There are a lot of posts out there and “ultimate guides” giving full spec breakdowns on new and old graphic cards, but a lot of people are just looking for a quick side by side comparison of the “new vs old” graphics cards. Newegg lets you do this, but only 5 at a time. Wikipedia kind of does this, but you can’t compare different series.
In this post I’m only going to compare CUDA cores, memory, MSRP, mem type, base clock speed, and power. I realize there is a whole wealth of other specifications that really matter when making a purchase, but these are always the first thing I look at when weighing a new graphics card.
Last updated August 2019 / September 2020 May 2025. If I made any mistakes or you think I should add anything tweet me at @pmkoom.
| Card | CUDA Cores | Memory | MSRP | Mem Type | Clock | Power |
|---|---|---|---|---|---|---|
| GTX 1060 | 1280 | 6GB | $249 | GDDR5 | 1506 MHz | 120 watts |
| GTX 1070 | 1920 | 8GB | $379 | GDDR5 | 1506 MHz | 150 watts |
| GTX 1070 Ti | 2432 | 8GB | $449 | GDDR5 | 1607 MHz | 180 watts |
| GTX 1080 | 2560 | 8GB | $499 | GDDR5 | 1607 MHz | 180 watts |
| GTX 1080 Ti | 3584 | 11GB | $669 | GDDR5 | 1480 MHz | 250 watts |
| GTX 1650 | 896 | 4GB | $179 | GDDR6 | 1485 MHz | 75 watts |
| GTX 1660 | 1408 | 6GB | $229 | GDDR6 | 1785 MHz | 120 watts |
| GTX 1660 Ti | 1536 | 6GB | $349 | GDDR6 | 1770 MHz | 120 watts |
| RTX 2060 | 1920 | 6GB | $349 | GDDR6 | 1365 MHz | 160 watts |
| RTX 2060 Super | 2176 | 6GB | $399 | GDDR6 | 1470 MHz | 175 watts |
| RTX 2070 | 2304 | 8GB | $499 | GDDR6 | 1410 MHz | 175 watts |
| RTX 2070 Super | 2560 | 8GB | $499 | GDDR6 | 1605 MHz | 215 watts |
| RTX 2080 | 2944 | 8GB | $699 | GDDR6 | 1650 MHz | 225 watts |
| RTX 2080 Super | 3072 | 8GB | $699 | GDDR6 | 1650 MHz | 250 watts |
| RTX 2080 Ti | 4352 | 11GB | $999 | GDDR6 | 1350 MHz | 260 watts |
| RTX 3050 | 2560 | 8GB | $249 | GDDR6 | 1552 MHz | 130 watts |
| RTX 3060 | 3584 | 12GB | $349 | GDDR6 | 1320 MHz | 180 watts |
| RTX 3060 Ti | 4864 | 8GB | $399 | GDDR6 | 1600 MHz | 200 watts |
| RTX 3070 | 5888 | 8GB | $499 | GDDR6 | 1500 MHz | 220 watts |
| RTX 3070 Ti | 6144 | 8GB | $599 | GDDR6X | 1575 MHz | 290 watts |
| RTX 3080 | 8704 | 10GB | $699 | GDDR6X | 1440 MHz | 320 watts |
| RTX 3080 Ti | 10240 | 12GB | $1199 | GDDR6X | 1365 MHz | 350 watts |
| RTX 3090 | 10496 | 24GB | $1499 | GDDR6X | 1400 MHz | 350 watts |
| RTX 3090 Ti | 10752 | 24GB | $1999 | GDDR6X | 1560 MHz | 450 watts |
| RTX 4050 | 3072 | 6GB | $199 | GDDR6 | 2460 MHz | 110 watts |
| RTX 4060 | 3072 | 8GB | $299 | GDDR6 | 2460 MHz | 115 watts |
| RTX 4060 Ti | 4352 | 8GB | $399 | GDDR6 | 2310 MHz | 160 watts |
| RTX 4060 Ti 16GB | 4352 | 16GB | $499 | GDDR6 | 2310 MHz | 160 watts |
| RTX 4070 | 5888 | 12GB | $549 | GDDR6X | 1920 MHz | 200 watts |
| RTX 4070 Super | 7168 | 12GB | $599 | GDDR6X | 1980 MHz | 220 watts |
| RTX 4070 Ti | 7680 | 12GB | $799 | GDDR6X | 2310 MHz | 285 watts |
| RTX 4070 Ti Super | 8448 | 16GB | $799 | GDDR6X | 2340 MHz | 285 watts |
| RTX 4080 | 9728 | 16GB | $1199 | GDDR6X | 2205 MHz | 320 watts |
| RTX 4080 Super | 10240 | 16GB | $999 | GDDR6X | 2295 MHz | 320 watts |
| RTX 4090 | 16384 | 24GB | $1599 | GDDR6X | 2235 MHz | 450 watts |
| RTX 5060 | 4352 | 8GB | $399 | GDDR7 | 2000 MHz | 170 watts |
| RTX 5060 Ti | 5632 | 12GB | $499 | GDDR7 | 2100 MHz | 200 watts |
| RTX 5070 | 6144 | 12GB | $549 | GDDR7 | 2230 MHz | 220 watts |
| RTX 5070 Ti | 7680 | 16GB | $749 | GDDR7 | 2300 MHz | 260 watts |
| RTX 5080 | 10752 | 16GB | $999 | GDDR7 | 2350 MHz | 360 watts |
| RTX 5090 | 21760 | 32GB | $1999 | GDDR7 | 2300 MHz | 575 watts |
The 20 series uses “Ray Tracing” which was largely debated when it was first released it is geared towards gaming graphics.
The 16 series uses the Turing architecture. They marketed the 16 series towards live streamers.
Is there anything that I’m missing that absolutely needs to be added? Memory Clock? Ports? If so, let me know! Thanks for checking this out!
