cancel
Showing results for 
Search instead for 
Did you mean: 

General Discussions

RX 5500 Benchmarked: Slower than GTX 1660, Consumes More Power

And this is the low power OEM edition drawing 133w, so there's no telling what the aftermarket versions will draw...

https://www.wccftech.com/amd-radeon-rx-5500-graphics-benchmarks-tested-on-par-with-rx-580/

4 Replies
hardcoregames_
Big Boss

Re: RX 5500 Benchmarked: Slower than GTX 1660, Consumes More Power

AMD's cards are a tad expensive for the level of performance offered. 

I have a bookmark for NewEgg filtered for 8GB VRAM or more. 4 pages of 96 cards so there are lots of cards out there but prices are brutal for a card with some major performance.

0 Likes
leyvin
Miniboss

Re: RX 5500 Benchmarked: Slower than GTX 1660, Consumes More Power

OEM != Reference

Typically speaking OEM Variants are neither the 'Best Quality' Silicon., and how they're Optimised is entirely up to the OEM Manufacturer...

I mean you do recall the Alienware RTX 2080 OEM, correct? That thing is an absolute freaking disaster.

Now as a note., you are aware that 'D' (Desktop OEM) and 'M' (Mobile OEM) are Cut-Down Versions of what will be the Retail Variants.

They always have been, and always will be. 

Typically speaking, while the overall ASIC Quality tends to be Higher than Retail Avg. in every other respect they tend to have fewer enabled CU., Lower Clocks., etc. 

Now the other weird thing here is that they're using the RX 580 Nitro+ (which is an Overclocked Variant) and 1660 OC as their "Baseline" to compare it against.

Yet, WCCF (as bloody usual) aren't listing their Test Rig(s) … nor are they providing any proper data points.

Am I expecting the RX 5500 to be a "Power Consumption Champion"? Well., frankly no.

I mean it's no secret that the RTX 2070 SUPER tends to peak at around 170w., while the RX 5700 XT (in general) hits 195 - 225w., and they're pretty comparable in terms of performance.

Don't get me wrong., I'm VERY sceptical that the RX 5500D is actually using 133w when AMD have rated them for 110w; as generally speaking AMD will list a "Peak" that is achieved under Over-Clocking. 

Take the RX 580 as an example... it's listed as a 175w GPU., but under it's 'Base' Clocks will typically hit 160w Peak. 

The same is true with the RX 480... listed as a 150w GPU., but under it's 'Base' Clocks (1266/1300MHz) will typically hit 135w. 

So for a 110w Rated Card to bit hitting 133w., well that should be raising eyebrows as to how exactly the Card was being tested. I'd wait for a Review from Gamers Nexus; who won't actually review the RX 5500D/M but instead will only test the RX 5500 and/or RX 5500 XT when they're released.

I have a feeling that the RX 5500 (which I suspect will have 24 RCU) will likely be listed as 125w., while the RX 5500 XT (which I suspect will have 30 RCU) will likely be listed as 150w.

And I'd be quite surprised if the Reference Cards ACTUALLY hit those Power Targets under Normal Usage., rather than Overclocking. 

And don't get me wrong... it's good to see Overclocking showcased as part of a Review for Enthusiasts, but in General most Consumers will NEVER Overclock their Hardware; they just use the Out-of-the-Box settings. 

I mean the whole "Review" from WCCF looks sketchy as their Articles always look. 

Don't know what the point is comparing OC Cards to an OEM Reference is., other than to make the OEM Reference just look worse than it actually is. The Power Consumption is seriously suspicious. 

They haven't even listed which OEM they've taken the card from... Dell, HP, Acer? 

Heck they're using Stock Reference Images from AMD; not Images of the actual OEM Card that they supposedly have. 

Are they testing all of them in the same Case / Rig?

You see what I mean., it's a bit too soon to jump to certain conclusions; especially given the source. 

0 Likes
leyvin
Miniboss

Re: RX 5500 Benchmarked: Slower than GTX 1660, Consumes More Power

So, I did a little digging., discovered that WCCF are citing a German Tech Website; who had frankly an abysmal Review Process.

What I did come across was the perhaps more reliable TechPowerUp Review

AMD Radeon RX 5500 Review | TechPowerUp 

If you notice in their Charts., they ONLY peak at 134w when running Furmark; which is renown for making GPU Power Hungry. 

In Avg. Gaming you're looking at 110 - 120w Power Utilisation... this is however above listed TDP., and TechPowerUp do list this as a PCPartner Manufactured Board.

This as a note is Highly Unusual., as Sapphire make all of AMD's Reference Designs. 

With PCPartner typically making all of AMD's Asian (Chinese) Market Boards., which tend to use Lower Quality Parts and change the AMD Reference Designs that are usually "Over-Engineered" and not exactly 'Cheap' to Produce En-Masse.

With this said., there is a much larger selection of Benchmarked Titles and Resolutions; which provides a much better picture of Performance. 

While obviously the OEM Version has no Price Point (as Consumers can't buy them) … I'd wager if this was a Retail Product it'd likely be between $150 - 165., where-as I believe the Retail Versions will range from $175 - 250 (RX 5500 4GB > RX 5500 XT 8GB); and as I said before I'm fairly confident in saying that the Retail Versions will both have more Unlocked RCU as well as likely Higher Sustainable Clocks. 

While Mobile OEM Parts tend to be Very High Quality Silicon (but deliberately cut-down and underclocked to be Power Efficient)., Desktop OEM Parts usually are the dregs of the Production Line that can't even be sold as Retail Versions.

As a great example of what I mean... most RX 580X (OEM Variants) were essentially clocked at Reference RX 480., and could rarely "overclock"; which was typically a 5-8% Performance Loss over a Reference., let alone an AIB Factory OC; that tended to push a further 5-8% Clock Increase.

I'm not going to lie... I really wish AMD wouldn't keep following the example of NVIDIA and return to how ATI handled OEM Branding. 

What I'd like to see would be for OEM Variants to have their own clear designation.

Given we know that they're going with the whole 3 / 5 / 7 / 9 schema; why not make the even Numbers the OEM / Mobile?

So for example:

RX 5200 / 14 RCU / 1810MHz / 90w TDP (OEM)

RX 5300 / 15 RCU / 1880MHz / 100w TDP

RX 5300 XT / 20 RCU / 1840MHz / 130w TDP

RX 5400 / 22 RCU / 1670MHz Sustained / 150w TDP (OEM / Mobile)

RX 5500 / 24 RCU / 1690MHz Sustained / 150w TDP

RX 5500 XT / 30 RCU / 1755MHz Sustained / 150w TDP

RX 5600 / 32 RCU / 1645MHz Sustained / 175w TDP (OEM / Mobile)

RX 5700 / 36 RCU / 1625MHz Sustained / 180w TDP

RX 5700 XT / 40 RCU / 1905MHz Sustained / 225w TDP

I'd wager that the Ryzen/Athlon with Graphics will likely use Navi 5 / 10 / 15 / 20 (Ryzen 4000 / Ryzen 4200 / Ryzen 4400 / Ryzen 4600) 

0 Likes

Re: RX 5500 Benchmarked: Slower than GTX 1660, Consumes More Power

It won't sell for $175-$250, it's too slow and power hungry compared to the GTX 1660, and if the rumors of Ampere are to be believed, its replacement will be in the realm of RTX 2060 performance while retaining the same power rating, which frankly is very believable for a simple generational improvement. That's not saying AMD won't PRICE it at that, it just won't shift at that price considering nVidia's advantages.

As for the naming convention, I believe it'd be much easier, simple, and easily understood to name desktop chips RX, OEM chips RO, and mobile chips RM.

0 Likes