Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

headless: Add some new and passing tests #16046

Merged
merged 2 commits into from
Sep 18, 2022
Merged

headless: Add some new and passing tests #16046

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
c65880f
headless: Add some new and passing tests.
unknownbrackets Sep 18, 2022
ca248e1
softgpu: Fix s8 primitives in throughmode.
unknownbrackets Sep 18, 2022
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
14 changes: 3 additions & 11 deletions GPU/Common/VertexDecoderArm.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -872,22 +872,14 @@ void VertexDecoderJitCache::Jit_NormalFloat() {
STMIA(scratchReg, false, 3, tempReg1, tempReg2, tempReg3);
}

// Through expands into floats, always. Might want to look at changing this.
void VertexDecoderJitCache::Jit_PosS8Through() {
Copy link
Owner

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Should this be renamed PosInvalid too?

Copy link
Collaborator Author

@unknownbrackets unknownbrackets Sep 18, 2022 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Well, "invalid" I used for when no position is specified at all, and the verts are all culled.

However, when a position is specified, as s8 in through, it is not culled. It decodes to zero. So it's kinda invalid, but different from "invalid position in throughmode (no bits set)" which has different behavior.

That is to say:

  • (vertType & GE_VTYPE_POS_MASK) == 0 -> invalid, no drawing occurs.
  • (vertType & GE_VTYPE_POS_MASK) == GE_VTYPE_POS_8BIT && (vertType & GE_VTYPE_THROUGH) != 0 ->
    "valid", but decodes to zero.
  • (vertType & GE_VTYPE_POS_MASK) == GE_VTYPE_POS_8BIT && (vertType & GE_VTYPE_THROUGH) == 0 -> valid, and decodes to sensible values.

(this behavior is all demonstrated in the "invalidprim" test.)

-[Unknown]

Copy link
Collaborator Author

@unknownbrackets unknownbrackets Sep 18, 2022 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

As to Arm/Arm64/etc. - I figured might as well keep these jitted, although they're hopefully rarely being hit. I didn't think it was worth bothering to newly jit invalid, though.

-[Unknown]

Copy link
Owner

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Ok, cool. Thanks for the clarifications!

DEBUG_LOG_REPORT_ONCE(vertexS8Through, G3D, "Using S8 positions in throughmode");
_dbg_assert_msg_(fpScratchReg + 1 == fpScratchReg2, "VertexDecoder fpScratchRegs must be in order.");
_dbg_assert_msg_(fpScratchReg2 + 1 == fpScratchReg3, "VertexDecoder fpScratchRegs must be in order.");

// TODO: SIMD
LDRSB(tempReg1, srcReg, dec_->posoff);
LDRSB(tempReg2, srcReg, dec_->posoff + 1);
LDRB(tempReg3, srcReg, dec_->posoff + 2);
static const ARMReg tr[3] = { tempReg1, tempReg2, tempReg3 };
static const ARMReg fr[3] = { fpScratchReg, fpScratchReg2, fpScratchReg3 };
// 8-bit positions in throughmode always decode to 0, depth included.
VEOR(neonScratchReg, neonScratchReg, neonScratchReg);
VEOR(neonScratchReg2, neonScratchReg, neonScratchReg);
ADD(scratchReg, dstReg, dec_->decFmt.posoff);
VMOV(neonScratchReg, tempReg1, tempReg2);
VMOV(neonScratchReg2, tempReg3, tempReg3);
VCVT(F_32 | I_SIGNED, neonScratchRegQ, neonScratchRegQ);
VST1(F_32, neonScratchReg, scratchReg, 2, ALIGN_NONE);
}

Expand Down
12 changes: 4 additions & 8 deletions GPU/Common/VertexDecoderArm64.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -668,15 +668,11 @@ void VertexDecoderJitCache::Jit_PosFloat() {
}

void VertexDecoderJitCache::Jit_PosS8Through() {
LDRSB(INDEX_UNSIGNED, tempReg1, srcReg, dec_->posoff);
LDRSB(INDEX_UNSIGNED, tempReg2, srcReg, dec_->posoff + 1);
LDRB(INDEX_UNSIGNED, tempReg3, srcReg, dec_->posoff + 2);
fp.SCVTF(fpScratchReg, tempReg1);
fp.SCVTF(fpScratchReg2, tempReg2);
fp.SCVTF(fpScratchReg3, tempReg3);
// 8-bit positions in throughmode always decode to 0, depth included.
fp.EOR(fpScratchReg, fpScratchReg, fpScratchReg);
STR(INDEX_UNSIGNED, fpScratchReg, dstReg, dec_->decFmt.posoff);
STR(INDEX_UNSIGNED, fpScratchReg2, dstReg, dec_->decFmt.posoff + 4);
STR(INDEX_UNSIGNED, fpScratchReg3, dstReg, dec_->decFmt.posoff + 8);
STR(INDEX_UNSIGNED, fpScratchReg, dstReg, dec_->decFmt.posoff + 4);
STR(INDEX_UNSIGNED, fpScratchReg, dstReg, dec_->decFmt.posoff + 8);
}

void VertexDecoderJitCache::Jit_PosS16Through() {
Expand Down
33 changes: 19 additions & 14 deletions GPU/Common/VertexDecoderCommon.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -773,14 +773,20 @@ void VertexDecoder::Step_PosFloatSkin() const
Vec3ByMatrix43(pos, fn, skinMatrix);
}

void VertexDecoder::Step_PosS8Through() const
{
void VertexDecoder::Step_PosInvalid() const {
// Invalid positions are just culled. Simulate by forcing invalid values.
float *v = (float *)(decoded_ + decFmt.posoff);
const s8 *sv = (const s8 *)(ptr_ + posoff);
const u8 *uv = (const u8 *)(ptr_ + posoff);
v[0] = sv[0];
v[1] = sv[1];
v[2] = uv[2];
v[0] = std::numeric_limits<float>::infinity();
v[1] = std::numeric_limits<float>::infinity();
v[2] = std::numeric_limits<float>::infinity();
}

void VertexDecoder::Step_PosS8Through() const {
// 8-bit positions in throughmode always decode to 0, depth included.
float *v = (float *)(decoded_ + decFmt.posoff);
v[0] = 0;
v[1] = 0;
v[2] = 0;
}

void VertexDecoder::Step_PosS16Through() const
Expand Down Expand Up @@ -1023,35 +1029,35 @@ static const StepFunction nrmstep_morphskin[4] = {
};

static const StepFunction posstep[4] = {
&VertexDecoder::Step_PosS8,
&VertexDecoder::Step_PosInvalid,
&VertexDecoder::Step_PosS8,
&VertexDecoder::Step_PosS16,
&VertexDecoder::Step_PosFloat,
};

static const StepFunction posstep_skin[4] = {
&VertexDecoder::Step_PosS8Skin,
&VertexDecoder::Step_PosInvalid,
&VertexDecoder::Step_PosS8Skin,
&VertexDecoder::Step_PosS16Skin,
&VertexDecoder::Step_PosFloatSkin,
};

static const StepFunction posstep_morph[4] = {
&VertexDecoder::Step_PosS8Morph,
&VertexDecoder::Step_PosInvalid,
&VertexDecoder::Step_PosS8Morph,
&VertexDecoder::Step_PosS16Morph,
&VertexDecoder::Step_PosFloatMorph,
};

static const StepFunction posstep_morph_skin[4] = {
&VertexDecoder::Step_PosS8MorphSkin,
&VertexDecoder::Step_PosInvalid,
&VertexDecoder::Step_PosS8MorphSkin,
&VertexDecoder::Step_PosS16MorphSkin,
&VertexDecoder::Step_PosFloatMorphSkin,
};

static const StepFunction posstep_through[4] = {
&VertexDecoder::Step_PosS8Through,
&VertexDecoder::Step_PosInvalid,
&VertexDecoder::Step_PosS8Through,
&VertexDecoder::Step_PosS16Through,
&VertexDecoder::Step_PosFloatThrough,
Expand Down Expand Up @@ -1224,9 +1230,8 @@ void VertexDecoder::SetVertexType(u32 fmt, const VertexDecoderOptions &options,
bool reportNoPos = false;
if (!pos) {
reportNoPos = true;
pos = 1;
}
if (pos) { // there's always a position
if (pos >= 0) { // there's always a position
size = align(size, posalign[pos]);
posoff = size;
size += possize[pos];
Expand Down
1 change: 1 addition & 0 deletions GPU/Common/VertexDecoderCommon.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -433,6 +433,7 @@ class VertexDecoder {
void Step_PosS16MorphSkin() const;
void Step_PosFloatMorphSkin() const;

void Step_PosInvalid() const;
void Step_PosS8Through() const;
void Step_PosS16Through() const;
void Step_PosFloatThrough() const;
Expand Down
7 changes: 1 addition & 6 deletions GPU/Common/VertexDecoderX86.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1345,14 +1345,9 @@ void VertexDecoderJitCache::Jit_NormalFloatSkin() {

// Through expands into floats, always. Might want to look at changing this.
void VertexDecoderJitCache::Jit_PosS8Through() {
DEBUG_LOG_REPORT_ONCE(vertexS8Through, G3D, "Using S8 positions in throughmode");
// SIMD doesn't really matter since this isn't useful on hardware.
XORPS(fpScratchReg, R(fpScratchReg));
for (int i = 0; i < 3; i++) {
if (i == 2)
MOVZX(32, 8, tempReg1, MDisp(srcReg, dec_->posoff + i));
else
MOVSX(32, 8, tempReg1, MDisp(srcReg, dec_->posoff + i));
CVTSI2SS(fpScratchReg, R(tempReg1));
MOVSS(MDisp(dstReg, dec_->decFmt.posoff + i * 4), fpScratchReg);
}
}
Expand Down
4 changes: 1 addition & 3 deletions GPU/Software/TransformUnit.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -503,10 +503,8 @@ void TransformUnit::SubmitPrimitive(const void* vertices, const void* indices, G
if (gstate_c.skipDrawReason & SKIPDRAW_SKIPFRAME) {
return;
}
// Throughmode never draws 8-bit primitives, maybe because they can't fully specify the screen?
if ((vertex_type & GE_VTYPE_THROUGH_MASK) != 0 && (vertex_type & GE_VTYPE_POS_MASK) == GE_VTYPE_POS_8BIT)
return;
// Vertices without position are just entirely culled.
// Note: Throughmode does draw 8-bit primitives, but positions are always zero - handled in decode.
if ((vertex_type & GE_VTYPE_POS_MASK) == 0)
return;

Expand Down
2 changes: 1 addition & 1 deletion headless/Headless.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -126,7 +126,7 @@ int printUsage(const char *progname, const char *reason)
fprintf(stderr, " -m, --mount umd.cso mount iso on umd1:\n");
fprintf(stderr, " -r, --root some/path mount path on host0: (elfs must be in here)\n");
fprintf(stderr, " -l, --log full log output, not just emulated printfs\n");
fprintf(stderr, " --debugger=PORT enable websocket debugger and break at start\n");
fprintf(stderr, " --debugger=PORT enable websocket debugger and break at start\n");

fprintf(stderr, " --graphics=BACKEND use a different gpu backend\n");
fprintf(stderr, " options: gles, software, directx9, etc.\n");
Expand Down
12 changes: 8 additions & 4 deletions test.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -147,6 +147,7 @@ def target():
"gpu/commands/blend",
"gpu/commands/blend565",
"gpu/commands/blocktransfer",
"gpu/commands/fog",
"gpu/commands/material",
"gpu/displaylist/alignment",
"gpu/dither/dither",
Expand All @@ -157,7 +158,11 @@ def target():
"gpu/ge/enqueueparam",
"gpu/ge/queue",
"gpu/primitives/indices",
"gpu/primitives/invalidprim",
"gpu/primitives/trianglefan",
"gpu/primitives/trianglestrip",
"gpu/primitives/triangles",
"gpu/rendertarget/copy",
"gpu/rendertarget/depal",
"gpu/signals/pause",
"gpu/signals/pause2",
Expand Down Expand Up @@ -269,6 +274,7 @@ def target():
"threads/mutex/refer",
"threads/mutex/try",
"threads/mutex/unlock",
"threads/mutex/unlock2",
"threads/semaphores/semaphores",
"threads/semaphores/cancel",
"threads/semaphores/create",
Expand Down Expand Up @@ -394,21 +400,19 @@ def target():
"gpu/ge/get",
"gpu/primitives/bezier",
"gpu/primitives/continue",
"gpu/primitives/invalidprim",
"gpu/primitives/immediate",
"gpu/primitives/lines",
"gpu/primitives/linestrip",
"gpu/primitives/points",
"gpu/primitives/rectangles",
"gpu/primitives/spline",
"gpu/primitives/trianglefan",
"gpu/primitives/trianglestrip",
"gpu/reflection/reflection",
"gpu/rendertarget/copy",
"gpu/rendertarget/rendertarget",
"gpu/signals/continue",
"gpu/signals/jumps",
"gpu/signals/simple",
"gpu/simple/simple",
"gpu/textures/size",
"gpu/triangle/triangle",
"gpu/vertices/colors",
"gpu/vertices/texcoords",
Expand Down