NVIDIA’s new RTX 6000 has been tested in 3DMark
Reddit has a thread featuring NVIDIA’s recently launched RTX 6000 ADA graphics card. This is thus far the most powerful variant of AD102 GPU on the market.
The RTX 6000 ADA is not a gaming card, though. NVIDIA has a separate line of workstation models designed specifically for that purpose. For this reason, this card actually has significantly lower TDP (300W) compared to RTX 4090 (450W).
NVIDIA RTX 6000 ADA, Source: Healthy-Blood-54
Redditor claims that the card runs 85°C with fans at around 70%, it is not louder than RTX A6000 (the predecessor). Although it may seem high, one must remember that this is a blower-type cooler and these cards are designed to work under heavy workloads for a prolonged period.
As a reminder, RTX A6000 ADA has 18176 CUDA cores, so 1792 cores or 10% more than RTX 4090. Higher core count and lower TDP are not the only differences. This workstation card is equipped with 48GB of GDDR6 memory clocked at 20 Gbps, so the effective bandwidth is just 48 GB/s lower than RTX 4090.
NVIDIA RTX 6000 ADA, Source: Healthy-Blood-54
The only screenshot that was posted shows the card scoring 30518 points in 3DMark TimeSpy Graphics, which is 15% slower than RTX 4090. However, it could reportedly hit 36K points before a Windows update. That said, the 3DMark performance is currently all over the place, and we do not know if the user was using the latest drivers.
The 6000 ADA should at least match the RTX 4090 benchmarks, but given how the TDP is limited, it may indeed have trouble being faster in more demanding workloads. With rumors about potential RTX 4090 Ti featuring just as many cores, this may explain why this upcoming flagship gaming GPU may require close to 600W.
Source: Reddit via @harukaze5719
