Hi
There are differences in architecture of cpu's as new generations are released
This is a quick interesting read
https://www.nvidia.com/en-us/about-nvidia/corporate-timeline/
The best way to decide on a GPU is to first decide on what you will be using to for.
Games. Stills. Day to day. Video. Visual effects.
Once you decide on where you will be spending most of your time, look up on reviews on say "Best Nvidia cpu for visual effects"
This will narrow the field down and also get you used to some of the technical features.
You will then be left with a choice of roughly 5-8 GPU's from a manufacturer.
Then look at your current rig set up and ask:
1. Do I have a strong enough power supply (PSU)
2. Is the case large enough to accommodate a large card
3. Is there sufficient cooling
Etc
This should then narrow you down to 4-6 card options and now budget comes into play.
So you have finally decided on a card that your PSU and case can manage and it costs $150
You are all excited, you buy it and its performs like a snail that ate a sleeping tablet lol
This is when you realize that you need to overhaul your entire rig.
The more you expect out of hardware the more it will cost. It's very difficult finding a happy medium at the rate that technology is changing. Unless you are doing the same edits and using the same footage resolution for the next 5 years, you will need to upgrade.
Here's a quick tip: Theres no need to go for the big cards with 16gig VRAM and 4 fans etc etc. Leave that for the gaming fraternity.
Most software does NOT and I will repeat this, does NOT use the GPU to render.
As an example After Effects uses the GPU to render on cpu accelerated effects and those are a handful but growing.
The BIGGEST investment you can make is you CPU and RAM. Heres where the speed comes into play
Trust this assists...even a little
Mo
I agree with that. But I would also like to add that one should NOT cheap out on the GPU in a video editing system: If you go too low on the GPU, you run the risk of the GPU memory throughput falling below that of the main system RAM. And that WILL prevent the CPU and RAM from getting anywhere close to being fully utilized even if the workload requires it, which may result in a given high-end system performing no better than a 10-year-old entry-level PC even for everyday tasks.
By the way, I would not spend much if any money on a four- to five-year-old CPU platform at this point: The discussion starter's planned GPU upgrade will come very close to becoming too expensive for such an aging platform. Now, had he purchased such an upgrade 20 months ago, then that purchase might have been justified. But current prices for the GTX 1060 6 GB are equal to or higher than they were 20 months ago, which makes those cards becoming too close to being not worth the cost for their own good on such an aging CPU platform.
Therefore, if the thread starter MUST get a discrete GPU with 4 GB of video RAM (required for CC 2019 to use MPE GPU acceleration), then he should go for a GTX 1050 Ti and be done with that - at least until he can save up enough money for a newer, more powerful system with newer, up-to-date components. Don't go below that; otherwise, he will end up with a GPU with either an insufficient amount of VRAM or an outdated, no-longer-supported GPU generation (most extant 4 GB GT 730s are actually 8-year-old GT 430s with different BIOS firmware).