
Tesla K8
Popular choices:

Radeon HD 8990
Popular choices:
Performance Spectrum - GPU
About G3D Mark
G3D Mark is a standard benchmark that measures graphics performance in real-world gaming scenarios. It simplifies comparing cards from different brands, where higher scores directly correlate with better fps and smoother gaming experiences.
Value Upgrade Path
This is the official ChipVERSUS Value Rating, comparing raw performance (G3D Mark) per dollar. The Tesla K8 is positioned at rank #208 in our cost-efficiency ranking, representing a Lower cost-benefit for your build. Components placed above yours deliver better value for money.
Avg price is the current average price collected from markets across the web.
Performance Per Dollar Tesla K8
Performance Per Dollar
Performance Comparison
About G3D Mark🏆 Chipversus Verdict
🚀 Performance Leadership
The Radeon HD 8990 is the superior choice for raw performance. It leads with a 2% higher G3D Mark score. However, the Tesla K8 offers more VRAM, which may be beneficial for texture-heavy scenarios at higher resolutions.
| Insight | Tesla K8 | Radeon HD 8990 |
|---|---|---|
| Performance | ❌Lower raw frame rates (-2%) | ✅Leading raw performance (+2%) |
| Longevity | 🛑Obsolete Architecture (2014 / Kepler (2012−2018)) | 🛑Obsolete Architecture (2013 / GCN 1.0 (2012−2020)) |
| Ecosystem | Supports FSR Upscaling | Supports FSR Upscaling |
| VRAM | ✅ More VRAM (+33.3%) | ❌ Less VRAM capacity |
| Efficiency | 💡 Excellent Perf/Watt | ⚡ Higher Power Consumption |
| Case Fit | — | — |
💎 Value Proposition
The Tesla K8 offers a compelling cost-to-performance ratio. While both GPUs are considered legacy components by modern standards, the Tesla K8 holds the technical lead. Priced at $100 (vs $999), it costs 90% less, resulting in a 879.3% higher cost efficiency score.
| Insight | Tesla K8 | Radeon HD 8990 |
|---|---|---|
| Cost Efficiency | ✅Better overall value (+879.3%) | ❌Lower cost efficiency |
| Upfront Cost | ✅More affordable ($100) | ⚠️Higher upfront cost ($999) |
Performance Check
Real-world benchmarks and performance projections based on comprehensive hardware analysis and comparative metrics. Values represent expected performance on High/Ultra settings at 1080p, 1440p, and 4K. Modeled using a Ryzen 7 7800X3D reference profile to minimize specific CPU bottlenecks.
Note: Performance behavior can vary per game. Specific architectures may perform better or worse depending on game engine optimizations and API implementation.
Technical Specifications
Side-by-side comparison of Tesla K8 and Radeon HD 8990

Tesla K8
The Tesla K8 is manufactured by NVIDIA. It was released in September 16 2014. It features the Kepler architecture. The core clock ranges from 693 MHz to 811 MHz. It has 1536 shading units. The thermal design power (TDP) is 100W. Manufactured using 28 nm process technology. G3D Mark benchmark score: 5,111 points.

Radeon HD 8990
The Radeon HD 8990 is manufactured by AMD. It was released in April 24 2013. It features the GCN 1.0 architecture. The core clock ranges from 950 MHz to 1000 MHz. It has 2048 ×2 shading units. The thermal design power (TDP) is 375W. Manufactured using 28 nm process technology. G3D Mark benchmark score: 5,214 points. Launch price was $999.
Graphics Performance
The Tesla K8 scores 5,111 and the Radeon HD 8990 reaches 5,214 in the G3D Mark benchmark — just a 2% difference, making them near-identical in rasterization performance. The Tesla K8 is built on Kepler while the Radeon HD 8990 uses GCN 1.0, both on a 28 nm process. Shader units: 1,536 (Tesla K8) vs 2,048 (Radeon HD 8990). Raw compute: 2.491 TFLOPS (Tesla K8) vs 4.096 TFLOPS ×2 (Radeon HD 8990). Boost clocks: 811 MHz vs 1000 MHz.
| Feature | Tesla K8 | Radeon HD 8990 |
|---|---|---|
| G3D Mark Score | 5,111 | 5,214+2% |
| Architecture | Kepler | GCN 1.0 |
| Process Node | 28 nm | 28 nm |
| Shading Units | 1536 | 2048 ×2+33% |
| Compute (TFLOPS) | 2.491 TFLOPS | 4.096 TFLOPS ×2+64% |
| Boost Clock | 811 MHz | 1000 MHz+23% |
| ROPs | 32 | 32 ×2 |
| TMUs | 128 | 128 ×2 |
| L1 Cache | 128 KB | 512 KB+300% |
| L2 Cache | 512 KB | 768 KB+50% |
Advanced Features (DLSS/FSR)
| Feature | Tesla K8 | Radeon HD 8990 |
|---|---|---|
| Upscaling Tech | FSR 1.0 (Software) | FSR 1.0 (Software) |
| Frame Generation | Not Supported | Not Supported |
| Ray Reconstruction | No | No |
| Low Latency | Standard | AMD Anti-Lag |
Video Memory (VRAM)
The Tesla K8 comes with 4 GB of VRAM, while the Radeon HD 8990 has 3 GB. The Tesla K8 offers 33.3% more capacity, crucial for higher resolutions and texture-heavy games. Bus width: 128-bit vs 384-bit x2. L2 Cache: 512 KB (Tesla K8) vs 768 KB (Radeon HD 8990) — the Radeon HD 8990 has significantly larger on-die cache to reduce VRAM reliance.
| Feature | Tesla K8 | Radeon HD 8990 |
|---|---|---|
| VRAM Capacity | 4 GB+33% | 3 GB |
| Memory Type | GDDR6 | GDDR5 |
| Bus Width | 128-bit | 384-bit x2+200% |
| L2 Cache | 512 KB | 768 KB+50% |
Power & Dimensions
The Tesla K8 draws 100W versus the Radeon HD 8990's 375W — a 115.8% difference. The Tesla K8 is more power-efficient. Recommended PSU: 350W (Tesla K8) vs 750W (Radeon HD 8990). Power connectors: PCIe-powered vs 2x 8-pin.
| Feature | Tesla K8 | Radeon HD 8990 |
|---|---|---|
| TDP | 100W-73% | 375W |
| Recommended PSU | 350W-53% | 750W |
| Power Connector | PCIe-powered | 2x 8-pin |
| Perf/Watt | 51.1+268% | 13.9 |
Value Analysis
The Tesla K8 launched at $1000 MSRP and currently averages $100, while the Radeon HD 8990 launched at $999 and now averages $999. The Tesla K8 costs 90% less ($899 savings) at current market prices. Performance per dollar (G3D Mark / price): 51.1 (Tesla K8) vs 5.2 (Radeon HD 8990) — the Tesla K8 offers 882.7% better value. The Tesla K8 is the newer GPU (2014 vs 2013).
| Feature | Tesla K8 | Radeon HD 8990 |
|---|---|---|
| MSRP | $1000 | $999 |
| Avg Price (30d) | $100-90% | $999 |
| Performance per Dollar | 51.1+883% | 5.2 |
| Codename | GK104 | Malta |
| Release | September 16 2014 | April 24 2013 |
| Ranking | #515 | #412 |
Top Performing GPUs
The most powerful gpus ranked by G3D Mark benchmark scores.

















