And it also says they may choose to favour lower power consumption over a pure speed boost so it may be even less than 45% at the end of the day.
Like I said, we'll have to wait and see.
-
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.
7xxx series cards
- Thread starter Compound
- Start date
And it also says they may choose to favour lower power consumption over a pure speed boost so it may be even less than 45% at the end of the day.
Like I said, we'll have to wait and see.
From what I've heard about the 7xxx they're mostly looking at power efficiency. I saw something about 69xx speeds with 50% power consumption - though I'm unsure where that slots in, likely to be 78xx or maybe even 77xx. I'd say more likely looking ~69xx performance on the 78xx and then a small step up for the 79xx.
No definite source on anything though, it's all small tidbits of information.
No definite source on anything though, it's all small tidbits of information.
Soldato
- Joined
- 30 Sep 2005
- Posts
- 16,738
not bad in all honesty, will hopefully mean a good bit of overclocking, lower temps, less heat in the case and quieter fans. Wonder if there's any improvements for crossfire?
or.....they might be able to make a proper dual card with two full fat gpus on
or.....they might be able to make a proper dual card with two full fat gpus on
Caporegime
- Joined
- 18 Oct 2002
- Posts
- 33,188
Do people really learn nothing from the same news every generation, I mean really nothing?
HD7000 series is NOT focused on power saving, EVERY generation the midrange can offer roughly the same performance for half the power and roughly double the performance(its usually 70-85%) for the same power usage. This is entirely no indication of aiming for power saving, the only gen this hasn't happened is hd6000, and that's entirely down to the 32nm process being canceled.
Likewise, when an article states a speed improvement and is talking about a PROCESS, it has NOTHING to do with the performance of any parts made on that process, its talking about the process.
Intel has said 22nm will be up to 37% faster than 32nm, yet Ivy bridge will only be 10% faster and only 5% of that is from clock speed.
Process speed and actual chip speed aren't comparable, have no link to each other and you can't read anything in to it.
As with anything else, that 45% will be very much best case scenario, on their lowest power chips, and won't be remotely relevant to high end 250-350mm2 gpu's.
Look at Intel's graphs for 22nm, they are available all over, the 37% is how fast the transistors can switch at their lowest voltage. IE idle speeds can improve with same power usage, and low end mobile chips, their ULV stuff that at the moment is now going in mental powerbooks will gain a decent amount of speed at low voltage. At high voltage the improvement in speed is 20-25% lower than best case scenario and that still has very little actual effect on the chips themselves.
I can see a new better process helping Nvidia a little more than AMD as its rumoured around 80-100W of the 480gtx/580gtx power usage is purely from leakage. The problem comes from the fact that, the midrange is essentially the same transistor count as the previous high end chips, so you have similar characteristics, when you get 30-40% less leakage and a smaller chip you can drop power usage in half. When you double the transistor count, which is roughly what happens with the high end, there might be less leakage per transistor........ but you've got double the number of transistors.
New process's and the 45% faster and the 40-70% less leakage really only balances out the doubling of transistor count, the doubled number of transistors that get leaky and the minor bump in clock speed.
Thing is, this stuff is the same for every new process, be it Intel, AMD, GloFo, Tsmc, IBM, Samsung or just about anyone else.
32nm from Glofo will have the same marketing guff, hugely lower leakage, much faster speeds, so will 32nm from Intel, its really not hard to read between the lines just by remembering the marketing guff from previous launches of processes, and then seeing what chips you eventually get.
HD7000 series is NOT focused on power saving, EVERY generation the midrange can offer roughly the same performance for half the power and roughly double the performance(its usually 70-85%) for the same power usage. This is entirely no indication of aiming for power saving, the only gen this hasn't happened is hd6000, and that's entirely down to the 32nm process being canceled.
Likewise, when an article states a speed improvement and is talking about a PROCESS, it has NOTHING to do with the performance of any parts made on that process, its talking about the process.
Intel has said 22nm will be up to 37% faster than 32nm, yet Ivy bridge will only be 10% faster and only 5% of that is from clock speed.
Process speed and actual chip speed aren't comparable, have no link to each other and you can't read anything in to it.
As with anything else, that 45% will be very much best case scenario, on their lowest power chips, and won't be remotely relevant to high end 250-350mm2 gpu's.
Look at Intel's graphs for 22nm, they are available all over, the 37% is how fast the transistors can switch at their lowest voltage. IE idle speeds can improve with same power usage, and low end mobile chips, their ULV stuff that at the moment is now going in mental powerbooks will gain a decent amount of speed at low voltage. At high voltage the improvement in speed is 20-25% lower than best case scenario and that still has very little actual effect on the chips themselves.
I can see a new better process helping Nvidia a little more than AMD as its rumoured around 80-100W of the 480gtx/580gtx power usage is purely from leakage. The problem comes from the fact that, the midrange is essentially the same transistor count as the previous high end chips, so you have similar characteristics, when you get 30-40% less leakage and a smaller chip you can drop power usage in half. When you double the transistor count, which is roughly what happens with the high end, there might be less leakage per transistor........ but you've got double the number of transistors.
New process's and the 45% faster and the 40-70% less leakage really only balances out the doubling of transistor count, the doubled number of transistors that get leaky and the minor bump in clock speed.
Thing is, this stuff is the same for every new process, be it Intel, AMD, GloFo, Tsmc, IBM, Samsung or just about anyone else.
32nm from Glofo will have the same marketing guff, hugely lower leakage, much faster speeds, so will 32nm from Intel, its really not hard to read between the lines just by remembering the marketing guff from previous launches of processes, and then seeing what chips you eventually get.
Soldato
- Joined
- 22 Nov 2007
- Posts
- 3,758
- Location
- Ayr, Scotland
To be honest, I dont care about all that guff drunkenmaster spouts (no offense intended ) about processes etc and i do understand some of you guys get off on that stuff (again no offense intended ) all I'm interested in is a decent performance boost from the previous generation (bearing in mind the dissappointing 'boost' in the previous change 400 to 500 series for Nvidia and 5*** to 6*** for Amd ) and value for money. If the 'new' mid range 7*** series are better than the current 'high' end and priced correctly ie £180-£220, then I'll be happy.
Soldato
- Joined
- 22 Nov 2007
- Posts
- 3,758
- Location
- Ayr, Scotland
Sorry this came to me after my last post. This 'revised' 560Ti from Nvidia, where does it fit in exactly. With 'allegedly' near 570 performance and power rquirements and current 560Ti still costing anything from £170-£230 and then the 570 priced from £235- £300 what will this card be pitched at price wise. Has to be cheaper than than the 570 at say no more than £220 but more expensive than the current 560Ti. What will this card do for current 560\570 sales. Can it be a 2Gb version or is that not possible given it appears to be based on 570 architecture, sorry if this is off the original op.