AMD Ryzen AI HX 170/370 in first leaked benchmarks
AMD was quite reluctant to share particulars on the RDNA3.5 in the course of the official keynote. What we all know is that it makes use of an up to date structure and options extra cores than last-gen merchandise. Both can contribute to greater efficiency.
Over at Geekbench, one can discover three scores that includes the Ryzen AI 9 HX 170 processor in CPU and GPU exams. As we defined in the earlier submit, AMD modified the naming schema for its Strix Point processors shortly earlier than bulletins, main some corporations to show laptops with older SKU names throughout their demos. Apparently, anyone received their fingers on certainly one of these items, which is probably going how the leak surfaced.
RYZEN STRIX POINT SCORES, Source: Geekbench
The Ryzen AI 9 HX 170/370 is a 12-core Zen5 processor with a most enhance functionality of 5.1 GHz. AMD has not but confirmed the precise clocks for the ‘denser’ Zen5c cores that this processor has, however we do know that it’s going to characteristic 12 RDNA 3.5 Compute Units which can enhance as much as 2.9 GHz.
What we have no idea is what the TDP was set for these exams, however we do know exams have been carried out on an ASUS ProArt A16 laptop computer, which is an APU-only laptop computer with out discrete graphics, so the APU wattage could be greater. AMD is simply saying that the default TDP is 28W, however it could be in the 15W to 54W vary.
Additionally, the Strix Point CPU core didn’t enhance to 5.10 GHz; as an alternative, the detailed end result web page reveals a max enhance of 4.25 GHz, which means it didn’t attain the very best frequency. This could possibly be as a result of a decrease TDP or the truth that the HX 170 isn’t the ultimate retail SKU, however an engineering pattern for demonstration functions.
So, for this submit, we’ll deal with the graphics, which are much less susceptible to TDP modifications as soon as the APU hits 28W. The GPU scores 41,995 factors in OpenCL exams, which is hardly a very good benchmark when gaming is taken into account, however that is all we have now for now.
| Mobile IGPU efficiency (Official Ranking) | |
|---|---|
| VideoCardz.com | OpenCL Score (Geekbench) |
| GeForce GTX 1650 Ti DESKTOP | |
| AMD Radeon RX 480 DESKTOP | |
| AMD Radeon 890M IGPU | |
| GeForce GTX 1650 Ti Max-Q MOBILE | |
| GeForce GTX TITAN X DESKTOP | |
| AMD Radeon 780M IGPU | |
| AMD Radeon 680M IGPU | |
In quick, which means that the 890M (listed as 880M right here) is 39% faster than the Radeon 780M (RDNA3) and 59% faster than the 680M (RDNA2). It additionally outperforms the GTX 1650 Ti Mobile, which was a well-liked advertising and marketing level for the 780M graphics when Mini-PC corporations in contrast uncooked TFLOPS. Now, it’s an actual artificial rating.
Of course, OpenCL exams is hardly one thing we might use for GPU comparisons, however it is a good signal that core depend replace apart, there may be clearly some architectural enchancment. Also, price noting that each 890M and 780M can enhance as much as 2.9 GHz.
Ultimately, this might lead to an entire reshaping of the entry-level desktop and cellular GPU market. These playing cards make little sensible sense anymore until they are particularly added to techniques for his or her encoding capabilities. Integrated GPUs are now faster and way more power-efficient, so it’s positively thrilling to see AMD bringing these APUs to desktops.
| Radeon IGPUs | |||
|---|---|---|---|
| VideoCardz.com | Radeon 890M | Radeon 780M | Radeon 680M |
| Architecture | RDNA 3.5 | RDNA 3 | RDNA 2 |
| Compute Units | 16 | 12 | 8 |
| Stream Processors | 1024 | 768 | 512 |
| Max GPU Clock | 2.9 GHz | 2.8 GHz | 2.4 GHz |
| Raw TFLOPS | 5.9 TFLOPS (Dual-Issue) | 4.3 TFLOPS (Dual-Issue) | 1.2 TFLOPS |
| Best SKU | Ryzen AI 9 HX 370 | Ryzen 9 8945HS | Ryzen 9 6980HX |
Source: Geekbench #1, #2, #3 by way of BenchLeaks