AMD Announces The Radeon RX 480 Starting At Just $199

razor1 said:
mGPU optimizations is done on the developer end, if I'm not mistake, AOTS uses explicit implementation, that means its 100% done on developer end. AMD drivers should have very little do with this, outside of functionality.
Click to expand...

Yeah, we're talking about the single gpu number we derived from the mgpu performance though
 
ah ok yeah I'm a bit drunk right now lol, hmm time to go to bed hehe
 
It's disappointing the amount of arguing going on this this thread, especially after a moderator's warning. The RX 480 is great news if you're an AMD or an NV fan.

The key takeaway from the RX 480 announcement is the potential for (not confirmed until actual reviews) a 60% boost in performance within the $200 segment. Nvidia won't be able to get away with the 15-25% boost they've been doing for their x60 segment if they want to compete with this card. A rising tide lifts all ships.

At that level of performance, the card falls within 25-40% of the GTX 1070 (not the 10-20% that some AMD fans are fantasizing over, but we'll see in a month or so). If NV cuts a GP104 to make a GTX 1060/ti to compete with the RX 480 at $200, they won't be able to justify the 1070 at $380.
 
Daniel_Chang said:
It's disappointing the amount of arguing going on this this thread, especially after a moderator's warning. The RX 480 is great news if you're an AMD or an NV fan.

The key takeaway from the RX 480 announcement is the potential for (not confirmed until actual reviews) a 60% boost in performance within the $200 segment. Nvidia won't be able to get away with the 15-25% boost they've been doing for their x60 segment if they want to compete with this card. A rising tide lifts all ships.

At that level of performance, the card falls within 25-40% of the GTX 1070 (not the 10-20% that some AMD fans are fantasizing over, but we'll see in a month or so). If NV cuts a GP104 to make a GTX 1060/ti to compete with the RX 480 at $200, they won't be able to justify the 1070 at $380.
Click to expand...

Has there ever been a GPU release without arguing?
 
Last edited:
Daniel_Chang said:
At that level of performance, the card falls within 25-40% of the GTX 1070 (not the 10-20% that some AMD fans are fantasizing over, but we'll see in a month or so). If NV cuts a GP104 to make a GTX 1060/ti to compete with the RX 480 at $200, they won't be able to justify the 1070 at $380.
Click to expand...

Does it? 34fps vs around 54 on a 1070, that's 38% slower than it.

The 1070 is 50% faster than it. At least in AotS, going by AMD's own numbers. A 390x is faster than this. Something ain't right
 
Ieldra said:
Does it? 34fps vs around 54 on a 1070, that's 38% slower than it.

The 1070 is 50% faster than it. At least in AotS, going by AMD's own numbers. A 390x is faster than this. Something ain't right
Click to expand...

The main issue is that we don't have a good 1:1 benchmark.

For example, the most recent 480 benchmark is a 1080p Standard at 59.1 fps. There are only two 390X that can be compared to the system, one at 64 fps and another at 60 fps. Yet there's things we still don't know, such as is the 64 fps running at stock speed or is it overclocked? Would there any benefits going from 0.94.18610.0 to 1.12.19928.0 for the latter benchmark?
 
Last edited:
Ieldra said:
Does it? 34fps vs around 54 on a 1070, that's 38% slower than it.

The 1070 is 50% faster than it. At least in AotS, going by AMD's own numbers. A 390x is faster than this. Something ain't right
Click to expand...

Alright check this out. Any possibility they were running this demo at 1085mhz? Because we are all assuming 1266 clocks that have been confirmed through other leaks. You think AMD running this with 1085 in crossfire with non finalized clocks? I really think that might be it, them running it around 5tflops then those numbers make sense.

I think I may have something here lol.
 
  • Like
Reactions: N4CR
like this
Ieldra said:
The 51% thing is just wrong, read the last posts from last page of this thread :p it will be very clear very quickly
Click to expand...

Just saw that, thanks. Very weird for them to say 1.5 then 1.8.... something isn't adding up then.. that's even less than 390 performance.

razor1 said:
4 is a given I think all people in tech should be anyways lol
Click to expand...
LMFAO

On point! Now go to bed, you drunken ass mofo.



NKD, you might be on to something... but doesn't make total sense for them to do that still!



And I actually want to play AOTS lol. Big total annihilation fan. Best RTS ever made in my books.. modded the crap out of it with units and it still runs 2560x1440 and looks good even today.. 1000 different units is not something you see in a RTS much if at all. All that from a title with interlaced VESA intros.. go figure.
 
Ieldra said:
Oh yes. I should apologize to you, because you posted correct information and I challenged it, right ? Oh wait. No. It's the other way round.




I feel pretty justified in getting angry when you tell me I miss the whole thing spelled out in the thread like I'm some kind of dimwit. I made the effort to read and understand before posting here. If anything, in the future, before accusing me of shit, GO READ THE DAMNED THREAD.

How many superfluous posts did I have to make, pointing out the obvious, hm? How many times did I post the same damn quote, in bold and underlined, font size 7, hm ?
How many times did I post the screenshot of the reply ?

Why did it take 7 or 8 posts for you to finally concede you were wrong ?
Click to expand...
look at this link AMD Announces The Radeon RX 480 Starting At Just $199

Your edit or addition of the edit isn't there. And you attacked me first by labeling me and acting like I had an agenda. It is obvious from the facts that you were the aggressor and even now are too full of yourself to admit you were wrongly accusing me even when I stated I didn't know about the edit.

It is obvious you have no desire to be a reasonable human being and that's ok, tells me all about the kind of person you are and that I would rather not have anything to do with you.
 
NKD said:
So has to be one of the following
1. the drivers are really bad at this point and they know it and got another month to do some optimization for this game.
2. may be they were using the 4gb cards and the 4gb memory is choking at 8x MSAA and crazy settings
3. Crossfire scaling is not 183% and they are just throwing a nice number out.
4. AMD fucked up the dx12 architecture with new front end lol. Which I highly doubt is the case.

I believe 1 to 3 are most likely than the last part.
Click to expand...

It still has not been confirmed whether Polaris with the 480 is based upon Hawaii or Tonga.
If it is Tonga that may explain the results, partially anyway.

Cheers
 
variant said:
You could compare a dual 480 to a dual 390X.

Here's a 1080p Crazy dual 390X and luckily, this same person has a 1080p Crazy with 8x MSAA.
Click to expand...

How do you k or they aren't dual 390s, possibly OC?

It's an old game version as well. Ideally we want to see a recent 1080/1070 review and hope to find a 390x 1080p Crazy preset result. Either that or we can ask someone on the forum to test for us

I could try downclocking my card as close to 1050mhz as possible and test that way
 
JustReason said:
Your edit or addition of the edit isn't there.
Click to expand...

I guess all the people who posted a screenshot of the comment, myself included, are playing an elaborate prank on you JustReason. We were hired by Jen-Hsun Huang to mess with you.

Even Robert Hallock is un on it. Every time you check the reddit thread, he edits his comment back to the initial '51% scaling comment', just to confuse you.

I love how you say 'your addition of the edit' as if I hacked into Robert Hallock's account and edited the comment myself, or maybe you think I, and the others who posted it, edited the screenshot lol.
 
Last edited:
Ieldra said:
How do you k or they aren't dual 390s, possibly OC?

It's an old game version as well. Ideally we want to see a recent 1080/1070 review and hope to find a 390x 1080p Crazy preset result. Either that or we can ask someone on the forum to test for us

I could try downclocking my card as close to 1050mhz as possible and test that way
Click to expand...

I think there is something to your point about 390x dual would be much higher than 1080FE.
I just went through various reviews where they have a reference 390x and applied the 1.83 crossfire figure.
When doing that a Xfire 390x in AoTS is anywhere between 20%-27% faster than a standard 1080FE.
So one should expect the 480 to have a similar trend rather than just matching 1080FE, considering this is the ideal game application for Polaris.

And I feel these are a better indicator as each individual review will have a better apples-to-apples comparison than what we have seen recently of trying to compare diverse-random results in the online Ashes table.
Also just to say the results skew if not using a reference 390x but a custom AIB as the performance gap increases a fair % (which one should expect).
Cheers
 
CSI_PC said:
I think there is something to your point about 390x dual would be much higher than 1080FE.
I just went through various reviews where they have a reference 390x and applied the 1.83 crossfire figure.
When doing that a Xfire 390x in AoTS is anywhere between 20%-27% faster than a standard 1080FE.
So one should expect the 480 to have a similar trend rather than just matching 1080FE, considering this is the ideal game application for Polaris.

And I feel these are a better indicator as each individual review will have a better apples-to-apples comparison than what we have seen recently of trying to compare diverse-random results in the online Ashes table.
Also just to say the results skew if not using a reference 390x but a custom AIB as the performance gap increases a fair % (which one should expect).
Cheers
Click to expand...

I'll be honest I've never seen a confirmed dual 390x result, but I very clearly remember the performance numbers from when aots first launched, and a 390x was on par with 980ti (approximately) in 1440p

Thats around 38fps with async enabled for both.

So *something* must be very wrong if two cards with 390x performance perform <10% higher than my OC 980ti.

Unless of course I'm getting 183% performance scaling from my overclock :p
 
Ieldra said:
I'll be honest I've never seen a confirmed dual 390x result, but I very clearly remember the performance numbers from when aots first launched, and a 390x was on par with 980ti (approximately) in 1440p

Thats around 38fps with async enabled for both.

So *something* must be very wrong if two cards with 390x performance perform <10% higher than my OC 980ti.

Unless of course I'm getting 183% performance scaling from my overclock :p
Click to expand...

Ah good point and a reason why older mGPU 390x results should possibly be ignored for comparisons when using AoTS, because that does mean the dual 480 has better performance than dual 390x, specifically mGPU in AoTS
I guess your point was translating that back to a single card.

Would be interesting to see if they got dual 390x working well now in mGPU with AotS.
I would be annoyed if I owned a 390/390x and this now only applies to Polaris with improved mGPU.
In theory 390x mGPU should be reasonably high and above 1080 in this game
Cheers
 
Daniel_Chang said:
It's disappointing the amount of arguing going on this this thread, especially after a moderator's warning. The RX 480 is great news if you're an AMD or an NV fan.
Click to expand...

Anyone who is an AMD or Nvidia fanboy is an idiot.

The RX 480 isn't very good news because it shows how noncompetitive AMD's next-gen chips are, 30-40% worse at the same TDP is pretty noncompetitive. We were at least hoping that it would show they were competitive by say, being at least as efficient, especially since that's what AMD has been hyping. That's why people are disappointed here. The Radeon RX 480 isn't all bad because it lowers the cost for the "value performance" segment, at $200 it will effectively replace the Geforce 950 as the "cheapest card worth buying". So, while there is nothing wrong with the pricing or positioning, I can totally understand why people are disappointed in AMD.
 
The weird thing is that in those old semi-leaked benches, the scaling in mGPU for rx480 was only about 1.5-1.7x (~40 to 60/66.6 in best case), never getting close to figure AMD guy mentioned.

And now he comes out and mentions 1.83x. Whoa?
 
  • Like
Reactions: N4CR
like this
CSI_PC said:
It still has not been confirmed whether Polaris with the 480 is based upon Hawaii or Tonga.
If it is Tonga that may explain the results, partially anyway.

Cheers
Click to expand...

I would imagine Polaris is based on Polaris (with Fiji being the most recent predecessor, not Hawaii or Tonga)
 
ecmaster76 said:
I would imagine Polaris is based on Polaris (with Fiji being the most recent predecessor, not Hawaii or Tonga)
Click to expand...

You realize Tonga and Fiji are one and the same right ?

I think what CSI was saying is, depending on whether the perf/w improvements were compared to Tonga or Hawaii, the performance we are seeing will either make sense or it will not
 
Quix said:
Anyone who is an AMD or Nvidia fanboy is an idiot.

The RX 480 isn't very good news because it shows how noncompetitive AMD's next-gen chips are, 30-40% worse at the same TDP is pretty noncompetitive.
Click to expand...

If you're comparing to the 1070 ...then the TDP is not the same. Unless AMD have managed to squeeze 8 pin juice through a 6 pin! The difference in performance (you claim 30/40% ...yet to be confirmed) is offset somewhat by the 50% difference in price ($200/+$400).

Trolling isn't your strong suite :(
 
fingerbob69 said:
If you're comparing to the 1070 ...then the TDP is not the same. Unless AMD have managed to squeeze 8 pin juice through a 6 pin! The difference in performance (you claim 30/40% ...yet to be confirmed) is offset somewhat by the 50% difference in price ($200/+$400).

Trolling isn't your strong suite :(
Click to expand...

No they both have 150W TDP as far as we know

TDP != power consumption
 
480X will max out at 150Watts power consumed. It's TDP has to be slightly less. No electric heater is 100% efficient. There are things like fans and LEDs etc....that don't generate heat.
 
fingerbob69 said:
Yes ...about 47w under load compared to the asus strix
GIGABYTE GTX 980Ti G1 & XTREME GAMING Review - Power, Temperatures, Acoustics

Although lets try and keep it all reverent to the cards at hand rather than exotic OC centric AIB specials.
Click to expand...

And you think this is because it has additional power inputs, and not because of the different clocks right ?

You can have 15 8-pin PCI-E power connectors on a single board, and power consumption will not change by one iota compared to a regular single 8-pin config
 
Ieldra said:
You realize Tonga and Fiji are one and the same right ?

I think what CSI was saying is, depending on whether the perf/w improvements were compared to Tonga or Hawaii, the performance we are seeing will either make sense or it will not
Click to expand...
Yeah and also:
It depends what the Polaris is an evolution of; either it is evolution of Hawaii or Tonga.
No-one seems to know yet.
And like you say Fiji is based upon Tonga, which was the last updated GCN released and 3rd generation, where as 390/390x is based upon Hawaii/2nd generation.

So when anyone says Polaris is based upon Polaris, its foundation/core has to be derived from either Tonga or Hawaii.
And this has implications when doing benchmark comparisons/expectations/modelling/etc.

Cheers
 
As an objective, dispassionate observer, I'm going to wait until some benchmarks come out and analyze the price vs. performance ratio and make a decision on that. I believe my 6950 is dying and was originally just going to replace it with a 970 and then decided to wait for the 1070 and do a Skylake system build but considering the amount of gaming I actually do nowadays (being an adult student has its drawbacks), I might not be able to justify the cost, especially when I'll more than likely only be doing 1080p gaming.

So if it's a solid performer and $200 or $230 for the rumored 8GB variant, it can fit my modest budget and regardless of the number of FPS deficit to the current nVidia card(s), it may still proudly find a home in my machine simply due to being an adequate performer for the price. Either way, I see this little price niche as a pretty good thing.

Time will tell at least...hopefully my ego doesn't get the better of me because that always gets expensive.
 
Ieldra said:
No they both have 150W TDP as far as we know

TDP != power consumption
Click to expand...
I fully disagree there. AMD didn't put exact specs on it but they mentioned TDP 150W. So we can't have card running at 5tflop and then 5.8 tflop have the same TDP. 5.8 has to higher than 150w if we are assuming 5tflop is 150w. So that shows that they are mentioning max TDP here depending on where the card is clocked. Kinda tired People saying that card at its lowest configuration will have 150w TDP
 
NKD said:
I fully disagree there. AMD didn't put exact specs on it but they mentioned TDP 150W. So we can't have card running at 5tflop and then 5.8 tflop have the same TDP. 5.8 has to higher than 150w if we are assuming 5tflop is 150w. So that shows that they are mentioning max TDP here depending on where the card is clocked. Kinda tired People saying that card at its lowest configuration will have 150w TDP
Click to expand...

So they mentioned tdp of highest clocked card but demoed performance of a lower/cut gpu?
 
You must log in or register to reply here.
Back
Top