AMD Radeon RX 5500

[jabbadap]

No clue mah dude, this is purely my speculation. If AMD want to cash in on these hypothetical SKUs they should drop them before Black Friday or before the Holidays.

Polaris stocks seem to be dwindling here in Hungary, so sooner or later they'll need to bring something to the market in place of those...

Seriously doubt they will make it to Black Friday, which is next Friday. Even just released gtx1650S might have hard time to get in the stocks for that. But yeah Q4 2019 is the time frame that AMD gave on RX 5500 announcement for AIB cards, so that in mind it should be out very soon.

 

[Freelancer]

I might be interested in a 5500XT, 5600, or 5600XT, given my current card is RX580 already. Just wanna have pretty good performance on 1080p with less noise and heat.

My Sapphire Nitro+ often directly go 100% on fan speed by itself when under load, even if the temperature is still in 60s celsius (yes I'm already using afterburner). Probably a bug or something.

 

[eidairaman1]

I might be interested in a 5500XT, 5600, or 5600XT, given my current card is RX580 already. Just wanna have pretty good performance on 1080p with less noise and heat.

My Sapphire Nitro+ often directly go 100% on fan speed by itself when under load, even if the temperature is still in 60s celsius (yes I'm already using afterburner). Probably a bug or something.

Depends on your ambient room temperature for blowing hot air on a already hot part it just makes it hotter

 

[bug]

I might be interested in a 5500XT, 5600, or 5600XT, given my current card is RX580 already. Just wanna have pretty good performance on 1080p with less noise and heat.

My Sapphire Nitro+ often directly go 100% on fan speed by itself when under load, even if the temperature is still in 60s celsius (yes I'm already using afterburner). Probably a bug or something.

Why not 1660 Super?

 

[Freelancer]

Why not 1660 Super?

Let's say I'm biased towards the A camp

 

[bug]

Let's say I'm biased towards the A camp

Ah, that. Fair enough.

 

[Freelancer]

Speaking of display card brands, I have started to lose faith in Sapphire. Which brand(s) are considered more reliable? Asus? Gigabyte?

 

[bug]

Speaking of display card brands, I have started to lose faith in Sapphire. Which brand(s) are considered more reliable? Asus? Gigabyte?

If you can't trust Sapphire, you probably can't trust anyone.

 

[Mouth of Sauron]

Depends on your ambient room temperature for blowing hot air on a already hot part it just makes it hotter

Hey, be realistic - nobody has like 60-80C room temperature - it's more like sauna environment. Normally, cooling air (or water) needs to be at least 20C lower to make decently efficient cooling (other factors apply, too).

On example, during the hot summer days of +35C, my office is cooled with chilled water @ 7-10C (yeah, I said 20C, but the surface of radiator and blower capacity are not nearly what the dimensions require). So, I get ~23-25C, minding the bracketed comment. If chilled water goes to 14C (and it does, my preciouss, it does) - then I have a nice working condition of ~30C - oh, given the company dress code and that The Master insist on this helmet, it's not nice at all...

Not meaning to drag this too long, but benefits of open-loop in water cooling systems (PC ones, no chilled water) are great, except nobody uses them A bitch to implement, probably...

Closed water loop (read: everything on the market) uses great heat transfer capability of water to cool efficiently. At the beginning... As water reaches 40, and then 60C and onward, its efficiency goes down and good air-blower (say Noctua) performs the same. Blowers on the water-cooling radiators are not enough, and we again get a solution that depends fully on air-blowers - only this time they're on radiator and not the CPU/GPU/both, whatever the setup is.

[And 65W cooling system is totally worth the energy, some will recognize this and their past mistakes, made in vanity and pride ]

Since I personally like relatively cool hardware, and not mind the noise, my custom profile for both CPU and GPU actually insists on doing 100% on temperatures reaching 50-60C, and though GPU has no-fan option I turned it off forever - you're a f-ing blower and you're going to blow no matter what, no free ride

That aside, I feel strange, unholy attraction to names such as 5600, 5600XT and they associate me strongly on 200-250g budget, emphasizing on 1440p. And I need them NOW. Not in undisclosed term in 2020, where they will be already announced to be replaced by RDNA2 parts...

 

[jabbadap]

Hmm finally had time to do little thinkering... so Voltages(1.125V) of this card is on the middle of the RX 5700(0.987V) and RX 5700 XT(1.2V) -> core clocks are in the middle too. So AMD could ease the binning by bumbing the voltage to 1.2V and gain 100MHz more core clock. Thus could probably equal 1650S perf in cost of power consumption. So there is some lee way there, but not by much. @W1zzard I assume voltages are locked on Wattman. But could you give some estimation, if 100MHz more core clock for this is enough for equal 1650S at the 1080p? After all your OC for RX 5500 is quite there with stock 5700xt clocks at higher voltages.

Nvidia surely messed AMDs plan with 1650S launch, which makes me wonder is AMD going to compete with price or trying to eek more perf out of navi 14.

 

[StaticVapour]

Honestly i'm surprised how well it performed in tests, almost feels like it's the real successor to the good old GTX 750 Ti if the price is right

 

[yeeeeman]

For a brand new uArch, 7nm, GDDR6 card, this is quite bad from all points, price, performance, efficiency. Navi is still lagging behind Turing, even with the process advantage.

 

[rruff]

For a brand new uArch, 7nm, GDDR6 card, this is quite bad from all points, price, performance, efficiency. Navi is still lagging behind Turing, even with the process advantage.

They won't lag on performance/price... because they can't. Nvidia has made this segment very competitive with the 1650S, which is great for anyone who wants to buy a card in this range. AMD will have a 5500XT 8GB variant that will probably beat it handily on performance, especially at higher resolution, and maybe beat the 1660 as well. Efficiency is still behind, but so much better than the last generation.

 

[bug]

They won't lag on performance/price... because they can't. Nvidia has made this segment very competitive with the 1650S, which is great for anyone who wants to buy a card in this range. AMD will have a 5500XT 8GB variant that will probably beat it handily on performance, especially at higher resolution, and maybe beat the 1660 as well. Efficiency is still behind, but so much better than the last generation.

5700 is slightly more efficient than 2060. The 5700XT tanks, but 5500(XT) might still surprise.

 

[nguyen]

5700 is slightly more efficient than 2060. The 5700XT tanks, but 5500(XT) might still surprise.

All the Non-RTX Turing GPUs have higher efficiency than 5700 though

 

[bug]

All the Non-RTX Turing GPUs have higher efficiency than 5700 though

I said "might", not "will"

 

[jabbadap]

I said "might", not "will"

Well if it is more efficient, then it would be slower than this OEM version...

But that would be actually okay, if the price is good. I.E. $120 for RX 5500 faster than gtx1650, $150 for RX 5500 xt with full navi14 chip and faster/equal to 1650S.

 

[Powl]

For a brand new uArch, 7nm, GDDR6 card, this is quite bad from all points, price, performance, efficiency. Navi is still lagging behind Turing, even with the process advantage.

Stop focusing on raw specs, look at actual perf/efficiency. As the review shows, the card is competitive in these aspects. Who cares what node it was on or what memory it uses? If the actual results hold up, they hold up.

 

[jgraham11]

for reference, it was this card reviewed here ; https://www.techpowerup.com/review/msi-gtx-1060-gaming-x/
yeah my bad it was not built in 14nm tech, but 16nm! so ancient! 4 years ago!

Nvidia does have an advantage in the efficiency department, the main culprit seems to be the patents that were sold to Qualcomm back in 2008. In those days AMD didn't think that division was important so they sold it off to keep the company afloat, now that Nvidia uses similar patents to reduce power on their cards, AMD can't keep up. A process node advantage only seems to get them on par with Nvidia. Now AMD is trying a different approach in their strategy to reduce power and increase efficiency, AMD Chill and that new anti-lag tech, with those two AMD seems to be trying to reinvent the metric for how a graphics card works. Its all about smoothness and not about high FPS anymore. The fact that these benchmarks don't incorporate AMD chill doesn't do AMD any favours!

 

[EarthDog]

Nvidia does have an advantage in the efficiency department, the main culprit seems to be the patents that were sold to Qualcomm back in 2008. In those days AMD didn't think that division was important so they sold it off to keep the company afloat, now that Nvidia uses similar patents to reduce power on their cards, AMD can't keep up. A process node advantage only seems to get them on par with Nvidia. Now AMD is trying a different approach in their strategy to reduce power and increase efficiency, AMD Chill and that new anti-lag tech, with those two AMD seems to be trying to reinvent the metric for how a graphics card works. Its all about smoothness and not about high FPS anymore. The fact that these benchmarks don't incorporate AMD chill doesn't do AMD any favours!

what Qualcomm parts/patents from 2008 are the culprit today?

I'm happy that general benchmarks arent run with chill. The point of testing these cards is to see how they perform out of the box against other cards out of the box. Default if you will. I'm curious to see the performance hit and power savings... but not mixed in with anything else.

 

[bug]

Nvidia does have an advantage in the efficiency department, the main culprit seems to be the patents that were sold to Qualcomm back in 2008. In those days AMD didn't think that division was important so they sold it off to keep the company afloat, now that Nvidia uses similar patents to reduce power on their cards, AMD can't keep up. A process node advantage only seems to get them on par with Nvidia. Now AMD is trying a different approach in their strategy to reduce power and increase efficiency, AMD Chill and that new anti-lag tech, with those two AMD seems to be trying to reinvent the metric for how a graphics card works. Its all about smoothness and not about high FPS anymore. The fact that these benchmarks don't incorporate AMD chill doesn't do AMD any favours!

The missing piece is tiled based rendering. Something Nvidia did back when TSMC failed to transition to 20nm, while AMD adamantly kept their computing power up. I don't think tiled based rendering was covered by those patents. But I could be wrong.