NVIDIA clarifies RTX PRO 5000 Blackwell specs after 176 ROP confusion
NVIDIA has confirmed that the RTX PRO 5000 Blackwell ships with 160 ROPs, not 176 as some databases and third-party listings had suggested. The confusion started after a Reddit user reported that his 48 GB card was showing 160 ROPs in GPU-Z and other utilities, while older spec pages elsewhere listed 176. NVIDIA’s own product page and datasheet do not publish an official ROP figure, which likely helped the wrong number spread.
That also explains why we never covered the original “missing ROPs” claim. The earlier reports were based on unconfirmed spec entries to begin with (from TPU), not on an NVIDIA specification. Still, it is useful that NVIDIA has now cleared it up, because the Blackwell launch cycle already had real ROP-related issues on some GeForce cards and this workstation part did not need more confusion around it.
NVIDIA has confirmed for TechPowerUP that this is an error and that the card officially comes with 160 ROPs, as detected by our GPU-Z software.
TechPowerUP
RTX PRO 5000 Blackwell , Source: NVIDIA
According to NVIDIA’s official materials, the RTX PRO 5000 Blackwell uses the Blackwell architecture with 14,080 CUDA cores, 48 GB or 72 GB of GDDR7 ECC memory, a 512-bit memory interface, 1,344 GB/s of bandwidth, PCIe 5.0 x16, and a 300 W total board power. Those are the published specs, and none of them include a 176 ROP figure.
The card offers a 10 GPC (Graphics Processing Cluster) configuration, which lines up with a 160 ROP count. In other words, the tools were likely reporting the card correctly all along, while the expected 176 figure came from a bad assumption about how the RTX PRO 5000 was cut down from GB202.
Source: TechPowerUP
