Ivy Bridge to be up to 199% faster than Sandy Bridge

[deadrats]

http://legitreviews.com/news/11958/

it seems that the cpu cores will be about 15-25% faster clock for clock, the quick sync engine will be about 50% faster and the integrated gpu will be about twice as fast; all in all that's pretty impressive.

when you consider that i can play the batman arkham asylum demo using just a core i3 2100 with the integrated hd2000 gpu, albeit in dx9 mode (it uses a modified unreal 3 engine) and with decent quality settings, one would expect the i7 3770 to be enough to satisfy the casual gamer without needing a discrete graphics card.

[aedipuss]

i'm still flabbergasted the 1155 series chips died so quickly........

[chowmein]

Isn't the i7 3770 use 1155 socket?

[jagabo]

Isn't the i7 3770 use 1155 socket?

Yes.

http://www.cpu-world.com/info/Intel/Intel_Core_i7.html

Later in the year they will come out with socket 2011 versions.

[Heywould3]

If something is 100% faster then technically that would be twice as fast right? if you are moving at 60mph and you are moving 100% faster that would be 120mph, correct?

so actually 199% faster should be 4 times faster or 440mph. if we use time in the equation, it takes sandy bridge 60 seconds to complete then 100% faster would be 30 seconds, if 199% then 7.5 seconds to complete..? correct?

not that this was the point of the topic but just an observation and question.

[jagabo]

so actually 199% faster should be 4 times faster or 440mph.

No, 199 percent faster would be nearly three times faster.

[Heywould3]

so actually 199% faster should be 4 times faster or 440mph.

No, 199 percent faster would be nearly three times faster.

Yes, of course, sorry, was in a hurry. I was making the comment/asking the question because someone said twice as fast..

thanks for the correction.

[dragonkeeper]

The benchmarks are nice esp for hardcore gammers, but what type of performance boost does x264 receive?

[jagabo]

The benchmarks are nice esp for hardcore gammers, but what type of performance boost does x264 receive?

From the numbers posted, my guess is about 15 percent, at the same clock speed.

[poisondeathray]

so actually 199% faster should be 4 times faster or 440mph.

No, 199 percent faster would be nearly three times faster.

Yes, of course, sorry, was in a hurry. I was making the comment/asking the question because someone said twice as fast..

thanks for the correction.

It's hard to do math when staring at your avatar

[deadrats]

so actually 199% faster should be 4 times faster or 440mph.

No, 199 percent faster would be nearly three times faster.

how do you arrive at that conclusion? saying something is 199% faster is the equivalent of saying that something is 398% as fast or nearly 4 times as fast:

assume 2 competitors, A and B:

A can run the 100 meter dash in 10 seconds, if B can do likewise we say that B is 100% as fast as A OR equivalently B is 0% faster than A.

if B can run the 100 meter dash in 5 seconds we say that B is 200% as fast as A OR equivalently B is 100% faster than A.

if B can run the 100 meter dash in 2.5 seconds we say that B is 400% as fast as A or equivalently B is 200% faster than B.

thus saying ivy bridge is going to be up to 199% faster than sandy bridge means that IV will be up to almost 4 times faster than SB (albeit with regards to integrated gpu).

[dragonkeeper]

From the numbers posted, my guess is about 15 percent, at the same clock speed.

Very nice, but what is the actual cost of the net gain? How much more is the chip than it's sibling? Does the Ivy Bridge require a new motherboard and ram? My current rig is almost 4 yrs old so I'm really due for an upgrade, but to me from a price\performance standpoint the 1100T seems to be the way to go, granted a quad core Ivy bridge is faster but at an additional $400 (estimated cost of building similar PCs one 1100T based the other Ivy Bridge) IMO not really worth it unless I want bragging rights.
All the heavy lifting done on my PC is done while I'm either sleep or away from home. Finishing a project 30 min faster is not that important I'll never see the benefit of it since i'm away from the PC in either case. Am I missing something here?

[poisondeathray]

Did you factor in performance/watt calculations ? electricty costs kw/h vary by region quite a bit but can make a big difference in operating costs over a time period like a year

[jagabo]

From the numbers posted, my guess is about 15 percent, at the same clock speed.

Very nice, but what is the actual cost of the net gain? How much more is the chip than it's sibling?

Prices have not been announced.

Does the Ivy Bridge require a new motherboard and ram?

It depends on what motherboard and RAM you have.

My current rig is almost 4 yrs old so I'm really due for an upgrade, but to me from a price\performance standpoint the 1100T seems to be the way to go, granted a quad core Ivy bridge is faster but at an additional $400 (estimated cost of building similar PCs one 1100T based the other Ivy Bridge) IMO not really worth it unless I want bragging rights. All the heavy lifting done on my PC is done while I'm either sleep or away from home. Finishing a project 30 min faster is not that important I'll never see the benefit of it since i'm away from the PC in either case. Am I missing something here?

I have a 17 year old truck. It gets me where I want to go. So there's no reason to manufacture new trucks.

[johns0]

I have a 17 year old truck. It gets me where I want to go. So there's no reason to manufacture new trucks.

What if your truck breaks down or gets in a bad accident?

[dragonkeeper]

I have a 17 year old truck. It gets me where I want to go. So there's no reason to manufacture new trucks.

There 's jagabo taking it to the extreme

In my post i mentioned i have a 4 year old rig; I'm over due for an upgrade. So i will be soon be building a new rig, my primary concern video encoding performance vs price. An 1100T can encode better than realtime, It's i7 counterpart can encode even faster but cost over $200 more. It also appears that between similarly spec'd motherboards Intel boards cost slightly more on average than AMD boards.
From a cost\performance standpoint the 1100T or perhaps even an i5 seems to be the better choice, the quad core i7 is almost twice the price but does not offer twice the performance . In numerous video forums i've read post from users who are eagerly awating the Ivy Bridge so they can upgrade their i7. To me $400 dollars, is extreme to net a 15% increase. Hence the question am i missing something here?

[usually_quiet]

I'm with dragonkeeper. If an AMD product or less expensive Intel product is gets the job done and costs $200 less, there isn't much reason to buy an Ivy Bridge CPU. Its like buying a Ferrari when you know you are only going to be driving it on surface streets at 45 mph maximum. What is the point? It isn't like many of us can't find enough places to spend the money that would be saved.

[jagabo]

An 1100T can encode better than realtime, It's i7 counterpart can encode even faster but cost over $200 more. It also appears that between similarly spec'd motherboards Intel boards cost slightly more on average than AMD boards. From a cost\performance standpoint the 1100T or perhaps even an i5 seems to be the better choice

Again, you're using your needs, and a single benchmark, to make that judgement.

Core i7 2600K: $320
Phenom 2 X6 1100T: $200

http://www.tomshardware.com/charts/desktop-cpu-charts-2010/compare,2423.html?prod%5B47...d%5B4757%5D=on

the quad core i7 is almost twice the price but does not offer twice the performance.

What if you make $100 per "job" and the i7 lets you bill one more job a week? Even one more job a month? Which is a better deal now?

In numerous video forums i've read post from users who are eagerly awating the Ivy Bridge so they can upgrade their i7. To me $400 dollars, is extreme to net a 15% increase.

To some people it's not.

[jagabo]

A can run the 100 meter dash in 10 seconds... if B can run the 100 meter dash in 2.5 seconds we say that B is 400% as fast as A.

"as fast" != "faster". In your example A runs at 10 m/s. B runs at 40 m/s. So A is 1/4 as fast as B. But when you say "B is x% faster than A" you mean "in addition to the speed of A". So B is 300 percent faster than A. Just like 11 m/s is 10 percent faster than 10 m/s.

If you make $100 a day and your boss said he's going to pay you 10 percent "morer" wouldn't you expect to get $110? If he said 100 percent morer wouldn't you expect to get $200? So 300 percent morer would be $400.

[poisondeathray]

People often only look at initial capital expenditure, but forget operating costs

Here is an example. I'm sure jagabo will fix my math because it sux

When you factor in operational costs, and cooling costs (less a factor for single node home systems, but a huge factor for large offices and datacenters - it makes using AMD systems almost infeasible).

Lets assume constant load, and use SB figures. IB should be about 10-15% lower for power consumption by all reports, largely because of 22nm tech

e.g. i7-2600k load 156w , 1100T load 207w . The delta is 51 w , or 0.051kw
http://www.bit-tech.net/hardware/cpus/2011/01/03/intel-sandy-bridge-review/11

Average price of electricity in LA Oct2011 $0.199/kw.h
http://www.bls.gov/ro9/cpilosa_energy.htm

$0.199 / kw.h * 0.051 kw = $0.010149 / h difference

multiply that by 24 h/ day and 365 days/ year = $88.91 difference per year electrical costs

If we assume IB is 10% more efficient, it would consume 156*0.9 = 140.4 w, so the delta would be 66.6 w, or $116.10 per year difference in electric bill

Now, there are a lot of assumptions, like usage pattern , load vs idle, methodology for measurements , test configurations, etc... but you get the basic idea , it's just a ballpark figure

So how long until you surpass that larger intial capital expenditure difference ? How many years do you plan to own the system before next upgrade? Then you have to factor in the time value of money (interest) .

[aedipuss]

forget the power cost. it would only apply if the cpu was running 100% all the time. here is what an i7 set to run overclocked at 4GHZ constantly uses for power most of the time.

[poisondeathray]

forget the power cost. it would only apply if the cpu was running 100% all the time. here is what an i7 set to run overclocked at 4GHZ constantly uses for power most of the time.


[Attachment 9970 - Click to enlarge]

You're not measuring system draw (from the wall) . If you turn on the computer, other components need power too - you need power for HDD, monitor, motherboard, memory,etc... Those aren't solar powered in most computers . Use a kill-a-watt meter at idle and compare to load, that's what the electric company is charging you
http://en.wikipedia.org/wiki/Kill_A_Watt

At stock, its 77w in that weblink for the i7, 108w for the 1100T for idle values. So the delta is smaller, but you have to factor in usage pattern. e.g. if you finish tasks faster, you spend more time in idle vs. load . Of course if you overclock, any system will consume more power. Or maybe you shut off the computer in the daytime when you're at work.... etc...

[poisondeathray]

In case it wasn't clear, that previous calculation was the difference in electrical savings between 1100T and i7-2600k at stock speeds, when at 100% load

If you want the electrical cost for i7 only, turned on, but idle doing nothing for 365 days it would be (77w or 0.077kw)

$0.199 /kw.h * 0.077kw * 24h/day * 365 day/year = $134.23

Similarly for the 1100T, turned on, but idle

$0.199 / kw.h * 0.108kw * 24h/day * 365 day/year = $188.27

[aedipuss]

yes but if i pull it out and put in an ivy bridge the only change is the cpu and i don't think it will save appreciable amount of energy over a sandy bridge that draws 10 watts most of the time. yes some would be saved during encodes as long as the ivy bridge doesn't require more power than a s.b. to overclock to 4GHZ.

did you read the first sentence?

we disable all power-saving technology

i don't and CPUID HW monitor pro is pretty accurate as far as i can tell. i wouldn't argue with the 10 watt idle figure.

[poisondeathray]

yes but if i pull it out and put in an ivy bridge the only change is the cpu and i don't think it will save appreciable amount of energy over a sandy bridge that draws 10 watts most of the time. yes some would be saved during encodes as long as the ivy bridge doesn't require more power than a s.b. to overclock to 4GHZ.

Well the TDP is 95w vs. 77w. Yes, TDP is misleading, but IB's main selling point is higher performance per watt .

did you read the first sentence?

we disable all power-saving technology

No, I missed that, nice catch. I just linked to the 1st link in google. But other reviews have similar numbers for idle ~70W
for system power consumption

CPUID HW monitor pro is pretty accurate as far as i can tell. i wouldn't argue with the 10 watt idle figure.

I agree. I'm not arguing with the 10 watt idle figure . That values in the screenshot are correct for CPU only. The screenshot you posted of CPUID HW doesn't measure total consumption from the wall.

[dragonkeeper]

Again, you're using your needs, and a single benchmark, to make that judgement.
Core i7 2600K: $320
Phenom 2 X6 1100T: $200

In video forums most of us are more concerned with the one benchmark.
I mis-quoted the price, but the i7 2600K is still considerably more than the 1100T. A mild overclock on the 1100T will exceed the performance of the i7.

What if you make $100 per "job" and the i7 lets you bill one more job a week? Even one more job a month? Which is a better deal now?

For me most of my encoding is done after hours, all my assets are encoded over night for use the following day (seems to be the most efficient way of doing things). If i do get a rush job i have a second PC that can encoded assets, while i continue to work on other projects

In numerous video forums i've read post from users who are eagerly awaiting the Ivy Bridge so they can upgrade their i7. To me $400 dollars, is extreme to net a 15% increase.

To some people it's not.

Agreed, but one can get +10% performance gain through overclocking with a aftermarket cooler for about $60. So back to my earlier statement of people seeking bragging rights.

[jagabo]

In video forums most of us are more concerned with the one benchmark.

Not me. I'm also concerned with how fast various AviSynth filters run. I also use MPEG 2 and Xvid encoders on occasion. I suspect most people here are interested in more than just video encoding. And for that matter the 1100T doesn't win all video encoding benchmarks. Not even those that use the x264 encoder. What about people that use Mainconcept (2600K 58 seconds vs. 1100T 69 seconds)? Adobe Premiere CS5 (2600K 224 seconds vs. 1100T 284 seconds)? Handbrake (2600K 122 seconds vs. 1100T 143 seconds)?

http://www.tomshardware.com/charts/desktop-cpu-charts-2010/compare,2423.html?prod[4785]=on&prod[4757]=on

A mild overclock on the 1100T will exceed the performance of the i7.

And a mild overclock of the i7 2600K will have it outperforming the overclocked 1100T. That's a useless argument.

For me most of my encoding is done after hours

Again, that's for you, not everybody. And if your encoding is done after hours why do you care how fast it goes? (Within reason, obviously.)

It's fine that you prefer an X6 1100T. But other people can have have different priorities.

[deadrats]

Very nice, but what is the actual cost of the net gain? How much more is the chip than it's sibling? Does the Ivy Bridge require a new motherboard and ram? My current rig is almost 4 yrs old so I'm really due for an upgrade, but to me from a price\performance standpoint the 1100T seems to be the way to go, granted a quad core Ivy bridge is faster but at an additional $400 (estimated cost of building similar PCs one 1100T based the other Ivy Bridge) IMO not really worth it unless I want bragging rights.
All the heavy lifting done on my PC is done while I'm either sleep or away from home. Finishing a project 30 min faster is not that important I'll never see the benefit of it since i'm away from the PC in either case. Am I missing something here?

i would guess that the ivy bridge cpu's will be comparably priced, if not cheaper than their sandy bridge siblings, as was the case with conroe/penryn or bloomfield/lynnfield. as the yields go up from the superior manufacturing process i would expect the prices to go down.

in terms of the price/performance of the 1100T a similar argument could be made with the X4 840, a quad core that can be had for $60.

you need to keep in mind that in the reviews posted online we only get a single brief look at a cpu's encoding performance, with just 1 or 2 sample test files and with certain reviewer chosen settings. just because an 1100T performs reasonably close to a SB in those tests doesn't mean that if you crank up the bit rate or the quality settings that the performance gap as a percentage will remain the same.

honestly, the IB seems like the best bang for the buck when it comes out; you will get the fastest x86 architecture available, DX11 graphics, low power consumption and a dedicated video encoding engine, to me that seems like it is the superior price/performance option.

[dragonkeeper]

And a mild overclock of the i7 2600K will have it outperforming the overclocked 1100T. That's a useless argument.

My point was that by overclocking I can get better performance for less money.
I acknowledge i mis-spoke when i commented that in a video forum most users are most concerned with the one benchmark. I should have said in a video forum most users are most concerned with video benchmarks. When seeking advice on a best chip for gaming I go to a gaming forum.

It's fine that you prefer an X6 1100T.

I don't prefer the 1100T, I just singled it out as seeming to be the better bang for the buck for video encoding (aside from cost of ownership that is), when compared to Sandy Bridge and Ivy Bridge. And my question is, Am i missing the big picture here?

[aedipuss]

if we are concerned about power consumption then it's best to start nearest the wall socket. i think my best purchase building this system was the 1000watt 80Plus Gold certification psu, which means that the power at 50% load more than 90% efficient. At 20% and 100% load, the power at least 87% effective. saves on power for the entire system compared to an unrated psu that's 75% or less efficient. also much less heat is generated and it only draws 0.5 watts when on but no load.