AMD pits the Radeon RX 5500 against Nvidia GTX 1650

We’ve been waiting for details on AMD’s Radeon RX 5500 GPU for a while. Details keep leaking out, and now we have some leaked marketing materials direct from AMD (via TechPowerUp) to give us a better idea of where the company plans to position its upcoming chip.

Radeon RX 5500: The hard specs

The RX 5500 is built on AMD’s 7-nm Navi 14 architecture. The chip has 22 RDNA compute units with 64 stream processors each for 1,408 stream processors in total. The design has a 128-bit wide GDDR6 memory bus, and will be available in 4- and 8-GB variants with 22 GB/s memory bandwidth. AMD lists the GPU clocks as 1670 GHz for “gaming” and 1845 GHz for boost. It doesn’t mention any standard clocks. The board will take its power through a single 8-pin PCIe plug and pull 108W of power.

Versus the 1650

AMD is apparently positioning this chip against , which goes for around $150 to $160 right now. Compared to the 1650, AMD quotes performance improvements from 20% to 85% depending on the game.

In a lineup of its own cards, AMD drops the card between the RX 5700 and RX 560; note the absence of the 570, 580, and 590 in that list. AMD is imagining the cards a 1080p gaming workhorse meant to play games at high settings with that resolution. The card will also feature technology like AMD’s anti-lag tech and Radeon Image Sharpening, as well as FreeSync compatibility.

AMD hasn’t revealed a hard price point or release date for the card, though the comparison to the GTX 1650 is telling about where we can expect the RX 5500 to land price-tag wise.

avatar
3 Comment threads
5 Thread replies
0 Followers
 
Most reacted comment
Hottest comment thread
5 Comment authors
derFunkensteinKrogothThe lost catJustAnEngineerDragontamer5788 Recent comment authors
  Subscribe  
newest oldest most voted
Notify of
The lost cat
Guest
The lost cat

Wouldn’t it be nice if they release prices so we had any idea if we should even care?

Dragontamer5788
Guest
Dragontamer5788

The design has a 128-bit wide GDDR6 memory bus, and will be available in 4- and 8-GB variants with 22 GB/s memory bandwidth. That’s awfully low, I think you’re missing a “4” in that memory bandwidth spec! (224 GB/s). The NAVI architecture is extremely interesting to me, but is not yet available for ROCm. It seems like the ROCm developers are focusing on Vega still, so I’ll be skipping out on this Navi chip personally. 224GB/s bandwidth on 22-compute units is pretty good on memory-bandwidth per compute-unit. (Vega64 was 480GB/s on 64-compute units), so this Rx 5500 is a more… Read more »

derFunkenstein
Guest

Maybe it’s just really high compute-to-bandwidth, considering the clock speeds:

1670 GHz

JustAnEngineer
Guest
JustAnEngineer

Man, that 7 nm fabrication node is magical! My Radeon RX Vega64 runs toasty hot at just 1.6 GHz.

Krogoth
Guest
Krogoth

TSMC 7nm process isn’t that magical though. Remember that Vega 64 is much more massive and has to share the same thermal space as the HBM2 chips. Radeon VIIs have been hitting ~1.7Ghz on boost without too much trouble

derFunkenstein
Guest

I think you missed the joke and the 3 orders of magnitude difference that’s still, days later, in the original article.

derFunkenstein
Guest

Apple just released new MBPs with this chip, so obviously magic is a big part of it. LOL

derFunkenstein
Guest

Considering how much more well-endowed the GeForce GTX 1650 Super is over its predecessor, my guess is that the RX 5500 will be roughly equivalent to the new low-end Turing.

Pin It on Pinterest

Share This