AMD has announced the Radeon RX 5500 graphics cards targeted primarily at 1080p gaming. The RX 5500 will be available in both desktop and laptop (M) variants by the end of this month. AMD has also partnered up with leading laptop manufacturers like HP, Lenovo, Acer and MSI to usher in the new 5500M GPU. One thing to keep in mind though: These Graphics Cards are OEM-only. As of now, there’s no info as to when the retail versions will be available.
MSI Alpha 15 is a new chapter for us, and we’re excited to partner with AMD
CHARLES CHIANG, CEO, MSI
The Radeon RX 5500 and 5500M are based on the 7nm process and the RDNA microarchitecture. They will succeed the RX 560X and will offer 1.6X more performance than the GCN based cards. The 5500 is touted to be up to 37% faster as compared to NVIDIA’s offerings. AMD claims a 60+ FPS in most games with up to 90+ FPS in eSports titles.
| Model | Compute Units | Stream Processors | Memory Size (GDDR6) | Boost Clock | Memory Interface |
| RX 5500 (Desktop) | 22 | 1408 | 8 GB | 1845 MHz | 128-bit |
| RX 5500M (Notebooks) | 22 | 1408 | 4 GB | 1645 MHz | 128-bit |
Compared to NVIDIA’s offerings, the RX 5500 absolutely destroys the GTX 1650 in terms of raw specs alone. Even when compared against the GTX 1660Ti, NVIDIA’s flagship GTX card, the 5500 can go head-to-head with it. With up to 8GB GDDR6 memory and a Boost Clock of 1845 MHz, AMD has seriously turned up the competition. Moreover, it also includes PCIe 4.0 support.
The Radeon RX 5500 will also include support for AMD’s proprietary gaming-specific optimizations at launch:
- Radeon Image Sharpening: Which upscales visuals to provide more clarity and crispiness
- AMD FidelityFX: Open-Source toolkit to allow support for different post-processing effects
- Radeon Anti-Lag: This severely reduces input lag and display response times. As per AMD, this makes Borderlands 3 up to 23% more responsive.
With the purchase of an eligible 5500 product, you will be eligible for a free copy of either Borderlands 3 or Tom Clancy’s Ghost Recon Breakpoint. The first of these products will be launched by MSI. The MSI Alpha 15-inch will launch later this month, followed by the HP Omen Obelisk and Pavilion desktops and the Lenovo Legion T530 and IdeaCentre T540 gaming PCs. The desktop RX 5500 will be available in Acer Nitro PCs starting December.
Further Reading:
Looking at the graphic on Anandtech shows a 3 x 4 grid of CU pairs, or 24 CU’s, so this is a lower bin part with one CU pair fused off. Expect a RX 5500 XT with 24 CU’s. It also looks like a one shader engine part. Navi 10 has two shader engines. This opens the possibility of a double Navi 14 part with two shader engines and 48 CU’s.
Could this be what Navi 12 (RX 5900 XT) is? 48 CU’s, 256 bit memory bus.
Can’t say for sure. It’s possible
Note AMD have used XL extensions alongside XT extensions with Vega. XT for top bin parts and XL for lowest bin parts. Could there be a RX 5700 XL with 32 CU’s? The RX 5700 fuses off one CU pair (WGP’s) per shader engine. Each shader engine has two groups of 5 WGP’s. Odds of have one fault in each of four groups (or three of the groups) would be quite low so I don’t think many chips would end up in the XL bin. AMD could be waiting to accumulate sufficient parts.
We don’t know yet what the RX 5300 is. At this point I’m thinking re-branded Polaris GPU.
7 Navi CU’s should exceed 11 Vega CU’s in an APU but would make it a harder sell as it would seem like a backwards step. DDR5 would allow for 14 Navi CU’s or perhaps a less memory constrained 11 Navi CU’s (for +60% performance). It also make sense not to port Navi to an APU for an outgoing memory system but for the incoming memory system especially when the memory performance is so important to APU performance. Since APU’s are for a very cost sensitive market this would require waiting until DDR5 becomes established and costs decline. I would say a year after DDR5 is on a mainstream CPU and we know Zen 3 will be AM4 socket and hence not DDR5. This means Zen 4 at the earliest or almost 2 cycles plus the usual APU delay.
It wasn’t launched, it got announced. If its already launched then give a price quote.
If the RX 5500 beats the GTX 1650 by 37% (AMD’s own numbers) and the GTX 1660 beats it by 37% then the RX 5500 should be equivalent to the GTX 1660 right? So why does AMD put the RX 5500 on the same teir as the GTX 1650? May be they wanted to show their card against one they could beat by a considerable amount in everything rather than winning some and losing some. It certainly lowers the expected price bracket. 150W TDP is disapointing.
RX 5500 (22 CU’s) versis GTX 1660 (22 SM’s)
RX 5500 XT (24 CU’s) versis GTX 1660 Ti (24 SM’s)
RX 5700 (36 CU’s) versis RTX 2070 (36 SM’s) -6%
RX 5700 XT (40 CU’s) versis RTX 2070 Super (40 SM’s) +1%
Curious thing is RX 5500 was to be a 110 W card became a 150 W card but should have been a 75 W card if it was a true GTX 1650 rival. Recommended PSU is 550 W which is high even for a 150 W card. RX 580 is a 185 W card recommends a smaller 500 W PSU. Mobile part will be dead on arrival (DOA) if it uses anything like this much power so it wont. Something doesn’t add up. It could be AMD pushing clocks on the non mobile part to try and get performance up to its obvious design rival the GTX 1660. Normally the lower bin part is clocked lower to aid performance differentiation especially when the difference between 22 CU’s and 24 CU’s is only about 9%.
It looks to me like the 6 CU pairs (WGP’s) per half shader engine it uses might be sub-optimal compared to the 5 CU pairs per half shader engine used in Navi 10. The RX 5500 with 22 CU’s already has one half shader engine reduced to 5 CU pairs. When real cards surface and are bench-marked this would be informed one way or another. This could inform architectural choices for big Navi and Navi in APU’s.