AMD today released its Radeon RX 7600 graphics card, and you'll see a rare overlap of launch-day reviews between AMD's RX 7600, and NVIDIA's RTX 4060 Ti on the same day. Both cards go on sale today. The RX 7600 is a mid-range offering from AMD's latest RDNA 3 family, and sees the company jump from the enthusiast-segment RX 7900 series straight to the most hotly contested market segment, the $300-ish mid-range, where the RX 7600 is expected to square off against NVIDIA's RTX 4060 series. The new RX 7600 is designed for AAA gaming at 1080p, with high to maxed-out details, you can take advantage of features such as FSR to further dial up display resolutions, and effectively game at 1440p.
The Radeon RX 7600 in this review is based on the same RDNA 3 graphics architecture that also powers the RX 7900 series, but understandably scaled down. What's unexpected, though, is that unlike the RX 7900 series, which use 5 nm GCDs, the Navi 33 silicon at the heart of the RX 7600 is a monolithic 6 nm chip, an entire generation older than 5 nm due to its DUV lithography, compared to the more advanced EUV. AMD probably calculates that for the target power and performance/Watt of the RX 7600, the 6 nm node fits the bill, and is cost-effective for the company, giving it headroom for future price-wars against NVIDIA in this segment.
Since it's based on the latest RDNA 3 graphics architecture, the RX 7600 comes with advanced Dual Issue-rate Compute Units with over 17% IPC improvement over the previous RDNA 2 CUs, second generation Ray Accelerators with a claimed 50% increase in ray intersection performance; and for the first time on an AMD GPU, hardware acceleration for AI in the form of two AI Accelerator units per CU. RDNA 3 also introduces hardware-accelerated AV1 video encoding, and the new Radiance Display Engine, with support for the latest DisplayPort 2.1 and HDMI 2.1a ports, and advanced 12-bpc color formats.
AMD intends for the RX 7600 to be seen as a successor to the RX 6600, and not the RX 6600 XT or RX 6650 XT, and so although the Navi 33 silicon has numerically the same number of shaders as the Navi 23 powering the RX 6600 series, there is a numerical increase in shaders for the RX 7600 over the RX 6600, because it maxes out the chip. The Navi 33 features two shader engines, and 32 RDNA 3 compute units, which work out to 2,048 stream processors, 64 AI Accelerators, 32 Ray Accelerators, 128 TMUs, and 64 ROPs. The memory sub-system sees minor updates—the same 8 GB GDDR6 over a 128-bit memory interface, but clocked faster at 18 Gbps, and backed by a faster Infinity Cache memory.
AMD has a reference-design "Made by AMD" (MBA) graphics card design for the Radeon RX 7600, which it intends to sell directly on the AMD website, as well as through its board partners, with minimal re-branding. The company is setting $269 as the baseline MSRP for this card, with board partners expected to come out with overclocked premium non-reference designs.
Radeon RX 7600 Market Segment Analysis
Price
Cores
ROPs
Core Clock
Boost Clock
Memory Clock
GPU
Transistors
Memory
RX 5500 XT
$170
1408
32
1717 MHz
1845 MHz
1750 MHz
Navi 14
6400M
4 GB, GDDR6, 128-bit
RX 5600 XT
$190
2304
64
1375 MHz
1560 MHz
1500 MHz
Navi 10
10300M
6 GB, GDDR6, 192-bit
RX 6500 XT
$150
1024
32
2685 MHz
2825 MHz
2248 MHz
Navi 24
5400M
4 GB, GDDR6, 64-bit
RTX 2060
$180
1920
48
1365 MHz
1680 MHz
1750 MHz
TU106
10800M
6 GB, GDDR6, 192-bit
RX Vega 64
$320
4096
64
1247 MHz
1546 MHz
953 MHz
Vega 10
12500M
8 GB, HBM2, 2048-bit
RX 5700 XT
$180
2560
64
1605 MHz
1755 MHz
1750 MHz
Navi 10
10300M
8 GB, GDDR6, 256-bit
RTX 3050
$260
2560
32
1552 MHz
1777 MHz
1750 MHz
GA106
12000M
8 GB, GDDR6, 128-bit
RTX 2070
$230
2304
64
1410 MHz
1620 MHz
1750 MHz
TU106
10800M
8 GB, GDDR6, 256-bit
Arc A750
$250
3584
112
2050 MHz
N/A
2000 MHz
ACM-G10
21700M
8 GB, GDDR6, 256-bit
RX 6600
$210
1792
64
2044 MHz
2491 MHz
1750 MHz
Navi 23
11060M
8 GB, GDDR6, 128-bit
RX 6600 XT
$250
2048
64
2359 MHz
2589 MHz
2000 MHz
Navi 23
11060M
8 GB, GDDR6, 128-bit
RTX 3060
$300
3584
48
1320 MHz
1777 MHz
1875 MHz
GA106
12000M
12 GB, GDDR6, 192-bit
RTX 4060
$300
3072
32
1830 MHz
2460 MHz
2125 MHz
AD107
unknown
8 GB, GDDR6, 128-bit
RX 7600
$270
2048
64
2250 MHz
2625 MHz
2250 MHz
Navi 33
13300M
8 GB, GDDR6, 128-bit
Arc A770
$290
4096
128
2100 MHz
N/A
2187 MHz
ACM-G10
21700M
16 GB, GDDR6, 256-bit
RTX 2080
$260
2944
64
1515 MHz
1710 MHz
1750 MHz
TU104
13600M
8 GB, GDDR6, 256-bit
RTX 3060 Ti
$320
4864
80
1410 MHz
1665 MHz
1750 MHz
GA104
17400M
8 GB, GDDR6, 256-bit
RTX 4060 Ti
$400
4352
48
2310 MHz
2535 MHz
2250 MHz
AD106
22900M
8 GB, GDDR6, 128-bit
RX 6700 XT
$320
2560
64
2424 MHz
2581 MHz
2000 MHz
Navi 22
17200M
12 GB, GDDR6, 192-bit
RTX 2080 Ti
$400
4352
88
1350 MHz
1545 MHz
1750 MHz
TU102
18600M
11 GB, GDDR6, 352-bit
RTX 3070
$350
5888
96
1500 MHz
1725 MHz
1750 MHz
GA104
17400M
8 GB, GDDR6, 256-bit
RTX 3070 Ti
$420
6144
96
1575 MHz
1770 MHz
1188 MHz
GA104
17400M
8 GB, GDDR6X, 256-bit
RX 6800
$470
3840
96
1815 MHz
2105 MHz
2000 MHz
Navi 21
26800M
16 GB, GDDR6, 256-bit
Architecture
AMD Radeon RX 7600 is based on the Navi 33 silicon that the company chose to built on the older 6 nm (DUV) foundry node that significantly lower transistor density than the 5 nm EUV node that the Navi 31 powering the RX 7900 series is based on. The chip has a die-area of 204 mm², and transistor count of 13.3 billion (for reference, NVIDIA is able to cram 22.9 billion transistors into a 190 mm² silicon using 5 nm, with the AD106 powering the RTX 4060 Ti). The GPU has essentially the same component hierarchy as the previous-generation Navi 23. Its host interface is PCI-Express 4.0 x8, and power is drawn from a single 8-pin PCIe power connector, which is sufficient, given the gaming total board power for the RX 7600 of 169 W. The GPU features a 128-bit wide GDDR6 memory interface, and 8 GB is the standard memory size for the RX 7600.
AMD hasn't put out a block diagram of the Navi 33, but it features two Shader Engines (compared to six on the Navi 31). Each of these has 8 RDNA 3 Dual-Compute Units (16 CU), which share Raster machinery, and Render Backends. The GPU features Centralized Geometry Processors, asynchronous compute engines, and geometry processors shared among the two Shader Engines. Given its CU count of 32, we arrive at 2,048 stream processors, 32 Ray Accelerators, 64 AI Accelerators, 128 TMUs, and 64 ROPs. The GPU uses a second generation 32 MB Infinity Cache memory that cushions data access to the memory.
Much of the architectural innovation is this generation is with the RDNA 3 Dual-Compute Unit (or Compute Unit pair). The "Navi 33" GPU physically features 32 compute units spread across two Shader Engines. AMD claims that at the same engine clocks, the RDNA 3 CU offers a 17.4% IPC increase over the RDNA 2 CU.
The new RDNA 3 CU introduces multi-precision capability for the 64 stream processors per CU: operating either as 1x SIMD64 or 2x SIMD32 units. The Vector Unit that houses these SIMD units can either function as a SIMD execution mechanism, or as a Matrix execution unit, thanks to the new AI Matrix Accelerator, which provides a 2.7x matrix multiplication performance uplift versus conventional SIMD execution. Also added are support for the Bfloat16 instruction-set, and SIMD8 execution. The GPU hence enjoys AI hardware-acceleration that can be leveraged in future feature-additions relevant to gamers, such as FSR 3.0. Game developers will also look for ways to exploit accelerated AI, now that all three brands feature it (NVIDIA Tensor cores and Intel XMX cores).
AMD's first-generation Ray Accelerator, introduced with the RDNA 2 architecture, was the result of a hasty effort to catch up to NVIDIA with a DirectX 12 Ultimate GPU, where they developed a fixed-function hardware to calculate ray intersections, and offloaded a large chunk of RT processing to the generationally-doubled SIMD resources. With RDNA 3, they've refined the Ray Accelerator to achieve an 80% ray tracing performance uplift over the previous generation, when you add up the Ray Accelerator count, their higher engine clocks, and other hardware-level optimizations, such as early sub-tree culling, specialized box sorting modes, and reduced traversal iterations.
There is a 50% ray intersection capacity improvement for RDNA 3 thanks to these optimizations, and cycles-per-ray reduction. Besides these, AMD has also made several improvements to the geometry- and pixel-pipes, with the introduction of the new multi-draw indirect accelerator (MDIA), which reduces CPU API and driver-level overheads by gathering and parsing of multi-draw command data. At the hardware-level 12 primitives per clock is now supported compared to 8 per clock on RDNA 2, thanks to culling. The core-configuration overall enables 50% more rasterized performance per clock.
AMD has significantly improved the Display Engine of "Navi 33" over the previous-generation in terms of connectivity. The new Radiance Display Engine comes with native support for DisplayPort 2.1, which enables 8K output at up to 165 Hz refresh-rate, or 4K at up to 480 Hz, with a single cable. AMD has refined its FSR 2 algorithm to support 8K (i.e. render at a lower resolution with FSR-enhanced upscaling), to make it possible to enjoy the latest AAA titles at playable frame-rates on 8K displays. The RX 7600 gets two full-size DP 2.1 connectors, besides an HDMI 2.1b, and a USB-C with DP 1.2 passthrough. The "Navi 33" silicon receives full hardware-accelerated AV1 encode and decode capabilities. With this generation, AMD is also introducing SmartAccess Video, a feature that lets the AMD driver leverage the hardware encoders of the RDNA 2 iGPU of Ryzen 7000 desktop processors, for additional encoding performance.
Packaging
The Card
The Radeon RX 7600 follows the design theme established by the company's previous Radeon RX 7000 cards. The main color is black with a few small red highlights. The three red painted cooler fins stand for "3" in RDNA 3—a nice touch. This cooler is all-metal, both the main heatsink and the backplate.
Dimensions of the card are 20.5 x 11.5 cm, and it weighs 758 g.
Installation requires two slots in your system.
Display connectivity includes two standard DisplayPort 2.1 ports (RDNA 2 had 1.4a) and two HDMI 2.1a (same as RDNA 2).
AMD has upgraded their encode/decode setup. It now comes with two independent hardware units that can encode and decode two streams of video in parallel, or one stream at double the FPS rate. There's support for VP9, H.264, H.265 and AV1 decode, and encoding is supported for H.264, H.265 and AV1.
The card uses a single 8-pin power input, rated for 150 W maximum power, plus 75 W over the PCIe slot.
8-Pin won't Connect
When I first installed the RX 7600 I noticed that plugging the power cable in didn't feel 100% right (easy after doing it 10,000+ times in my life). After taking a second look I realized that the PCIe 8-pin didn't go all the way in. It took me a few minutes to figure out that the problem is that the backplate is expanding too far to the PCIe connector area.
Due the way some 6+2 pins are designed they won't fit, because there's a plastic lip that makes contact with the backplate, so the connection can't be made properly. With some other cables in my labs there's no issue—these don't have the raised lip. Roughly 20% of the cables I have here are affected. This is clearly an AMD issue—I've never encountered it on any other card before.
One solution is to remove the backplate and run the card without it, now the plug goes in all the way. Another option is to buy an 8-pin PCIe extension cable without the lip, or try to cut it off from your connector.
Teardown
AMD's thermal solution uses a copper baseplate and two heatpipes to keep the card cool. The main cooler also provides cooling for the memory chips and VRM circuitry.
The backplate is made of metal and protects the card against damage during installation and handling.