mirror of
https://github.com/horizon-rust/horizon-site.git
synced 2026-11-07 10:05:58 -10:00
Benchmarks: Mac mini and desktop measured again with Nightly 2026-11-04 (#58)
* Benchmarks: the Mac mini and the desktop measured again with Nightly 2026-11-04 - Mac mini (M4 Pro) and desktop (RTX 4080): Horizon Nightly 2026-11-04 (67dad0d, the bundle players get, resources from its own horizon-prepare) against vanilla 26.3, render distance 12 and 32, 3 games each, tools/perf_bench.py (seed 1234, 1280x720, VSync off), the 10-22 recipe. - What differs from 10-22, said in each machine's note: the Mac's session was locked, so Horizon drew offscreen (no longer held at 144 Hz); the desktop ran at load 28-30 at each start, 40-70 during the runs (CI and builds), and vanilla went through sway's Wayland (Xwayland timed out each frame). - The ThinkPad keeps its 10-22 runs (no benchmarks on it any more); its note says so, and that players with the same HD Graphics 620 see less in real play (opt-in telemetry, aggregates only). - summary.json and public/benchmarks.json rebuilt with the old site's scripts/benchmarks.mjs (now outside the repo), which reproduces the previous summary exactly; the Mac's Horizon entries are allowed offscreen. - "Measured on" shows the span of dates, since the machines were measured on different days. Co-Authored-By: Claude Opus 5.5 <[email protected]> * Benchmarks: keep the Mac mini's 2026-10-22 runs The 2026-11-04 Mac mini runs had Horizon offscreen against a windowed Java Edition, which skips presenting and isn't a fair comparison with the page's "1280x720 window" method. The Mac mini keeps its 10-22 files, and its note says its runs are older, as the ThinkPad's does. summary.json and public/benchmarks.json rebuilt with the same script. Co-Authored-By: Claude Opus 5.5 <[email protected]> --------- Co-authored-by: Claude Opus 5.5 <[email protected]>
This commit is contained in:
Ludovic MarieandClaude Opus 5.5 authored and GitHub committed
1 parent
8dcc740829
commit
c8a185aff4
8 files changed
+2748
-2426
No files matched your search
@@ -60,7 +60,9 @@ const COMMAND = "horizon-client --benchmark standard --render-distance 12";
|
||||
export default async function BenchmarksPage() {
|
||||
const { machines, rows } = await getBenchmarks();
|
||||
const lost = losses(machines);
|
||||
const measured = machines[0]?.measured;
|
||||
// Machines can be measured on different days (the ThinkPad's runs are older than the others'): give the span.
|
||||
const days = machines.map((m) => m.measured).filter(Boolean).sort();
|
||||
const measured = days.length ? (days[0] === days.at(-1) ? `on ${days[0]}` : `between ${days[0]} and ${days.at(-1)}`) : null;
|
||||
const trace = machines.find((m) => m.id === "macmini")?.runs["12"];
|
||||
|
||||
return (
|
||||
@@ -168,7 +170,7 @@ export default async function BenchmarksPage() {
|
||||
<p>
|
||||
VSync off, no frame cap, a 1280x720 window, simulation distance 12, everything else at vanilla's defaults. Java Edition 26.3 runs as
|
||||
players get it, with the official launcher's Java options. Each figure is the median over the runs
|
||||
{measured ? `, measured on ${measured}` : ""}.
|
||||
{measured ? `, measured ${measured}` : ""}.
|
||||
</p>
|
||||
<p>
|
||||
The full method and its caveats are in{" "}
|
||||
|
||||
File diff suppressed because it is too large.
Load diff
File diff suppressed because it is too large.
Load diff
File diff suppressed because it is too large.
Load diff
File diff suppressed because it is too large.
Load diff
@@ -6,7 +6,7 @@
|
||||
],
|
||||
"version": "v2026.10.22.04",
|
||||
"capped_fps": 144,
|
||||
"note": "On macOS this Horizon build presents its frames in step with the 144 Hz display even with VSync off, so its average frame rate here is capped at 144: compare the 1% lows and frame times instead. The Mac mini was also running other jobs during the runs (CI, simulators; 1-minute load average 20 to 56 on 14 threads), which costs both games, vanilla more at render distance 32."
|
||||
"note": "Measured on 2026-10-22 with Horizon v2026.10.22.04 and not measured again on 2026-11-04: that day the Mac mini's session was locked, so Horizon could only draw offscreen, which isn't a fair comparison with Java Edition's window. On macOS this Horizon build presents its frames in step with the 144 Hz display even with VSync off, so its average frame rate here is capped at 144: compare the 1% lows and frame times instead. The Mac mini was also running other jobs during the runs (CI, simulators; 1-minute load average 20 to 56 on 14 threads), which costs both games, vanilla more at render distance 32."
|
||||
},
|
||||
{
|
||||
"machine": "Mac mini (M4 Pro)",
|
||||
@@ -34,7 +34,7 @@
|
||||
"thinkpad/horizon-thinkpad-rd12.json"
|
||||
],
|
||||
"version": "v2026.10.22.04",
|
||||
"note": "A 2016 laptop: 2 cores / 4 threads and Intel integrated graphics, Linux with Mesa 26.1.8 (Horizon on Vulkan, vanilla on OpenGL). It has no screen attached, so both games drew to a GPU compositor (sway on the Intel GPU) in a 1280x720 window. Horizon ran two flights per scene, vanilla two separate games of one. Render distance 32 is not shown: in 5 minutes neither game finished loading the spawn, and given 25, Horizon kept only about 900 to 1,400 chunk columns resident on this machine instead of the 3,200 a full view holds, so the two games would not have drawn the same world."
|
||||
"note": "Measured on 2026-10-22 with Horizon v2026.10.22.04 and not measured again on 2026-11-04: this laptop now runs other work and no longer runs benchmarks. A 2016 laptop: 2 cores / 4 threads and Intel integrated graphics, Linux with Mesa 26.1.8 (Horizon on Vulkan, vanilla on OpenGL). It has no screen attached, so both games drew to a GPU compositor (sway on the Intel GPU) in a 1280x720 window. Horizon ran two flights per scene, vanilla two separate games of one. Render distance 32 is not shown: in 5 minutes neither game finished loading the spawn, and given 25, Horizon kept only about 900 to 1,400 chunk columns resident on this machine instead of the 3,200 a full view holds, so the two games would not have drawn the same world. Players see less on this kind of hardware: in the last 30 days of real play (opt-in telemetry), the 480 installs with an HD Graphics 620 had a middle frame of about 16 ms (61 FPS, many of them with VSync on) and their slowest 1% of frames at about 34 ms (29 FPS), against the 39 FPS 1% low measured here; Intel's built-in graphics as a whole, about a quarter of all play hours, spent a third of their hours under 60 FPS."
|
||||
},
|
||||
{
|
||||
"machine": "ThinkPad (i7-7500U, HD Graphics 620)",
|
||||
@@ -48,15 +48,15 @@
|
||||
"reports": [
|
||||
"desktop/horizon-desktop-rd12.json"
|
||||
],
|
||||
"version": "v2026.10.22.13",
|
||||
"note": "Linux (NixOS) with the NVIDIA 595 driver. The desktop is in daily use, so both games drew to a separate GPU compositor (a headless sway rendering on the RTX 4080) in a 1280x720 window, never on its screen: Horizon through Wayland, vanilla through Xwayland, both on Vulkan. Other jobs were running during the runs (1-minute load average 18 to 25 on 24 threads), which costs both games, chunk loading most."
|
||||
"version": "Nightly 2026-11-04 (67dad0d)",
|
||||
"note": "Measured on 2026-11-04 with Horizon's Nightly of that day. Linux (NixOS) with the NVIDIA 595 driver. Both games drew to a separate GPU compositor (a headless sway rendering on the RTX 4080) in a 1280x720 window, never on its screen, both through Wayland on Vulkan. The desktop also runs Horizon's CI and other builds: each game started at a 1-minute load average of 28 to 30 on 24 threads, and it rose to 40 to 70 during the runs (18 to 25 on 2026-10-23). That costs both games, most in the 1% lows and the start-up times; the first start of each series also read the game from a cold disk. Java Edition at render distance 32 crashed twice under that load (a GPU timeout waiting for the next frame) before a third series finished."
|
||||
},
|
||||
{
|
||||
"machine": "Desktop (RTX 4080, i7-13700KF)",
|
||||
"reports": [
|
||||
"desktop/horizon-desktop-rd32.json"
|
||||
],
|
||||
"version": "v2026.10.22.13"
|
||||
"version": "Nightly 2026-11-04 (67dad0d)"
|
||||
},
|
||||
{
|
||||
"machine": "Desktop (RTX 4080, i7-13700KF)",
|
||||
|
||||
@@ -1,7 +1,7 @@
|
||||
{
|
||||
"format": "horizon-benchmark-summary",
|
||||
"version": 1,
|
||||
"generated": "2026-10-23T01:09:54Z",
|
||||
"generated": "2026-11-04T16:30:52Z",
|
||||
"min_installs": 3,
|
||||
"groups": [
|
||||
{
|
||||
@@ -21,7 +21,7 @@
|
||||
],
|
||||
"repeats": 3,
|
||||
"load_avg_1m": 55.691,
|
||||
"note": "On macOS this Horizon build presents its frames in step with the 144 Hz display even with VSync off, so its average frame rate here is capped at 144: compare the 1% lows and frame times instead. The Mac mini was also running other jobs during the runs (CI, simulators; 1-minute load average 20 to 56 on 14 threads), which costs both games, vanilla more at render distance 32.",
|
||||
"note": "Measured on 2026-10-22 with Horizon v2026.10.22.04 and not measured again on 2026-11-04: that day the Mac mini's session was locked, so Horizon could only draw offscreen, which isn't a fair comparison with Java Edition's window. On macOS this Horizon build presents its frames in step with the 144 Hz display even with VSync off, so its average frame rate here is capped at 144: compare the 1% lows and frame times instead. The Mac mini was also running other jobs during the runs (CI, simulators; 1-minute load average 20 to 56 on 14 threads), which costs both games, vanilla more at render distance 32.",
|
||||
"capped_fps": 144,
|
||||
"medians": {
|
||||
"avg_fps": 144,
|
||||
@@ -228,7 +228,7 @@
|
||||
],
|
||||
"repeats": 2,
|
||||
"load_avg_1m": 0.988,
|
||||
"note": "A 2016 laptop: 2 cores / 4 threads and Intel integrated graphics, Linux with Mesa 26.1.8 (Horizon on Vulkan, vanilla on OpenGL). It has no screen attached, so both games drew to a GPU compositor (sway on the Intel GPU) in a 1280x720 window. Horizon ran two flights per scene, vanilla two separate games of one. Render distance 32 is not shown: in 5 minutes neither game finished loading the spawn, and given 25, Horizon kept only about 900 to 1,400 chunk columns resident on this machine instead of the 3,200 a full view holds, so the two games would not have drawn the same world.",
|
||||
"note": "Measured on 2026-10-22 with Horizon v2026.10.22.04 and not measured again on 2026-11-04: this laptop now runs other work and no longer runs benchmarks. A 2016 laptop: 2 cores / 4 threads and Intel integrated graphics, Linux with Mesa 26.1.8 (Horizon on Vulkan, vanilla on OpenGL). It has no screen attached, so both games drew to a GPU compositor (sway on the Intel GPU) in a 1280x720 window. Horizon ran two flights per scene, vanilla two separate games of one. Render distance 32 is not shown: in 5 minutes neither game finished loading the spawn, and given 25, Horizon kept only about 900 to 1,400 chunk columns resident on this machine instead of the 3,200 a full view holds, so the two games would not have drawn the same world. Players see less on this kind of hardware: in the last 30 days of real play (opt-in telemetry), the 480 installs with an HD Graphics 620 had a middle frame of about 16 ms (61 FPS, many of them with VSync on) and their slowest 1% of frames at about 34 ms (29 FPS), against the 39 FPS 1% low measured here; Intel's built-in graphics as a whole, about a quarter of all play hours, spent a third of their hours under 60 FPS.",
|
||||
"medians": {
|
||||
"avg_fps": 209,
|
||||
"low_1pct_fps": 39,
|
||||
@@ -321,8 +321,8 @@
|
||||
"game": "horizon",
|
||||
"source": "us",
|
||||
"machine": "Desktop (RTX 4080, i7-13700KF)",
|
||||
"measured": "2026-10-23",
|
||||
"version": "v2026.10.22.13",
|
||||
"measured": "2026-11-04",
|
||||
"version": "Nightly 2026-11-04 (67dad0d)",
|
||||
"backend": "vulkan",
|
||||
"os": "linux",
|
||||
"window": [
|
||||
@@ -330,39 +330,39 @@
|
||||
720
|
||||
],
|
||||
"repeats": 3,
|
||||
"load_avg_1m": 17.934,
|
||||
"note": "Linux (NixOS) with the NVIDIA 595 driver. The desktop is in daily use, so both games drew to a separate GPU compositor (a headless sway rendering on the RTX 4080) in a 1280x720 window, never on its screen: Horizon through Wayland, vanilla through Xwayland, both on Vulkan. Other jobs were running during the runs (1-minute load average 18 to 25 on 24 threads), which costs both games, chunk loading most.",
|
||||
"load_avg_1m": 28.408,
|
||||
"note": "Measured on 2026-11-04 with Horizon's Nightly of that day. Linux (NixOS) with the NVIDIA 595 driver. Both games drew to a separate GPU compositor (a headless sway rendering on the RTX 4080) in a 1280x720 window, never on its screen, both through Wayland on Vulkan. The desktop also runs Horizon's CI and other builds: each game started at a 1-minute load average of 28 to 30 on 24 threads, and it rose to 40 to 70 during the runs (18 to 25 on 2026-10-23). That costs both games, most in the 1% lows and the start-up times; the first start of each series also read the game from a cold disk. Java Edition at render distance 32 crashed twice under that load (a GPU timeout waiting for the next frame) before a third series finished.",
|
||||
"medians": {
|
||||
"avg_fps": 2993,
|
||||
"low_1pct_fps": 311,
|
||||
"p95_ms": 0.5,
|
||||
"world_ms": 17973,
|
||||
"chunk_load_ms": 1797,
|
||||
"rss_peak_mb": 1414,
|
||||
"rss_steady_mb": 1202,
|
||||
"cpu_p50_ms": 0.3,
|
||||
"avg_fps": 2224,
|
||||
"low_1pct_fps": 308,
|
||||
"p95_ms": 1,
|
||||
"world_ms": 74754,
|
||||
"chunk_load_ms": 4403,
|
||||
"rss_peak_mb": 1546,
|
||||
"rss_steady_mb": 1348,
|
||||
"cpu_p50_ms": 0.2,
|
||||
"gpu_p50_ms": 0.1
|
||||
},
|
||||
"scenes": {
|
||||
"village": {
|
||||
"avg_fps": 2594,
|
||||
"low_1pct_fps": 254,
|
||||
"chunk_load_ms": 1260
|
||||
"avg_fps": 2959,
|
||||
"low_1pct_fps": 368,
|
||||
"chunk_load_ms": 5571
|
||||
},
|
||||
"forest": {
|
||||
"avg_fps": 3427,
|
||||
"low_1pct_fps": 328,
|
||||
"chunk_load_ms": 1592
|
||||
"avg_fps": 2334,
|
||||
"low_1pct_fps": 337,
|
||||
"chunk_load_ms": 3318
|
||||
},
|
||||
"ocean": {
|
||||
"avg_fps": 3277,
|
||||
"low_1pct_fps": 388,
|
||||
"chunk_load_ms": 2660
|
||||
"avg_fps": 2193,
|
||||
"low_1pct_fps": 345,
|
||||
"chunk_load_ms": 2463
|
||||
},
|
||||
"mountain": {
|
||||
"avg_fps": 2673,
|
||||
"low_1pct_fps": 275,
|
||||
"chunk_load_ms": 1675
|
||||
"avg_fps": 1412,
|
||||
"low_1pct_fps": 182,
|
||||
"chunk_load_ms": 6261
|
||||
}
|
||||
}
|
||||
},
|
||||
@@ -373,8 +373,8 @@
|
||||
"game": "horizon",
|
||||
"source": "us",
|
||||
"machine": "Desktop (RTX 4080, i7-13700KF)",
|
||||
"measured": "2026-10-23",
|
||||
"version": "v2026.10.22.13",
|
||||
"measured": "2026-11-04",
|
||||
"version": "Nightly 2026-11-04 (67dad0d)",
|
||||
"backend": "vulkan",
|
||||
"os": "linux",
|
||||
"window": [
|
||||
@@ -382,38 +382,38 @@
|
||||
720
|
||||
],
|
||||
"repeats": 3,
|
||||
"load_avg_1m": 22.354,
|
||||
"load_avg_1m": 29.665,
|
||||
"medians": {
|
||||
"avg_fps": 672,
|
||||
"low_1pct_fps": 89,
|
||||
"p95_ms": 3.6,
|
||||
"world_ms": 8999,
|
||||
"chunk_load_ms": 4677,
|
||||
"rss_peak_mb": 1957,
|
||||
"rss_steady_mb": 1646,
|
||||
"cpu_p50_ms": 1.1,
|
||||
"gpu_p50_ms": 0.3
|
||||
"avg_fps": 800,
|
||||
"low_1pct_fps": 140,
|
||||
"p95_ms": 2.7,
|
||||
"world_ms": 86586,
|
||||
"chunk_load_ms": 17579,
|
||||
"rss_peak_mb": 2051,
|
||||
"rss_steady_mb": 1811,
|
||||
"cpu_p50_ms": 0.8,
|
||||
"gpu_p50_ms": 0.4
|
||||
},
|
||||
"scenes": {
|
||||
"village": {
|
||||
"avg_fps": 686,
|
||||
"low_1pct_fps": 93,
|
||||
"chunk_load_ms": 2702
|
||||
"avg_fps": 854,
|
||||
"low_1pct_fps": 130,
|
||||
"chunk_load_ms": 11083
|
||||
},
|
||||
"forest": {
|
||||
"avg_fps": 590,
|
||||
"low_1pct_fps": 76,
|
||||
"chunk_load_ms": 4823
|
||||
"avg_fps": 898,
|
||||
"low_1pct_fps": 179,
|
||||
"chunk_load_ms": 15020
|
||||
},
|
||||
"ocean": {
|
||||
"avg_fps": 868,
|
||||
"low_1pct_fps": 113,
|
||||
"chunk_load_ms": 3629
|
||||
"avg_fps": 735,
|
||||
"low_1pct_fps": 135,
|
||||
"chunk_load_ms": 12399
|
||||
},
|
||||
"mountain": {
|
||||
"avg_fps": 544,
|
||||
"low_1pct_fps": 73,
|
||||
"chunk_load_ms": 7552
|
||||
"avg_fps": 713,
|
||||
"low_1pct_fps": 118,
|
||||
"chunk_load_ms": 31816
|
||||
}
|
||||
}
|
||||
},
|
||||
@@ -424,7 +424,7 @@
|
||||
"game": "vanilla",
|
||||
"source": "us",
|
||||
"machine": "Desktop (RTX 4080, i7-13700KF)",
|
||||
"measured": "2026-10-23",
|
||||
"measured": "2026-11-04",
|
||||
"version": "26.3",
|
||||
"backend": "vulkan",
|
||||
"os": "linux",
|
||||
@@ -433,38 +433,38 @@
|
||||
720
|
||||
],
|
||||
"repeats": 3,
|
||||
"load_avg_1m": 25.007,
|
||||
"load_avg_1m": 28.788,
|
||||
"medians": {
|
||||
"avg_fps": 2600,
|
||||
"low_1pct_fps": 312,
|
||||
"p95_ms": 0.5,
|
||||
"world_ms": 30470,
|
||||
"chunk_load_ms": 10837,
|
||||
"rss_peak_mb": 3078,
|
||||
"rss_steady_mb": 2883,
|
||||
"avg_fps": 1824,
|
||||
"low_1pct_fps": 201,
|
||||
"p95_ms": 1.1,
|
||||
"world_ms": 191064,
|
||||
"chunk_load_ms": 20512,
|
||||
"rss_peak_mb": 3108,
|
||||
"rss_steady_mb": 2985,
|
||||
"cpu_p50_ms": null,
|
||||
"gpu_p50_ms": null
|
||||
},
|
||||
"scenes": {
|
||||
"village": {
|
||||
"avg_fps": 2482,
|
||||
"low_1pct_fps": 385,
|
||||
"chunk_load_ms": 8003
|
||||
"avg_fps": 2162,
|
||||
"low_1pct_fps": 163,
|
||||
"chunk_load_ms": 18724
|
||||
},
|
||||
"forest": {
|
||||
"avg_fps": 2368,
|
||||
"low_1pct_fps": 209,
|
||||
"chunk_load_ms": 9654
|
||||
"avg_fps": 1842,
|
||||
"low_1pct_fps": 255,
|
||||
"chunk_load_ms": 8852
|
||||
},
|
||||
"ocean": {
|
||||
"avg_fps": 2727,
|
||||
"low_1pct_fps": 247,
|
||||
"chunk_load_ms": 10442
|
||||
"avg_fps": 1525,
|
||||
"low_1pct_fps": 191,
|
||||
"chunk_load_ms": 21216
|
||||
},
|
||||
"mountain": {
|
||||
"avg_fps": 2824,
|
||||
"low_1pct_fps": 407,
|
||||
"chunk_load_ms": 15249
|
||||
"avg_fps": 1769,
|
||||
"low_1pct_fps": 196,
|
||||
"chunk_load_ms": 33255
|
||||
}
|
||||
}
|
||||
},
|
||||
@@ -475,7 +475,7 @@
|
||||
"game": "vanilla",
|
||||
"source": "us",
|
||||
"machine": "Desktop (RTX 4080, i7-13700KF)",
|
||||
"measured": "2026-10-23",
|
||||
"measured": "2026-11-04",
|
||||
"version": "26.3",
|
||||
"backend": "vulkan",
|
||||
"os": "linux",
|
||||
@@ -484,38 +484,38 @@
|
||||
720
|
||||
],
|
||||
"repeats": 3,
|
||||
"load_avg_1m": 21.652,
|
||||
"load_avg_1m": 28.799,
|
||||
"medians": {
|
||||
"avg_fps": 849,
|
||||
"low_1pct_fps": 112,
|
||||
"p95_ms": 3.1,
|
||||
"world_ms": 58554,
|
||||
"chunk_load_ms": 31819,
|
||||
"rss_peak_mb": 3484,
|
||||
"rss_steady_mb": 3419,
|
||||
"avg_fps": 696,
|
||||
"low_1pct_fps": 64,
|
||||
"p95_ms": 4.1,
|
||||
"world_ms": 235049,
|
||||
"chunk_load_ms": 79051,
|
||||
"rss_peak_mb": 3372,
|
||||
"rss_steady_mb": 3291,
|
||||
"cpu_p50_ms": null,
|
||||
"gpu_p50_ms": null
|
||||
},
|
||||
"scenes": {
|
||||
"village": {
|
||||
"avg_fps": 881,
|
||||
"low_1pct_fps": 110,
|
||||
"chunk_load_ms": 19818
|
||||
"avg_fps": 526,
|
||||
"low_1pct_fps": 44,
|
||||
"chunk_load_ms": 79828
|
||||
},
|
||||
"forest": {
|
||||
"avg_fps": 1072,
|
||||
"low_1pct_fps": 165,
|
||||
"chunk_load_ms": 28927
|
||||
"avg_fps": 599,
|
||||
"low_1pct_fps": 62,
|
||||
"chunk_load_ms": 61173
|
||||
},
|
||||
"ocean": {
|
||||
"avg_fps": 894,
|
||||
"low_1pct_fps": 142,
|
||||
"chunk_load_ms": 19246
|
||||
"avg_fps": 1175,
|
||||
"low_1pct_fps": 120,
|
||||
"chunk_load_ms": 61762
|
||||
},
|
||||
"mountain": {
|
||||
"avg_fps": 551,
|
||||
"low_1pct_fps": 33,
|
||||
"chunk_load_ms": 59283
|
||||
"avg_fps": 486,
|
||||
"low_1pct_fps": 29,
|
||||
"chunk_load_ms": 113440
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
+93
-93
@@ -1,7 +1,7 @@
|
||||
{
|
||||
"format": "horizon-benchmark-summary",
|
||||
"version": 1,
|
||||
"generated": "2026-10-23T01:09:54Z",
|
||||
"generated": "2026-11-04T16:30:52Z",
|
||||
"min_installs": 3,
|
||||
"groups": [
|
||||
{
|
||||
@@ -21,7 +21,7 @@
|
||||
],
|
||||
"repeats": 3,
|
||||
"load_avg_1m": 55.691,
|
||||
"note": "On macOS this Horizon build presents its frames in step with the 144 Hz display even with VSync off, so its average frame rate here is capped at 144: compare the 1% lows and frame times instead. The Mac mini was also running other jobs during the runs (CI, simulators; 1-minute load average 20 to 56 on 14 threads), which costs both games, vanilla more at render distance 32.",
|
||||
"note": "Measured on 2026-10-22 with Horizon v2026.10.22.04 and not measured again on 2026-11-04: that day the Mac mini's session was locked, so Horizon could only draw offscreen, which isn't a fair comparison with Java Edition's window. On macOS this Horizon build presents its frames in step with the 144 Hz display even with VSync off, so its average frame rate here is capped at 144: compare the 1% lows and frame times instead. The Mac mini was also running other jobs during the runs (CI, simulators; 1-minute load average 20 to 56 on 14 threads), which costs both games, vanilla more at render distance 32.",
|
||||
"capped_fps": 144,
|
||||
"medians": {
|
||||
"avg_fps": 144,
|
||||
@@ -228,7 +228,7 @@
|
||||
],
|
||||
"repeats": 2,
|
||||
"load_avg_1m": 0.988,
|
||||
"note": "A 2016 laptop: 2 cores / 4 threads and Intel integrated graphics, Linux with Mesa 26.1.8 (Horizon on Vulkan, vanilla on OpenGL). It has no screen attached, so both games drew to a GPU compositor (sway on the Intel GPU) in a 1280x720 window. Horizon ran two flights per scene, vanilla two separate games of one. Render distance 32 is not shown: in 5 minutes neither game finished loading the spawn, and given 25, Horizon kept only about 900 to 1,400 chunk columns resident on this machine instead of the 3,200 a full view holds, so the two games would not have drawn the same world.",
|
||||
"note": "Measured on 2026-10-22 with Horizon v2026.10.22.04 and not measured again on 2026-11-04: this laptop now runs other work and no longer runs benchmarks. A 2016 laptop: 2 cores / 4 threads and Intel integrated graphics, Linux with Mesa 26.1.8 (Horizon on Vulkan, vanilla on OpenGL). It has no screen attached, so both games drew to a GPU compositor (sway on the Intel GPU) in a 1280x720 window. Horizon ran two flights per scene, vanilla two separate games of one. Render distance 32 is not shown: in 5 minutes neither game finished loading the spawn, and given 25, Horizon kept only about 900 to 1,400 chunk columns resident on this machine instead of the 3,200 a full view holds, so the two games would not have drawn the same world. Players see less on this kind of hardware: in the last 30 days of real play (opt-in telemetry), the 480 installs with an HD Graphics 620 had a middle frame of about 16 ms (61 FPS, many of them with VSync on) and their slowest 1% of frames at about 34 ms (29 FPS), against the 39 FPS 1% low measured here; Intel's built-in graphics as a whole, about a quarter of all play hours, spent a third of their hours under 60 FPS.",
|
||||
"medians": {
|
||||
"avg_fps": 209,
|
||||
"low_1pct_fps": 39,
|
||||
@@ -321,8 +321,8 @@
|
||||
"game": "horizon",
|
||||
"source": "us",
|
||||
"machine": "Desktop (RTX 4080, i7-13700KF)",
|
||||
"measured": "2026-10-23",
|
||||
"version": "v2026.10.22.13",
|
||||
"measured": "2026-11-04",
|
||||
"version": "Nightly 2026-11-04 (67dad0d)",
|
||||
"backend": "vulkan",
|
||||
"os": "linux",
|
||||
"window": [
|
||||
@@ -330,39 +330,39 @@
|
||||
720
|
||||
],
|
||||
"repeats": 3,
|
||||
"load_avg_1m": 17.934,
|
||||
"note": "Linux (NixOS) with the NVIDIA 595 driver. The desktop is in daily use, so both games drew to a separate GPU compositor (a headless sway rendering on the RTX 4080) in a 1280x720 window, never on its screen: Horizon through Wayland, vanilla through Xwayland, both on Vulkan. Other jobs were running during the runs (1-minute load average 18 to 25 on 24 threads), which costs both games, chunk loading most.",
|
||||
"load_avg_1m": 28.408,
|
||||
"note": "Measured on 2026-11-04 with Horizon's Nightly of that day. Linux (NixOS) with the NVIDIA 595 driver. Both games drew to a separate GPU compositor (a headless sway rendering on the RTX 4080) in a 1280x720 window, never on its screen, both through Wayland on Vulkan. The desktop also runs Horizon's CI and other builds: each game started at a 1-minute load average of 28 to 30 on 24 threads, and it rose to 40 to 70 during the runs (18 to 25 on 2026-10-23). That costs both games, most in the 1% lows and the start-up times; the first start of each series also read the game from a cold disk. Java Edition at render distance 32 crashed twice under that load (a GPU timeout waiting for the next frame) before a third series finished.",
|
||||
"medians": {
|
||||
"avg_fps": 2993,
|
||||
"low_1pct_fps": 311,
|
||||
"p95_ms": 0.5,
|
||||
"world_ms": 17973,
|
||||
"chunk_load_ms": 1797,
|
||||
"rss_peak_mb": 1414,
|
||||
"rss_steady_mb": 1202,
|
||||
"cpu_p50_ms": 0.3,
|
||||
"avg_fps": 2224,
|
||||
"low_1pct_fps": 308,
|
||||
"p95_ms": 1,
|
||||
"world_ms": 74754,
|
||||
"chunk_load_ms": 4403,
|
||||
"rss_peak_mb": 1546,
|
||||
"rss_steady_mb": 1348,
|
||||
"cpu_p50_ms": 0.2,
|
||||
"gpu_p50_ms": 0.1
|
||||
},
|
||||
"scenes": {
|
||||
"village": {
|
||||
"avg_fps": 2594,
|
||||
"low_1pct_fps": 254,
|
||||
"chunk_load_ms": 1260
|
||||
"avg_fps": 2959,
|
||||
"low_1pct_fps": 368,
|
||||
"chunk_load_ms": 5571
|
||||
},
|
||||
"forest": {
|
||||
"avg_fps": 3427,
|
||||
"low_1pct_fps": 328,
|
||||
"chunk_load_ms": 1592
|
||||
"avg_fps": 2334,
|
||||
"low_1pct_fps": 337,
|
||||
"chunk_load_ms": 3318
|
||||
},
|
||||
"ocean": {
|
||||
"avg_fps": 3277,
|
||||
"low_1pct_fps": 388,
|
||||
"chunk_load_ms": 2660
|
||||
"avg_fps": 2193,
|
||||
"low_1pct_fps": 345,
|
||||
"chunk_load_ms": 2463
|
||||
},
|
||||
"mountain": {
|
||||
"avg_fps": 2673,
|
||||
"low_1pct_fps": 275,
|
||||
"chunk_load_ms": 1675
|
||||
"avg_fps": 1412,
|
||||
"low_1pct_fps": 182,
|
||||
"chunk_load_ms": 6261
|
||||
}
|
||||
}
|
||||
},
|
||||
@@ -373,8 +373,8 @@
|
||||
"game": "horizon",
|
||||
"source": "us",
|
||||
"machine": "Desktop (RTX 4080, i7-13700KF)",
|
||||
"measured": "2026-10-23",
|
||||
"version": "v2026.10.22.13",
|
||||
"measured": "2026-11-04",
|
||||
"version": "Nightly 2026-11-04 (67dad0d)",
|
||||
"backend": "vulkan",
|
||||
"os": "linux",
|
||||
"window": [
|
||||
@@ -382,38 +382,38 @@
|
||||
720
|
||||
],
|
||||
"repeats": 3,
|
||||
"load_avg_1m": 22.354,
|
||||
"load_avg_1m": 29.665,
|
||||
"medians": {
|
||||
"avg_fps": 672,
|
||||
"low_1pct_fps": 89,
|
||||
"p95_ms": 3.6,
|
||||
"world_ms": 8999,
|
||||
"chunk_load_ms": 4677,
|
||||
"rss_peak_mb": 1957,
|
||||
"rss_steady_mb": 1646,
|
||||
"cpu_p50_ms": 1.1,
|
||||
"gpu_p50_ms": 0.3
|
||||
"avg_fps": 800,
|
||||
"low_1pct_fps": 140,
|
||||
"p95_ms": 2.7,
|
||||
"world_ms": 86586,
|
||||
"chunk_load_ms": 17579,
|
||||
"rss_peak_mb": 2051,
|
||||
"rss_steady_mb": 1811,
|
||||
"cpu_p50_ms": 0.8,
|
||||
"gpu_p50_ms": 0.4
|
||||
},
|
||||
"scenes": {
|
||||
"village": {
|
||||
"avg_fps": 686,
|
||||
"low_1pct_fps": 93,
|
||||
"chunk_load_ms": 2702
|
||||
"avg_fps": 854,
|
||||
"low_1pct_fps": 130,
|
||||
"chunk_load_ms": 11083
|
||||
},
|
||||
"forest": {
|
||||
"avg_fps": 590,
|
||||
"low_1pct_fps": 76,
|
||||
"chunk_load_ms": 4823
|
||||
"avg_fps": 898,
|
||||
"low_1pct_fps": 179,
|
||||
"chunk_load_ms": 15020
|
||||
},
|
||||
"ocean": {
|
||||
"avg_fps": 868,
|
||||
"low_1pct_fps": 113,
|
||||
"chunk_load_ms": 3629
|
||||
"avg_fps": 735,
|
||||
"low_1pct_fps": 135,
|
||||
"chunk_load_ms": 12399
|
||||
},
|
||||
"mountain": {
|
||||
"avg_fps": 544,
|
||||
"low_1pct_fps": 73,
|
||||
"chunk_load_ms": 7552
|
||||
"avg_fps": 713,
|
||||
"low_1pct_fps": 118,
|
||||
"chunk_load_ms": 31816
|
||||
}
|
||||
}
|
||||
},
|
||||
@@ -424,7 +424,7 @@
|
||||
"game": "vanilla",
|
||||
"source": "us",
|
||||
"machine": "Desktop (RTX 4080, i7-13700KF)",
|
||||
"measured": "2026-10-23",
|
||||
"measured": "2026-11-04",
|
||||
"version": "26.3",
|
||||
"backend": "vulkan",
|
||||
"os": "linux",
|
||||
@@ -433,38 +433,38 @@
|
||||
720
|
||||
],
|
||||
"repeats": 3,
|
||||
"load_avg_1m": 25.007,
|
||||
"load_avg_1m": 28.788,
|
||||
"medians": {
|
||||
"avg_fps": 2600,
|
||||
"low_1pct_fps": 312,
|
||||
"p95_ms": 0.5,
|
||||
"world_ms": 30470,
|
||||
"chunk_load_ms": 10837,
|
||||
"rss_peak_mb": 3078,
|
||||
"rss_steady_mb": 2883,
|
||||
"avg_fps": 1824,
|
||||
"low_1pct_fps": 201,
|
||||
"p95_ms": 1.1,
|
||||
"world_ms": 191064,
|
||||
"chunk_load_ms": 20512,
|
||||
"rss_peak_mb": 3108,
|
||||
"rss_steady_mb": 2985,
|
||||
"cpu_p50_ms": null,
|
||||
"gpu_p50_ms": null
|
||||
},
|
||||
"scenes": {
|
||||
"village": {
|
||||
"avg_fps": 2482,
|
||||
"low_1pct_fps": 385,
|
||||
"chunk_load_ms": 8003
|
||||
"avg_fps": 2162,
|
||||
"low_1pct_fps": 163,
|
||||
"chunk_load_ms": 18724
|
||||
},
|
||||
"forest": {
|
||||
"avg_fps": 2368,
|
||||
"low_1pct_fps": 209,
|
||||
"chunk_load_ms": 9654
|
||||
"avg_fps": 1842,
|
||||
"low_1pct_fps": 255,
|
||||
"chunk_load_ms": 8852
|
||||
},
|
||||
"ocean": {
|
||||
"avg_fps": 2727,
|
||||
"low_1pct_fps": 247,
|
||||
"chunk_load_ms": 10442
|
||||
"avg_fps": 1525,
|
||||
"low_1pct_fps": 191,
|
||||
"chunk_load_ms": 21216
|
||||
},
|
||||
"mountain": {
|
||||
"avg_fps": 2824,
|
||||
"low_1pct_fps": 407,
|
||||
"chunk_load_ms": 15249
|
||||
"avg_fps": 1769,
|
||||
"low_1pct_fps": 196,
|
||||
"chunk_load_ms": 33255
|
||||
}
|
||||
}
|
||||
},
|
||||
@@ -475,7 +475,7 @@
|
||||
"game": "vanilla",
|
||||
"source": "us",
|
||||
"machine": "Desktop (RTX 4080, i7-13700KF)",
|
||||
"measured": "2026-10-23",
|
||||
"measured": "2026-11-04",
|
||||
"version": "26.3",
|
||||
"backend": "vulkan",
|
||||
"os": "linux",
|
||||
@@ -484,38 +484,38 @@
|
||||
720
|
||||
],
|
||||
"repeats": 3,
|
||||
"load_avg_1m": 21.652,
|
||||
"load_avg_1m": 28.799,
|
||||
"medians": {
|
||||
"avg_fps": 849,
|
||||
"low_1pct_fps": 112,
|
||||
"p95_ms": 3.1,
|
||||
"world_ms": 58554,
|
||||
"chunk_load_ms": 31819,
|
||||
"rss_peak_mb": 3484,
|
||||
"rss_steady_mb": 3419,
|
||||
"avg_fps": 696,
|
||||
"low_1pct_fps": 64,
|
||||
"p95_ms": 4.1,
|
||||
"world_ms": 235049,
|
||||
"chunk_load_ms": 79051,
|
||||
"rss_peak_mb": 3372,
|
||||
"rss_steady_mb": 3291,
|
||||
"cpu_p50_ms": null,
|
||||
"gpu_p50_ms": null
|
||||
},
|
||||
"scenes": {
|
||||
"village": {
|
||||
"avg_fps": 881,
|
||||
"low_1pct_fps": 110,
|
||||
"chunk_load_ms": 19818
|
||||
"avg_fps": 526,
|
||||
"low_1pct_fps": 44,
|
||||
"chunk_load_ms": 79828
|
||||
},
|
||||
"forest": {
|
||||
"avg_fps": 1072,
|
||||
"low_1pct_fps": 165,
|
||||
"chunk_load_ms": 28927
|
||||
"avg_fps": 599,
|
||||
"low_1pct_fps": 62,
|
||||
"chunk_load_ms": 61173
|
||||
},
|
||||
"ocean": {
|
||||
"avg_fps": 894,
|
||||
"low_1pct_fps": 142,
|
||||
"chunk_load_ms": 19246
|
||||
"avg_fps": 1175,
|
||||
"low_1pct_fps": 120,
|
||||
"chunk_load_ms": 61762
|
||||
},
|
||||
"mountain": {
|
||||
"avg_fps": 551,
|
||||
"low_1pct_fps": 33,
|
||||
"chunk_load_ms": 59283
|
||||
"avg_fps": 486,
|
||||
"low_1pct_fps": 29,
|
||||
"chunk_load_ms": 113440
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
Reference in new issue
Block a user