Launching alongside the RX 9070 XT, today we also have a full review of the slightly cheaper RX 9070 (non-XT). This cut-down Navi 48 GPU targets a $549 MSRP and, as such, is going head-to-head with Nvidia's RTX 5070 which we reviewed only yesterday. But with just a $50 gap between the RX 9070 and the XT model, does AMD's own product positioning make sense? All that and more is explored in this analysis.
Timestamps:
00:00 Intro
00:41 Recap and spec overview
01:48 Test setup
03:02 Alan Wake 2
03:41 Black Myth: Wukong
04:11 Cyberpunk 2077
04:14 Final Fantasy XVI
05:14 Forza Horizon 5
05:47 Ghost of Tsushima
06:21 Horizon Forbidden West
06:54 The Last of Us Part 1
07:20 Plague Tale: Requiem
07:42 Senua’s Saga: Hellblade 2
08:09 Starfield
08:36 Total War: Warhammer III
08:55 12-game average results
10:57 Cost per frame analysis
11:55 RT Alan Wake II
12:28 RT Black Myth: Wukong
12:57 RT Cyberpunk 2077
13:19 RT F1 24
13:40 RT Ratchet & Clank
14:00 RT Returnal
14:12 RT Shadow of the Tomb Raider
14:23 RT Star Wars Outlaws
14:43 RT 8-game average
15:18 FSR 4 analysis
18:56 Meet the Sapphire Pulse & ASRock Taichi
20:19 Thermals and acoustics
21:24 Power draw and efficiency
23:18 Closing thoughts
In lieu of an official AMD reference (or MBA) card, today we are using the Sapphire RX 9070 Pulse for our testing. This card still comes at reference specifications, however, while there's a range of factory overclocked models set to launch at retail tomorrow.
It's also worth noting that while this is a standalone review for the RX 9070, we do have a day one review for the RX 9070 XT, and you can find that HERE.

|
RX 9070 XT |
RX 9070 |
RX 7900 GRE |
RX 7800 XT |
RX 7700 XT |
Architecture |
RDNA 4 |
RDNA 4 |
RDNA 3 |
RDNA 3 |
RDNA 3 |
Manufacturing Process |
TSMC N4 |
TSMC N4 |
TSMC N5 GCD + N6 MCD |
TSMC N5 GCD + N6 MCD |
TSMC N5 GCD + N6 MCD |
Transistor Count |
53.9 billion |
53.9 billion |
57.7 billion |
28.1 billion |
28.1 billion |
Die Size |
357 mm² |
357 mm² |
300 mm² GCD
220 mm² MCD |
200 mm² GCD
150 mm² MCD |
200 mm² GCD
150 mm² MCD |
Compute Units |
64 |
56 |
80 |
60 |
54 |
Ray Accelerators |
64 |
56 |
80 |
60 |
54 |
Stream Processors |
4096 |
3584 |
5120 |
3840 |
3456 |
Game GPU Clock |
2400 MHz |
2070 MHz |
1880 MHz |
2124 MHz |
2171 MHz |
Boost GPU Clock |
Up to 2970 MHz |
Up to 2520 MHz |
Up to 2245 MHz |
Up to 2430 MHz |
Up to 2544 MHz |
ROPs |
128 |
128 |
192 |
96 |
96 |
AMD Infinity Cache |
64MB |
64MB |
64MB |
64MB |
48MB |
Memory |
16GB GDDR6 |
16GB GDDR6 |
16GB GDDR6 |
16GB GDDR6 |
12GB GDDR6 |
Memory Data Rate |
20 Gbps |
20 Gbps |
18 Gbps |
19.5 Gbps |
18 Gbps |
Memory Bandwidth |
645 GB/s |
645 GB/s |
576 GB/s |
624 GB/s |
432 GB/s |
Memory Interface |
256-bit |
256-bit |
256-bit |
256-bit |
192-bit |
Board Power |
304W |
220W |
260 W |
263W |
245W |

First, let's take a quick look at the specs. The RX 9070 is built on the new Navi 48 die, measuring 357mm², and it's worth pointing out this is a monolithic chip, so AMD is not using a chiplet-based design as per RDNA 3. In total, Navi 48 silicon packs in 53.9 billion transistors.
While a full Navi 48 GPU packs in 64 Compute Units, the 9070 is cut down to 56 CUs, and each CU houses 64 Steam Processors, for a total of 3584 shaders. There's also 56 Ray Accelerators – one per CU – and 128 ROPs.
As for clock speed, the RX 9070 runs notably slower than its sibling, the RX 9070 XT, and instead it sports a rated game clock of 2070MHz and a boost clock of up to 2520MHz.
The memory configuration is the same between both GPUs though, with 16GB of GDDR6 memory clocked at 20Gbps, operating over a 256-bit memory interface, for total memory bandwidth of 645 GB/s. 64MB of Infinity Cache is also present.
Power draw for the RX 9070 rated at 220W Total Board Power (TBP), but we are using our updated GPU power testing methodology in this review, so read on for our most detailed power and efficiency testing yet.
The post
AMD RX 9070 Review ft. Sapphire first appeared on
KitGuru.