Skip the Quadro RTX 5000 Deal | Generated by AI
Question: What is this Walmart product listing?
Answer:
This is a used/refurbished NVIDIA Quadro RTX 5000 (the Turing-generation workstation GPU, not a consumer GeForce card — the listing title is misleading with “GeForce”) sold for $799 by a third-party seller (SYNNEX Ltd) on Walmart.
Key specs:
- GPU: Quadro RTX 5000 (Turing, TU104 die) — same die as consumer RTX 2080
- VRAM: 16 GB GDDR6 (vs 8 GB on consumer RTX 2080)
- Interface: PCIe x16 3.0
- Outputs: 4× DisplayPort 1.4 + 1× USB-C
- TDP: 265W (needs PCIe power)
- Condition: Grade A “Like New”, 30-day return, no accessories
- Dell OEM part:
7XK5H— pulled from a Dell workstation
Is it worth it for you?
Honestly, no — at least not for ML training. Here’s the comparison:
| Card | VRAM | Tensor TFLOPS | Generation | Price |
|---|---|---|---|---|
| Quadro RTX 5000 (this) | 16 GB GDDR6 | ~57 (FP16) | Turing (2018) | $799 used |
| RTX 4070 Ti | 12 GB GDDR6X | ~285 (FP16) | Ada (2023) | ~$650 |
| RTX 3090 | 24 GB GDDR6X | ~142 (FP16) | Ampere (2020) | ~$500 used |
The Quadro RTX 5000 is a 2018-era professional card — the workstation branding gives it 16 GB VRAM (good), but the raw compute is ancient compared to modern consumer cards. Tensor core performance is ~4-5x worse than current Ada gen cards.
The only real reason to buy this would be if you specifically need 16 GB on a tight budget and can’t get a used RTX 3090 (24 GB, ~$500). For your use case — model training, nanoGPT, fine-tuning — a used RTX 3090 at ~$500 would be far superior: more VRAM, significantly faster Tensor cores, better cuDNN support maturity.
Pass on this one.