• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

Sapphire Radeon RX 470D NITRO Pictured, Beats GTX 1050 Ti

btarunr

Editor & Senior Moderator
Staff member
Joined
Oct 9, 2007
Messages
47,670 (7.43/day)
Location
Dublin, Ireland
System Name RBMK-1000
Processor AMD Ryzen 7 5700G
Motherboard Gigabyte B550 AORUS Elite V2
Cooling DeepCool Gammax L240 V2
Memory 2x 16GB DDR4-3200
Video Card(s) Galax RTX 4070 Ti EX
Storage Samsung 990 1TB
Display(s) BenQ 1440p 60 Hz 27-inch
Case Corsair Carbide 100R
Audio Device(s) ASUS SupremeFX S1220A
Power Supply Cooler Master MWE Gold 650W
Mouse ASUS ROG Strix Impact
Keyboard Gamdias Hermes E2
Software Windows 11 Pro
It turns out that the new SKU based on the Polaris 10 "Ellesmere" silicon AMD is working on, to fill the gap between the Radeon RX 460 and RX 470, could be named the "Radeon RX 470D," after all. It is also being reported that AMD is planning to launch this SKU only in China, for now. It is further cut down from the RX 470, in featuring 1,792 stream processors, by enabling just 28 out of 36 GCN compute units physically present on the chip. The SKU further features 112 TMUs, 32 ROPs, and 4 GB of GDDR5 memory across a 256-bit wide memory interface.

AMD is clocking the RX 470D slightly higher than the conventional RX 470. Its core ticks at speeds of up to 1266 MHz (Boost clock), and its memory at 7.00 GHz (GDDR5-effective), compared to the 1206 MHz (Boost clock) and 6.60 GHz (GDDR5-effective) speeds of the RX 470. The RX 470D will be built to a cost, and it shows, on the Sapphire RX 470D NITRO. The PCB is shorter, and the cooler leaner. Chinese tech publication PCOnline got its hands on a sample, and tested it to be faster than the GeForce GTX 1050 Ti, yet slower than the RX 470. The card is expected to be priced about 14 percent less than the RX 470. Find more benchmark figures in the PCOnline review linked below.



View at TechPowerUp Main Site
 
Why though

Would have been better off dropping the price on the current RX 470
 
The reason that AMD isn't cutting the price of the RX 470 any further is they don't want to undercut sales of the RX 480 by pricing the RX 470 too low. The other thing is that This new card is supposed to be China only according the the article
 
Beats GTX 1050 Ti in performance, still needs an extra 150W of power.
 
The reason that AMD isn't cutting the price of the RX 470 any further is they don't want to undercut sales of the RX 480 by pricing the RX 470 too low. The other thing is that This new card is supposed to be China only according the the article

That doesn't change the fact that 470 is priced too high relative to it's performance, there's no reason to buy it over 480.

Beats GTX 1050 Ti in performance, still needs an extra 150W of power.

Yes, that's not a beat in my book.
 
Beats GTX 1050 Ti in performance, still needs an extra 150W of power.

I don't think the TDP is released yet. The original Rx470 didn't need an extra 150W of power on top of the GTX 1050 Ti. The TDP of the 1050 Ti (reference design) is 75 Watts, the TDP of the Rx470 is 150 Watts. 150 Watts extra in addition to the 1050 Ti's TDP would be 75 Watts + 150 Watts = 225 Watts, which is incorrect.
 
Beats GTX 1050 Ti in performance, still needs an extra 150W of power.
My RX 480 @1400MHz doesn't even draw 225W. I'd guess it draws about 100W
 
Beats GTX 1050 Ti in performance, still needs an extra 150W of power.

I am pretty sure in your dreams all AMD cards draw more than 300 WATTS isn't it.
power_peak.png
 
Last edited:
We're not talking about idle power consumption here.

https://www.techpowerup.com/reviews/MSI/RX_480_Gaming_X/21.html

According to this:

https://www.techpowerup.com/reviews/MSI/GTX_1050_Ti_Gaming_X/25.html

The Rx470 uses about 126W while the GTX 1050 Ti is at 75W, peak gaming. If the Rx470D is anything like the Rx470, the power difference is about 50W. If the Rx470D uses less power because it is slower overall than the non-D version, then that difference is going to be smaller than 50W.
 
According to this:

https://www.techpowerup.com/reviews/MSI/GTX_1050_Ti_Gaming_X/25.html

The Rx470 uses about 126W while the GTX 1050 Ti is at 75W, peak gaming. If the Rx470D is anything like the Rx470, the power difference is about 50W. If the Rx470D uses less power because it is slower overall than the non-D version, then that difference is going to be smaller than 50W.

I agree that RX 470 shouldn't need anything near 225W, but if that's the case, why are the manufacturers provisioning these cards with 8-pin PCIe power connectors? It's almost like a reversal of the RX 480, where that card did need up to 225W but was only provisioned for 150W. I just don't understand AMD and/or the AIBs' strategy in this regard.
 
I agree that RX 470 shouldn't need anything near 225W, but if that's the case, why are the manufacturers provisioning these cards with 8-pin PCIe power connectors? It's almost like a reversal of the RX 480, where that card did need up to 225W but was only provisioned for 150W. I just don't understand AMD and/or the AIBs' strategy in this regard.

That's a good question that we can only guess the answers to. If it was me designing these cards and they were going to feature a PCI-e connector, I might just spring for the 8 pin over the 6 pin for a few reasons: it gives me/users more power capacity to OC, it allows me to change the VRM design for something less efficient and get away with it, it is simpler than 2x6 pin plugs, it looks more marketable towards "enthusiasts".

I strongly suspect the VRM designs are not OEM suggested types (my 2nd idea above). I don't have an exact reference on the Rx-series of current, but when AMD introduced digital VRM controls, I think with the HD2900 series (featuring much higher frequency phases and surface mount tantalum polymer capacitors), none of the AIBs followed that trend with their custom cards. We frequently see these VRM design changes from digital to VRM since that time and if you check the prices, the digital parts are more expensive, as are the capacitors they use, in exchange you get better regulation and finger voltage controls and you can design a flatter-profile cooler that is very close to PCB height. During the HD4870 era, a lot of AMD's top notch cards had their digital VRMs damaged due to Furmark tests (was a big issue) and a lot of the AIBs marketed more robust VRM systems that used the older analog VRM designs.
 
Sounds like a solid card. The 1050ti would still be more appealing for power or space constrained builds though
 
Beats GTX 1050 Ti in performance, still needs an extra 150W of power.

LMAO ok buddy. When I had my 470 Nitro 8GB (8000 GHz memory), it used 120w and beat a stock 480/1060.


Yes the 470D will use more energy than a 1050 Ti, it is also incredibly more powerful.


Not that the difference in power usage in these cards matters at all. It's not like the situation with the GTX 480 vs 5870. Now THAT was pathetic.
 
LMAO ok buddy. When I had my 470 Nitro 8GB (8000 GHz memory), it used 120w and beat a stock 480/1060.


Yes the 470D will use more energy than a 1050 Ti, it is also incredibly more powerful.


Not that the difference in power usage in these cards matters at all. It's not like the situation with the GTX 480 vs 5870. Now THAT was pathetic.

120W 470? I think you're mistaking chip power with total card power. ;)
 
It will really depend on binning of the cards. If they are bad 470's cut and clocked power consumption will be junk.

But remember amd has 95w complete board and memory TDP figures for the rx480 based embedded solutions. They have only shown the tip of what Polaris can do. Unluckily it looks like glofo let them down again with a poor manufacturing process.
 
It will really depend on binning of the cards. If they are bad 470's cut and clocked power consumption will be junk.

But remember amd has 95w complete board and memory TDP figures for the rx480 based embedded solutions. They have only shown the tip of what Polaris can do. Unluckily it looks like glofo let them down again with a poor manufacturing process.

Bingo.

It's pretty obvious to me that the first Polaris samples were simply early junk yields that AMD was buying for bargain prices. 14nm was clearly a few months behind 16nm in maturation, and frankly it should be considering it is 9% more dense. But it's clear that the process is catching up, and we should see 100w 485's at 1400MHz+ very soon.
 
Bingo.

It's pretty obvious to me that the first Polaris samples were simply early junk yields that AMD was buying for bargain prices. 14nm was clearly a few months behind 16nm in maturation, and frankly it should be considering it is 9% more dense. But it's clear that the process is catching up, and we should see 100w 485's at 1400MHz+ very soon.

There already are 120W RX480's at 1400mhz, hell my card isn't exactly the best clocker and does 1485
 
LMAO ok buddy. When I had my 470 Nitro 8GB (8000 GHz memory), it used 120w and beat a stock 480/1060.

Cool story bro. Care to post any tests from independent websites substantiating your claim?

Not that the difference in power usage in these cards matters at all. It's not like the situation with the GTX 480 vs 5870. Now THAT was pathetic.

Of course it matters. It's the difference between having the card entirely bus-powered, and hence suitable to be dropped into any machine with a PCIe x16 slot, versus requiring external PEG power.

There already are 120W RX480's at 1400mhz, hell my card isn't exactly the best clocker and does 1485

Have you measured this? If so how, and under what conditions (e.g. is your card under air or water)?
 
Have you measured this? If so how, and under what conditions (e.g. is your card under air or water)?

I'm closer to 180W under stock air with a nitro+

The lower wattage ones are the xfx they appear to have some preferential binning deal.
 
Back
Top