What does Intel sacrifice.....to get 65w?

[Sardonicus]

What does Intel sacrifice on the i7 3770S to get it to run at 65 watts?
Intel® Core

The reason why I ask, is because my current i7 2600 CPU has a TDP of 95 watts, and if I can get a quad-core processor that will knock the TDP down to 65 watt, it's worth it to me to switch. The 30 watt power savings, I'd enjoy that.

I was also looking at the other i7 processors. Compared to the i7 3770 and the i7 3770K, I really don't see much difference other than the TDP. So if anyone can see a disadvantage of using the "S" type processor, please let me know.
Compare Intel® Products

 

[ignatzatsonic]

Those TDP figures are maximums, aren't they?

I don't know, I am asking, but have always assumed they were maximums.

If so, I'd take that to mean the actual power savings when not running at a very high load would be noticeably less than 30 watts?

Accepting the 30 watts as accurate, at my "hours on" per day and power rates (13 cents per KWH), that comes out to about 17 bucks per year.

Might be worthwhile----not that you need an economic justification to upgrade.

 

[essenbe]

A review on the i7-3770K at Tom's Hardware.

 

[DeaconFrost]

I would think it would take a very very long time to make up the cost of the new processor with a possible 30 watts of power savings.

 

[tritous]

it would take around 15 years, far longer than the processor would last.

However, I can answer your question in full.

The new processor is using the Tri-Gate transistor technology: the first major change in the construction of how a transistor is made in over 50 years. The entire transistor design is geared to faster switching, less leakage, lower power consumption...and on top of that they have produced it at 22nm (something feared to be impossible for regular MOSFET construction).

The new processors I have to say disappoint me hugely. They have the potential to be 50% less power, 30% faster, all the while fitting on more cores. Intel has done a low risk move because of lack of competition: the new i7 with the new transistors is almost identical in design except smaller and with a better graphics section. All it means is that they can pump out more processors from each silicon crystal (it's the area that costs) so they make more profit. The size of the die on the processor could easily take another 12 cores, and probably still keep within 100w. They could have clocked it up to past 5GHz and still kept the heat down. Instead they kept it as is and kept the power down: something we shouldn't be caring about for a PC: the i7 is basically a laptop chip anyway.

I'd recommend waiting for Intel to get it together and put the new tech to real use, but they are not very interested in serious desktop processors any more. I'd also recommend not worrying about the power consumption for a PC, not when we are putting in 300+w of graphics cards, double that for some people. In my opinion a desktop's concern is heat, not power, and 90w is handled even with some of the cheaper heatsinks out there. the only time 60w could be seen as important is if you are using the intel stock cooler, hammering it hard all the time, and refuse (or cannot fit) a 3rd party heatsink for £15.

Not that savings for nothing is a bad thing: but when turning off your pc for 1 hour per day saves you twice that, we know who the blame is with. On that note: measurements for my old Q9400 show that my actual power consumption is between 5w and 21w depending on usage: this is on a 95w rated CPU.

 

[pparks1]

I'd recommend waiting for Intel to get it together and put the new tech to real use, but they are not very interested in serious desktop processors any more. I'd also recommend not worrying about the power consumption for a PC, not when we are putting in 300+w of graphics cards, double that for some people.

Well, if Intel doesn't care about serious desktop performance anymore...then we don't really have anything to choose from...do we? I mean, I seriously haven't even considered an AMD chip since the days of the Athlon 64 X2 back when they were kicking butt on the P4. Since the Core 2 release, Intel has been the clear leader.

Personally, I'm still using a Q9550 quad core myself as I haven't a demonstrated need for anything more powerful. This chip handles BF3 beautifully and everything else I throw at it. < So, I guess the point I am making is that maybe we are not at the point where we need 12+ cores and over 5Ghz of performance.

Obviously, the manta among the tech enthusiasts is always more, more more and faster, faster, faster. Often it's to the point of obsession. We see it with CPU(s) obviously. We see it with power supplies (often way overpowered for the sake of "what-if's). We see it with cooling and case fans (even though we are well within heat specs, we try to drive it down for the belief that it's going to make our hardware last so much longer. We see it with benchmarks (we want to improve our numbers for the sake of benig better, even if it doesn't really equate to any real-world performance).

Now, the latest craze power consumption. I'm all for lower power consumption. But as others have said, to upgrade mainly for the power savings isn't wise. I've heard of people moving from standard hard drives to SSD's for power savings. The amount of power saved is really small. It would never pay for itself. An SSD should be purchased for it's speed and extremely low access times. But if somebody were just building a new PC, then I wouldn't hesitate to evaluate the lower power options. I'm getting ready to retire my 6 year old file server and am looking at the Core i3-2100t for it, as it runs at 35w max. That's good for power and heat and still provides way more CPU power than a file server would need.

 

[tritous]

Well, if Intel doesn't care about serious desktop performance anymore...then we don't really have anything to choose from...do we? I mean, I seriously haven't even considered an AMD chip since the days of the Athlon 64 X2 back when they were kicking butt on the P4. Since the Core 2 release, Intel has been the clear leader.

A complex question. as far as processors go: AMD has been left behind at 55nm, the Core 2 yorkbridge (you have just a die batch newer than mine) isn't bad. As you said you've not felt the need for better. Since then...not a huge amount has really changed. Some power management, onboard graphics processing (not a comparison to discrete graphics still), etc. AMD simply can't keep up with 45 and 33nm techs. at 22, not even ARM can compete with Intel.

The problem is the i3, i5, and i7 are....mobile processors. They could have kept giving us better speeds, but instead stalled the technology world because they no longer had to race. so they made the same chips for cheaper and cheaper. Made it more power efficient for mobile devices, made it more integrated for mobile devices, then stuck it in a different case and called it a desktop processor.

ok, we don't need 5ghz processors and 16 cores....but wouldn't you like a 6 core for the same price as a current 4? Then you'd like an 8. make it 10% faster at no extra cost. We could have been getting this for years. They could quite easily make a 16 core 5ghz processor but have held it back because there is no one forcing them to: people are happy enough paying as it is. Instead we just go get a drink from the kitchen while we wait for our pc to start up, or go to the loo while it logs in. We could have been at speeds where we barely have time to shift in our seat before its ready. Of course...that's provided there was RAM speeds for it: or they would need to define an industry standard for "quad channel".

And that is within the bounds of "sensible". if you want to look at the overclockers world I couldn't even guess would could be done with the currents these new transistors are capable of. I don't like big coolers either, the big towers terrify me when i hear reviews saying the board starts bending. That's why when my stock intel fan started to go and the heatsink was getting clogs in places i couldn't reach with a tooth brush i went for the Arctic cooling freezer 7 pro: it was 15% heavier than the intel heat sink, wasn't a huge amount bigger (although still broke some mental barriers), and was very cheap. I have to say it's impressively easy and light, it's noticeably quieter (than a dying fan at least), and it's very cool (gone from a peak of 47C down to 40C).

On that light we may have missed one point in this discussion: cooling = lifetime. I've always maintained that th thermal cycling of electronics DOES affect it's lifespan and increases the risk of degredation. By that standard a lower power CPU or a better cooling is a worthwhile investment. Of course I wouldn't go spending £100s on a cpu or £70 on a cooler, but when it's the difference between a £10 stock cooler and a £15 super cooler it makes you start to think.

 

[ignatzatsonic]

They could have kept giving us better speeds, but instead stalled the technology world because they no longer had to race..........but wouldn't you like a 6 core for the same price as a current 4.............They could quite easily make a 16 core 5ghz processor but have held it back because there is no one forcing them to............We could have been at speeds where we barely have time to shift in our seat before its ready.

Who might force Intel?

Given the current tasks asked of processors, I'd say the prodding would have to come from purchasers, rather than AMD.

How much market share is to be gained by the next 50% increase in CPU speed?

Does anyone have stats on the rate at which hardware is refreshed compared to say 8 or 10 years ago?

I'm not sure the average buyer (consumer or enterprise) is demanding more power as incessantly as a decade ago.

There is such a thing as "good enough", barring new more-intensive tasks.

 

[tritous]

a decade ago there was competition. Actually the average obsolescence age of a computer part was under a month, and within 6 months there would be at least 8-9 items that completely blew it out of the water. a 50% increase in performance between two produces a few months apart was mandatory even to maintain market share, let alone gain it. Frustrating days because things moved so fast, but exciting.

"How much market share is to be gained by the next 50% increase in CPU speed?". I think you hit the nail on the head there. We can ask all we like, we'll take what we are given. There is basically no alternative, and so there is no market gain at all if Intel ups the quality: they are practically at 100% anyway.

There may be such a thing as "good enough", but until we are at the point where things are instantly perfect we should always keep aiming for better than now. Once we are satisfied all progress stops.

 

[pparks1]

On that light we may have missed one point in this discussion: cooling = lifetime. I've always maintained that th thermal cycling of electronics DOES affect it's lifespan and increases the risk of degredation.

Sure, but these electronic devices are intended to get warm too. My Q9550 is overclocked slightly, from 2.83Ghz to 3.40Ghz and I'm using the stock Intel cooler. It's about 3 years old now, stable as can be and no problems. Sure, I could go with liquid cooling and run it even cooler, but is there a point? Let's say this CPU lasts 7 years right now instead of 11. If I'm not going to use it longer then 4 years, who cares if it could have lasted longer. But some people will go to great lengths to run it cooler in an effort to make it last longer.

 

[tritous]

ah yes, the old "I don't plan to keep more than 5-6 years" argument. There's no real counter to that, in fact by that measure i'm daft for not clocking my pc up a notch or two. It does seem to be headed now to the point where they are supporting overclocking so much that we are almost obliged to do it. Maybe that is the true future of desktop cpus: they are getting faster and faster as the cooling improves and the OC facilities improve, it's just that we don't get any warranty with that.

 

[ignatzatsonic]

There may be such a thing as "good enough", but until we are at the point where things are instantly perfect we should always keep aiming for better than now. Once we are satisfied all progress stops.

As a highly abstract ideal, I might agree. But capitalism (AMD, Intel, whoever) doesn't use that yardstick or share that opinion. The things have to be sold----at some price/volume that keeps the business going.

If you can convince/deceive buyers through marketing, so be it. That seems to work well for certain tech companies I won't name.