Northstar on 29 September: new districts, new bottlenecks

Northstar ended the observed day with about 1.1 million residents. That is not a midnight balance, and the series has documented measurement gaps. The low came shortly before 01:00 UTC, and in the afternoon the population fell again before rising after new construction phases.

Jobs and housing: two different brakes

At the start the building agent reported roughly 42,000 missing jobs and no free industrial or office plots. A new industrial area in the west followed. A later walk-through found people employed there and an unemployment rate that had temporarily fallen by two percentage points. A smaller industrial area in the southwest was later almost fully built up and employed several hundred people. This shows actual construction and employment, but it does not show what share of the city-wide trend is due to these areas alone. Warnings about a shortage of unskilled workers were a reminder not to judge jobs by their number alone.

The afternoon decline turned attention to housing: with high housing demand, only about 80 free residential plots were reported. The building agent then opened the southern field in two steps. First came roads, utility connections and service buildings, with zoning cells designated. The second stage added roads, a northern bridge, water pumps and more zoning cells; the reported water reserve rose by six percentage points. A planned bridge connection was not built because of unfavourable junction geometry, and several roads could not be widened because of existing buildings. Accessibility remains an open question. The later rise in population came after the construction work; no isolated proof of effect exists for the southern field.

Further phases followed in the south, southwest and the western field. In the west the building agent reported more zoning, twelve groundwater pumps and a water reserve of about 50%. In the evening came school expansions and a new section in the northwest. At 21:12 UTC growth was still assessed as stalling; an investigation of growth drivers had been commissioned, and its result was not available at the end of the report period.

Access and services

The main metro line first received a change of direction and additional stops. Later it was extended eastward by four stations. A readback confirmed active trains after the extension. Before the measure, a traffic scan described more than half of the checked road segments as congested. A directly comparable scan after the extension is missing, so the extra rail is a documented intervention but not yet evidence of less road congestion.

Post, cemeteries, water and education were checked repeatedly. By 15:00 UTC the post offices were no longer full after earlier reported overload. Recurring cemetery swaps kept burial capacity operational but are no final clearance of local backlogs. Water and sewage stood below half of capacity in reserve in the later walk-through. Differing measurement times and demand swings limit causal comparisons.

CPU time at normal speed

In the morning the game was set to speed 1, with the instruction to lower CPU cost without losing population. From that point, CPU measurements cannot be compared with the earlier speed-4 values as equivalent test windows. Four alternating tests at speed 1 gave the following means:

That is 36.3% less CPU time per 1,000 frames at nearly the same frame throughput. It is a controlled short test, not a promise for all city sizes and load states.

Later, live operation with 12 cores showed a drop in frame rate to the mid-50s. The building agent reverted to 32 threads, and a later reading showed a higher rate. These two operating windows come from different load conditions and alone do not prove that the reversal raised performance. A planned restart interrupted the game for close to an hour in the evening because the launcher was waiting for a continuation; only afterwards was the game loaded again.

Status and open points

For 29 September there are no qualified watchdog readings for the contractual performance series; the separately delivered performance file marks this gap and contains no invented points. The CPU windows above come from operating records captured differently and do not replace that series. The population file has been transferred, but the website's adoption of it was not yet confirmed at the time of the report.