Early AMD Radeon RX 5500 OEM Testing Shows RX 580-Level Performance With Less Power

News
By
published

AMD Radeon RX 5500 OEM sample fails to topple the GTX 1660.

Radeon RX 5500 OEM

(Image credit: Heise)

AMD formally announced the Radeon RX 5500 last month, but did not mention pricing or a release date. However, German media Heise reportedly got its hands on an OEM version of the component and posted early benchmark results of AMD's upcoming mid-range graphics card. 

The graphics card Heise reviewed came from inside a HP Pavilion Desktop (TP01-0004ng) PC that Heise requested for review. PC Partner reportedly produced Heise's Radeon RX 5500 graphics card. The name probably won't sound familiar to you, but the PC Partner's Zotac, Inno3D and Manli brands should. AMD's Radeon RX 5500 will be available with up to 8GB of GDDR6 memory, but the publication's sample came packing 4GB and adheres to AMD's reference specifications for the graphics card. 

The Radeon RX 5500 in question features a compact dual-slot design, plus a simple black plastic shroud and a single cooling fan. For an OEM product, it doesn't look that bad. However, we expect AMD's partners to dazzle us with more elaborate designs when custom third-party cards arrive. 

Heise noted that while the RX 5500 supports the PCIe 4.0 interface, the graphics card it testedcan only exploit eight of the 16 PCIe lanes. The RX 5500 is rated with a 150W TBP (typical board power), and, therefore, requires a single 8-pin PCIe power connector. The graphics card provides two DisplayPort outputs and one HDMI port for connecting displays.

AMD Radeon RX 5500 (OEM) Benchmark Results

Heise paired the RX 5500 with the Intel Core i7-8700K CPU, 32GB of RAM and an undisclosed motherboard. The publication ran the gaming benchmarks at a resolution of 1920 x 1080 with each game's built-in benchmark.

The GPU was compared against the Sapphire Nitro+ Radeon RX 580 and Gigabyte GTX 1660 OC 6G.

Swipe to scroll horizontally
Model3DMark Fire StrikeShadow of the Tomb RaiderFar Cry 5GPU IdleGPU Load
AMD Radeon RX 5500 (OEM)12,11159 fps72 fps7W133W
Sapphire Nitro+ Radeon RX 58012,74465 fps75 fps12W207W
Gigabyte GTX 1660 OC 6G12,52569 fps85 fps10W128W

The Radeon RX 5500 fell behind the RX 580 and GTX 1660 in Fire Strike by 5.2% and 3.4%, respectively.

When it came to Shadow of the Tomb Raider, the RX 5500 pumped out 10.2% lower framerates than the Radeon RX 580. In comparison to its Nvidia rival, the RX 5500 was up to 16.9% slower.

In Far Cry 5, the RX 5500 delivered 4.2% and 18.1% lower performance than the RX 580 and GTX 1660, respectively.

In regards to power consumption, the RX 5500 consumed 74W less than the RX 580 at full load. This is a significant improvement, as AMD graphics cards have been mocked in the past for running too hot. However, it seems like the RX 5500 still drew a bit more power than the GTX 1660 while also being fairly slower.

If Heise's results are realistic to how the card will actually perform upon release, AMD will have a worthy replacement for the RX 580 in the RX 5500. However, the RX 5500 doesn't seem ready to conquer the GTX 1660. Of course, these are early benchmarks; stay tuned for a full review from Tom's Hardware for a deep look at the card's true potential. 

See more GPUs News
Zhiye Liu
Zhiye Liu
News Editor and Memory Reviewer

Zhiye Liu is a news editor and memory reviewer at Tom’s Hardware. Although he loves everything that’s hardware, he has a soft spot for CPUs, GPUs, and RAM.

15 Comments Comment from the forums
  • King_V
    I see comments in the article about it failing to topple the 1660, but wasn't AMD's stated goal to outperform the 1650?

    From the link to the RX5500 announcement:
    Performance-wise, AMD puts this card up against a GTX 1650 in order to compare it to “similarly positioned products.” In AMD’s internal testing, the RX 5500 beats the GTX 1650 (both mobile and desktop cards) by a fair amount, but does so with notably higher power consumption -- which we’ll touch on shortly. According to AMD’s press materials, we can expect 90 fps in Gears 5, 82 fps in Borderlands 3, and 60 fps in Ghost Recon: Breakpoint, while eSports titles such as Fortnite, Apex Legends, PUBG, WoW, Overwatch, and Rainbow Six: Seige are all 90 fps or above.

    Was there any expectation that it would outperform the 1660? I mean, nice if does in final form, but I didn't think anyone was expecting it.

    Now, if it turns out that the 1650 Super performs on the level of the 1660 non-Super, then AMD has a bit of a problem on its hands, and the price is definitely going to have to undercut the 1650 Super.

    But, I suspect that, if they're offering near-1660 performance for 1650 money, then they're in a good place with this.

    Any which way, I'm definitely looking forward to seeing how the released version performs, relative to the Polaris cards, as well as to the 1650 Super. I imagine that, if previous patterns remain consistent, it's going to win on price/performance if not on outright performance in its tier.
    Reply
  • redgarl
    From a an early access from a video on youtube, the 5500 (^) is consuming between 85-90W in RDR2 while offering some honest performances.

    Not sure who is that guy and honestly, until we get real reviews, there is not much to talk about that we didn't know already.
    Reply
  • InvalidError
    King_V said:
    Now, if it turns out that the 1650 Super performs on the level of the 1660 non-Super, then AMD has a bit of a problem on its hands, and the price is definitely going to have to undercut the 1650 Super.
    Hardware Unboxed hacked GPU drivers to run its 1650S without waiting for review drivers and it is ~10% slower than the plain 1660 and ~5% slower than the RX580.

    Looks like the RX5500 and GTX1650S are going to be pretty close. The MSRP on 1650S won't be any higher than $165 (the price of UB's 1650S review sample), so AMD shouldn't be able to charge any more than that.
    Reply
  • King_V
    InvalidError said:
    Hardware Unboxed hacked GPU drivers to run its 1650S without waiting for review drivers and it is ~10% slower than the plain 1660 and ~5% slower than the RX580.

    Looks like the RX5500 and GTX1650S are going to be pretty close. The MSRP on 1650S won't be any higher than $165 (the price of UB's 1650S review sample), so AMD shouldn't be able to charge any more than that.


    Interesting - and if so, then pricing gets kind of crazy. Not only does AMD have to price the RX5500 appropriately, but then the 1660 becomes sort of overpriced. As it is, the 1660 Super makes the 1660Ti look like a bad value.

    And if all that happens, then the RX 570/580/590 cards should see their prices plummet. I'm still surprised the RX 560 is about the same price as the 570, and that the RX 550 isn't significantly cheaper than it currently is.

    I start to wonder if AMD anticipated the 1650 Super's performance when they said competing with the 1650... Any which way, when the 1650 Super and the RX5500 are both on the market, things should get very interesting - and I'm looking forward to it.
    Reply
  • InvalidError
    King_V said:
    I'm still surprised the RX 560 is about the same price as the 570, and that the RX 550 isn't significantly cheaper than it currently is.
    I'm not surprised. Between all of the unit costs of making complex products like GPUs and getting them to store / e-tailer shelves, there isn't much room for anything under $100 to be economically viable, so everything at the lower end gets squeezed between that ~$100 minimum viable MSRP and remnants of the crypto-mining bust.
    Reply
  • alextheblue
    King_V said:
    But, I suspect that, if they're offering near-1660 performance for 1650 money, then they're in a good place with this.
    Indeed. Also of (lesser) note is that the comparison is between an OEM 5500 and a pair of aftermarket designs. So final clocks of similar aftermarket 5500 designs will probably be a smidge higher.
    Reply
  • digitalgriffin
    It seems the 5500 seems closer to a replacement for the rx570.

    Those cards are commonly available for $125. I can still find rx580 for $160

    Im sure amd took a loss after the crypto crash. Getting prices back to $200 for entry level is a difficult proposition given the competition and excess glut of entry level cards.

    It also leaves a HUGE gap between $150 for a 5500 and $330 5700. That could leave room for a 5600 at $250. But that would have to beat a 1660ti
    Reply
  • InvalidError
    digitalgriffin said:
    Getting prices back to $200 for entry level is a difficult proposition
    Last I checked, entry-level (the minimum commercially viable GPU) was closer to $100. Once you get to $200, you are in mainstream territory where buyers expect actually useful gaming performance.

    The main reasons for mainstream GPUs pushing the $200 mark a few years ago was GDDR5 costing ~$20/GB so 8GB of memory accounted for ~$160 of manufacturing cost concurrently with the crypto-boom draining the market dry. Both justifications are no longer applicable, pressure to provide better performance at lower price points is coming back on and we're finally seeing some significant movement below the $200 mark..
    Reply
  • digitalgriffin
    InvalidError said:
    Last I checked, entry-level (the minimum commercially viable GPU) was closer to $100. Once you get to $200, you are in mainstream territory where buyers expect actually useful gaming performance.

    The main reasons for mainstream GPUs pushing the $200 mark a few years ago was GDDR5 costing ~$20/GB so 8GB of memory accounted for ~$160 of manufacturing cost concurrently with the crypto-boom draining the market dry. Both justifications are no longer applicable, pressure to provide better performance at lower price points is coming back on and we're finally seeing some significant movement below the $200 mark..

    I think the target window for main stream is high framerate (100fps) 1080p or mid frame rate 1440p. So I would consider a 1660 ti /2060/5700 mainstream now.

    We have been stuck at 1080p for a long long time. And if you look at steam stats youll see 1440p monitors are reaching the tipping point.

    Video games also keep advancing. If you cant play a AAA game at reasonable quality settings and 60fps, you are no longer main stream. For example a 580 and 1060 struggle with rdr2 at 1080p.

    If you take the stance that 1080p should be mainstream i think that effectively says gpu advancement has stopped.
    Reply
  • InvalidError
    digitalgriffin said:
    We have been stuck at 1080p for a long long time. And if you look at steam stats youll see 1440p monitors are reaching the tipping point.
    The latest Steam survey says 1440p has lost ground (-0.1% to 5.5%) while 1080p is still rising (+1.1% to 65.4%) so we're nowhere close to anything resembling a "tipping point" in the general public yet. Enthusiasts may talk about 1440p a lot but normal people don't.

    My prediction is that the same thing will happen to 1440p as has happened to all other resolutions near1080p: UHD is getting cheaper faster than 1440p is simply from being the next all-encompassing world-wide resolution standard and once UHD starts under-cutting everything else like 1080p did, most intermediate resolutions will get relegated to often overpriced niches like 1200p and 1600p did. If you look at what is happening in the TV space, FHD didn't gain massive momentum until it reached price parity with lower resolution TVs. Now, UHD has been undercutting FHD TVs for about a year and FHD is getting phased out of the market place. UHD is well on its way to becoming the de-facto standard for the unwashed masses as anything less no longer makes economic sense.

    Only thing missing is affordable GPUs to drive 4k at playable frame rates and with the RX5500/GTX1650S, we should be just about there for low-ish detail. It may not be good enough for you, but it'll be good enough for a large chunk of people playing Steam games.
    Reply