Add pixel comparison and NVDI note to ET4000 GemBench section

This commit is contained in:
2026-09-13 11:51:36 +02:00
parent 265dfa7541
commit 3bb90af124
+1 -1
View File
@@ -54,7 +54,7 @@ parent: "/retro/"
<tr><td>VDI Scroll</td><td>53%</td><td>Slower, scrolling is CPU-bound despite the card</td></tr>
<tr><td>Blitting</td><td>32%</td><td>Much slower, no hardware BitBLT on the ET4000</td></tr>
</table>
<p>The overall average sits at 100%, which is misleading. The picture is split: anything the ET4000 can accelerate (text, graphics drawing) runs significantly faster, while GEM-level operations that still go through the CPU (window management, scrolling, blitting) are dragged down by the overhead of driving a higher resolution through the ISA bus. In day-to-day use, the VDI gains are what you see and feel: text is crisp, menus pop up quickly, and drawing applications are noticeably more responsive.</p>
<p>The overall average sits at 100%, which is misleading. The picture is split: anything the ET4000 can accelerate (text, graphics drawing) runs significantly faster, while GEM-level operations that still go through the CPU (window management, scrolling, blitting) are dragged down by the overhead of driving a higher resolution through the ISA bus. Running at 1024x768 with 256 colors means 7.5 times the pixel count and 15 times the data volume compared to ST High (640x400 in 2 colors). The Nova driver without NVDI squeezes the maximum out of this particular card and driver combination. In day-to-day use, the VDI gains are what you see and feel: text is crisp, menus pop up quickly, and drawing applications are noticeably more responsive.</p>
<h2>The Context</h2>
<p>FPGA cores on MiSTer and MiST still only emulate the native ST Shifter, with no ET4000 emulation on the horizon given the FPGA resources it would require. So the real hardware approach remains the only way to get those higher resolutions, and now it works in the Mega ST 4 "High" machine with its dedicated <a href="/retro/atari-monitor-compatibility/">monitor</a>.</p>