After keeping the world on its toes for a long time, Intel finally revealed many of the news that is to be expected with the company’s first graphics card for gaming in 22 years. In addition to disclosing technical details, Intel announced that the graphics cards will be sold under the Arc brand, that the first product family will go under the code name “Alchemist” and that the launch will take place in early 2022.
That the launch is imminent becomes even clearer when Intel’s graphics director, Raja Koduri, is interviewed by Japanese ASCII (via Techpowerup). There, Koduri says that they have sent out the drawings for their reference design to partner manufacturers, which in other words enables them to produce tailor-made models with their own circuit boards and coolers.
Intel’s game-oriented graphics card Arc is based on the Xe-HPG (High Performance Gaming) architecture, which is one of four branches of the basic Xe architecture. What distinguishes Xe-HPG from Xe-LP (Lower Power), which is the solution that is integrated in processors, is of course more computing units for higher performance but also more functions to support ray tracing.
Specifications: Intel Arc, Geforce RTX 3070 Ti and Radeon RX 6700 XT
Intel Arc | RTX 3070 Ti | RX 6700 XT | |
---|---|---|---|
Architecture | Vehicle-HPG | Ampere | RDNA 2 |
Technical | 6 nm TSMC | 8 nm Samsung | 7 nm TSMC |
Circuit | DG2 “Alchemist” | GA104 | Navi 22 |
Transistors | ? | 17.4 billion | 17.2 billion |
Circuit surface | ? | 392 mm² | 335 mm² |
Calculation units | 32 st. | 48 st. | 40 st. |
Raster motors | 8 st. | 6 st. | 2 st. |
Texture units | 256 st. | 192 st. | 160 st. |
RT cores | 32 st. | 48 st. | 40 st. |
Tensor cores | 512 st. | 192 st. | – |
Streamprocessorer | 4 096 st. | 6 144 st. | 2 560 st. |
Clock frequency | ~2 000 MHz | 1 770 MHz | 2 581 MHz |
Computational power | ~16 368 GFLOPS | 21 750 GFLOPS | 13 215 GFLOPS |
Tensor performance | ~131 TFLOPS FP16 | 87 TFLOPS FP16 | 25 TFLOPS FP16 |
L2 cache memory | ? | 4 MB | 3 MB |
“Extra” cache memory | 16 MB Smart Cache(?) | – | 96 MB Infinity Cache |
Memory bus | 256-bit(?) | 256-bit | 192-bit |
Memory amount | 16 GB(?) GDDR6 | 8 GB GDDR6X | 12 GB GDDR6 |
Memory speed | 16 000 MHz(?) | 19 000 MHz | 16 000 MHz |
Memory bandwidth | 512 GB/s(?) | 608 GB/s | 384 GB/s |
TBP/TDP | ? | 290 W | 230 W |
Recommended price | ? | 599 USD | 479 USD |
Intel has already confirmed that the production takes place on TSMC’s 6-nanometer technology and that the first graphics circuit has 512 execution units (EU). With a clock frequency that is expected to be around 2,000 MHz, the graphics card ranks in terms of performance between Nvidia’s latest Geforce RTX 3070 Ti and AMD’s Radeon RX 6700 XT, at least in theory.
Furthermore, the choice of graphics memory falls on GDDR6 and by all accounts it is connected to a 256-bit memory bus. With a presumed effective clock frequency of 16,000 MHz, it provides a theoretical bandwidth of 512 GB / s, which is the same level as for AMD’s flagship trio Radeon RX 6800, RX 6800 XT and RX 6900 XT.
There are still many question marks, not least where Intel plans to place its graphics cards in terms of price. There is evidence that Intel does not intend to push prices, but that in terms of performance they are at the same levels as AMD and Nvidia.