FPS & Latency Power Plans Analysis

1.

POWER PLAN BENCHMARK
FPS & Latency Analysis
39 Windows power plans tested by Alchemy Tweaks on a controlled Fortnite + DPC/ISR latency
setup.
POWER PLANS
39
Same rig, same OS image
FPS RUNS
195
Measured by
Primary verdict
5 captures per plan
Alchemy Tweaks
Split winner
LATENCY RUNS
In-house tools · documented hardware · repeat
captures
Best FPS and best latency are not the
same objective.
117
3 ETWs per run
FRAMES ANALYZED
2.68M
FPS workbook total
Source: Alchemy Tweaks FPS + Latency analysis workbooks

2.

The practical answer: choose the plan by the performance goal
EXECUTIVE VERDICT
The FPS winner and the latency winner optimize different metrics, treating one plan as universal would overstate the data.
BEST FOR FPS
BEST FOR DPC P99 LATENCY
BEST BALANCED LATENCY
Arsenza INTEL
Kizzimo Extreme LL
amit's low latency
242.0 Avg FPS · 192.6 1% low · CV 0.4%
115.2 µs DPC p99 · CV 0.1%
171.8 µs · 1.24% combined CPU busy
Summary verdict
1 For FPS: Arsenza INTEL leads, with EndGame statistically
interchangeable.
Important caveat
2 For worst-case latency: Kizzimo is the clear DPC p99
winner.
For a safer latency profile: amit's low
3 latency trades peak p99 for
consistency and lower overhead.
This is a controlled single-rig benchmark: the result is strong for this hardware, OS build and scene, but viewers should validate on their own systems
before treating it as universal.
Alchemy Tweaks · Power plan benchmark summary
Source: Alchemy Tweaks in-house Excel workbooks · Slide 02

3.

METHODOLOGY
The benchmark isolates power-plan behavior on one controlled Windows build
Only the Windows power plan changed between captures, the OS image, hardware and test scene remained fixed.
CPU
GPU
DISPLAY
OS
i5-10400F
RX 5700 XT
1080p · 144 Hz
Win 11 26200
Hyperthreading OFF
FPS capture terminology
Latency capture terminology
• Fortnite custom 1v1 static scene
• Capture-FPS.ps1 v2.0.0
• Intel PresentMon 2.4.1
• 5 captures per plan · 65 s each
• first 5 s discarded → 60 s analyzed
• Same fixed scene and system
• Measure-Latency.ps1 v1.1.0
• WPT xperf kernel ETW traces
• 3 captures per plan · ~40 s traces
• DPC/ISR metrics, lower is better
Method integrity
Repeated captures, documented tools and a fixed static scene reduce gameplay randomness; the Excel files remain the backup evidence for raw runs.
Alchemy Tweaks · Power plan benchmark summary
Source: Alchemy Tweaks in-house Excel workbooks · Slide 03

4.

DATA QUALITY
The dataset is stable enough for a summary verdict
The run counts and integrity checks support a credible comparison without putting raw tables on the slides.
FPS COVERAGE
FRAMES
FPS CV
DROPPED RUNS
195 runs
2.68M
0.41%
0
39 plans × 5 captures
total frames analyzed
median per-plan CV
all failed captures replaced/excluded
LATENCY COVERAGE
LOST ETW EVENTS
LATENCY CV
CPU CLOCK
117 traces
0
0.92%
~2.89 GHz
39 plans × 3 captures
clean trace set
median DPC p99 CV
no plan throttled
Interpretation boundary
Differences are strong within this controlled setup, but absolute rankings can change with CPU, GPU driver, Windows build, game scene
and background services.
Alchemy Tweaks · Power plan benchmark summary
Source: Alchemy Tweaks in-house Excel workbooks · Slide 04

5.

FPS RESULTS
Arsenza INTEL leads FPS, with EndGame effectively interchangeable
Top community plans cluster tightly at the top, Microsoft stock references trail the field leaders.
Average FPS, higher is better
1 Arsenza INTEL
242.0 FPS
2 EndGame
240.8 FPS
3 Prodazin
239.3 FPS
4 AMIT
239.0 FPS
FPS verdict
Winner
Arsenza INTEL
EndGame is statistically tied with
the leader (p=0.09), so the
practical FPS gap is not
meaningful in this run set.
Stock gap
5 LLG Parking + E-core Fix
238.6 FPS
31 MSFT High Performance
226.5 FPS
35 MSFT Ultimate Performance
222.2 FPS
Alchemy Tweaks · Power plan benchmark summary
MSFT High Performance sits
−6.4% vs the top plan; Ultimate
Performance sits −8.2%.
Source: Alchemy Tweaks in-house Excel workbooks · Slide 05

6.

FPS SMOOTHNESS
Low-frame metrics confirm the winner, but highlight a specialist plan
Arsenza remains the best all-round FPS plan, LLG Parking + E-core Fix has the strongest deepest-dip metric.
0.1% Low FPS, higher is better
1 Arsenza INTEL
2 EndGame
168.9 FPS
How to interpret smoothness
169.8 FPS
Average FPS decides the FPS winner. Low-frame metrics
show whether that speed holds under the worst
rendered moments.
5 LLG Parking + E-core Fix
170.4 FPS
3 Prodazin
166.8 FPS
4 AMIT
150.3 FPS
Bottom line
Specialist pick
LLG Parking + E-core Fix
Best 0.1% low (170.4 FPS), while still staying near the
average-FPS leaders.
Use Arsenza INTEL for the best FPS verdict, mention LLG as a smoothness specialist, not as the headline FPS winner.
Alchemy Tweaks · Power plan benchmark summary
Source: Alchemy Tweaks in-house Excel workbooks · Slide 06

7.

FPS VERDICTS
Most plans are rejected because they do not add a measurable FPS advantage
The verdict table is summarized here, the raw per-plan details remain in the workbook.
KEEP
TEST AGAIN
REJECT
5
4
30
winner + close specialists
promising but noisy
slower with no smoothness edge
FPS action list
Run as default
Arsenza INTEL
242.0 Avg FPS, top mean and excellent repeatability.
Fallback
EndGame
Statistically tied with Arsenza on average FPS, 240.8 Avg FPS.
Smoothness specialist
LLG Parking + E-core Fix
170.4 FPS 0.1% low, best deepest-dip metric.
Alchemy Tweaks · Power plan benchmark summary
Source: Alchemy Tweaks in-house Excel workbooks · Slide 07

8.

LATENCY RESULTS
Kizzimo is the clear DPC p99 latency winner
Latency metrics are lower-is-better, all plans remain healthy in absolute terms, but Kizzimo separates clearly from the field.
DPC p99 latency, lower is better
1 Kizzimo Extreme LL
115.2 µs
2 Lowest Latency
161.0 µs
3 Reticle
167.1 µs
4 amit's low latency
171.8 µs
Latency verdict
Pure DPC p99
Kizzimo
Kizzimo reaches 115.2 µs, the next
tested plan is 161.0 µs and much
noisier.
Reality check
9 CactusOS Absolute
175.2 µs
10 MSFT High Performance
176.3 µs
12 MSFT Ultimate Performance
177.4 µs
Alchemy Tweaks · Power plan benchmark summary
The workbook notes all plans are in
a healthy absolute DPC range, the
comparison is about optimization,
not fixing a broken system.
Source: Alchemy Tweaks in-house Excel workbooks · Slide 08

9.

LATENCY TRADE-OFF
Balanced latency is not just the lowest p99 number
Kizzimo wins peak p99, amit's and VTRL are better choices when consistency or low overhead matters.
A · Responsiveness
B · Overhead
DPC p99, lower is better
Combined DPC+ISR CPU busy, lower is better
Kizzimo Extreme LL
115.2 µs
Kizzimo Extreme LL
1.88%
amit's low latency
171.8 µs
amit's low latency
1.24%
CactusOS Absolute
175.2 µs
CactusOS Absolute
1.30%
MSFT High Performance
176.3 µs
MSFT High Performance
1.31%
VTRL Optimized
177.7 µs
VTRL Optimized
1.19%
ARSENZA INTEL Latency
186.6 µs
ARSENZA INTEL Latency
1.17%
Decision rule
Use Kizzimo when minimum DPC p99 is the priority. Use amit's for a safer low-latency profile. Use VTRL / ARSENZA when minimal interrupt overhead matters most.
Alchemy Tweaks · Power plan benchmark summary
Source: Alchemy Tweaks in-house Excel workbooks · Slide 09

10.

LATENCY VERDICTS
Latency recommendations split into three use cases
The workbook verdicts distinguish pure p99 latency, balanced consistency and lowest interrupt overhead.
KEEP
TEST AGAIN
REJECT
5
5
29
latency picks worth using
low number but unstable
no latency advantage
Objective
Recommended plan
Why this is the right choice
Lowest DPC p99
Kizzimo Extreme LL
115.2 µs DPC p99, CV 0.1%.
Balanced latency
amit's low latency
171.8 µs with excellent repeatability and lower combined busy.
Low overhead
VTRL / ARSENZA
VTRL at 1.19% and ARSENZA INTEL at 1.17% combined busy.
Stock reference
MSFT High Performance
Competitive latency (176.3 µs), but not the FPS winner.
Alchemy Tweaks · Power plan benchmark summary
Source: Alchemy Tweaks in-house Excel workbooks · Slide 10

11.

CROSS-METRIC DECISION
There is no universal winner across FPS and latency
A single power plan should be chosen based on the metric the user actually wants to optimize.
Objective
Recommended plan
Why this is the right choice
Max FPS
Arsenza INTEL
242.0 Avg FPS, top mean, excellent repeatability.
FPS fallback
EndGame
Statistically tied with Arsenza on average FPS (p=0.09).
Deepest FPS dips
LLG Parking + E-core Fix
Best 0.1% low in the FPS workbook (170.4 FPS).
Lowest DPC p99
Kizzimo Extreme LL
115.2 µs DPC p99; best worst-case latency result.
Latency + safety
amit's low latency
Consistent and low-overhead latency profile, not just a one-run spike.
Decision rule
The best FPS plan is not the same as the lowest DPC p99 plan, the honest recommendation depends on the target metric.
Alchemy Tweaks · Power plan benchmark summary
Source: Alchemy Tweaks in-house Excel workbooks · Slide 11

12.

INTERPRETATION RULES
How to read the results without over-claiming
The deck is a summary; the Excel files remain the place for raw runs, formulas and per-plan backup evidence.
1
2
3
Separate metrics
Respect statistics
Keep scope visible
Average FPS, low-frame FPS and
DPC/ISR latency do not measure the
same thing. A plan can win one
objective and lose another.
When results are statistically tied or
noisy, present them as tied or re-test
candidates rather than forcing a false
ranking.
The result is valid for this hardware, OS
build and fixed scene. Encourage
viewers to treat the Excel files as
backup evidence.
Evidence boundary
PowerPoint = verdict summary, Excel workbooks = raw evidence, formulas and detailed per-plan backup.
Alchemy Tweaks · Power plan benchmark summary
Source: Alchemy Tweaks in-house Excel workbooks · Slide 12

13.

FINAL VERDICT
Final recommendation: optimize for the bottleneck, not for the plan name
The data supports a split recommendation: one FPS winner, one latency winner and one balanced latency option.
RUN FOR FPS
RUN FOR DPC P99
RUN FOR BALANCED LATENCY
Arsenza INTEL
Kizzimo Extreme LL
amit's low latency
242.0 Avg FPS · best overall FPS verdict
115.2 µs DPC p99 · clear latency leader
171.8 µs · low overhead · consistent
Final interpretation
Power plans are not magic switches. In this dataset, the best plan depends on whether you care most about FPS throughput, low-frame
smoothness, DPC latency or interrupt overhead.
Alchemy Tweaks
Alchemy Tweaks · Power plan benchmark summary
Benchmark summary deck · FPS and latency analysis
Source: Alchemy Tweaks in-house Excel workbooks · Slide 13