All statements to the contrary.AMD does have a history of exaggerating things, y'know. If the entry point for VR is 970ish performance, then I suppose he could be talking about a decently cut P10 that's slightly worse than 970 performance for something less than what the 970 goes for.
Well, DP 1.3, HDMI 2.0a with HDCP 2.2 compliance, Full h265 hardware decoding is also there to look forward to.Even if Polaris 10 performed identically to my current 390 I'd still get one.
To me, the likely price difference between those two after a 390 resale is worth half the power usage (plus any potential performance increase).
AMD does have a history of exaggerating things, y'know. If the entry point for VR is 970ish performance, then I suppose he could be talking about a decently cut P10 that's slightly worse than 970 performance for something less than what the 970 goes for.
Nvidia's not launching a full Pascal chip for 1080Ti / Titan or something like that at the same time. Does that mean Nvidia's abandoning the high end? Or do you think it has more to do with process, and both companies slowly moving to bigger chips.I did say "up to". So far I'm not wrong. I certainly won't count a paper launch.
I wasn't expecting AMD to abandon the high end all together. I guess they need to focus on consoles.
It is said that AMD is planning a big announcement of Polaris architecture between May 26th to 31st. The event would take place in ‘Las Vegas of Asia’, which is Macau. This event could take place shortly before Computex.
AMD has allegedly already sent invitation to media outlets, however many have been omitted. This event would probably begin with a briefing for the press, and end with a press conference similar to Capsaicin.
So there’s a good chance Radeon 400 series based on Polaris 10 and 11 would make its first appearance at Computex presented by both AMD and its partners.
Well, I did not expect AMD to hold a separate Polaris presentation so close to Computex itself.
Depending on how optimistic one would want to be about the 400 series, it might mean anything from empty PR spinning hot air to a one-two punch of reveal/release through AIB channels to cut the legs from under the 1070.
The joys of rumour season.
I mean Nvidia jumped mainstream buyers who wouldn't have bought a 970. Now, Nvidia will try to get those people again with a 380 1070. This isn't Nvidia banking on getting people to purchase based on price range but on marketing.The 1070 and 1080 are now effectively $449 and $699 cards until AIBs prove otherwise and have good supply of $379 1070s. That means AMD has the entire $99-$329 market all to itself. Mainstream buyers don't just magically go from a $249-299 card to a $449 card. That's not even the same market segment.
The 1070 and 1080 are now effectively $449 and $699 cards until AIBs prove otherwise and have good supply of $379 1070s. That means AMD has the entire $99-$329 market all to itself. Mainstream buyers don't just magically go from a $249-299 card to a $449 card. That's not even the same market segment.
I think a more interesting question is if AMD raises Polaris 10 prices to fill some of that gap. The rumors and information we have suggest it should within spitting distance of the 1070, and unless they expect wide availability of the 1070 within 1 month instead of just limited amounts Founder Edition cards, they're leaving a lot of money on table if they only charge $329.
If there's enough availability I can easily see a lot of customized Polaris 10 GPUs that will be filling in that gap.
Exactly. AMD selling cheaper might still mean similar margins. Cheaper coolers also.Nvidia is using a nearly 40% larger chip though.
Nvidia is using a nearly 40% larger chip though.
I mean Nvidia jumped mainstream buyers who wouldn't have bought a 970. Now, Nvidia will try to get those people again with a 380 1070. This isn't Nvidia banking on getting people to purchase based on price range but on marketing.
I think that in general on an overall basis the 1070 will be better, but not by so much that it justifies a ~$120 price difference between itself and Polaris 10. Eventually when non-reference 1070's become available, about half of that gap vanishes and maybe looks more reasonable, but if the 1070 is only 10% better than Polaris 10 on average, it's a hard sell at $450.
Nvidia's not launching a full Pascal chip for 1080Ti / Titan or something like that at the same time. Does that mean Nvidia's abandoning the high end? Or do you think it has more to do with process, and both companies slowly moving to bigger chips.
You're being very vague. Let's talk die size as that is the only constant, terms like "high end" "mid range" constantly get redefined to suit a poster's agenda.
AMD was forced abandon 300mm2+ dies because they lacked engineering resources.
Going back to 5870, 6970, 7970, 290x, AMD has ALWAYS had a chip in the 300mm range to start off a new gen. Likewise, they have ALWAYS had a chip down in the lower 200mm2 range ready to go.
They never had to make this "choice". Its pretty obvious they had to develop only 1 chip and so they chose Polaris to focus on OEM/laptop/under $300 gaming crowd. You better believe if they had the money they would have had Vega out this June, yields are most definitely not a problem at ~350mm2 right now.
10%? They can be 25-30% slower and still have way better price/performance.
$299 = 100% base
$449 = 51% more expensive for 25-30% more performance is still worse value
I think you and I will have to agree to disagree on this point. I do not consider $380-450 videocards a mainstream market segment. At first, the hype that you can get a $380-450 videocard with Titan X performance will sound amazing, but then you realize for 1080p 60Hz gaming, a $200 card with ~ R9 390 level of performance is fast enough, so why spend the extra $?
Despite how popular 970/980 were on online forums, 970 managed to garner 5.10% on Steam in 1.5 years, while 980 and 980Ti sit at less than 1% each. 970 seems to be the exception, not the rule since the most popular cards are all in the sub-$300 price ranges per Steam:
http://store.steampowered.com/hwsurvey
For example, here in Canada if 1070 reference is $449, it'll show up as $449 x 1.30 = $584 + 13% tax = $660 CAD.
You think mainstream buyers spend that kind of $ on a videocard?
10%? They can be 25-30% slower and still have way better price/performance.
$299 = 100% base
$449 = 51% more expensive for 25-30% more performance is still worse value
$150 is still a lot of $ to spend. It's like going from an i3-6100 to almost an i7-6700! Someone building a new rig would be stupid as hell to get an i3-6100 + 1070 over an i5-6600K OC / i7 6700 + Polaris 10 for 1080p gaming.
The vast majority of PC gamers are also reluctant to admit that older CPUs are a severe bottleneck for 1080p with 980Ti level of GPU performance.
AMD was forced abandon 300mm2+ dies because they lacked engineering resources.
Within the present climate, then not true.No one reasonably expects value to scale linearly though. Everyone should know that if they buy at the high end the cost goes up more than the performance does. Not everyone buys that way, but you can't deny that if the performance gap is only 10%, that AMD is leaving money on the table at $330.
Anyone has an idea on HBM2 availability?I think Polaris was the focus because of Nintendo and maybe Sony. They are both purchasing a large number of chips. Assuming that Polaris is the basis for one or both of those consoles it is like printing money. I think that Sony wants to upgrade their console quickly and own the VR market, if they succeed they'll be essentially first to market and may reap benefits for generations from it. They are also putting chips into their VR headsets. How many sales does AMD have from these deals?
People are saying things about Vega being so far behind but if the performance of Pascal we are given is after the overclock then honestly they're not in much trouble there. Of course real world performance remains to be seen.
I have already made the choice to stick with AMD this round. I decided I want to replace my 3 monitors and I wanted freesync/gsync. The pricing difference in getting two monitors was in excess of 350 bucks more to go with Nvidia - without the GPU. Making Pascal an expensive proposition for me.
No one reasonably expects value to scale linearly though. Everyone should know that if they buy at the high end the cost goes up more than the performance does. Not everyone buys that way, but you can't deny that if the performance gap is only 10%, that AMD is leaving money on the table at $330.
That's where you and I differ. You think gamers will use logic and time and time again we see they vote with emotion.
Look at how the Nvidia marketing machine is positioning already before Polaris launch. I just don't see amd out playing amd, especially when the argument means gamers have to use logic.
The 970 proved that ~$300 is a sweet spot in the market and AMD may be able to capture a lot of sales by targeting that segment.
That would be hilarious if Vega [small] came out around the time 1080 really started to ramp.