Numbers don't lie, its widely reported market for gtx 970 and 290 cards is about 7.5 million at that price point. So you can only push vr to 7.5 million people at this time if you were to saturate that base. Now you have faster performance at 200 (if thats true) which one would you pick? Now you have all those other people that don't want to spend the kind of money on 970 or 1070. they have an option at 200. That is an insane amount of market share. All those people that are cash strapped will now be able to get faster performance at 200. It just makes sense. Imagine dell, hp, and other manufacturer putting these cards in and slapping vr ready logo on there.. for cheap.
![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
AMD Radeon RX 480 Specifications Leaked?
- Thread starter HardOCP News
- Start date
Numbers don't lie, its widely reported market for gtx 970 and 290 cards is about 7.5 million at that price point. So you can only push vr to 7.5 million people at this time if you were to saturate that base. Now you have faster performance at 200 (if thats true) which one would you pick? Now you have all those other people that don't want to spend the kind of money on 970 or 1070. they have an option at 200. That is an insane amount of market share. All those people that are cash strapped will now be able to get faster performance at 200. It just makes sense. Imagine dell, hp, and other manufacturer putting these cards in and slapping vr ready logo on there.. for cheap.
Sorry, fucking analogies are a pain in the ass, so lets skip them.
This card was built to win in the mid-range. What's your exact complaint? That they're choosing to compare it to last gen's high end cards? There are no new mid-range cards from nvidia yet, so nothing to compare to there. Using all the rumored info, If the "high end" polaris 10 manages to beat Fury and 980, and does it for $199 and with 150w, what should they compare it to, right now?
I guess I'm not clear on what your issue is.
edit:
Mid-range is 390(x) and 970, in my opinion, if you're well ahead of those, there's nothing to be gained by comparing to them.
lol don't worry about it man. People don't see the price they are just trying to stand up for their side. if its 200 and close to fury, thats is WTF performance per dollar right there.
Dude, cards priced like the 980/980ti are niche products. Cards priced at 150-250 are mainstream. If this can compete with the 980 and 980ti for 200-250, it's going to sell very well (unless Nvidia has a faster card at the same price or they sell the 980s at an even lower price.
In the videocardz benchmarks it showed around Fury level performance, above 390x and 980...
A Fury isnt that far behind 980ti/TitanX level, and the vram and likely overclocking would be equal this time around. It will be behind for sure, but at the $200-$400 segment, is it worth the extra $130-$180?
Last edited:
rat
Supreme [H]ardness
- Joined
- Apr 16, 2008
- Messages
- 4,915
Right on the front page:
So nVidia went right into the "new revision number upgrade" gimmick as well. Remains to be seen if they will get as much unnecessary shit for it as AMD did.
arthur_tuxedo
Gawd
- Joined
- Jan 8, 2003
- Messages
- 559
Yes, I didn't mean to imply that the higher-end segment doesn't matter, just that the <$250 segments represents the majority of sales.
The 970 is a bit of an outlier since second-tier cards are usually priced near $400 but it debuted at $330 with performance that was closer than expected to the top-end 980. When a card like that comes out, it's easier to justify stretching your budget, so the 970 absorbed a larger portion of sales than the 770 or 670.
arthur_tuxedo
Gawd
- Joined
- Jan 8, 2003
- Messages
- 559
Wow I thought I was behind the times with my GTX 460! I actually upgraded from a 9600 GT back in 2010, which was my first PCI-E card. I'm also considering stepping up to the latest AMD 4850 equivalent if it's all it's cracked up to be. According to the latest Linus Tech Tips video it seems like it will at least hit the $200 price point.
I don't see it that way at all. I see a card that's about 40% slower (think 390x performance) while operating at the same TDP, which indicates the card won't OC worth a damn. AMD also has almost no room to drop prices while still making a profit, so they will get squeezed hard by the 1060 when it launches.
Honestly, Polaris is not going to bring in a lot of money to AMD, and that's a major problem for the company going forward.
lolfail9001
[H]ard|Gawd
- Joined
- May 31, 2016
- Messages
- 1,496
Amount of people overrating and underrating the card is kinda hilarious.
I mean, i understand a whole summer month is a long time and knowing AMD's ability to get stuff to market i would not be surprised if it went longer cause of OEM needs, but at least wait for reviews. For now i expect something between 390-390x, as 3dmark leak suggests.
Just like 1070's, just in case.
I mean, i understand a whole summer month is a long time and knowing AMD's ability to get stuff to market i would not be surprised if it went longer cause of OEM needs, but at least wait for reviews. For now i expect something between 390-390x, as 3dmark leak suggests.
Just like 1070's, just in case.
viscountalpha
2[H]4U
- Joined
- Oct 16, 2011
- Messages
- 2,618
I'm predicting a price gouge for awhile because demand will outpace supply for awhile in Nvidia.
Manny Calavera
2[H]4U
- Joined
- Jun 10, 2004
- Messages
- 3,986
USD isn't what it was. I think this card(480) will sell reasonably well|as more then most are far more concerned with $/buying power, then anything else given the horrific state of the world's economy.
My last 3 cards (including my current card) were all nVidia. I don't know what my next one will be, but this card is roughly what I've paid for my last 4 cards and if the benchmarks are true, it's an incredible value. How can you bitch about a 200-250 card that competes with a card that sold for 600 a few months earlier?
nVidia may just lower the price of the 980s to compete until they put out new midrange cards, but that doesn't detract at all from this release (so long as AMD is making money at that price).
Arcygenical
Fully [H]
- Joined
- Jun 10, 2005
- Messages
- 25,066
Because some people don't care about value.
Because some people care about raw performance and generational increases.
Because some people desire the excitement of large speed jumps of the first unified shader cards.
Because some people are worried that AMDs forward looking architectures may fail their expectations, the way Polaris has.
Because some people are tired of waiting for a truly 4k capable card to come to market, without multi carding (myself included).
Because some people are worried about AMDs profit margins on a 200$ card (which have to be near nothing).
Because some people (such as myself) are greatly concerned with the cost cutting measures AIB partners, who are already shoddy partners, will implement to make more on a 200$ card.
Really, there are lots of reasons.
Ieldra
I Promise to RTFM
- Joined
- Mar 28, 2016
- Messages
- 3,539
There's also the fact that the multi-gpu scaling numbers AMD provided for the AotS lead to you the horrifying conclusion that the 480 is significantly slower than a 390x.
it's DX11 performance isn't very impressive either, and it should have been improved due to the new command processor
Ashes of the Singularity
Last edited:
People who are buying this card don't give a flying fuck about 4k. If you want that, you're not in the market for a $200 card.
People who buy this card have something like a 960 or slower and that means this isn't an incremental upgrade.
But I get your point. Nobody wants or buys a Toyota Camry, when for less than 3x-4x the price they can get a Panamera
horrorshow
Lakewood Original
- Joined
- Dec 14, 2007
- Messages
- 9,447
People who are buying this card don't give a flying fuck about 4k. If you want that, you're not in the market for a $200 card.
People who buy this card have something like a 960 or slower and that means this isn't an incremental upgrade.
But I get your point. Nobody wants or buys a Toyota Camry, when for less than 3x-4x the price they can get a Panamera
Well said.
And for chuckles, I own both a 960 and a Camry....
If you are attempting to scale the frame rate based on GPU utilization, no it doesn't. It would put it at Fury levels.
Ieldra
I Promise to RTFM
- Joined
- Mar 28, 2016
- Messages
- 3,539
1.83x mgpu scaling = 34 fps single gpu
No. 34 fps would be 62.5/1.83 and that's not how it works. You are attempting to take a frame rate that is already severely CPU bottlenecked and then divide it by 1.83? He said a dual GPU is the performance of 1.83x of a single GPU, a single GPU doesn't suffer only 51% GPU utilization under Single Batch.
My numbers are very specific based solely on GPU utilization.
Last edited:
lolfail9001
[H]ard|Gawd
- Joined
- May 31, 2016
- Messages
- 1,496
He said we saw the 1.83x the performance of single GPU, however, so 1.83x is not the "theoretical" scaling, it's what we saw, according to Hallock. And that's the main problem if i have with your math, admittedly, it actually goes against PR's guy words.
Unless he is really obscure in wording. And frankly, for AMD's sake, i hope you are correct and Hallock messed up the message.
Ieldra
I Promise to RTFM
- Joined
- Mar 28, 2016
- Messages
- 3,539
51% isn't gpu utilization.
that is how it works. if a single gpu gets 100fps, then 1.83x scaling means 183 fps mgpu
His exact words word, "All together for the entire test, there is 1.83X the performance of a single GPU in what users saw on YouTube."
If what you say is true, that 1.83X performance includes the CPU bottleneck which would make the performance of a dual GPU with explicit multi-adapter even more efficient once you remove the CPU bottlenecks.
Yes, it's GPU utilization.
"The single batch GPU utilization is 51% (CPU-bound), medium is 71.9% utilization (less CPU-bound) and heavy batch utilization is 92.3% (not CPU-bound)."
Source
Also, you just used a fancy way of saying 62.5/1.8=34 fps.
horrorshow
Lakewood Original
- Joined
- Dec 14, 2007
- Messages
- 9,447
You've racked up over a thousand posts in two months time.
I cannot say more without getting a demerit.
lolfail9001
[H]ard|Gawd
- Joined
- May 31, 2016
- Messages
- 1,496
Well, NVIDIA actually got SLI scaling to almost 100% in Pascal (above 100% in synthetics), i doubt EMA without CPU bottlenecks would be that much worse. I mean, heck, even Crossfire scaled ~2x in some cases ([H]'s review of RotR performance).
And btw, i have opinion that what was meant with that quote is not GPU utilization but amount of frames GPU bound, and that's different ballpark entirely. For all we know it could have near 100% GPU utilization at all times, but only 51% of frames with GPU bound.
Ieldra
I Promise to RTFM
- Joined
- Mar 28, 2016
- Messages
- 3,539
It is amount of frames cpu bound, which makes the image he posted all the more amusing
No matter how cpu bound it might be in mgpu, if it's 1.83x of single gpu, it's 1.83x of single gpu. It being cpu bound in mgpu has no bearing on it's single gpu performance.
What's fancy about this ?
1.83x mgpu scaling = 34 fps single gpu
You've racked up over a thousand posts in two months time.
I cannot say more without getting a demerit.
Here's to a thousand more posts
You mad ?
Last edited:
Ieldra
I Promise to RTFM
- Joined
- Mar 28, 2016
- Messages
- 3,539
horrorshow
Lakewood Original
- Joined
- Dec 14, 2007
- Messages
- 9,447
Here's to a thousand more posts
You mad ?
Not at all sir.
It would just be nice to click on a "thread of interest" without having to scroll through so many posts between you and variant.
It feels more like a chat log than a thread.
Ieldra
I Promise to RTFM
- Joined
- Mar 28, 2016
- Messages
- 3,539
Would you rather scroll through a thread brimming with misinformation?
Are you really attempting to say that despite AMD saying, "single batch GPU utilization is 51% (CPU-bound), medium is 71.9% utilization (less CPU-bound) and heavy batch utilization is 92.3% (not CPU-bound)", that what he really meant was that it's really 91% GPU utilization?
horrorshow
Lakewood Original
- Joined
- Dec 14, 2007
- Messages
- 9,447
Petty squabbles between children are of no interest to me.
lolfail9001
[H]ard|Gawd
- Joined
- May 31, 2016
- Messages
- 1,496
You see, the problem here is that the way these numbers are presented it really does sound like these numbers are taken straight from AotS benchmark run. And they are listed as "frames GPU-bound" on benchmark screen itself. I mean, do you believe AMD would have some sort of GPU-load tracking software running, tracking batch timers and then presenting averaged utilization figures, for a petty demonstration of how 2 $199-249 GPUs are more cost efficient than $700 GPU?
Maybe, maybe not, but does that actually change anything other than make it more likely for a linear fps scaling? Wouldn't 51% frame GPU-bound be 51% of the possible frames the GPU can output?
I think they got the 1.83x from an average or just a rough estimate of the inefficiency of the explicit multi-adapter. Heavy Batch without being CPU bound is 92.3%, which is comparable to the 91% GPU utilization that 1.83x would indicate.
Ieldra
I Promise to RTFM
- Joined
- Mar 28, 2016
- Messages
- 3,539
I think they got 1.83x scaling by calculated the ratio of mgpu:single gpu performance. As one does.
51% of frames gpu bound could mean the gpu utilization is 95% when the frame is cpu bound.
I've looked at a number of AoS benchmarks, many at 98% GPU bound. It just seems like a strange way of saying GPU utilization. I doubt AMD made the mistake three times, from two different people.
If a batch frame rate is 95% GPU bound, it means it is 5% bound by something else, either explicit multi-adapter inefficiency, driver overhead, or CPU.
Last edited:
horrorshow
Lakewood Original
- Joined
- Dec 14, 2007
- Messages
- 9,447
N4CR
Supreme [H]ardness
- Joined
- Oct 17, 2011
- Messages
- 4,947
Money is a band aid fix for AMD. Even if e.g. they released Polaris and it shit on 1070, Vega comes out early and takes out 1080 and 1080taxi edition etc (best case).
That cash influx is nearly irrelevant with their debts. They need market share, long term stability and profitability to turn things around. They need to grow the PC market base for more awareness. They own the consoles already.
Predicted it and said it many times: Polaris is/will be a loss leader. It's how business has to work sometimes.
A halo card that's a fraction of a percent of users is irrelevant to profitability, more mostly relevant to marketing.. but of course Lelzor1 will claim mainstream market is tiny and some bs like 99% of the market is enthusiast performance cards for the 20th time. Because everyone drives hypercars to work.
It's quite funny seeing the dynamics of certain members here. Someone posts something remotely positive for AMD in whatever thread, incomes the usual suspects 'NO THATS WRONG ITS IMPOSSIBLE BAIT AND SWITCH AND DETRACT'. It's like they're paid to do it. 2k posts in a few months says it all.
Unemployed, retired or paid to post?
Assuming it's a loss leader, what's it a loss leader for? If I buy an AMD card this summer, when are they going to make a profit from me? I'm not going to replace that card anytime soon (more than likely) and even if I do, there's no guarantee that I'll buy another AMD card. Obviously some here are very loyal to nVidia or AMD, but I have no allegiance. I don't even have an allegiance to Intel, other than the fact that AMD hasn't been as fast/cool as Intel since the Core2 release (and at this point, I don't think they ever will)
2000? Wow. I think I post a lot and I have not more than 160 posts in the last month (prior to this one
)