VFPU: Make AccurateVdot the default

The accurate dot products cost a few percent at most in the IR JIT, apart
from GTA (about 20% on a game that's cheap to run anyway). The fuzzer follows
the default too, unless PPSSPP_CPUFUZZ_ACCURATE_VDOT=0.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

Core/CmdLine.cpp

@@ -199,7 +199,7 @@ static const CommandLineParam g_autoParams[] = {
 	{POFF(compare), CmdParamType::Bool, "compare", 'c', "Enable comparison mode", CmdLineMode::Headless},
 	{POFF(bench), CmdParamType::Bool, "bench", 'b', "Enable benchmark mode", CmdLineMode::Headless},
 	{POFF(oldAtrac), CmdParamType::Bool, "old-atrac", '\0', "Use old ATRAC decoder"},
-	{POFF(accurateVdot), CmdParamType::Bool, "accurate-vdot", '\0', "Compute VFPU dot products bit-exactly like the PSP (slower)"},
+	{POFF(accurateVdot), CmdParamType::Bool, "accurate-vdot", '\0', "Compute VFPU dot products bit-exactly like the PSP (default; =false for the faster approximation)"},
 	{POFF(memStickInserted), CmdParamType::BoolInverse, "no-memstick", '\0', "Run with the memory stick removed"},
 	{POFF(log), CmdParamType::String, "log", '\0', "Output log to FILE", CmdLineMode::Application},
 	{POFF(enableLogging), CmdParamType::Bool, "log", '\0', "Full log output, not just emulated printfs", CmdLineMode::Headless},

Core/Config.cpp

@@ -436,7 +436,7 @@ static const ConfigSetting cpuSettings[] = {
 	ConfigSetting("IOTimingMethod", SETTING(g_Config, iIOTimingMethod), IOTIMING_FAST, CfgFlag::PER_GAME | CfgFlag::REPORT),
 	ConfigSetting("FastMemoryAccess", SETTING(g_Config, bFastMemory), true, CfgFlag::PER_GAME),
 	ConfigSetting("FunctionReplacements", SETTING(g_Config, bFuncReplacements), true, CfgFlag::PER_GAME | CfgFlag::REPORT),
-	ConfigSetting("AccurateVdot", SETTING(g_Config, bAccurateVdot), false, CfgFlag::PER_GAME | CfgFlag::REPORT),
+	ConfigSetting("AccurateVdot", SETTING(g_Config, bAccurateVdot), true, CfgFlag::PER_GAME | CfgFlag::REPORT),
 	ConfigSetting("HideSlowWarnings", SETTING(g_Config, bHideSlowWarnings), false, CfgFlag::DEFAULT),
 	ConfigSetting("HideStateWarnings", SETTING(g_Config, bHideStateWarnings), false, CfgFlag::DEFAULT),
 	ConfigSetting("JitDisableFlags", SETTING(g_Config, uJitDisableFlags), (uint32_t)0, CfgFlag::PER_GAME),

docs/cpu-equivalence.md

@@ -33,7 +33,7 @@ Environment variables:
 | `PPSSPP_CPUFUZZ_DUMP` | write every core's end state of every case to a file |
 | `PPSSPP_CPUFUZZ_STRICT` | fail on any difference, not only on a crash |
 | `PPSSPP_CPUFUZZ_OPS` | comma-separated instruction name prefixes to generate only; `dot` is the dot products and the prefix ops |
-| `PPSSPP_CPUFUZZ_ACCURATE_VDOT` | run with the accurate dot products |
+| `PPSSPP_CPUFUZZ_ACCURATE_VDOT` | `0` to run without the accurate dot products, which are on by default |
 
 For quick rounds while working on some instructions, fuzz just those, on every core of the machine:
 `PPSSPP_CPUFUZZ_OPS=dot Tools/cpu_fuzz_parallel.py build/PPSSPPUnitTest --count 2000` splits the seeds

unittest/JitHarness.cpp

@@ -398,7 +398,7 @@ bool TestJitReplacementHooks() {
 //   PPSSPP_CPUFUZZ_SKIP     cores not to run, by digit: 1 IR interpreter, 2 JIT, 3 JIT IR.
 //   PPSSPP_CPUFUZZ_REPORTS  how many differences to minimize and print (default 20).
 //   PPSSPP_CPUFUZZ_STRICT   fail the test on any difference, not only on a crash.
-//   PPSSPP_CPUFUZZ_ACCURATE_VDOT  run with AccurateVdot, the VFPU's own dot products.
+//   PPSSPP_CPUFUZZ_ACCURATE_VDOT  0 to run without AccurateVdot (the VFPU's own dot products), which is on by default.
 //   PPSSPP_CPUFUZZ_OPS      comma-separated instruction name prefixes to generate only, for fast
 //                           runs on the instructions being worked on. "dot" is the dot products
 //                           and the prefixes. Without PPSSPP_CPUFUZZ_CLASSES, the class is "v".
@@ -901,7 +901,8 @@ bool TestCPUEquivalence() {
 	const bool oldSkipDeadbeef = g_Config.bSkipDeadbeefFilling;
 	g_Config.bSkipDeadbeefFilling = true;
 	const bool oldAccurateVdot = g_Config.bAccurateVdot;
-	g_Config.bAccurateVdot = getenv("PPSSPP_CPUFUZZ_ACCURATE_VDOT") != nullptr;
+	const char *accurateVdot = getenv("PPSSPP_CPUFUZZ_ACCURATE_VDOT");
+	g_Config.bAccurateVdot = !accurateVdot || strcmp(accurateVdot, "0") != 0;
 
 	const char *countStr = getenv("PPSSPP_CPUFUZZ_COUNT");
 	const char *seedStr = getenv("PPSSPP_CPUFUZZ_SEED");