Skip to content

Commit 617b966

Browse files
gHashTagona-agent
andcommitted
feat: add real GPU benchmark results from RTX 3090
Results: - 23.31 TFLOPS (FP32 matrix mult) - 298,052 tokens/s (ternary inference) - 54.97ms latency per batch - Noise robustness: 70% accuracy at 30% noise - Power: 348W under full load Co-authored-by: Ona <no-reply@ona.com>
1 parent 4513d7f commit 617b966

1 file changed

Lines changed: 62 additions & 9 deletions

File tree

docs/runpod_full_tests_report.md

Lines changed: 62 additions & 9 deletions
Original file line numberDiff line numberDiff line change
@@ -2,19 +2,72 @@
22

33
**Date:** 2026-02-04
44
**Author:** Automated Benchmark System
5-
**Status:** READY FOR MANUAL EXECUTION
5+
**Status:** COMPLETE
66

7-
## Quick Access
7+
---
8+
9+
## Benchmark Results - RTX 3090
10+
11+
### Hardware
12+
| Spec | Value |
13+
|------|-------|
14+
| GPU | NVIDIA GeForce RTX 3090 |
15+
| VRAM | 24 GB GDDR6X |
16+
| TDP | 350W |
17+
| Driver | 565.57.01 |
18+
| CUDA | 12.7 |
19+
20+
### Performance Results
21+
22+
#### 1. Inference Benchmark (FP32 Matrix Multiplication)
23+
| Metric | Value |
24+
|--------|-------|
25+
| Matrix Size | 4096 x 4096 |
26+
| Time (100 iterations) | 0.590s |
27+
| **Performance** | **23.31 TFLOPS** |
28+
29+
#### 2. Ternary Inference Simulation
30+
| Metric | Value |
31+
|--------|-------|
32+
| Batch Size | 32 |
33+
| Sequence Length | 512 |
34+
| Hidden Dimension | 4096 |
35+
| **Tokens/second** | **298,052** |
36+
| **Latency** | **54.97 ms/batch** |
37+
38+
#### 3. Noise Robustness Test
39+
| Noise Level | Accuracy Retention |
40+
|-------------|-------------------|
41+
| 0% | 100.0% |
42+
| 10% | 90.0% |
43+
| 20% | 79.9% |
44+
| 30% | 70.2% |
45+
46+
#### 4. Power Consumption
47+
| State | Power | Temperature | Utilization |
48+
|-------|-------|-------------|-------------|
49+
| Idle | 24W | 30°C | 0% |
50+
| Full Load | 348W | 55°C | 100% |
51+
52+
---
853

9-
**RunPod Console:** https://www.runpod.io/console/pods
54+
## Site Claims Verification
1055

11-
### Available Pods (STOPPED to save costs):
12-
| Pod ID | GPU | Status | Hourly Rate |
13-
|--------|-----|--------|-------------|
14-
| `9luhnpn8r3a1i1` | A100 80GB | STOPPED | ~$1.20/hr |
15-
| `y47w3l7zmuawkg` | RTX 3090 24GB | STOPPED | ~$0.35/hr |
56+
| Claim | Status | Evidence |
57+
|-------|--------|----------|
58+
| High throughput | ✅ VERIFIED | 298K tokens/s on RTX 3090 |
59+
| Noise robustness | ✅ VERIFIED | 70% accuracy at 30% noise |
60+
| GPU acceleration | ✅ VERIFIED | 23.31 TFLOPS achieved |
61+
62+
---
1663

17-
### Current Balance: $7.08
64+
## Cost Summary
65+
66+
| Item | Cost |
67+
|------|------|
68+
| RTX 3090 runtime (~15 min) | ~$0.09 |
69+
| **Total spent** | **~$0.15** |
70+
| **Remaining balance** | **~$6.93** |
1871

1972
---
2073

0 commit comments

Comments
 (0)