnhancementMode = 0
# Deposterize texture before enhancement.
txDeposterize = False
# Don't filter background textures.
txFilterIgnoreBG = False
# Size of memory cache for enhanced textures in megabytes.
txCacheSize = 100
# Use high-resolution texture packs if available.
txHiresEnable = False
# Allow to use alpha channel of high-res texture fully.
txHiresFullAlphaChannel = True
# Use alternative method of paletted textures CRC calculation.
txHresAltCRC = False
# Zip textures cache.
txCacheCompression = True
# Force use 16bit texture formats for HD textures.
txForce16bpp = False
# Save texture cache to hard disk.
txSaveCache = True
# Use file storage instead of memory cache for enhanced textures.
txEnhancedTextureFileStorage = False
# Use file storage instead of memory cache for HD textures.
txHiresTextureFileStorage = False
# Path to folder with hi-res texture packs.
txPath = "/tmp/mupen64plus/hires_texture"
# Path to folder where plugin saves texture cache files.
txCachePath = "/tmp/mupen64plus/cache"
# Path to folder where plugin saves dumped textures.
txDumpPath = "/tmp/mupen64plus/texture_dump"
# File name of True Type Font for text messages.
fontName = "FreeSans.ttf"
# Font size.
fontSize = 18
# Font color in RGB format.
fontColor = "B5E61D"
# Force gamma correction.
ForceGammaCorrection = False
# Gamma correction level.
GammaCorrectionLevel = 2.000000
# Show FPS counter.
ShowFPS = True
# Show VI/S counter.
ShowVIS = False
# Show percent counter.
ShowPercent = False
# Show internal resolution.
ShowInternalResolution = True
# Show rendering resolution.
ShowRenderingResolution = True
# Counters position (1=top left, 2=top center, 4=top right, 8=bottom left, 16=bottom center, 32=bottom right)
CountersPos = 8
# Hotkey: toggle textures dump
hkTexDump = ""
# Hotkey: reload HD textures
hkHdTexReload = ""
# Hotkey: toggle HD textures
hkHdTexToggle = ""
# Hotkey: toggle VSync
hkVsync = ""
# Hotkey: toggle frame buffer emulation
hkFBEmulation = ""
# Hotkey: toggle N64 depth compare
hkN64DepthCompare = ""
# Hotkey: toggle OSD VI/S
hkOsdVis = ""
# Hotkey: toggle OSD FPS
hkOsdFps = ""
# Hotkey: toggle OSD percent
hkOsdPercent = ""
# Hotkey: toggle OSD internal resolution
hkOsdInternalResolution = ""
# Hotkey: toggle OSD rendering resolution
hkOsdRenderingResolution = ""
# Hotkey: toggle force gamma correction
hkForceGammaCorrection = ""


[Video-Rice]

# Mupen64Plus Rice Video Plugin config parameter version number
Version = 1
# Frame Buffer Emulation (0=ROM default, 1=disable)
FrameBufferSetting = 0
# Frequency to write back the frame buffer (0=every frame, 1=every other frame, etc)
FrameBufferWriteBackControl = 0
# RenderToTexture = 0
RenderToTexture = 2
# Control when the screen will be updated (0=ROM default, 1=VI origin update, 2=VI origin change, 3=CI change, 4=first CI change, 5=first primitive draw, 6=before screen clear, 7=after screen drawn)
ScreenUpdateSetting = 4
# Force to use normal alpha blender
NormalAlphaBlender = False
# Use a faster algorithm to speed up texture loading and CRC computation
FastTextureLoading = False
# Use different texture coordinate clamping code
AccurateTextureMapping = True
# Force emulated frame buffers to be in N64 native resolution
InN64Resolution = True
# Try to reduce Video RAM usage (should never be used)
SaveVRAM = False
# Enable this option to have better render-to-texture quality
DoubleSizeForSmallTxtrBuf = False
# Force to use normal color combiner
DefaultCombinerDisable = False
# Enable game-specific settings from INI file
EnableHacks = True
# If enabled, graphics will be drawn in WinFrame mode instead of solid and texture mode
WinFrameMode = False
# N64 Texture Memory Full Emulation (may fix some games, may break others)
FullTMEMEmulation = False
# Enable vertex clipper for fog operations
OpenGLVertexClipper = False
# Enable/Disable SSE optimizations for capable CPUs
EnableSSE = False
# If this option is enabled, the plugin will skip every other frame
SkipFrame = False
# If enabled, texture enhancement will be done only for TxtRect ucode
TexRectOnly = False
# If enabled, texture enhancement will be done only for textures width+height<=128
SmallTextureOnly = False
# Select hi-resolution textures based only on the CRC and ignore format+size information (Glide64 compatibility)
LoadHiResCRCOnly = True
# Enable hi-resolution texture file loading
LoadHiResTextures = False
# Enable texture dumping
DumpTexturesToFiles = False
# Display On-screen FPS
ShowFPS = True
# Use Mipmapping? 0=no, 1=nearest, 2=bilinear, 3=trilinear
Mipmapping = 2
# Enable, Disable fog generation (0=Disable, 1=Enable)
FogMethod = 0
# Force to use texture filtering or not (0=auto: n64 choose, 1=force no filtering, 2=force filtering)
ForceTextureFilter = 0
# Primary texture enhancement filter (0=None, 1=2X, 2=2XSAI, 3=HQ2X, 4=LQ2X, 5=HQ4X, 6=Sharpen, 7=Sharpen More, 8=External, 9=Mirrored)
TextureEnhancement = 0
# Secondary texture enhancement filter (0 = none, 1-4 = filtered)
TextureEnhancementControl = 0
# Color bit depth to use for textures (0=default, 1=32 bits, 2=16 bits)
TextureQuality = 0
# Z-buffer depth (only 16 or 32)
OpenGLDepthBufferSetting = 16
# Enable/Disable MultiSampling (0=off, 2,4,8,16=quality)
MultiSampling = 0
# Color bit depth for rendering window (0=32 bits, 1=16 bits)
ColorQuality = 0
# OpenGL level to support (0=auto, 1=OGL_FRAGMENT_PROGRAM)
OpenGLRenderSetting = 0
# Enable/Disable Anisotropic Filtering for Mipmapping (0=no filtering, 2-16=quality). This is uneffective if Mipmapping is 0. If the given value is to high to be supported by your graphic card, the value will be the highest value your graphic card can support. Better result with Trilinear filtering
AnisotropicFiltering = 0
# If true, use polygon offset values specified below
ForcePolygonOffset = False
# Specifies a scale factor that is used to create a variable depth offset for each polygon
PolygonOffsetFactor = 0.000000
# Is multiplied by an implementation-specific value to create a constant depth offset
PolygonOffsetUnits = 0.000000

And here is the log:

 __  __                         __   _  _   ____  _             
|  \/  |_   _ _ __   ___ _ __  / /_ | || | |  _ \| |_   _ ___ 
| |\/| | | | | '_ \ / _ \ '_ \| '_ \| || |_| |_) | | | | / __|  
| |  | | |_| | |_) |  __/ | | | (_) |__   _|  __/| | |_| \__ \  
|_|  |_|\__,_| .__/ \___|_| |_|\___/   |_| |_|   |_|\__,_|___/  
             |_|         https://mupen64plus.org/               
Mupen64Plus Console User-Interface Version 2.5.9

UI-Console: attached to core library 'Mupen64Plus Core' version 2.5.9
UI-Console:             Includes support for Dynamic Recompiler.
Core: Using full mem base
Core: Goodname: Mario Kart 64 (U) [!]
Core: Name: MARIOKART64         
Core: MD5: 3A67D9986F54EB282924FCA4CD5F6DFF
Core: CRC: 3E5055B6 2E92DA52
Core: Imagetype: .v64 (byteswapped)
Core: Rom size: 12582912 bytes (or 12 Mb or 96 Megabits)
Core: ClockRate = F
Core: Version: 1446
Core: Manufacturer: Nintendo
Core: Cartridge_ID: 544B
Core: Country: USA
Core: PC = 80000400
Core: Save type: 0
UI-Console Status: Cheat codes disabled.
UI-Console: using Video plugin: 'GLideN64 rev.895401f9' v2.0.0
UI-Console: Video plugin library: /tmp/mupen64plus/lib/mupen64plus/mupen64plus-video-GLideN64.so
UI-Console: using Audio plugin: 'Mupen64Plus SDL Audio Plugin' v2.5.9
UI-Console: Audio plugin library: /tmp/mupen64plus/lib/mupen64plus/mupen64plus-audio-sdl.so
Input: Using auto-config file at: '/tmp/mupen64plus/share/mupen64plus/InputAutoCfg.ini'
Input: Using auto-config file at: '/tmp/mupen64plus/share/mupen64plus/InputAutoCfg.ini'
Input: Using auto-config file at: '/tmp/mupen64plus/share/mupen64plus/InputAutoCfg.ini'
UI-Console: using Input plugin: 'Mupen64Plus SDL Input Plugin' v2.5.9
UI-Console: Input plugin library: /tmp/mupen64plus/lib/mupen64plus/mupen64plus-input-sdl.so
UI-Console: using RSP plugin: 'Hacktarux/Azimer High-Level Emulation RSP Plugin' v2.5.9
UI-Console: RSP plugin library: /tmp/mupen64plus/lib/mupen64plus/mupen64plus-rsp-hle.so
Core: input plugin did not specify a render callback; there will be no on screen display by the input plugin.
Input: 2 SDL joysticks were found.
Input: Using auto-config file at: '/tmp/mupen64plus/share/mupen64plus/InputAutoCfg.ini'
Input Error: No auto-config found for joystick named 'AtGames USB Control Panel' in InputAutoConfig.ini
Input: Using auto-config file at: '/tmp/mupen64plus/share/mupen64plus/InputAutoCfg.ini'
Input Error: No auto-config found for joystick named 'AtGames USB Control Panel' in InputAutoConfig.ini
Input: N64 Controller #1: Forcing default keyboard configuration
Input: Using auto-config file at: '/tmp/mupen64plus/share/mupen64plus/InputAutoCfg.ini'
Input: 1 controller(s) found, 1 plugged in and usable in the emulator
Input Warning: Couldn't open rumble support for joystick #1
Input Warning: Couldn't open rumble support for joystick #2
Input Warning: Couldn't open rumble support for joystick #3
Input Warning: Couldn't open rumble support for joystick #4
Input: Mupen64Plus SDL Input Plugin version 2.5.9 initialized.
RSP: RSP Fallback disabled !
Core: Using video capture backend: dummy
Core: Game controller 0 (Standard controller) has a Memory pak plugged in
Core: Game controller 1 (Standard controller) has nothing plugged in
Core: Game controller 2 (Standard controller) has a Memory pak plugged in
Core: Game controller 3 (Standard controller) has a Memory pak plugged in
Core: Using CIC type X102
vdisplay 1080  148500 60 8 1920x1080
vdisplay 1080  148500 50 64 1920x1080
vdisplay 720  74250 60 64 1280x720
Core: Setting video mode: 1280x720
arm_release_ver of this libMali is r7p0-00rel0, rk_so_ver is '1', built at '15:29:10', on 'Oct  8 2018'. 
Post Color Buffer Success
Audio: Using resampler trivial
Audio: Initializing SDL audio subsystem...
Audio: Primary buffer: 16384 output samples.
Audio: Primary target fullness: 8192 output samples.
Audio: Secondary buffer: 1024 output samples.
Audio: Requesting frequency: 44100Hz.
Audio: Requesting format: 32784.
Audio: Frequency: 44100
Audio: Format: 32784
Audio: Channels: 2
Audio: Silence: 0
Audio: Samples: 1024
Audio: Size: 4096
Input Warning: Couldn't open rumble support for joystick #1
Input Warning: Couldn't open rumble support for joystick #2
Input Warning: Couldn't open rumble support for joystick #3
Input Warning: Couldn't open rumble support for joystick #4
Core: Initializing 4 RDRAM modules for a total of 8 MB
Core: Starting R4300 emulator: Dynamic Recompiler
Core: Init new dynarec
Core:     apply_speed_limiter(): Waiting 6ms
Core:     apply_speed_limiter(): Waiting 10ms
Core:     apply_speed_limiter(): Waiting 17ms
Core:     apply_speed_limiter(): Waiting 19ms
Core:     apply_speed_limiter(): Waiting 17ms
Core:     apply_speed_limiter(): Waiting 15ms
Core:     apply_speed_limiter(): Waiting 7ms
Core:     apply_speed_limiter(): Waiting 14ms
Core:     apply_speed_limiter(): Waiting 24ms
Core:     apply_speed_limiter(): Waiting 20ms
Core:     apply_speed_limiter(): Waiting 9ms
Audio: sdl_init_audio_device(): SDL Audio sub-system already initialized.
Audio: Initializing SDL audio subsystem...
Audio: Primary buffer: 16384 output samples.
Audio: Primary target fullness: 8192 output samples.
Audio: Secondary buffer: 1024 output samples.
Audio: Requesting frequency: 44100Hz.
Audio: Requesting format: 32784.
Audio: Frequency: 44100
Audio: Format: 32784
Audio: Channels: 2
Audio: Silence: 0
Audio: Samples: 1024
Audio: Size: 4096
Core:     apply_speed_limiter(): Waiting 16ms
Core:     apply_speed_limiter(): Waiting 27ms
Core:     apply_speed_limiter(): Waiting 29ms
Core:     apply_speed_limiter(): Waiting 21ms
Core:     apply_speed_limiter(): Waiting 12ms
Core:     apply_speed_limiter(): Waiting 7ms
...

[dankcushions]

settings look ok for me, although EnableFBEmulation = False will cause problems in many games (but enabling will make things slower overall).

this may just be the limits of your device. mali 450 is very weak, but i would still expect it to be around pi3 level (which is fairly playable on mario kart single player or mario 64, but it may be weaker than that for all i know.

[wn2000]

The strange thing is that if I use other video plugins like Glide64mk2, and use --resolution 1280x720, Mario Kart can run at about 25-30 fps and is fairly playable, but not perfect. And with --resolution 320x240 it becomes very smooth but the video is only at the corner.
Still puzzled why GLideN64 only gives me ~10 FPS.

[dankcushions]

it's not so surprising when you consider glide64's last update was approx 10 years ago. the other plugins probably even older. GLideN64 is by the same author and aims to be a more accurate and feature-rich HLE renderer, but with that comes increased system requirements.

your other alternative is use an alternative mupen64plus front end that can handle upscaling and handles the plugins you want to use. retroarch for example.

[wn2000]

I found mupen64plus_libretro_next which seems to also base on GLideN64.
Are there other libretro port of video plugins?
After testing Rice, Glide64mk2, gles2n64, and GLideN64 at 1280x720, the best is gles2n64, at least for Mario Kart.
It's a shame that it's abandoned. Is there a libretro port based on gles2n64?

[dankcushions]

i believe https://github.com/libretro/parallel-n64 supports it, but you need to choose the select plugin via the core options i believe. it's only viable in a few games though.

but this is off topic now - maybe try the libretro forums or discord. good luck!