New AMD Polaris based GPU

[Dribble]

I known a few really good gamers over the years, and generally they were young (teenagers or slightly older), generally they weren't very well off and had average or below kit but they completely dominated. This is in fast paced twitch shooters as well (going back go ut2004). Nice kit is nice, and does help slightly but it's not the be all and end all. Generally it's lag that gets you killed and most lag is not due to fps (60 vs 144 adds hardly any) but you can easily get 100ms+ if you internet sucks, and second after that is probably the monitor - it doesn't need to be 144hz but it needs low input lag (which most 144hz have), but many average monitors have 30ms+ and TV's often have 100ms+. Hell, the servers often have v low tick rates (30fps) and there's nothing you can do about that so *only* having 60fps is hardly the end of the world.

What barely matters is huge screens, huge resolutions, and high settings. That's all eye candy. Infact pro gamers often play on smaller screens at average resolutions and with the eye candy all turned off (i.e. pretty well min settings but with max draw distance).

Anyway I bet most of the people thrashing you at rainbow 6 siege don't have fancy kit at all, they are just good at the game.

 

[wilds]

I known a few really good gamers over the years, and generally they were young (teenagers or slightly older), generally they weren't very well off and had average or below kit but they completely dominated. This is in fast paced twitch shooters as well (going back go ut2004). Nice kit is nice, and does help slightly but it's not the be all and end all. Generally it's lag that gets you killed and most lag is not due to fps (60 vs 144 adds hardly any) but you can easily get 100ms+ if you internet sucks, and second after that is probably the monitor - it doesn't need to be 144hz but it needs low input lag (which most 144hz have), but many average monitors have 30ms+ and TV's often have 100ms+. Hell, the servers often have v low tick rates (30fps) and there's nothing you can do about that so *only* having 60fps is hardly the end of the world.

What barely matters is huge screens, huge resolutions, and high settings. That's all eye candy. Infact pro gamers often play on smaller screens at average resolutions and with the eye candy all turned off (i.e. pretty well min settings but with max draw distance).

Anyway I bet most of the people thrashing you at rainbow 6 siege don't have fancy kit at all, they are just good at the game.

Oh yeah they do have 144hz. It is very easy to tell someone's aim when they are at 60hz and 144hz; embarrassingly so.

The difference is night in day! But why even argue at this point when 1080p144hz is often under $150? I was skeptical until I got off 60hz. It is such a big advantage.

For single player games, and CPU limited games, I still use my 4k60 monitor, but aiming is harder. In these games, an RX 580 and 1060 are plenty as the CPU is limited to 30fps due to so much AI. Even at 30fps, a high refresh rate still feels better, but nowhere near as nice as a high framerate + high refresh rate.

 

[PhonakV30]

Efficiency comparison between AMD Radeon RX 590 and RX 580 with interesting findings in undervoltage

so RX 590 at 1450 ~ 155w vs RX 580 at 1450 ~ 255 w , on other hand AMD ruined RX 590 by putting too much voltage....

 

[maddie]

Efficiency comparison between AMD Radeon RX 590 and RX 580 with interesting findings in undervoltage

so RX 590 at 1450 ~ 155w vs RX 580 at 1450 ~ 255 w , on other hand AMD ruined RX 590 by putting too much voltage....

There goes all the posturing on these being more power hungry by the usual suspects.

A while ago, didn't AMD announce the transfer of power efficiency engineers to assist the GPU group?

The Ryzen CPUs are very optimized in self regulation for varying loads and its now obvious that these skillsets were absent/neglected from the GPU group. Properly designed, there should be no need for these ruinous overvolts. Nvidia has been way ahead here and this alone probably accounts for a substantial percentage of their power efficiency advantage.

 

[coercitiv]

There goes all the posturing on these being more power hungry by the usual suspects.

The whole idea was criticizing AMD for pushing this chip to its limits on stock configurations.

The Aorus RX 580 card used as comparison point for power in the post above is a factory overclocked card. It was also used by Hardware Unboxed in their review and was only 5% slower than the XFX 590.

Spoiler: HWUB: 21 game average

In the end, unless the RX 590 cards match the price of the RX 580 Gigabyte Aorus or Saphhire Nitro+ cards, they'll just end up being more expensive cards with a 5% performance advantage, arriving on the market 18 months later after the first Polaris refresh.

 

[AtenRa]

According to this,

1 .The under-volted RX 590 is consuming ~100W less power for the same performance vs the under-volted RX 580.

Or,

2. The under-volted RX 590 is 10% faster consuming 24% less power (205W vs 255W) vs the under-volted RX 580.

 

[frozentundra123456]

Still is a power hungry card. Just less so if it is lovingly tweaked to the optimum voltage. Even then uses a lot more power than the 1060. Also begs the question of why AMD didnt optimize the card better.

 

[AtenRa]

Even then uses a lot more power than the 1060.

Really ???

From the RX 590 review

https://www.techpowerup.com/reviews/XFX/Radeon_RX_590_Fatboy/32.html

1. RX 580 performance = GTX 1060

From the RX 590 undervolting article

https://www.tomshw.de/2018/11/22/ef...n-erkenntnissen-beim-untervolting-igorslab/2/

2. RX 590 at the same performance as RX 580 = 100W less power

3. Now since the RX 580 is using ~82W more than the GTX 1060 6GB

From 1 & 2 & 3 =

RX 590 at the same performance as GTX 1060 = ~same power

 

[happy medium]

gtx 1060 undervolts/overclocks also.
gtx 1060 at 90 watts = 480

https://www.tomshardware.com/reviews/nvidia-geforce-gtx-1060-pascal,4679-7.html

the 590 is a power hog just like the 480, no way around it.

 

[coercitiv]

RX 590 at the same performance as GTX 1060 = ~same power

Welcome to the same conclusion some had pages ago: in order to get their sweet sales pitch of "faster than GTX 1060", AMD threw 12nm Polaris efficiency out the window.

They could have made a well balanced card for less money, slightly edging the GTX 1060 while finally joining it in the same power usage class.

RX 590 at the same performance as RX 580 = 100W less power

Now since the RX 580 is using ~82W more than the GTX 1060 6GB

You may want to keep in mind the RX 580 Aorus used in the reviews is a pre-overclocked card, hence using it for relative perfromance & power estimates will be quite tricky when also involving stock RX 580 numbers from another review.

Stock boost for RX 580 is 1350Mhz. At this frequency the undevolted RX 590 used 145W, while using auto voltage settings brings it to 165W. So even when undervolted and scaled down to lower frequency for ISO performance, Polaris still cannot match Pascal's 110-120W. But that would be fine either way, especially considering "overclocking" performance from such a low imaginary stock frequency.

 

[mohit9206]

I play a lot of R6 Siege, and playing below 144hz is a serious disadvantage. 60hz players are simply destroyed.

1080p144hz is getting seriously cheap now, and without the GSYNC tax, AMD is looking quite good. I also bring up R6 Siege because of how surprisingly well it runs on AMD hardware. I say this because the NVIDIA splashscreen appears on starting up R6.

The most popular card for 144hz on Siege appears to be the GTX 970. Most 144hz competitive players do not run max settings.

Don't exaggerate. You do not need 144hz to win. "serious disadvantage" i doubt it.

 

[Qwertilot]

Welcome to the same conclusion some had pages ago: in order to get their sweet sales pitch of "faster than GTX 1060", AMD threw 12nm Polaris efficiency out the window.

They could have made a well balanced card for less money, slightly edging the GTX 1060 while finally joining it in the same power usage class.

I really wish they'd leave the crazy overclocks to partner cards rather than specifying them as standard. This sort of very high power draw config doesn't really fly in the mainstream sectors. Maybe they wanted to have a bit of room for when it got paired against the eventual 2060?!

 

[tamz_msc]

Remember "up to 2.8x perf/watt."

 

[darkswordsman17]

This card seems like such a dud to me. If they'd optimized it better it'd be ok for $250 with a game bundle, but this like that Vega Nano/Mini card, seems kinda dead on arrival because the perf/$ and perf/w is just out of whack.

I really wish they'd leave the crazy overclocks to partner cards rather than specifying them as standard. This sort of very high power draw config doesn't really fly in the mainstream sectors. Maybe they wanted to have a bit of room for when it got paired against the eventual 2060?!

Yeah. Its just stupid as they could have that performance (actually better, especially usually in sustained performance) with less power use. There's simply no reason for AMD to be doing this kind of idiocy.

I really hope they're going to put extra emphasis on binning/optimizing on Navi and beyond. But it was already so bad that Microsoft took to doing their own optimization/binning for consoles. The stupid part is that they could sell different versions or upsell highest binned/optimized versions (both for efficiency and for performance). For reviews, I'd have a reference design that aims for efficiency with the "max performance" of it being more balanced, and then also have an AIB card that pushes it.

 

[wilds]

Don't exaggerate. You do not need 144hz to win. "serious disadvantage" i doubt it.

I thought the same before I started playing competitively. 60hz is a serious disadvantage against 144hz+ players.

 

[happy medium]

does 144hz make you a better gamer?
https://www.youtube.com/watch?v=6y--ZUHlyaU

 

[wilds]

does 144hz make you a better gamer?
https://www.youtube.com/watch?v=6y--ZUHlyaU

Not a great video, but I will bite. You have to be good or else getting a 144hz+ monitor won't really do much for you. It is an advantage over 60hz in R6 Siege.

In Siege, players often peek into a room to see of it is clear. Doing that at a high refresh rate allows more frames to be displayed in the few milliseconds before a gunfight ensues.

The concept of 'quick peeking' just doesn't work at 60hz. There's a reason high refresh rate is popular in games like this.

These monitors often hit about $1 per hz on sales, with Alienware's 240hz Freesync having hit $240 several times now. Feel free to ask some 'professional' players on why they do not use 60hz.

Back when Siege first started, 60hz was used by some 'pros', but they will not go back to 60hz for a reason.

I have a friend on 240hz that absolutely sucks at the game, but they always were always bad. High refresh rate is not going to make a mediocre player suddenly good.

 

[beginner99]

does 144hz make you a better gamer?
https://www.youtube.com/watch?v=6y--ZUHlyaU

While the video would proof my point of view I agree with ZGR, it's rather bad. Way too small sample size and the real issue is he goes from 60hz vsync to unlocked (well freesync actually but at >200 fps freesync doesn't matter). The real difference here most likley wasn't the refresh rate but the vsync which adds additional delay on top of the 60hz. He should try again with 60hz but without vsync.

 

[coercitiv]

At some point in time, 5ms sooner rather than later, do you think you'll realize that the refresh rate discussion belongs in a thread of it's own?

 

[lifeblood]

I think that as a 1080p gamer I would go for the RX 580. At $199 its a awesome deal. However, if I was going to buy a GPU with the intention of using it for 5 years, I would get the RX 590 and undervolt it. The standard 580/590 run too hot and the chances of it lasting 5 years is low. The 590 has the benefit of the 12nm process and when undervolted should run relatively cool thus limiting it's deterioration over time due to heat.

Of course, if I have any plans of stepping up to 1440p, then a gtx 1070 makes more sense, at least while their still available.

Unfortunately I have a daughter who figure skates so I can barely afford to pay attention, let alone buy a shiny new video card.

 

[Elfear]

The standard 580/590 run too hot and the chances of it lasting 5 years is low.

The 580 isn't that bad actually and should last just as long as other GPUs.

The 590 isn't too shabby either.

 

[Phynaz]

RX 590 at the same performance as GTX 1060 = ~same power

And it only costs $100 more than the GTX 1060!

 

[lifeblood]

And it only costs $100 more than the GTX 1060!

That was yesterday. Today on Newegg the cheapest (new) 1060 6Gb is $240 while the cheapest (new) RX 590 8Gb was $280. That's only a $40 difference. Of course that's today. Tomorrow may be something else entirely. Like the stock market you gotta play it by ear to get the best deals.

The 6Gb vs 8Gb of memory might be a big deal. I want to run high def texture packs for games like Skyrim and such. I don't know if the 6Gb would be a problem vs 8Gb. I'd need to research that some.

 

[coercitiv]

The 580 isn't that bad actually and should last just as long as other GPUs.

Especially considering it's running at virtually the same temps as GTX 1060 / 1070

 

[AtenRa]

Evidence that's a Black Friday price?

You have none.

Todays GTX 1070 prices start at $380

That is $100 more than the RX 590 which comes with a 3 games bundle