Video AI includes a Benchmark, which is pretty helpful to read other people’s experiences. It might not be 100% exact for the own case with other input, but at least it gives a general idea about the performance on a certain platform. There are many tests from different users, so one can also get a feeling for the precision of the test.
I think it would be a good idea to get something similar in Gigapixel as well.
I’m actually just asking myself whether to invest another $1200 to buy a TX 5090 or to go for a 5080 in my next computer - so I’d love to have some comparison.
Clusters by CPU/GPU Performance: Show the 23-subtest results aggregated by hardware clusters (e.g., “High,” “Medium,” “Low”).
Version Comparison: Track relative performance within clusters between TVAI versions for the same hardware to detect anomalous drops or gains in subtest scores by test version.
Flag Anomalies: Identify cases where, for example, a usually top-performing GPU (e.g., RTX 5090) drops unexpectedly in rank in a specific test, signaling a possible code regression or hardware-specific issue.
For End Users: Upgrade Decision Support
Cluster Mapping: Place the user’s machine into the performance cluster their actual benchmark scores warrant, based on objective results, not solely hardware configuration.
Post-Clustering Hardware Review: Once the machine is clustered by performance, examine the CPU, GPU, RAM, and VRAM versus the modal or average hardware in that cluster.
Bottleneck Identification: Indicate whether it’s the CPU, GPU, RAM, or VRAM (by deviations from median cluster scores or hardware patterns) restricting performance or prompting further investigation.
Upgrade Pathways: Recommend a cost-effective upgrade path (e.g., increase RAM, swap GPU, replace both CPU and GPU) based on their position and desired target cluster, noting component-specific fixes when hardware lags behind typical cluster standards.
End User Report: Cluster Placement and Hardware Fit
User
Cluster
CPU
GPU
RAM (GB)
VRAM (GB)
Artemis 1X
Rhea 4X
RAM vs Median
VRAM vs Median
Is Anomaly
Median CPU
Median GPU
A
Low
i7-8700K
RTX 3060
64
12
9.47
1.15
-32
-2
No
i7-8700K
RTX 3060
B
High
i9-14900KF
RTX 4090
64
24
37.10
4.84
+16
+4
Yes
Ryzen 9 7950X3D
RTX 4090
C
High
Ryzen 9 7950X3D
RTX 5080
32
16
39.65
4.75
-16
-4
Yes
Ryzen 9 7950X3D
RTX 4090
Example Use for End Users
Given “Machine A” in Low: Their machine fits the cluster, upgrading RAM/VRAM alone would not move them up a cluster; new hardware (CPU/GPU) would be required.
For “Machine B” and “Machine C” in High: Their performance matches the cluster, but RAM or GPU differences from the median may suggest possible performance tuning or upgrades could yield further gains.
