From baea0ee1effba2e92857de19f7dbf87f12c38818 Mon Sep 17 00:00:00 2001
From: Dwayne Robinson <dwayner@microsoft.com>
Date: Thu, 30 May 2024 08:51:42 -0700
Subject: [PATCH 1/8] Add MLDeviceType npu

---
 index.bs | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/index.bs b/index.bs
index 78e0b4a9..9aa597ec 100644
--- a/index.bs
+++ b/index.bs
@@ -672,7 +672,7 @@ Unlike WebGPU, this API does not intrinsically support custom shader authoring;
 
 The WebGPU API identifies <a href="https://gpuweb.github.io/gpuweb/#privacy-machine-artifacts">machine-specific artifacts</a> as a privacy consideration. Similarly, the WebNN API's compute unit scheduling may under certain circumstances introduce a fingerprint. However, similarly to WebGPU, such fingerprints are identical across most or all of the devices of each vendor, mitigating the concern. Furthermore, software implementations can be used to further eliminate such artifacts.
 
-The WebNN API defines two developer-settable preferences to help inform [[#programming-model-device-selection]] and allow the implementation to better select the most appropriate underlying execution device for the workload. [=Device type=] normatively indicates the kind of device and is either {{MLDeviceType/"cpu"}} or {{MLDeviceType/"gpu"}}. If this type cannot be satisfied, an "{{OperationError}}" {{DOMException}} is thrown, thus this type can in some cases add two bits of entropy to the fingerprint. [=Power preference=] indicates preference as related to the power consumption and is considered a hint only and as such does not increase entropy of the fingerprint.
+The WebNN API defines two developer-settable preferences to help inform [[#programming-model-device-selection]] and allow the implementation to better select the most appropriate underlying execution device for the workload. [=Device type=] normatively indicates the kind of device is one of: {{MLDeviceType/"cpu"}}, {{MLDeviceType/"gpu"}}, {{MLDeviceType/"npu"}}. If this type cannot be satisfied, an "{{OperationError}}" {{DOMException}} is thrown, and thus this type can in some cases add two bits of entropy to the fingerprint. [=Power preference=] indicates preference as related to the power consumption and is considered a hint only and as such does not increase entropy of the fingerprint.
 
 If a future version of this specification introduces support for new a  [=device type=] that can only support a subset of {{MLOperandDataType}}s, that may introduce a new fingerprint.
 
@@ -782,7 +782,8 @@ WorkerNavigator includes NavigatorML;
 <script type=idl>
 enum MLDeviceType {
   "cpu",
-  "gpu"
+  "gpu",
+  "npu"
 };
 
 enum MLPowerPreference {
@@ -958,6 +959,8 @@ The <dfn>device type</dfn> indicates the kind of device used for the context. It
 <dd>Provides the broadest compatibility and usability across all client devices with varying degrees of performance.</dd>
 <dt>"<dfn enum-value>gpu</dfn>"</dt>
 <dd>Provides the broadest range of achievable performance across graphics hardware platforms from consumer devices to professional workstations.</dd>
+<dt>"<dfn enum-value>npu</dfn>"</dt>
+<dd>Provides narrower functionality on dedicated machine learning devices which can have performance in between CPU and GPU, favoring other factors over raw performance like lower power usage for longer battery life or lower latency for results.</dd>
 </dl>
 
 The <dfn>power preference</dfn> indicates preference as related to power consumption. It is one of the following:

From 70d34651eb71512b51e9481452d7dccea1bb2f32 Mon Sep 17 00:00:00 2001
From: Dwayne Robinson <dwayner@microsoft.com>
Date: Thu, 30 May 2024 08:53:29 -0700
Subject: [PATCH 2/8] Minor wording

---
 index.bs | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/index.bs b/index.bs
index 9aa597ec..9f70a144 100644
--- a/index.bs
+++ b/index.bs
@@ -960,7 +960,7 @@ The <dfn>device type</dfn> indicates the kind of device used for the context. It
 <dt>"<dfn enum-value>gpu</dfn>"</dt>
 <dd>Provides the broadest range of achievable performance across graphics hardware platforms from consumer devices to professional workstations.</dd>
 <dt>"<dfn enum-value>npu</dfn>"</dt>
-<dd>Provides narrower functionality on dedicated machine learning devices which can have performance in between CPU and GPU, favoring other factors over raw performance like lower power usage for longer battery life or lower latency for results.</dd>
+<dd>Provides narrower functionality on dedicated machine learning devices which can perform between CPU and GPU, favoring other factors over raw performance like lower power usage for longer battery life or lower latency for results.</dd>
 </dl>
 
 The <dfn>power preference</dfn> indicates preference as related to power consumption. It is one of the following:

From 8574b6e0fb5611a7477c10ee5be94992cf1beac3 Mon Sep 17 00:00:00 2001
From: Dwayne Robinson <dwayner@microsoft.com>
Date: Thu, 30 May 2024 09:05:09 -0700
Subject: [PATCH 3/8] Add note about fallback

---
 index.bs | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/index.bs b/index.bs
index 9f70a144..4473ccb5 100644
--- a/index.bs
+++ b/index.bs
@@ -960,7 +960,7 @@ The <dfn>device type</dfn> indicates the kind of device used for the context. It
 <dt>"<dfn enum-value>gpu</dfn>"</dt>
 <dd>Provides the broadest range of achievable performance across graphics hardware platforms from consumer devices to professional workstations.</dd>
 <dt>"<dfn enum-value>npu</dfn>"</dt>
-<dd>Provides narrower functionality on dedicated machine learning devices which can perform between CPU and GPU, favoring other factors over raw performance like lower power usage for longer battery life or lower latency for results.</dd>
+<dd>Provides narrower functionality on dedicated machine learning devices which can perform between CPU and GPU, favoring other factors over raw performance like lower power usage for longer battery life or lower latency for results. The backend implementation may fall back to other devices for certain operators and parts of the graph.</dd>
 </dl>
 
 The <dfn>power preference</dfn> indicates preference as related to power consumption. It is one of the following:

From c003e950f98184a03c4137d5e398fc2ddc8b8a49 Mon Sep 17 00:00:00 2001
From: Dwayne Robinson <dwayner@microsoft.com>
Date: Thu, 30 May 2024 09:18:39 -0700
Subject: [PATCH 4/8] Remove 'can' regarding performance

---
 index.bs | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/index.bs b/index.bs
index 4473ccb5..5eea0e22 100644
--- a/index.bs
+++ b/index.bs
@@ -960,7 +960,7 @@ The <dfn>device type</dfn> indicates the kind of device used for the context. It
 <dt>"<dfn enum-value>gpu</dfn>"</dt>
 <dd>Provides the broadest range of achievable performance across graphics hardware platforms from consumer devices to professional workstations.</dd>
 <dt>"<dfn enum-value>npu</dfn>"</dt>
-<dd>Provides narrower functionality on dedicated machine learning devices which can perform between CPU and GPU, favoring other factors over raw performance like lower power usage for longer battery life or lower latency for results. The backend implementation may fall back to other devices for certain operators and parts of the graph.</dd>
+<dd>Provides narrower functionality on dedicated machine learning devices which perform between CPU and GPU, favoring other factors over raw performance like lower power usage for longer battery life or lower latency for results. The backend implementation may fall back to other devices for certain operators and parts of the graph.</dd>
 </dl>
 
 The <dfn>power preference</dfn> indicates preference as related to power consumption. It is one of the following:

From fb9865bf898a1d8fc1f7c4faed2da44125079ec3 Mon Sep 17 00:00:00 2001
From: Dwayne Robinson <dwayner@microsoft.com>
Date: Sun, 16 Jun 2024 23:17:22 -0700
Subject: [PATCH 5/8] Review feedback

---
 index.bs | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/index.bs b/index.bs
index e6eac9ab..58a06b8d 100644
--- a/index.bs
+++ b/index.bs
@@ -726,14 +726,17 @@ interface ML {
 
 ### {{MLContextOptions}} ### {#api-mlcontextoptions}
 
+Issue(623): MLContextOptions is under active development, and the design is expected to change, informed by further implementation experience and new use cases from the wider web community. The Working Group is considering additional API controls to allow the definition of a fallback device, multiple devices in a preferred order, or an exclusion of a specific device. Other considerations under discussion include error handling, ultimate
+fallback, and quantized operators. Feedback is welcome on any of these design considerations from web developers, library authors, OS and hardware vendors, and other stakeholders via GitHub:
+
 The <dfn dfn-for=MLContextOptions dfn-type=dict-member>deviceType</dfn> option is an <dfn dfn-type=enum>MLDeviceType</dfn> and indicates the application's preference for the kind of device used for the context. It is one of the following:
 <dl dfn-for="MLDeviceType">
 <dt>"<dfn enum-value>cpu</dfn>"</dt>
 <dd>Provides the broadest compatibility and usability across all client devices with varying degrees of performance.</dd>
 <dt>"<dfn enum-value>gpu</dfn>"</dt>
-<dd>Provides the broadest range of achievable performance across graphics hardware platforms from consumer devices to professional workstations.</dd>
+<dd>Provides the broadest range of achievable performance across graphics hardware platforms from consumer devices to professional workstations. The backend implementation may fall back to other devices for certain operators and parts of the graph.</dd>
 <dt>"<dfn enum-value>npu</dfn>"</dt>
-<dd>Provides narrower functionality on dedicated machine learning devices which perform between CPU and GPU, favoring other factors over raw performance like lower power usage for longer battery life or lower latency for results. The backend implementation may fall back to other devices for certain operators and parts of the graph.</dd>
+<dd>Provides power efficiency for sustained workloads across hardware platforms with purpose-built accelerators. The backend implementation may fall back to other devices for certain operators and parts of the graph.</dd>
 </dl>
 
 The <dfn dfn-for=MLContextOptions dfn-type=dict-member>powerPreference</dfn> option is an <dfn dfn-type=enum>MLPowerPreference</dfn> and indicates the application's preference as related to power consumption. It is one of the following:

From bba0a4822792603f4d5bf8dbf8432fab894545bf Mon Sep 17 00:00:00 2001
From: Dwayne Robinson <dwayner@microsoft.com>
Date: Sun, 16 Jun 2024 23:23:14 -0700
Subject: [PATCH 6/8] Fix funky line break in issue text

---
 index.bs | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/index.bs b/index.bs
index 58a06b8d..f341381d 100644
--- a/index.bs
+++ b/index.bs
@@ -726,8 +726,7 @@ interface ML {
 
 ### {{MLContextOptions}} ### {#api-mlcontextoptions}
 
-Issue(623): MLContextOptions is under active development, and the design is expected to change, informed by further implementation experience and new use cases from the wider web community. The Working Group is considering additional API controls to allow the definition of a fallback device, multiple devices in a preferred order, or an exclusion of a specific device. Other considerations under discussion include error handling, ultimate
-fallback, and quantized operators. Feedback is welcome on any of these design considerations from web developers, library authors, OS and hardware vendors, and other stakeholders via GitHub:
+Issue(623): MLContextOptions is under active development, and the design is expected to change, informed by further implementation experience and new use cases from the wider web community. The Working Group is considering additional API controls to allow the definition of a fallback device, multiple devices in a preferred order, or an exclusion of a specific device. Other considerations under discussion include error handling, ultimate fallback, and quantized operators. Feedback is welcome on any of these design considerations from web developers, library authors, OS and hardware vendors, and other stakeholders via GitHub:
 
 The <dfn dfn-for=MLContextOptions dfn-type=dict-member>deviceType</dfn> option is an <dfn dfn-type=enum>MLDeviceType</dfn> and indicates the application's preference for the kind of device used for the context. It is one of the following:
 <dl dfn-for="MLDeviceType">

From 74c368e44de3d155d3faeeda7e80ed692454dbb5 Mon Sep 17 00:00:00 2001
From: Dwayne Robinson <dwayner@microsoft.com>
Date: Mon, 17 Jun 2024 13:44:37 -0700
Subject: [PATCH 7/8] Review feedback

---
 index.bs | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)

diff --git a/index.bs b/index.bs
index f341381d..71673a0a 100644
--- a/index.bs
+++ b/index.bs
@@ -615,6 +615,8 @@ The WebGPU API identifies <a href="https://gpuweb.github.io/gpuweb/#privacy-mach
 
 The WebNN API defines two developer-settable preferences to help inform [[#programming-model-device-selection]] and allow the implementation to better select the most appropriate underlying execution device for the workload. An {{MLDeviceType}} normatively indicates the kind of device and is one of: {{MLDeviceType/"cpu"}}, {{MLDeviceType/"gpu"}}, {{MLDeviceType/"npu"}}. If this type cannot be satisfied, an "{{OperationError}}" {{DOMException}} is thrown, thus this type can in some cases add two bits of entropy to the fingerprint. An {{MLPowerPreference}} indicates preference as related to the power consumption and is considered a hint only and as such does not increase entropy of the fingerprint.
 
+Issue(623): {{MLContextOptions}} is under active development, and the design is expected to change, informed by further implementation experience and new use cases from the wider web community.
+
 If a future version of this specification introduces support for a new {{MLDeviceType}} that can only support a subset of {{MLOperandDataType}}s, that may introduce a new fingerprint.
 
 In general, implementers of this API are expected to apply <a href="https://gpuweb.github.io/gpuweb/#privacy-considerations">WebGPU Privacy Considerations</a> to their implementations where applicable.
@@ -726,16 +728,16 @@ interface ML {
 
 ### {{MLContextOptions}} ### {#api-mlcontextoptions}
 
-Issue(623): MLContextOptions is under active development, and the design is expected to change, informed by further implementation experience and new use cases from the wider web community. The Working Group is considering additional API controls to allow the definition of a fallback device, multiple devices in a preferred order, or an exclusion of a specific device. Other considerations under discussion include error handling, ultimate fallback, and quantized operators. Feedback is welcome on any of these design considerations from web developers, library authors, OS and hardware vendors, and other stakeholders via GitHub:
+Issue(623): {{MLContextOptions}} is under active development, and the design is expected to change, informed by further implementation experience and new use cases from the wider web community. The Working Group is considering additional API controls to allow the definition of a fallback device, multiple devices in a preferred order, or an exclusion of a specific device. Other considerations under discussion include error handling, ultimate fallback, and quantized operators. Feedback is welcome on any of these design considerations from web developers, library authors, OS and hardware vendors, and other stakeholders via GitHub:
 
 The <dfn dfn-for=MLContextOptions dfn-type=dict-member>deviceType</dfn> option is an <dfn dfn-type=enum>MLDeviceType</dfn> and indicates the application's preference for the kind of device used for the context. It is one of the following:
 <dl dfn-for="MLDeviceType">
 <dt>"<dfn enum-value>cpu</dfn>"</dt>
 <dd>Provides the broadest compatibility and usability across all client devices with varying degrees of performance.</dd>
 <dt>"<dfn enum-value>gpu</dfn>"</dt>
-<dd>Provides the broadest range of achievable performance across graphics hardware platforms from consumer devices to professional workstations. The backend implementation may fall back to other devices for certain operators and parts of the graph.</dd>
+<dd>Provides the broadest range of achievable performance across graphics hardware platforms from consumer devices to professional workstations. The underlying platform implementation may fall back to other devices for certain operators and parts of the graph.</dd>
 <dt>"<dfn enum-value>npu</dfn>"</dt>
-<dd>Provides power efficiency for sustained workloads across hardware platforms with purpose-built accelerators. The backend implementation may fall back to other devices for certain operators and parts of the graph.</dd>
+<dd>Provides power efficiency for sustained workloads across hardware platforms with purpose-built accelerators. The underlying platform implementation may fall back to other devices for certain operators and parts of the graph.</dd>
 </dl>
 
 The <dfn dfn-for=MLContextOptions dfn-type=dict-member>powerPreference</dfn> option is an <dfn dfn-type=enum>MLPowerPreference</dfn> and indicates the application's preference as related to power consumption. It is one of the following:

From 3b834661d237f2b0db7f2e684e19b8c1e10168ae Mon Sep 17 00:00:00 2001
From: Dwayne Robinson <dwayner@microsoft.com>
Date: Mon, 17 Jun 2024 14:09:16 -0700
Subject: [PATCH 8/8] Review feedback from Joshua

---
 index.bs | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/index.bs b/index.bs
index 71673a0a..03b13009 100644
--- a/index.bs
+++ b/index.bs
@@ -898,7 +898,7 @@ When the {{MLContext/[[contextType]]}} is set to [=context type/default=] with t
 </details>
 
 ### {{MLContext/compute()}}  ### {#api-mlcontext-compute}
-Asynchronously carries out the computational workload of a compiled graph {{MLGraph}} on a separate timeline, either on a worker thread for the CPU execution, or on a GPU timeline for the submission of GPU workload on the command queue. The asynchronous nature of this call avoids blocking the calling thread while the computation for result is ongoing. This method of execution requires an {{MLContext}} created with {{MLContextOptions}}. Otherwise, it [=exception/throws=] an "{{OperationError}}" {{DOMException}}.
+Asynchronously carries out the computational workload of a compiled graph {{MLGraph}} on a separate timeline, either on a worker thread for the CPU execution, or on a GPU/NPU timeline for submitting a workload onto the command queue. The asynchronous nature of this call avoids blocking the calling thread while the computation for result is ongoing. This method of execution requires an {{MLContext}} created with {{MLContextOptions}}. Otherwise, it [=exception/throws=] an "{{OperationError}}" {{DOMException}}.
 
 <div class="note">
 In accordance with the [=ArrayBufferView/write|Web IDL warning=], to prevent the calling thread from modifying the input and output resources while the computation is ongoing, this method [=MLNamedArrayBufferViews/transfer|transfers=] the input and output {{MLNamedArrayBufferViews}} to new views that share the same backing memory allocations. The transferred views are returned to the caller via the promise fulfillment with the computation result written into the backing memory of the output views.