CPU

Builds on Common Methodology

Measures integer compute speed with 7-Zip's built-in compression and decompression benchmark, on one thread and on every vCPU.

Agents that compile code, run test suites, compress build artifacts, or process data spend much of their runtime bound on integer CPU execution. Two providers selling a 2 vCPU / 4 GiB sandbox can deliver very different compute speeds depending on the underlying host processor generation, clock frequency, cache hierarchy, and hypervisor scheduling.

The CPU tab measures integer compute speed using 7-Zip's built-in LZMA benchmark in two configurations: 7-Zip, 1 Thread (single-core speed) and 7-Zip, All vCPUs (multi-core scaling across the requested vCPUs).

View the Procedure View CPU Results Download Raw Data View the Parameters

Metrics

7-Zip, 1 Thread

7-Zip is the combined compression and decompression Tot rating of single-core integer compute speed on 7-Zip's internal LZMA benchmark scale, reported on the Avr summary line of ./7zzs b -mmt1.

• 7-Zip, 1 Thread: Total Rating
• 7-Zip, 1 Thread: Compression Rating
• 7-Zip, 1 Thread: Decompression Rating

7-Zip, All vCPUs

7-Zip, All vCPUs is the same measurement as 7-Zip, 1 Thread, run with one thread per vCPU (2 threads).

• 7-Zip, All vCPUs: Total Rating
• 7-Zip, All vCPUs: Compression Rating
• 7-Zip, All vCPUs: Decompression Rating

Timed Part	Starts	Ends
7-Zip, 1 Thread	7-Zip's benchmark begins.	7zzs b -mmt1 exits; its rating is 7-Zip's own total MIPS over compress and decompress, by its own clock.
7-Zip, All vCPUs	7-Zip's benchmark begins.	7zzs b, run with one thread per requested vCPU, exits; its rating is 7-Zip's own total MIPS over compress and decompress, by its own clock.

R_{\mathrm{total},\,i} = \dfrac{R_{\mathrm{compress},\,i} + R_{\mathrm{decompress},\,i}}{2}

• R_{\mathrm{total},\,i} Total rating in MIPS of trial i: the mean of its compression and decompression ratings.
• R_{\mathrm{compress},\,i} Compression rating in MIPS of trial i.
• R_{\mathrm{decompress},\,i} Decompression rating in MIPS of trial i.

7-Zip runs its built-in LZMA dictionary compression and decompression passes in memory and prints Avr: ratings for compression (CompMIPS), decompression (DecompMIPS), and their combined total (Tot:), which we record directly as the total rating. With -mmt1 it runs on a single thread; with -mmt2 it spawns 2 worker threads, sized to the requested vCPU count rather than nproc.

Measurement

Each trial uses a sandbox created from BoxRouter's pinned standard image, running Igor Pavlov's official static Linux x86_64 build of 7-Zip 25.01 (7zzs, verified by SHA-256).

1. Create sandbox and run readiness check.
2. Record CPU model, flags, cache size, and cgroup limits, then upload and verify static 7zzs and stream binaries.
3. 7-Zip, 1 Thread: Total Rating Run ./7zzs b -mmt1 and ./7zzs b -mmt2 (sized to the requested vCPUs, never nproc) as polled background scripts in shuffled order alongside the memory test. Timed
4. Destroy sandbox.

The built-in 7zzs b benchmark compresses and decompresses synthetic data in memory with LZMA, exercising integer arithmetic, branching, lookup tables, and cache speed without disk or network I/O. Flag -mmt1 isolates single-vCPU throughput in 7-Zip, 1 Thread, while -mmt2 drives both requested vCPUs in 7-Zip, All vCPUs to measure multi-core scaling, CPU quota throttling, and hyperthread contention.

The hardware probe records visible CPU count (nproc), processor model name, instruction flags (SHA-NI, AVX2, AVX-512), L3 cache capacity, cgroup CPU quota (cpu.max), and available memory.

Scope Limits and Caveats

• Evaluates integer compression, branching, and cache lookup operations rather than floating-point or SIMD vector math; sandboxes on a single provider can land on different host CPU models across trials.
• The workload exercises no floating-point or wide SIMD instructions, though every trial records whether the sha_ni, avx2 and avx512f flags are present.
• Sandboxes on one provider can land on different CPU models, and each trial records the processor model, so a bimodal spread or a low-end tail can be traced to the models across the 40 trials.
• When a provider exposes a higher host core count through nproc, the CPU-count check raises an informational note while the multi-thread test still runs 2 threads.

Failures and Exclusions

Outcome classifications, top-up rounds, and outlier rules follow Common Methodology; the first failed step ends the trial without in-trial retries.

A trial succeeds only when the 7zzs binary SHA-256 matches the pinned SHA-256, both 7zzs b runs exit 0 with a positive Avr: MIPS rating, the guest clock passes the twenty-millisecond /proc/uptime gate, and the shared sandbox is confirmed destroyed.

Step	Time Limit
Create	5 min
CPU facts	30 s
7-Zip, 1 Thread	5 min
Destroy	5 min

Statistics

Based On

Common Methodology

Per-unit percentiles, bootstrap confidence intervals, and tail rules follow Common Methodology.

A provider is ranked once it has at least 5 successful trials.

Sandbox and Trials

Image Digest

sha256

Image Version

0.3.11

Image Registry Page

View on the Registry

Pull the Image

docker pull ghcr.io/boxrouter/standard-image@sha256:89d4c97048dbdfb8b6d85ad91833962ff38d1c334a5f47e6de80a9762310191b Copy

Lifetime

20.0 min

Idle Connection Time

5.0 min

From Common Methodology

Idle Connection Exceptions

Provider	Rule	File Transfer Idle Time
Daytona	The SDK has no idle-timeout setting. Its command channel keeps its connection open, and its file-transfer channel reconnects after 5 s of idle time.	5.0 s

From Common Methodology

Network Path Attempts

10

From Common Methodology

Mode

Single

Sandboxes per Trial

1

Parallel Trials

10

Concurrent Sandboxes

1

At a Glance

Metrics

• 7-Zip, 1 Thread: Total Rating
• 7-Zip, 1 Thread: Compression Rating
• 7-Zip, 1 Thread: Decompression Rating
• 7-Zip, All vCPUs: Total Rating
• 7-Zip, All vCPUs: Compression Rating
• 7-Zip, All vCPUs: Decompression Rating

Quantity

Rating

Unit

Million Instructions per Second (MIPS)

Direction

Higher Is Better

Statistic

Median

Clock

Sandbox clock

From Common Methodology

Machine Size

2 vCPU / 4 GiB

Trials

40