PNY Tesla M60 16GB GDDR5 workstation
Equipped with 4096 CUDA cores and 16GB of GDDR5 memory on a 256-bit bus, this dual-GPU card delivers substantial parallel processing power for compute-intensive tasks. Its active cooling and 300W TDP via a single 8-pin connector make it a straightforward drop-in upgrade for server environments with adequate airflow. This card is best for data scientists running neural network training or engineers performing complex simulations in a dedicated workstation.
Derzeit kein aktueller Händlerpreis
Überblick
The 30-Second Version
The Tesla M60 packs 16GB of GDDR5 and 4096 CUDA cores into a USD 93 package, which sounds great until you see the PassMark G3D score of 7002. That's bottom-fifth raster performance, and the 300W TDP makes it one of the least efficient cards we track. It's a decent budget AI card at 67th percentile, but a terrible gaming card at 11.7 out of 100 for 1080p.
Pros & Cons
Vorteile
- 16GB of GDDR5 VRAM is generous for AI and compute workloads 86th
- 4096 CUDA cores give it above-average AI performance (67th percentile) 67th
- Affordability ranks in the 86th percentile, so used prices are low
- PCI Express 3.0 x16 interface is still widely compatible
- USD 93 on Amazon is cheap for a 16GB CUDA card
Nachteile
- PassMark G3D score of 7002 puts raster performance in the bottom fifth
- 300W TDP with a single fan means hot, loud, and inefficient (12th percentile efficiency)
- 1080p gaming score of 11.7 out of 100 is nearly unusable
- Connectivity ranks in the 5th percentile with no modern display outputs
- Customer rating of 2 out of 5 from the single review is not encouraging
Die Fakten
Performance
The raw numbers here are a mixed bag. The 4096 CUDA cores and 16GB of GDDR5 sound beefy, but the 899 MHz clock speed is low by modern standards, and the 256-bit memory bus limits bandwidth. The PassMark G3D score of 7002 lands in the bottom fifth of all cards we track, which means even budget gaming cards from a few generations ago will outrun it in most raster workloads. The 1080p gaming score of 11.7 out of 100 is one of the weakest we've seen for a card with this much VRAM.
Where it does better is AI. The 67th percentile AI score puts it above average for machine learning tasks, and that 16GB frame buffer is genuinely useful for larger models that would choke on 8GB cards. The compute score is less encouraging at 27th percentile, so don't expect miracles in heavy FP64 or professional rendering. The thermal situation is also rough: 300W TDP with a single fan means this card runs hot and loud, and the efficiency score of 12th percentile reflects that. You're paying a lot in electricity for every unit of work.
Specifications
Full Specifications
Chipset
| GPU Chipset | Tesla M60 |
| Brand | NVIDIA |
| Class | workstation |
Memory
| VRAM | 16 GB |
| Memory Type | GDDR5 |
| Memory Bus | 256 |
Compute
| CUDA Cores | 4096 |
| Base Clock | 899 |
Performance
| PassMark G3D | 7002 |
| GPU Compute | 2530 |
Power & Cooling
| TDP | 300 |
| Power Connectors | 1x 8-pin |
| Cooler | Fan |
Connectivity
| Interface | PCI Express x16 3.0 |
| Max Resolution | 1080p Full HD |
Physical
| Weight | 0.1 kg / 0.3 lbs |
vs Competition
Stacked against the MSI GTX 1080 Ti GAMING X, this card looks rough. The 1080 Ti has a much higher PassMark score and will absolutely destroy the Tesla M60 in any gaming or raster workload, though it has less VRAM at 11GB. The ASRock Challenger B570 CL 10GO and XFX SPEEDSTER SWFT309 RX-67XLKWFD are both modern cards with far better efficiency and feature sets, and they'll run circles around the M60 in anything involving ray tracing or modern APIs. The Gigabyte Eagle GV-R66EAGLE-8GD has half the VRAM but much better raster performance. Where the M60 wins is raw memory capacity per dollar. If your workload needs 16GB and you don't care about gaming, none of these competitors can match it at this price point. But for everything else, they're better cards.
| Spec | PNY Tesla M60 | Gigabyte AORUS GV-N5080AORUS M-16GD | ASUS PRIME PRIME-RTX5070TI-O16G | MSI Suprim G5080-16SLS | ZOTAC SOLID SFF OC ZT-B50710J3-10P | Sapphire NITRO+ 11348-01-20G |
|---|---|---|---|---|---|---|
| Chipset | Tesla M60 | GeForce RTX 5080 | GeForce RTX 5070 Ti | GeForce RTX 5080 | GeForce RTX 5070 Ti | Radeon RX 9070 XT |
| Vram | 16GB GDDR5 | 16GB GDDR7 | 16GB GDDR7 | 16GB GDDR7 | 16GB GDDR7 | 16GB GDDR6 |
| Boost Clock Mhz | - | 2805 | 2527 | 2740 | 2482 | 3060 |
| Tdp W | 300 | 360 | 300 | 360 | 300 | 304 |
| Memory Bus Bits | 256 | 256 | 256 | 256 | 256 | 256 |
| Ray Tracing | - | true | true | true | true | true |
| GPU Class | workstation | consumer | consumer | consumer | consumer | consumer |
| Compare | Compare | Compare | Compare | Compare |
| Produkt | AI | Rt | Vram | Raster | Compute | Feature | Thermal | Efficiency | Connectivity | Nutzerresonanz | Affordability |
|---|---|---|---|---|---|---|---|---|---|---|---|
| PNY Tesla M60 | 66.5 | 15.5 | 51.2 | 17.3 | 27.5 | 11.1 | 16.9 | 11.9 | 5.4 | 0.8 | 85.6 |
| Gigabyte AORUS GV-N5080AORUS M-16GD Compare | 85.4 | 95.3 | 88.8 | 94.4 | 93.3 | 96.1 | 91.5 | 57.9 | 94.1 | 88.8 | 15.9 |
| ASUS PRIME PRIME-RTX5070TI-O16G Compare | 79.4 | 92.3 | 85.8 | 89 | 88.2 | 89.5 | 91.5 | 73 | 94.1 | 88.8 | 27.5 |
| MSI Suprim G5080-16SLS Compare | 85.4 | 97.6 | 88.8 | 94.4 | 93.3 | 96.1 | 100 | 57.9 | 94.1 | 73 | 19.2 |
| ZOTAC SOLID SFF OC ZT-B50710J3-10P Compare | 79.4 | 92.3 | 85.8 | 89 | 88.2 | 96.1 | 91.5 | 73 | 94.1 | 82.5 | 24.1 |
| Sapphire NITRO+ 11348-01-20G Compare | 89.5 | 84.9 | 74.1 | 75.9 | 90.8 | 96.1 | 91.5 | 46.2 | 88.8 | 88.8 | 36.2 |
Preis
Value & Pricing
At USD 93 on Amazon, this is a cheap way to get 16GB of CUDA-capable VRAM. The affordability percentile of 86 confirms it's priced well below most cards in our database. But the value story gets complicated in Europe. EUR prices range from EUR 203 to EUR 345 across four listings, which is a huge spread. If you're in Germany, Spain, France, or Italy, you'll want to shop around because paying EUR 345 for this card is a bad deal when the same thing is listed for EUR 203 elsewhere. The CAD 193 price in Canada is also reasonable, but the BRL 1,774 in Brazil is a harder sell. For the money, you're getting compute potential, not gaming performance, so make sure that's what you actually need.
Price History
Mehr erfahren
Overview
The PNY Tesla M60 is a weird one. It's a 16GB workstation card with 4096 CUDA cores that sounds impressive on paper, but the PassMark G3D score of 7002 tells a different story. That puts it in the bottom fifth for raster performance, which means this thing gets smoked by cards that cost less and draw way less power. The 300W TDP is a lot of heat for not a lot of frames.
But here's the thing: this isn't a gaming card. It's a repurposed datacenter part, and the 16GB of GDDR5 on a 256-bit bus gives it some life for AI and compute workloads. The AI score sits at 67th percentile, which is solidly above average. If you're looking for a cheap CUDA box for inference or light model work, the math changes a bit. Just don't expect it to be quiet or efficient about it.
Common Questions
Q: Can the Tesla M60 be used for gaming?
Not really. The 1080p gaming score of 11.7 out of 100 is one of the weakest in our database, and the PassMark G3D score of 7002 puts it in the bottom fifth for raster performance. The 899 MHz clock speed and lack of modern display outputs make it a poor fit for gaming.
Q: Is 16GB of VRAM enough for AI workloads?
For many inference tasks and smaller model fine-tuning jobs, yes. The 16GB of GDDR5 is the card's biggest strength, and the AI score of 67th percentile means it performs above average for machine learning compared to other cards we track. Just keep in mind the 256-bit memory bus limits bandwidth compared to newer cards.
Q: How much power does the Tesla M60 draw?
It has a 300W TDP and requires a single 8-pin supplementary power connector. That's a lot of power for the performance it delivers, and the efficiency score of 12th percentile reflects that. You'll want a decent power supply and good case airflow, especially since it uses a single fan for cooling.
Who Should Skip This
Skip this if you care about gaming, noise, or your electricity bill. The 1080p gaming score of 11.7 out of 100 is nearly unusable, and the 300W TDP with a single fan means this card will be loud and hot under load. The efficiency score of 12th percentile is a real letdown. If you need a general-purpose card or something for modern games, look at the ASRock Challenger B570 CL 10GO or the XFX SPEEDSTER SWFT309 RX-67XLKWFD instead. They're far more efficient and have actual feature sets. Also skip it if you need modern display outputs, since connectivity ranks in the 5th percentile.
Verdict
The PNY Tesla M60 is a niche buy. If you need 16GB of CUDA VRAM for AI inference or memory-hungry compute tasks and you're on a tight budget, the USD 93 price tag makes it tempting. The 67th percentile AI score backs that up. But the 300W power draw, single-fan cooler, and bottom-tier raster performance make it a poor choice for anything resembling gaming or general use. The 2 out of 5 customer rating, even with just one review, doesn't inspire confidence either. Buy it for the VRAM, not for the experience.
Usage Scores
- Gesamt24.4
- AI/LLM34.4
- Budget45.4
- Gaming 4k13.8
- Ray Tracing13
- Workstation31.6
- Gaming 1080p11.6
- Gaming 1440p13
- Content Creation21