Nah, I think just because it's the Titan and they can.
Plus it gives marketing a field day pushing HBM as an extra reason to buy the card. You know, the whole "OMG SO MANY GIGABYTES OF BANDWIDTH" thing.
-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
Radeon RX 480 "Polaris" Launched at $199
- Thread starter Kaapstad
- Start date
Nah, I think just because it's the Titan and they can.
Plus it gives marketing a field day pushing HBM as an extra reason to buy the card. You know, the whole "OMG SO MANY GIGABYTES OF BANDWIDTH" thing.
AMD have the advantage of knowing the exact performance to beat, so my guess full vega11 should be above 1080, unless AMD is going to a completely different performance range independently of nvidia's offering.
personally i believe there will be a gddr5x vega11 during Q4 2016.
because the faster they release, the more pressure they will put on nvidia to release GP102 to take performance crown back, which would give them a good idea to where vega 10 should be, actualy i think we might just see a 2 thousand dollar GPU from nvidia by 2018 with a GP100 at 600mm² as a response to vega 10.
thats just another benefit from AMD's target to mid range first, it turned to table compared to what used to happen to them during last generation, where nvidia had performance crown, with bigger chip in the pipiline waiting to over take AMD's newest offering performance.
they just needed to make the mid range attractive enough by agressive pricing and overall good performance, OC is the cerise on top of the gateau.
Last edited:
Soldato
- Joined
- 28 May 2007
- Posts
- 10,207
The full on GP100 has a TDP of 300 and I think that is using HBM 2. This is a professional card so will be running at lowish clocks compared to what it's capabe of. If the GP102 is not using HBM 2 and is a relative beast power usage could well be high.
300w at 1.6ghz and it is denser than gp104.
GP102 won't have the DP cores though, so it'll be a completely different chip in terms of Tflops per watt (single-precision). GP100 is actually complete trash for single-precision Tflops per watt (for the process its on).
It'll essentially be a big 1080, so it could just be ~4500 cores at ~1.5 GHz and get away with no HBM2.
Good. Im glad we agree
I think the point was being made, with so few numbers, Nvidia hasnt really "launched" yet and just did it so they could be first to the market.
Can we get back to the 480 hype train now please?
Yes CONFIRMED nvidia 1080 1070 launch re stock utter rubbish
Now all back onto the train please destination AMD hype valley
Soldato
- Joined
- 28 May 2007
- Posts
- 10,207
Yes CONFIRMED nvidia 1080 1070 launch re stock utter rubbish
Now all back onto the train please destination AMD hype valley
Let's all get off this train for a bit. The way this hype is going the 480 could beat the 1080 and still be disappointing due to the hype that's built up to high. Take a breath then a step back.
Yes CONFIRMED nvidia 1080 1070 launch re stock utter rubbish
Now all back onto the train please destination AMD hype valley
Leaving the platform. News just in, 480 beats 1070 at stock, Choochoo.
I don't get that impression.
As far as I can see the hype is at stock the 480 will match/slighty beat a 980/390X, and then it MIGHT be able to OC quite well and MAYBE catch up with a stock 1070.
Doesn't seem that mad to me.
As there are cards out in the wild, Nvidia will know exactly how they perform. They'll have "aquired" some and fully examined them.
Soldato
- Joined
- 28 May 2007
- Posts
- 10,207
What about overclock past the gtx1070. A few pages back there was 1800 core and beating a 1080. The overclock to a 1070 could be possible but I won't be holding my breath.
Yes this for sure ...
Well by what we think we know, the 480 would be a bit ahead in IPC. So to beat a 1070 it needs to get near the 1070's clocks.
In other words, 1500 core 480 should roughly match a 1070 and 1600 core should beat it (stock 1070 obviously).
Trains certainly picking up speed again lets shove even more coal in ... at this rate we should be passing the 2500mhz mark any time soon
lol 2 weeks ago you were hoping for it to beat 390, now you are even considering the 1070 match up possibility
no one escapes the hype train
There's the 1060 that was rumoured for a July release.
Soldato
- Joined
- 28 May 2007
- Posts
- 10,207
lol 2 weeks ago you were hoping for it to beat 390, now you are even considering the 1070 match up possibility
no one escapes the hype train
No I said it's possible. I never said I am expecting it. I think you misunderstand. I am trying to get people off the hype train not further it. At 390x speeds or above it will be a good card for the money. Anything above is a bonus. If it overclocks well then that's an even bigger bonus.
Amd made it look slower than a 390 by putting out that vr test result. All the other leaks suggest better so I am hoping for 390x or above which I think it will achieve.
Last edited:
Caporegime
- Joined
- 18 Oct 2002
- Posts
- 33,188
But you do and you know they havent sold many?
You act like you know it all but we are both ******* in the wind here.
All i have said is that nvidia have released first and will have sold 1000's (at least as it is obvious)by now. Im not claiming anything else.
No I don't, too many take something simple, pointing out that them showing as in stock is zero indication of how many are in stock as me claiming the opposite, that i know there weren't any in stock.
Or that is how most people on here seem to argue. I said what I said, I didn't imply I knew there weren't 100s in stock just that being in stock wouldn't tell you if there were 11 of 11000 in stock, presuming more than the former with every bit of information out there is frankly ridiculous. There are dozens of reasons to believe there is extremely little stock out there, your one reason for believing there is lots of stock is a symbol which means 10+ stock and extracting that it means 100's were in stock at any given time.
Every single AIB has missed it's supposed due dates, pretty much by a month and only if the July 1st date is real which we have no reason to believe it is.
The first chips seen for GP104 were mid april, at the time I said based on everything I know about chip production, every other 'normal' launch that real supply would be around 3 months after that date presuming those chips go into full production within 2 weeks or so of that mid april date. It could be that the chips work but have such dire yields their only option is respin and could be longer than 3 months.
I also said AT THAT TIME that the other possibility is risk production and they could have anything from 10-100k chips(more is frankly too much money on wafers to risk even for Nvidia) ready within a couple of weeks. I literally identified that as a possibility before Nvidia announced the 1080 or the May 27th date or the first card was available.
Everything suggests tiny initial supply and no massive upshift in availability. Every single indication from every single source suggests such an initial low capacity and it's a well known and previously done method of jumping to market early at the expense of the risk involved, hence risk production. Producing 5 or 55million of wafers without actually having the first chips back and knowing they work is a risk, if it pays off you get an early single batch of chips and production of a new batch starting after you test these chips take 3 months to produce(maybe a little less if paying for hot lots which increases costs but gives your wafers priority and thus a little less waiting and faster through the production line).
Have you forgotten about the Fury launch? There were so few cards available on launch day it was complete joke.