Intel core 2 duo vs dual,core i5 or i7

I am curious. If we took,a core 2 duo and compare it with an i5 or i7 cpu with the same frequency and the same number of cores, are the i5 or i7'really much faster? Both being dual cores and same frequency, what would really make the newer cpus faster? I guess I am curious if I go from my 5 year old core2 to a i5 or i7 with 2 cores, will I notice much of a difference? Do I need to,step up to a quad core to see a measurable difference?

 

If mobility is important to you them lower TDPs will be of great help. I have moved on from Core 2 for a while and now am on ivy i3. The batter batt runtime is nice. Still cant make meself to move on to haswell so that slight feel of lag despite them figures. On top of that current thin and light build just feel a little too flimsy to me.

 

Mobility is not a huge concern for me. I do need something I can move from the office to the home one time a week. I need more of a mobile workstation. But getting back to my original questions above. What do you guys say?

 

An excellent resource for comparing CPU performance is the notebookcheck mobile CPU benchmark list.

The raw CPU performance at a given clock speed (as indicated by basic tests such as wPrime) has not increased much but the system performance may have improved significantly due to, for example, the incorporation of the memory controller in the CPU package (Sandy Bridge onwards?) and the availability of faster storage interfaces. Adding an SSD is the best way to make your computer feel faster in everyday usage.

John

 

The short answer is core for core and clock for clock, anything current gen (Haswell i3/5/7) is going to be significantly faster than even the last gen of C2D (Penryn).

 

/close thread

 

This.

10char

 

C2D is just obsolete now.. Even a Haswell ULV at 1.6GHz will be faster then my T9400 at 2.53GHz...

 

there's so much more to a cpu than clock speed. for example an old pentium 4 at 3ghz will probably perform at about 10-15% compared to modern cpu at the same clock speed, and that's on single threaded tasks.

clock speed can only be used for direct comparisons when comparing processors from the same line really.

 

The Fire Snake, I'd say (top) C2Ds can compete vs i3 only, and it's actually a good news they still can. They can't compete with anything that has TurboBoost because it usually means faster clocks than most C2Ds ever reach while per-MHz C2D performance is worse - not that much but still, and combined with higher actual frequency - yes, you will definitely notice the difference in demanding applications. However, If you're using your notebook for browsing, media playback, etc. light tasks - most C2Ds will do, just drop in a nice SSD and enough RAM.

 

I have a thinkpad T500 right now with a core 2 duo. I have already maxed out the ram to 8GB and have a SSD in it. Even with this the cpu is getting pegged at 100% use often and I get warnings. I just dont want to be in a situation where I spend a grand on a new machine with a dual cor ,and it winds up being almost as fast as,the old machine. So I was thinking I had to go with a quad core cpu.

 

The Fire Snake,

Even if your Core 2 Duo was actually introduced 5 years ago, the underlying technology is closer to a decade old.

Anything current you buy will make your C2D look like it is powered by steam... (yeah very, very hot H2O).

100% usage, often? Time to upgrade.

 

Have you checked if you have the most recent ati hd 3000 series drivers ? A friend of mine has a laptop with a hd 3000 series gpu (HP 6735B). The cpu was almost always 100% when watching hd video because he had installed only the drivers from hp and from windows update. Then we ran the Amd driver AutoDetect and installed the latest driver from amd. Now the cpu never reaches 35% and the laptop can even play hd 1080p in youtube.

 

Like Ratsey says, given that you could clock a core2duo to the same frequency as a newer "core" with hyperthreading and a decent l3 cache size, and had front bus speed and ram overclocked to at least match the transfer rate for the smaller ram fetches (the bus is a bit different now, allowing larger dma without "missing" a cpu cycle and suddenly doubling the execution time and things of that sort) -- then you would get something near a complete lack of any difference in the results whatsoever. I mean, you'd likely be as surprised as Anadtech at how few "real world" applications would even be different, and that the rest could be explained away by the memory bus.

Of course, that c2duo would then soon combust into a small puddle of a burned graphite stain, at the floor after melting through the table the laptop stood on. You're never going to be able to actually overclock one successfully, I think, even if you had enough cooling in theory. The construction just won't allow it. So going from a c2d to a "cut down" "mobile" haswell, that still has 3.4Ghz boost clocks, etc. That's in most of the situations that matter a difference similar to doubling the max frequency while reducing the power drain to a maximum of 1/8th or something like that.

But yes, the actual architecture is the same (even though the memory bus is "integrated", it's still a separate module -- it will stay this way until Intel goes out of business). Suggesting that if you are actually running low intensity, several threads, that have low response times, etc. There'd be very little benefit to moving to something newer.

 

T500 has GMA 4500 integrated graphics, and may or may not sport an ATi HD 3650 card along with it. When properly set it will run as a daily driver just fine.

Unfortunately, we know very little about OP's usage patterns and what exactly could be causing this 100% spike in CPU usage.

My $0.02 only...

 

Regarding video playback, Sandy Bridge onwards includes a video decoder built in to the CPU.

John

 

What CPU do you have in your computer? I have 8GB of RAM and a T9900 processor. The T9900 is at least equivalent to any Haswell low-power processor, which is fine for lots of things.

I too have a Lenovo T500 that is my daily driver. It has the switchable graphics. The most common video codecs are hardware accelerated by either of the GPU's. I watch 1080p bluray videos with low CPU usage. What does bog down the CPU is watching super high res HD videos on Youtube using flash. 1080p or anything lower while using flash is perfectly fine. HTML5 video works great though.

I use my T500 a lot for watching movies and TV shows, both streaming and not, for playing games that are at least a couple of years old (I have the GPU overclocked about 50%), running Folding@Home 24/7, editing photos with Adobe's Photoshop Lightroom, transcoding videos (which is the number one thing that would be helped by a brand new computer), and "heavy" web browsing/working online.

Several months ago I was visiting my dad and using his workstation desktop and laptop for the above tasks. I have to be honest in that I didn't feel much difference with them, and their specs absolutely blow my T500 away. The only two things that I feel that I want more from my T500 are an IPS screen (it has the best 1900x1200 15.4" LVDS screen already) and a more powerful graphics card. Games are just a waste of time, and I can always get a good external monitor. I have been thinking of buying an X9100 processor just for fun although it is rare for the stuff I do to bog down the CPU.

 

Games are a waste of time? Really? Good way to offend 99% of the world's population.. I would choose your words more carefully in the future!

 

99% of the population? Are you kidding me?

Try more like 20%. And there is nothing to be 'offended' about either.

 

Every time after I play computer games, I feel I've just completely wasted some of my precious time. =/
There are other, much worse, ways to waste it, however. =)

 

The problem comes if the board is only capable of C2D processor not of the i3---i7 mobile family. So trying to compare chip to chip and forgetting the board also limited as well is like comparing Apples to Oranges...same company but two different technologies from different times...of the family line...does neither processor justice.

Have to agree most of the laptop from i3---i7 are just browsing, media playback...and SSD/RAM upgrade will do far more to improve performance...

Must be dark in the cave there...You do know if it wasn't for the PC gaming market PC computers would not be where they are now or have the market shares they do now. Also Gaming shows off the high value and potential of what True Gaming is experienced like.

 

I have yet to see a laptop motherboard that can accommodate CPUs from both of these platforms...

 

You totally missed my reasoning....

 

No I didn't.

The problem here is that you've brought something that is a non-issue in this thread by wording your post the way you did.

What is being discussed here is the (lack of) power that OP is currently experiencing with his C2D-powered ThinkPad T500, and the upgrade options.

It's not chip vs. chip or board vs. board. It's platform vs. platform.

It's also not about doing a chip - or any component of the aforementioned platform(s) - justice, or failing to do so.

It is about what would - as a platform - present a meaningful upgrade from OP's current one in their usage scenario which we yet have to learn more about.

Are we in the clear now?

 

Missed to point clearly....OP can't use i3-i7 totally different sockets... so did you now get the point....trying to compare Apples to Oranges...does no service to either procesor families from iNtel... So this debate is basically no debate to be had. Old is old...and old will run into technology limits regardless of how you want to cherry pick or make it a rosy picture. Side by side i-series family iNtel will run circles around C2D processors. If we just talk about C2D comparison and what they can upgrade to that is different but trying to compare C2D to i-Seires will never work the processors socket couldn't be more different making a comparison not to be had. So rather doing a vs should be C2D vs C2D and what makes the best upgrade should be the topic that would better serve the OP request.

 

Really....a penryn core system with ample ram and a SSD will be fine for anything other than gaming. I even have a merom core T60P with the ATI v5250 and T7200 CPU that still works fine for most everything. Of course its totally obsolete for gaming, but as a backup for word processing and internet usage it performs just fine. Battery life with the 9 Cell and some undervolting of the CPU is around ~3 hours

 

Do you ever read what other people post before "replying"? Let alone think about it?

I wrote the following in the post # 22:

And then you go into this...

You obviously haven't comprehended what the OP was inquiring about, so let me translate it for you:

They are NOT interested in a C2D vs. C2D comparison.

They are interested in moving onto a newer platform, and debating whether to go with a quad core CPU or not depending on what the actual performance gain when compared with their current CPU would be...

 

They'll be drastically faster.

Modern CPUs have larger caches, meaning that they have to fetch things from RAM less often. Every time your CPU leaves the cache, your performance drops from 50-500x.

Another way to avoid that is what's called branch prediction. If in your code there is an "if" statement, or some condition that won't be known until runtime, the CPU can compute the most likely future branch, again avoiding unneeded computation or latency. Modern CPUs are far better at this than old CPUs.

Clock cycles for computation have improved a lot. All computation comes down to clock cycles, though the bottleneck is almost always latency.

Oh, and more cores/ hyperthreading means that more code can be run in parallel.
If I've misunderstood the question, let me know. I can try to answer more suitably.

 

ComradeQuestion, could you please provide some proof to back your drastically faster remark?

The Fire Snake, bottom line is, if you really want to feel the difference, go for a quad-core (Intel). Dual-cores will be faster than C2D, but not nearly as much as one'd expect computer technology to advance for 6 years.

 

Well, the proof is the rest of my post that explains each of the improvements to modern CPUs. If you want me to clarify any of it I can.

Computation-wise it's easier to get hard numbers because you can look at FLOPS, and I'm sure there's a graph showing exponential progression there. But FLOPS are rarely where your bottleneck is, typically it's latency, which is why my post is discussing that aspect in particular.

 

ComradeQuestion, I choose to believe in actual performance.

i3-4005U
T9900

p.s. couldn't find better match, there ain't many i3 tests around.

 

Not sure what your point is. I've stated only facts. You've posted two completely different CPUs with completely different cache sizes, architecture, TDP, threading, and a bunch of synthetic benchmarks.

What conclusion did you draw from those two articles that you'd like me to address?

If you state your point I'll have an easier time discussing this.

 

ComradeQuestion, my point is simple. Modern mobile i3 equals top mobile C2D in performance. Now, is modern mobile dual-core i5 or i7 really drastically faster than modern mobile i3?

 

Well, it more than equals it. That i3 is considerably more powerful for 1/3rd of the power. Quite drastic.

 

ComradeQuestion, say what? You were talking exponential growth in performance, c/d? Where did that TDP thing come from? That i3 doesn't match C2D produced YEARS ago in terms of performance. TDP means battery life, and battery life only - which obviously many people don't even care about. What everyone cares about, however, is price... So why don't you compare that instead?

 

Which benchmarks are you looking at? Specifically, which benchmark on each page.

 

ComradeQuestion, just open Comparison chart in each test of i3-4005U. For every benchmark which was run on C2Ds, it's somewhere around P9600 or worse.

 

Is it only on some benchmarks? I'm jus tlooking at ones that look meaningful so far and I haven't seen both of them show up. Can you point one out?

This site would really benefit from a basic unified test suite.

Of course, the comparison between these CPUs is completely flawed anyways. But I'm curious. Even if that i3 is 3x as efficient power wise, I'd imagine it's still faster.

 

You beat me to it

Test | T9900 | i3-4005
wPrime 2.0x - less=faster
1024m | 799,4 | 954,2
32m | 25,1 | 30,4
3DMark - more=faster
Vantage - P CPU | 6045 | 6042
06 - CPU | 2787 | 2211
Cinebench R10 - more=faster
Single 32Bit | 3573 | 2514
Multiple CPUs 32Bit | 7094 | 5680
Super Pi mod 1.5 XS - less=faster
1M | 15.3 | 21.9
2M | 37.1 | 49.7
32M | 971.5 | 1088.7

p.s. if you can't read that... I give up.

 

Let's try this again.

Mobile Processors - Benchmarklist - NotebookCheck.net Tech

 

octiceps, comparing C2D to something with that high turbo frequency is an instant overkill, thus pointless. =\

 

i7-3540M: 3.5 GHz
i5-4330M: 3.4 GHz
C2D T9900: 3.06 GHz

So the alternative is to compare completely different classes of processors TDP-wise (i3 ULV...are you kidding me?!) and be even more misleading?

 

Interesting. They seem quite close, despite the fact that one CPU uses 1/3rd of the energy. Still, I would have expected better, but I suspect these benchmarks are largely throughput.


octiceps appears to have posted a more relevant benchmark, with similar TDP's. Results are obviously in favor of modern CPUs.

Even still, the comparison is weak, because the largest factors are in latency. I'd like to see something that's difficult on cache locality, and really shows off the massive effects of modern CPU prefetching.

For example, if you have a giant segment of memory, which you read from indexed in predictable but >cache size locations, you'd see 50x performance because modern prefetching will predict the next offset, and old CPUs will have to replenish their cache.

Honestly, I find most benchmarks that aren't very very explicit in the technology they target, to be like reading a magic 8 ball.

If you understand how a CPU works, and how modern programs work, you'll appreciate that 1% of what a CPU does is calculate data, and 99% of it is fetching data. Benchmarks never seem to reflect that properly.

 

To conclude, I'm not in any way an adept of old tech junkie. Actually, I'm very disappointed by the fact that progress in computer industry has slowed down substantially. I believe that after 5 years even cheapest Atoms should have been faster than any C2D... but they're not nearly, as we see. And when I read marketing-inspired bla-bla-bla, I may get a bit carried away. Cheers.

 

Honestly, "clock speed" is not a constraint. Power consumption is. So it makes sense to benchmark around power consumption. That's why his results make sense - we're discussin efficiency of modern CPUs here. Efficiency is the relationship of power to cost. To discuss one thing that has a very high power with a very high cost to something that's number one goal is low cost, well, that should be clear.

For what it's worth, I don't think I've read any intel marketing... ever. Everything I've posted is based on computer architecture and performance oriented programming.

Intel atoms are perhaps not "faster" than an old core 2 duo, but they are more efficient. Without the power constraint they'd blow it out of teh water. I actually would expect them to be faster in many situations though, since an atom will probably have the same amount of cache as a high end C2D. But I dn't really keep up with these things, so perhaps they reduce it for cost rasons.

 

ComradeQuestion, you measure efficiency in performance per watt, I measure it in performance per $. As long as we're able to buy an 5+ year old notebook for $100 and it will destroy most new machines priced <500$ in terms of performance, something is terribly wrong with progress in the industry.

 

Performance per dollar is important too, but when discussing things from purely a technological standpoint ie: "which is more efficient" my go-to would be cost per watt.

 

The biggest issue here is that due to the actual properties of the materials used in CPU design, the CPU manufacturers are going straight for a ceiling between reliability of the transistor gates (that is whether what should be a 0 could be interpreted as 1 and vice versa due to decreasing the operating voltage too much), power consumption, clock speed and dissipated heat. That is one of the main reasons why we've seen Intel (and others) switch to instructions per clock, power efficiency and multi-core as their main design focuses. Increasing the clock speed simply can't be done as easily as it used to be in the old days where all you have to do was increase the clocks and lower the operating voltage with each architecture change/refresh. It's still possible to lower voltages and increase clocks, but the CPU manufacturers are seeing the wall getting close.

I'd say the fact that Intel has come to dominate the x86 market is also slowing down R&D efforts as well as causing the current CPU prices. In a market where AMD, Intel's main competition as far as x86 is concerned is moribund, it's no surprise that the performance per $ ratio isn't sweet at all.

Intel Turbo Boost doesn't help comparing performance at all. You can have a laptop with a ULV CPU that will be able to maintain a higher turbo due to better cooling and a laptop with the exact same CPU that will run at lower sustained clocks because of poorer cooling. These days, you have to take benchmarks with a grain of salt. All that being said, I'd still expect slightly better overall performance even from ULV Haswell core i CPUs.

EDIT:

Here's another comparison for the sake of it:
http://www.notebookcheck.net/Mobile...technology=0&architecture=0&64bit=0&daysold=0

Even a ULV i3 compared to a P8600 which was not the most powerful at its time, but was still a very core 2 duo (at the high end core i5 level if you were to compare it today) is faster. Not by much, but still slightly faster. I'd say it's best not to forget too that ULV CPUs were never designed with heaps of performance in mind and also that the i3 has no turbo which makes a huge difference. The core i3 is almost bottom of the barrel and we are comparing those to high end core 2 duo, yes they're old, but they were still high end for their time.

If you were to compare performance for $, I'd say go with a core 2 duo P8600 and a core i5-4200M, they have almost the same MSRP. As with anything, the ULV core 2 were also more expensive than the full voltage ones which is also the case for the current core i CPUs, albeit, core i is still somewhat more expensive than core 2 was.

Here's more on two ULV CPUs: http://www.notebookcheck.net/Mobile...technology=0&architecture=0&64bit=0&daysold=0
IT was as close as I could do it clock speed and price wise. The i3 is roughly 30$ more expensive than the core 2 according to ark.intel.com The i3 wins by about 25% in wPrime, 63% in multithreaded cinebench and close to twice as fast in 3DMark06 (no idea whether that one is single of multithreaded).

EDIT2: Yes, certain C2D can compete with very low end core i, but also consider that a new laptop will also include newer features like USB3.0 and so on.

 

Well ignoring everything else....
Comparing clock for clock , Core 2 Duo vs newer i3 processors. All are dual cores.

3DMark06 CPU scores

T8300 (2.40 GHz)------------- 2141

i3-370M (2.40 GHz)----------- 2578

i3-2370M (2.40 GHz)--------- 2869

i3-3110M (2.40 GHz)--------- 2988

i3-4000M (2.40 GHz)--------- 3098


Now if someone is bottlenecking with T8300, i think he should forget i3 processors altogether. And low voltage i5 aren't gonna be much improvement either.
4th Gen standard voltage i5 will be decent enough to upgrade to. For example
i5-4300M (2.60-3.30GHz)---------- 4147

And new i7 Quads will make you forget what bottleneck is offcourse.

 

Judging by this thread, the OP has chosen to go QC:

http://forum.notebookreview.com/wha...593-dell-precision-vs-hp-elitebook-zbook.html