Update README.md
Browse filesChange the performance to prefill 512 tokens.
README.md
CHANGED
|
@@ -24,7 +24,7 @@ To build the demo app from source, please follow the [instructions](https://gith
|
|
| 24 |
|
| 25 |
### Android
|
| 26 |
|
| 27 |
-
Note that all benchmark stats are from a Samsung S24 Ultra with 1280 KV cache size,
|
| 28 |
|
| 29 |
<table border="1">
|
| 30 |
<tr>
|
|
@@ -41,16 +41,16 @@ Note that all benchmark stats are from a Samsung S24 Ultra with 1280 KV cache si
|
|
| 41 |
<td rowspan="2">CPU</td>
|
| 42 |
<td><p style="text-align: right">45</p></td>
|
| 43 |
<td><p style="text-align: right">6</p></td>
|
| 44 |
-
<td><p style="text-align: right">
|
| 45 |
-
<td><p style="text-align: right">6,
|
| 46 |
<td><p style="text-align: right">7,124</p></td>
|
| 47 |
</tr>
|
| 48 |
<tr>
|
| 49 |
<td>dynamic_int8</td>
|
| 50 |
-
<td><p style="text-align: right">
|
| 51 |
<td><p style="text-align: right">23</p></td>
|
| 52 |
-
<td><p style="text-align: right">
|
| 53 |
-
<td><p style="text-align: right">1,
|
| 54 |
<td><p style="text-align: right">1,861</p></td>
|
| 55 |
</tr>
|
| 56 |
</table>
|
|
|
|
| 24 |
|
| 25 |
### Android
|
| 26 |
|
| 27 |
+
Note that all benchmark stats are from a Samsung S24 Ultra with 1280 KV cache size, 512 tokens prefill, 128 tokens decode.
|
| 28 |
|
| 29 |
<table border="1">
|
| 30 |
<tr>
|
|
|
|
| 41 |
<td rowspan="2">CPU</td>
|
| 42 |
<td><p style="text-align: right">45</p></td>
|
| 43 |
<td><p style="text-align: right">6</p></td>
|
| 44 |
+
<td><p style="text-align: right">8</p></td>
|
| 45 |
+
<td><p style="text-align: right">6,213</p></td>
|
| 46 |
<td><p style="text-align: right">7,124</p></td>
|
| 47 |
</tr>
|
| 48 |
<tr>
|
| 49 |
<td>dynamic_int8</td>
|
| 50 |
+
<td><p style="text-align: right">261</p></td>
|
| 51 |
<td><p style="text-align: right">23</p></td>
|
| 52 |
+
<td><p style="text-align: right">2 </p></td>
|
| 53 |
+
<td><p style="text-align: right">1,936 </p></td>
|
| 54 |
<td><p style="text-align: right">1,861</p></td>
|
| 55 |
</tr>
|
| 56 |
</table>
|