Skip to content

[SIT] Extend I/O precision support #33966

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

Open
atamas19 wants to merge 2 commits into
base: master
Choose a base branch
from
+50 −7
Open

[SIT] Extend I/O precision support #33966

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
1a18db2
Added new requested OV precisions to SIT
atamas19 Feb 3, 2026
3732b16
Implemented copilot suggestion
atamas19 Feb 4, 2026
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
57 changes: 50 additions & 7 deletions src/plugins/intel_npu/tools/single-image-test/main.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -69,8 +69,8 @@ DEFINE_string(compiled_blob, "", "Output compiled network file (compiled result
DEFINE_uint32(override_model_batch_size, 1, "Enforce a model to be compiled for batch size");
DEFINE_string(device, "", "Device to use");
DEFINE_string(config, "", "Path to the configuration file (optional)");
DEFINE_string(ip, "", "Input precision (default: U8, available: FP32, FP16, I32, I64, U8)");
DEFINE_string(op, "", "Output precision (default: FP32, available: FP32, FP16, I32, I64, U8)");
DEFINE_string(ip, "", "Input precision (default: U8, available: FP32, FP16, I32, I64, U8, U16, I16, U4, I4, U2, BF8, HF8)");
DEFINE_string(op, "", "Output precision (default: FP32, available: FP32, FP16, I32, I64, U8, U16, I16, U4, I4, U2, BF8, HF8)");
DEFINE_string(
il, "",
"Input layout for all inputs, or ';' separated list of pairs <input>:<layout>. Regex in <input> is supported");
Expand Down Expand Up @@ -2658,15 +2658,16 @@ static int runSingleImageTest() {
// 9. (For loadable networks) Compile model
// 10. Store compile model (if given)
// 11. Run inference / tests
const std::unordered_set<std::string> allowedPrecision = {"U8", "I32", "I64", "FP16", "FP32"};
const std::unordered_set<std::string> allowedPrecision = {"U8", "I32", "I64", "FP16", "FP32",
"U16", "I16", "U4", "I4", "U2", "BF8", "HF8"};
if (!FLAGS_ip.empty()) {
// input precision is U8, I32, I64, FP16 or FP32 only
// input precision: U8, U16, I16, I32, I64, FP16, FP32, U4, I4, U2, BF8, HF8
std::transform(FLAGS_ip.begin(), FLAGS_ip.end(), FLAGS_ip.begin(), ::toupper);
if (allowedPrecision.count(FLAGS_ip) == 0)
throw std::logic_error("Parameter -ip " + FLAGS_ip + " is not supported");
}
if (!FLAGS_op.empty()) {
// output precision is U8, I32, I64, FP16 or FP32 only
// output precision: U8, U16, I16, I32, I64, FP16, FP32, U4, I4, U2, BF8, HF8
std::transform(FLAGS_op.begin(), FLAGS_op.end(), FLAGS_op.begin(), ::toupper);
if (allowedPrecision.count(FLAGS_op) == 0)
throw std::logic_error("Parameter -op " + FLAGS_op + " is not supported");
Expand Down Expand Up @@ -2709,6 +2710,20 @@ static int runSingleImageTest() {
prc_in = ov::element::i32;
else if (FLAGS_ip == "I64")
prc_in = ov::element::i64;
else if (FLAGS_ip == "U16")
prc_in = ov::element::u16;
else if (FLAGS_ip == "I16")
prc_in = ov::element::i16;
else if (FLAGS_ip == "U4")
prc_in = ov::element::u4;
else if (FLAGS_ip == "I4")
prc_in = ov::element::i4;
else if (FLAGS_ip == "U2")
prc_in = ov::element::u2;
else if (FLAGS_ip == "BF8")
prc_in = ov::element::f8e5m2;
else if (FLAGS_ip == "HF8")
prc_in = ov::element::f8e4m3;
else
prc_in = ov::element::u8;

Expand Down Expand Up @@ -2770,6 +2785,20 @@ static int runSingleImageTest() {
prc_out = ov::element::i32;
else if (FLAGS_op == "I64")
prc_out = ov::element::i64;
else if (FLAGS_op == "U16")
prc_out = ov::element::u16;
else if (FLAGS_op == "I16")
prc_out = ov::element::i16;
else if (FLAGS_op == "U4")
prc_out = ov::element::u4;
else if (FLAGS_op == "I4")
prc_out = ov::element::i4;
else if (FLAGS_op == "U2")
prc_out = ov::element::u2;
else if (FLAGS_op == "BF8")
prc_out = ov::element::f8e5m2;
else if (FLAGS_op == "HF8")
prc_out = ov::element::f8e4m3;
else
prc_out = ov::element::u8;

Expand Down Expand Up @@ -2908,10 +2937,24 @@ static int runSingleImageTest() {
inputBinPrecisionForOneInfer[inferIdx][precisionIdx] = ov::element::i64;
} else if (strEq(precision, "U8")) {
inputBinPrecisionForOneInfer[inferIdx][precisionIdx] = ov::element::u8;
} else if (strEq(precision, "U16")) {
inputBinPrecisionForOneInfer[inferIdx][precisionIdx] = ov::element::u16;
} else if (strEq(precision, "I16")) {
inputBinPrecisionForOneInfer[inferIdx][precisionIdx] = ov::element::i16;
} else if (strEq(precision, "U4")) {
inputBinPrecisionForOneInfer[inferIdx][precisionIdx] = ov::element::u4;
} else if (strEq(precision, "I4")) {
inputBinPrecisionForOneInfer[inferIdx][precisionIdx] = ov::element::i4;
} else if (strEq(precision, "U2")) {
inputBinPrecisionForOneInfer[inferIdx][precisionIdx] = ov::element::u2;
} else if (strEq(precision, "BF8")) {
inputBinPrecisionForOneInfer[inferIdx][precisionIdx] = ov::element::f8e5m2;
} else if (strEq(precision, "HF8")) {
inputBinPrecisionForOneInfer[inferIdx][precisionIdx] = ov::element::f8e4m3;
} else {
std::cout << "WARNING: Unhandled precision '" << precision
<< "'! Only FP32, FP16, I32, I64 and U8 can be currently converted to the network's"
<< "input tensor precision.";
<< "'! Only FP32, FP16, I32, I64, U8, U16, I16, U4, I4, U2, BF8 and HF8 "
<< "can be currently converted to the network's input tensor precision.";
}
++precisionIdx;
}
Expand Down
Loading