Very poor benchmarks for 4060, why?

Why are my benchmarks so bad ? I get half the speed of people in the benchmark thread of those with a 3060TI which should be fairly similar in to speed to a 4060. Below are my results…

Topaz Video AI  v3.5.2
System Information
OS: Windows v10.22
CPU: AMD Ryzen 7 5700X 8-Core Processor               31.927 GB
GPU: NVIDIA GeForce RTX 4060  7.7842 GB
Processing Settings
device: 0 vram: 1 instances: 0
Input Resolution: 1920x1080
Benchmark Results
Artemis		1X: 	09.69 fps 	2X: 	06.35 fps 	4X: 	01.97 fps 	
Iris		1X: 	10.77 fps 	2X: 	05.28 fps 	4X: 	01.56 fps 	
Proteus		1X: 	08.76 fps 	2X: 	05.88 fps 	4X: 	02.01 fps 	
Gaia		1X: 	03.26 fps 	2X: 	02.20 fps 	4X: 	01.47 fps 	
Nyx		1X: 	04.03 fps 	
4X Slowmo		Apollo: 	13.00 fps 	APFast: 	42.09 fps 	Chronos: 	07.80 fps 	CHFast: 	11.92 fps 	

Memory bus is only 128Bit and it only uses x8 pcie lanes.

It’s a dreadful card.

1 Like

The 4060 does suck but 8 pcie lanes has nothing to do with it. They are PCIE gen4 lanes, it’s as fast as gen3x16 and since he’s running a 5700x he has a gen4 CPU and mobo. PCIE bandwidth is not high on TVAI I’m seeing 3GiB/s max in each direction running 2 720p->1080p upscales right now with a 4070.

Memory bandwidth is also not the issue, GDDR6 is plenty fast enough @ 128bit bus width for TVAI. The card just doesn’t have enough of a GPU core to be any good. The 4060ti runs about 50% faster than this guys 4060 (there is a benchmark on the forum). The 4060ti has the same 8xgen4 PCIE lanes, the same 128bit memory bus but it has almost 50% more CUDA cores and a faster clock speed.

My last TVAI setup was 2 1080ti’s, my motherboard had a bios bug that would run the 2nd PCIE slot with only 4 lanes even though it was a x8 electrical slot. I never even bothered updating the bios because that card ran TVAI just as fast as the card in the x16 slot. An updated bios fixed it recently.

1 Like