Desktop : Higher performance or higher efficiency ?

[Arkadrel]

The people talking about ITX have good points; I almost don't consider them to be in the desktop category, but one of their own that is in between desktop and mobile devices.

Small doesnt have to mean slow.

^ H80 cooler.

CPU: 2600K @ 4.6
RAM: 8Gb 1866 9-10-9-27
GPU: 670 4Gb
PSU: 500w
Cooling: H80 with push pull Gentle Typhoons AP-15

 

[EdgeOfDetroit]

Planetside 2 is my game of choice, and its 100% CPU-bound on my 2600K. I have a lot of CPU-hogging features turned off, and its still CPU bound. There is definitely room for faster CPUs in gaming, anyone who says otherwise just isn't playing the right games.

Therefore, the first option, more performance. Without question.

 

[crashtech]

Small doesnt have to mean slow.

^ H80 cooler.

That's pretty cool, thanks for sharing the photos. Certainly when I opined that ITX is in a different category, I didn't mean to imply that they were slow, just that the form factor opens up different possibilities than the traditional desktop, but faces different challenges, too. The more I see these nifty ITX builds, the more it makes me want to try one, but probably not with an overclocked i7. Good on you for making it happen!

 

[BrightCandle]

More performance. There are still plenty of games that are CPU limited to lower than 60fps and plenty of programming problems that take vast amounts of CPU and today are big slow batch jobs. Compilation is still too slow, I need faster, it will make me more productive.

 

[vipirius]

the need aspect yes.

The greed aspect no...
Everyone wants to be 1ghz higher then the next person who posted while holding the same voltage.

No, maybe every enthusiast wanys that, but the average joe that uses his pc to browse facebook and watch netflix probably doesnt know or care what his processor's clock speed is. Don't get me wrong, I would also love to have ultra fast cpu but let's gace it, we are a minority and it just isn't worh Intels resources to cater to us.

 

[jaqie]

I want both, with a slow reduction in power use over years of development/upgrades. Kind of inbetween the two poll options.

 

[ShintaiDK]

I want both, with a slow reduction in power use over years of development/upgrades. Kind of inbetween the two poll options.

Thats what most want and what we get. But the poll was biased to avoid that option.

 

[Idontcare]

I want both, with a slow reduction in power use over years of development/upgrades. Kind of inbetween the two poll options.

Given that this would describe Intel's roadmap for the past 6-7 yrs, there is probably a rather specific reason why it was not captured as a voting option by the poll. :whiste:

 

[Ajay]

Thats what most want and what we get. But the poll was biased to avoid that option.

I don't think that is what most enthusiasts want. It is what we have gotten though. When you look at Intel's xtor performance curves, increasing clocks decreases the switching speed advantage offered by their current processes.

I'd like a 4 GHz Haswell to make my 4 GHz Nehalem look like it was standing still. Sadly, that's not the case.

What ever I want, as CPUs shrink, they are going need lower total thermal output otherwise they will become self igniting. At the moment, physics sucks.

 

[crashtech]

I don't think that is what most enthusiasts want. It is what we have gotten though. When you look at Intel's xtor performance curves, increasing clocks decreases the switching speed advantage offered by their current processes.

I'd like a 4 GHz Haswell to make my 4 GHz Nehalem look like it was standing still. Sadly, that's not the case.

What ever I want, as CPUs shrink, they are going need lower total thermal output otherwise they will become self igniting. At the moment, physics sucks.

It may not be what we want, but as you say at the end, we don't really have a choice. Any big jump will have to wait for materials science to catch up, or immerse our systems in liquid helium.

 

[AtenRa]

Thats what most want and what we get. But the poll was biased to avoid that option.

Given that this would describe Intel's roadmap for the past 6-7 yrs, there is probably a rather specific reason why it was not captured as a voting option by the poll.

I can give you the same performance at half the power usage and you will jump all over the place and say that intel raised its performance/watt. Thats not what High-End desktop users want.

We want the same power usage(130W) but 50% more performance with every new node at the same price point. That still get you higher performance/watt at the same price point 😉

 

[Revolution 11]

AtenRa, I picked the performance option myself but let's face the truth. If Intel could cram 50% more performance at a given TDP, they would do so if only for the desktop market.

Instead, core number is limited by Amadahl's law, core size is limited by law of diminishing returns, and clock speeds are limited by both power usage and heat density barriers. Intel has dedicated a lot of funding and time into maximizing IPC and introducing new instructions when possible.

What more can they really do? All the other options have significant drawbacks. How would they deliver 50% more performance without either a increase in cores which only benefits multithreaded code (not ubiquitous) or increase in clock speeds (requires expensive cooling and power delivery).

I want what you want. I just don't see how it is possible with current CPU tech.

 

[AtenRa]

AtenRa, I picked the performance option myself but let's face the truth. If Intel could cram 50% more performance at a given TDP, they would do so if only for the desktop market.

Instead, core number is limited by Amadahl's law, core size is limited by law of diminishing returns, and clock speeds are limited by both power usage and heat density barriers. Intel has dedicated a lot of funding and time into maximizing IPC and introducing new instructions when possible.

What more can they really do? All the other options have significant drawbacks. How would they deliver 50% more performance without either a increase in cores which only benefits multithreaded code (not ubiquitous) or increase in clock speeds (requires expensive cooling and power delivery).

I want what you want. I just don't see how it is possible with current CPU tech.

Power users are not only high-end gamers, i can give you lots of applications where a 6-core has ~50% performance gain over a quad core. IPC is not the only metric in performance. If you cannot raise the IPC then raise the f....king frequency or raise the core count.

The problem is that Intel chose to give us lower power usage over higher performance. Well, i dont want lower power usage, I want higher performance at the same power usage i had before.

Thats what this poll is all about 😉

 

[Idontcare]

I can give you the same performance at half the power usage and you will jump all over the place and say that intel raised its performance/watt. Thats not what High-End desktop users want.

We want the same power usage(130W) but 50% more performance with every new node at the same price point. That still get you higher performance/watt at the same price point 😉

If you define a "High-End desktop user" as a person who wants what you want, then sure the numbers are going to work out to support your desired outcome. That is what bias confirmation is all about, and it infects and corrupts the person who is constructing a given poll or study which is why in medical trials the need to resort to so-called "double-blind" studies has been found to be necessary.

Confirmation bias (also called confirmatory bias or myside bias) is a tendency of people to favor information that confirms their beliefs or hypotheses.[Note 1][1] People display this bias when they gather or remember information selectively, or when they interpret it in a biased way. The effect is stronger for emotionally charged issues and for deeply entrenched beliefs.

I'm not here to tell you that you are biased, you are human therefore you are biased according to psychologists, and I am not here to tell you your poll is biased, you constructed it without assistance from others and therefore it is the definition of a biased poll according to sociologists, but I am here and so I may as well tell you that I do notice it is lacking some rather obvious choices in terms of voting options.

What you do with that feedback is up to you, just as it is up to everyone who reads your response to that feedback to make a decision for themselves regarding why you chose to concoct the voting options along specific paths as you have done.

 

[Gikaseixas]

I want both, with a slow reduction in power use over years of development/upgrades. Kind of inbetween the two poll options.

Feel the same way but given the two options we have i chose MORE POWER lol

If I want to make a small PC (Micro) that i can carry it around i would want to push the limits when selecting the hardware for it. Right now it would be Haswell and Kepler based.

I have a Cosmos II that houses my FX chip. Next year i want to upgrade again and i'll go Intel or AMD biggest and baddest for CPU. If something similar to Titan is available at that time, that's what i'm getting.

It's about what works best for you

 

[jaqie]

Imagine a PC that sounds like a vacuum cleaner, you have to exhaust it out a window, and you have to run it in a special outlet with a breaker just for that PC. That is where things are going with high performance PCs unless they start calming down and getting reasonable with their products.

I want a quiet PC with some large fraction of that power that I can easily plug in 5 of them to the same outlet without worrying about blowing a breaker.

 

[AtenRa]

Again, by having more performance at the same power usage will raise the performance/watt and it will still have a higher efficiency than before. I may have not made it clear in the poll but the first chose is all about having more performance at the higher power levels we had before, that is at 130W TDP CPUs.

When i say "I want" it refers to what the majority of Power users and High-End gamers care about. With graphics cards we get 50%+ performance at almost the same power usage with every new node. That’s also raises the performance/watt isn’t it ?? Well, we could have the same CPU performance gains with every new node.

Core 2 Quad 9xxx was a 95W TDP, Core i7 920 was 130W TDP at 45nm. The performance gains were substantial and nobody complained about the higher power usage.
Then SB came at 32nm and raised the performance lowering the power usage at the same time and everyone was happy. But that have changed at 22nm, we barely got a performance increase, instead we only got lower power usage and higher iGPU performance that power users and High-End gamers doesn’t care about.

That’s the point of this Poll, but if you or others feel that it is biased or it have a hidden agent that’s something I can’t do nothing about, we are a democratic forum after all and everyone’s opinion is free to be heard 😉

 

[aigomorla]

^ H80 cooler.

MINES A CUBE TOO!!!

uhh... maybe more of a box?
😛

But But But... how else am i to fit all this junk inside it?

80% of my hardware which generate heat has a water plate keeping it cool.

 

[Revolution 11]

Decreases in power usage suit the overall market. Cheaper cooling solutions and the enabling of more powerful chips in smaller form factors is the goal here. If you are ok with a higher TDP, simply increase the voltage and clockspeed and overclock. Intel lets you do it and even offers a cheap plan to cover damages.

22nm in particular offered a larger power decrease at a performance level compared to a performance increase at a power level. It makes sense that Intel would take advantage of the larger benefit. If Intel could offer a 10x performance increase for 10W more, they would do it. If Intel could offer Core 2 performance in a 10W TDP, they would do it.

 

[AtenRa]

Decreases in power usage suit the overall market. Cheaper cooling solutions and the enabling of more powerful chips in smaller form factors is the goal here. If you are ok with a higher TDP, simply increase the voltage and clockspeed and overclock. Intel lets you do it and even offers a cheap plan to cover damages.

22nm in particular offered a larger power decrease at a performance level compared to a performance increase at a power level. It makes sense that Intel would take advantage of the larger benefit. If Intel could offer a 10x performance increase for 10W more, they would do it. If Intel could offer Core 2 performance in a 10W TDP, they would do it.

I dont have enything against lower power Products. Actually i like small form factor PCs and those need lower power usage/TDPs than large high-end Gaming/work desktops. But Intel didnt provide us with a second choice here, they only gave us lower power usage at 22nm.

They(Intel) deleberately delayed the release of the 22nm socket 2011 CPUs because they know that power users and Gamers would buy those over socket 1155/1150.

 

[crashtech]

They(Intel) deleberately delayed the release of the 22nm socket 2011 CPUs because they know that power users and Gamers would buy those over socket 1155/1150.

So why haven't they released 2011 IB-E Xeons, then? That's a bigger market than enthusiast products.

 

[Vesku]

Well, look what you can get in the 130W range from Intel. It's just they are charging a premium for it. In terms of 2600K->4770K the main change is efficiency where as I don't think it's an unreasonable assumption that many of the people who are shopping in that category would have preferred to see 2 more cores tacked on. I think this does come down to the lack of competition from AMD, if they were more of a threat to Intel's market share in this segment there's a very good chance there would be a 6 core S1155 SKU with moderate pricing or at the very least that they'd open up the wattage a bit on K series and have them at higher default clocks.

Of course they also have to convince people to upgrade, so I'm expecting some missing information to be filled by official 4770K reviews that the early unofficial reviews haven't provided. Given the base clocks aren't that different I'm expecting Haswell K series to have some additional OC features from Intel as an enticement.

 

[Phynaz]

I may have not made it clear in the poll but the first chose is all about having more performance at the higher power levels we had before, that is at 130W TDP CPUs.

You can have half of that today with the 140W FX chips. For the performance part you need to go Intel, and you get half the power as a bonus. 😉

 

[Ajay]

I just hope Intel releases a 150W Haswell-E :twisted:
Then I'll be happy, for a while.

Of course, my next upgrade after that may be a BGA mobo/CPU D:
I'll be keeping that H-E as long as I can...

 

[Haserath]

I'd like a system with everything fanless except one 240mm fan to cool every component. The box would have vent holes on every side and the fan would be on the bottom.

An APU with a decent IGP at 65W TDP on either a matx board or itx board.

If they could keep most effects and fit enough power into the APU to render it, that would be good enough for me. Quad core Haswell with 7870 GPU.

A nice, quiet set-top box, and it could also be a coaster! Efficient and powerful(I know, not powerful enough for some).