“4090 efficiency at $549.”
That’s what Nvidia CEO Jensen Huang said of the GeForce RTX 5070 when he introduced the cardboard at CES in January. Because of AI, this new midrange GPU would be capable of match the body charges of what had been the quickest shopper GPU that had beforehand existed for round one-third the worth.
Let’s dispel that notion up entrance. No, the GeForce RTX 5070 isn’t as quick as an RTX 4090, not with out some very artistic evaluating of non-comparable numbers. Per standard for the 50-series, Nvidia is leaning on its AI-generated interpolated frames for the majority of its claimed efficiency enhancements. When it comes to precise rendering velocity, the 5070 isn’t whilst quick as a 4080 or a 4070 Ti. It’s barely quicker than final 12 months’s 4070 Tremendous, and it has disproportionately larger energy utilization.
Andrew Cunningham
The RTX 5070 Founders Version is so much smaller than the 5090/5080 design.
Andrew Cunningham
Andrew Cunningham
Just like the 5090/5080, the 5070 switches to a barely angled 12-pin energy connector.
Andrew Cunningham
The RTX 5070 Founders Version is so much smaller than the 5090/5080 design.
Andrew Cunningham
Just like the 5090/5080, the 5070 switches to a barely angled 12-pin energy connector.
Andrew Cunningham
For all that, it is nonetheless not essentially a unhealthy card—it is slightly quicker than the $599 4070 Tremendous on the similar $549 value that Nvidia used for the RTX 4070. It nonetheless represents Nvidia’s minimal viable 4K GPU, when you think about DLSS upscaling, and it’s nicely suited to individuals with 1440p screens.
But it surely’s a tough card to get enthusiastic about, and in contrast to each different 50-series GPU launched to date, it really has to compete towards one thing—AMD’s new Radeon RX 9070 sequence. I can not speak intimately about how these playing cards all stack up till later this week, however suffice it to say that Nvidia may use one thing a bit extra spectacular than the RTX 5070 at this value.
GeForce RTX 5070 specs
| RTX 5070 Ti | RTX 4070 Ti Tremendous | RTX 5070 | RTX 4070 Tremendous | RTX 4070 | |
|---|---|---|---|---|---|
| CUDA Cores | 8,960 | 8,448 | 6,144 | 7,168 | 5,888 |
| Increase Clock | 2,452 MHz | 2,610 MHz | 2,512 MHz | 2,475 MHz | 2,475 MHz |
| Reminiscence Bus Width | 256-bit | 256-bit | 192-bit | 192-bit | 192-bit |
| Reminiscence Bandwidth | 896 GB/s | 672 GB/s | 672 GB/s | 504 GB/s | 504 GB/s |
| Reminiscence dimension | 16GB GDDR7 | 16GB GDDR6X | 12GB GDDR7 | 12GB GDDR6X | 12GB GDDR6X |
| TGP | 300 W | 285 W | 250 W | 220 W | 200 W |
The 5070’s CUDA core depend falls proper in between the RTX 4070 and the 4070 Tremendous’s, the place the 5080 and 5070 Ti each included small will increase from their previous-generation counterparts. To spice up speeds over the 4070 Tremendous, this implies Nvidia is leaning on architectural enhancements in Blackwell, the reminiscence bandwidth enhance from GDDR7 (a reasonably important 33 p.c enhance), and elevated GPU clock speeds (the reminiscence change and clock velocity boosts probably clarify the upper energy consumption).