Is Core i7-4600U (2.1GHz) faster than Core 2 Duo T9900 (3.06GHz)?

My Core 2 Duo has been having issues so I'll be replacing it. (100% CPU usage issues which we can't resolve, but that's another matter....). Corporate IT quoted a Dell Latitude E5440 4th gen Intel Core i7-4600U Processor (2.1GHz, 4M Cache). They're both dual cores, so won't this i7 be slower than what I have now given the clock speeds? I know there are efficiencies w/ the newer processor, but 2.1GHz vs 3.06GHz seems like such as difference. I can't find one-to-one benchmarks online and I don't know enough about the processor architectures to know if the 'upgrade' is really an upgrade or not.

I mostly run MATLAB, a mail program, MS Office apps, Acrobat, etc. always while connected via WiFi and tunneling in to the company w/ VPN. Future projects may have me doing real-time controls within Matlab but that is TBD.

With travel, one of my criteria w/ my laptop is that the screen size be 14". So faster quad core options are currently out for me AFAIK.

 

Clock speed rarely is relevant today, so I wouldn't focus on the GHz. You have to remember that with each new generation, the CPU is getting more efficient per clock, so a lower-clock modern CPU can be just as powerful (if not more) than a higher-clock old CPU. Honestly, that much of a GHz different doesn't mean much with such a large generation gap (other than the C2D using a lot more power). You can check out CPU benchmarks to see that the newer ULV i7 should be competitive to that C2D.

 

No these new i7 are much better architecture and even with lower clock, there much much better... You also have hyper threading (so you have a virtual quad core) and turbo boost, which can go up to 3-3.5GHz depending on your model and needs... This will be much much better... I would take this without a hearbeat.

 

I guess the best point to make here is to have you understand that this 2.1GHz is the nominal frequency, which you should never take into account as you'll be running a different clock 95% of the time - idle and turbo mainly. So you're way more likely to see 800MHz clock when browsing/idling/moving stuff around the desktop, and between 3.0 and 3.3GHz when you're actually doing something more than opening Chrome, and the 2.1GHz just doesn't mean anything at all.

When trying to compare CPUs, you should always go the to manufacturer's website and have a closer look to the specs (Google is the fastest way to get exactly there). Then, while you're there, make sure you have a look at physical cores, as the core clock is to be multiplied at least by each physical core to really compare CPUs between them - and then added logical cores provided by Intel's Hyper Threading may potentially add more performance on top of that, in specific applications only. A quad core running 3GHz is likely to be faster than a dual core with HT running at 3.3GHz, except in apps/activities using only one or two threads, where the 300MHz difference won't mean much anyway. To have more cores is more important than to have more clock, unless the difference in clocks actually makes for another core or two, something you'll never see outside of overclocked processors.

In the present situation, you wouldn't have felt the need to be reassured if you had had a look and knew the CPU actually has a boost up to faster than your old CPU. I'm not saying this because you shouldn't post, just because you'll be much more efficient researching if you know to never trust CPU specs listed in a laptop's description without verifying what it really is. Most of the time these descriptions will list the nominal clock only, or both nominal and turbo clock. If you only see one clock listed, knowing that just about any CPU on the market right now actually has a turbo boost, then you know for sure you have to crosscheck the information.

Good luck, and if you want to understand all this better, don't hesitate to ask here for help!

 

ModRQC has a good point about turbo mode. Back in the T9900's heydey, turbo mode didn't really work that well. But now, it really does work. My work laptop has a Core i7 3520M. I have the power plan set to Maximum Performance all the time, and while plugged in, the minimum CPU power in Windows 7's power controls is set to 100%. Sure enough, even though it's 2.9 GHz nominally, it's almost always at the max turbo speed of 3.6 GHz, sometimes 3.4 GHz when it's using both cores heavily. The 4600U could theoretically get up to 3.3 GHz when using one core, or 3.0 GHz when using both, eliminating the frequency gap. And I suspect the E5540's cooling is good enough for that. My work laptop is an E5530, and it has never had cooling issues.

That said, since performance is the bottleneck now, I'd go for the 4600M (instead of 4600U) if possible. It is the same clocks as the 3520 in my work laptop (2.9 GHz base/3.4 dual turbo/3.6 single core turbo), and thus would be slightly faster. And according to Wikipedia, it should be slightly cheaper. I don't know if Dell offers it, but if they do, I'd go for it - I know I'm glad I have the faster, more power-hungry CPU when I'm waiting for it to finish something. Well, actually, I'm not glad, I'm wishing I had an even more powerful desktop, but you get the point.

ModRQC did mention "make sure you have a look at physical cores, as the core clock is to be multiplied at least by each physical core to really compare CPUs between them - and then added logical cores provided by Intel's Hyper Threading may potentially add more performance on top of that, in specific applications only". The in specific applications only part is important, and thus, I would not say that the core clock should be multiplied by "at least by each physical core". If everything you do is multithreaded, that works. However, most of us have at least some single-threaded bottlenecks, sometimes the vast majority of the bottlenecks. In that case single-core performance is all you really care about.

But since turbo actually works, the 4600U should indeed be faster than the T9900. Whether it will actually be fast enough to notice, I'm less sure of. There's a lot of other factors at play, too - my Core 2 Duo T7500 laptop feels faster than my Core i7 3520M work laptop. But that's largely because the software installed on it is less demanding (and there's less of it - no corporate-required bloatware), and also in part because the hard drive on my E5530 is amazingly slow for a 7200 RPM drive (it's a Seagate encrypted drive, and the seek time is horrendous). Half the time it's pegged on the CPU, and I'm wishing I had a desktop quad-core i5 or AMD FX octo-core instead of a mobile i7, and half the time it's waiting for the hard drive and I'm wondering how many times over an SSD would have paid for itself in time I wouldn't have been staring at the hard drive light, unable to do anything. So make sure the new laptop is at least as good as the old one in other areas, too, or it might not feel faster.

 

Yes, considerably faster.

 

If you are looking for performance and dont really care about battery life then hit up a non ULV processor, but depending on what you are doing you may never notice the performance hit and prefer the %50 battery life the ULV gives. Btw the 4600U scores almost twice as much as the core 2 on passmark.

 

See:
ARK | Compare Intel® Products


See:
PassMark - Intel Core2 Duo T9900 @ 3.06GHz - Price performance comparison

See:
PassMark - Intel Core i7-4600U @ 2.10GHz - Price performance comparison

See:
PassMark - Intel Core i7-4600M @ 2.90GHz - Price performance comparison



As everyone mentioned and backed up by the above links, the i7-4600U is greatly preferred over the T9900 not only in raw performance, but features, capabilities and less than half the TDP needed to cool it off too.


Something to keep in mind though is the fact that while all Intel 'u' cpu's are created equal, not all chassis' are (Dell is a big problem in this area as is Apple too in my experience) at keeping the cpu cool over extended work sessions or worse; by actually hampering the performance with BIOS limitations that may be impossible to work around (so in essence, you will be getting netbook class performance).


See:
http://forum.notebookreview.com/har...531329-throttlestop-guide-41.html#post9622581


In the link above the, unclewebb clearly shows his dislike for 'u' processors and how they can greatly limit performance (keep reading that thread and he'll also state that a non-'u' processor is just as efficient at idle but has greatly increased performance in a chassis/BIOS configuration that allows it to run at the TDP stated.



With all the above info, I would be hesitant about recommending a Dell with a 'u' processor - even though they should run circles around the old platform you are replacing.

An additional inch or two (15.6") notebook will offer not only better cooling and a true quad core inside with similar idle run times - but it will actually quadruple the performance you once had and be able to sustain that workload indefinitely.


Unless you can be assured/guaranteed that that particular model doesn't artificially limit the 'u' cpu's performance because of BIOS or cooling design concerns, I would stay well away from it.

If nothing else; don't try to fix it if it is ordered already; make sure you stay within the Dell return window and don't be stuck with it for another 3 or 4 years at subpar performance. (Subpar not only for the general field of platforms you can buy today, but for the specific platform too).


Hope this helps.


Good luck.

 

An i7-4600u with proper cooling (which the Dell E5530 seems to have) is roughly 35% faster than a T9900. I think it matches the Core 2 Quad Q9100 or even QX9300, actually.

 

Just curious...

 

The i7-4600u is going to be faster, generally. It is limited by a 15W TDP, so if you stress the video card at the same time, it will limit the processor's maximum clock. Also, it can only maintain 3.1Ghz for about 15 seconds, after which it settles around 2.6Ghz (this is if you are loading both cores 100%).

If your work could benifit from a faster processor (which it sounds like it could..) then ask IT for an E6440. That series has full voltage processors which will be faster than the ULV variant that the E5440 uses.

 

I took some interest in tillers info about ulv processors being limited a second time due to poor cooling conditions on a per laptop bases so I decided to drag out prime95 for my Q550lf laptop. A few interesting things happened. This is on an i7 4500u. That is 1.8ghz stock, 3ghz turbo.

1) This left me perplexed. Running a single prime thread showed %25 usage over all 4 processors (2 core, 2 hyper). How does this happen? Did intel magically make single threads multi threaded? Anyways while the laptop idles at 3ghz, it quickly dropped to 2.6ghz with loading all cores to %25.

2) 4 threads loaded up all 4 processors and the fan slowly ramped up. It held 2.6ghz for about 60 seconds before dropping down to 2.2ghz.

3) The processor held at 2.2ghz indefinitely even after the temps and fanspeeds dropped dramatically, only jumping to 3ghz after I stopped prime.

So it looks like my laptop has a really good cooling solution, and will keep the cpu running at a min of 2.2ghz, and for MOST games and applications will probably run 2.6ghz (I didn't do any integrated gpu test to go along side the cpu, the laptop comes with a dedicated 745m) all the time. In games even if at some point it drops to 2.2ghz, temps would drop and the next time the cpu load lets up it would jump back to 2.6.

It would probably run 3ghz more often but it seems to be awfully good at distributing loads over multiple processors, and that is probably what automatically drops it. Maybe this changes on low power modes when it wants to conserve power by shutting down cores? I'll have to humor myself later.

 

@ 1) It's not Intel, it's Windows scheduler. You can manually change that in Windows Task Manager/Processes/[right-click on a process]/Set Affinity.

 

The scheduler in Windows 8 is more efficient than in previous OSes and it splits up even single-threaded tasks more evenly among the available cores/threads. Which is why AMD FX CPU's, which rely heavily on multi-threading to achieve their peak performance, run better in Windows 8 than in Windows 7 unless you download a couple Windows 7 hotfixes from Microsoft which include a scheduler update and a core parking update.

 

Wow, looks like I'm gonna be getting windows 8 for my gaming desktop then. I downloaded some fixes labeled for the FX but I don't think any of them were magic to break up single threaded processes. Fall update can't come soon enough now. -_-

 

It's more efficient to keep a single threaded task on the same core, however depending on the dispatcher it may have something else running on there when the single threaded task comes around for it's next quantum so gets shoved onto a different core.

 

I've noticed that PlanetSide 2, which is a single-threaded game, mostly loads core 1 in Windows 7 but in Windows 8 it loads all four cores pretty evenly. However, overall CPU usage and performance is the same between the two OSes.

 

cpumark99 use only 1 thread 1 core.

 

It's worthwhile consulting Notebookcheck's mobile CPU benchmark list.

They put the i7-4600U at #93 while the T9900 is at #137 (in close company with the i5-4200U).

You can study the individual scores to see the differences.

John

 

Even easier, here they are narrowed down: Mobile Processors - Benchmarklist - NotebookCheck.net Tech

 

No matter if apps are single or multi threaded in Win 8. And I was supposing all along that user will probably be performing MORE THAN ONE task at the same time, like most people do. Then it matters even less if apps are single or multi, you're running multiple apps...

But I fully agree with everything else that has been said about how limited ULV processors are. I could never make do with one, personally. Great for battery... crappy for just about anything but browsing and editing docs in Office...

 

Do you have a source for that about Windows 8? For actually processes within Windows itself, I'm sure it is more efficient, and more multithreaded, and I do recall hearing something about AMD FX CPUs benefitting from Windows 8. However, for something that is explicitly single-threaded and will be using the same executable in Windows 8 as Windows 2000, I can't see Windows 8 magically speeding it up to an appreciable degree.

From what I've seen, it's relatively rare in regular usage to have a bottleneck on all cores at once, so I argue against multiplying cores by speed per core in general. It's more likely that someone who falls into the bottleneck on all cores category is aware that they'll benefit from more cores than someone who's in a bottleneck on one core category is aware that they won't benefit from more cores.

 

Here's a single threaded SuperPi run on W7

You can see the single thread being reassigned over different cores as it runs.


Here's a single threaded SuperPi run on W8.1

Also being reassigned but less so than W7 above.


How threads will be allocated with no specific affinity set will depend on what else is also running so results may vary somewhat.

Shame there isn't a performance benchmarklist that shows what the processors can be or can not be pushed to. For instance OC'ing the T9900 by a fair amount to improve performance which can not be done for the 4600U. The 4600U might give better figures with the TDP limit disabled. And so on.

The OP mentions Matlab which if taking advantage of the newer AVX instructions should see a huge advantage for the 4600U. Also mentions a 14" preference, there are 14" laptops with HSW quads but if you want better cooling and bigger battery then the bigger, heavier and more expensive laptops would IMO be of benefit.

 

I am interested in what you consider is anything else? My laptop has the i7 4500u, with the 745m gpu. I can handle every movie format, including 10bit h264 and h265 compression. I can also play any and every current gen game with medium or better settings.

The only thing I wouldn't do is video encoding, and heavy gaming. If you need either one of those in a laptop then you will pay through the nose.

 

To describe the 100% CPU issue more, at times it pegs at 100% of at least one core and the second core is at 75-100% and I'm very crippled to do anything but painfully wait to perform tasks. I've seen this with two different hard drives, w/ and w/o Lotus open, w/ and w/o Adobe Acrobat open, w/ Firefox or Chrome open, w/ no browser open, w/ and w/o my VPN connected. As I'm writing this it jumped from 50% to 100% on one core and the other sits at 75% for no known reason & w/o me doing anything new in other apps or in the background. All I can guess is that there is some issue w/ the memory or CPU that is messed up. It has been doing this since the machine was about 1.5- years old. It's now 4 years old and IT has finally said yes to a replacement.

To some of the other comments: thanks very much for the detailed technical input (a lot of which is over my head). I know there is Turbo on the new computer but wasn't thinking/understanding that I could rely on Turbo for single-threaded operations etc. and not knowing Turbo was something that worked well on the new machines. I had done my best to research my question before posting here by looking up the processor specifics -- sorry if my question seemed ignorant to some readers.

IT is now agreeing to an E6440. I'm going to get that. Battery life isn't a big factor for me either btw. I need to be portable as I work in many different locations, but I'm most often within reach of a power outlet. (so totally different criteria than I'd have for a personal laptop actually). Thanks everyone for your wealth of info!