Core i5-2400 Sandy Bridge versus i7-960 Bloomfield

Poppa Bear

Posts: 228   +9
I have searched Google and the Intel site to try and establish which CPU has the fastest data transfer rate in terms of opening programs, copying/moving data, etc. And I can't seem to find any definitive answer.

i5 is quad core while i7 is quad core with hyperthreading, having 2 threads in each core. This gives a total of 8 threads.
 
Opening programs and copying/moving data is typically going to be storage constrained vs CPU. I've got a machine with a 2400 in it and had a slightly overclocked 930 and to be honest there was very little difference between the two machines.
 
Hyper threading analogy

Opening programs and copying/moving data is typically going to be storage constrained vs CPU. I've got a machine with a 2400 in it and had a slightly overclocked 930 and to be honest there was very little difference between the two machines.
Thanks for the input LNC.

The reason I'm interested in comparing these two CPUs is because I have the i5-2400 Sandybridge CPU in my back-up computer which has a SATA3 SSD Hard Drive, and Vista Ultimate, In my flagship I've got the i7-960 Bloomfield CPU, with same SATA3 SSD HD and Win7 Ultimate.

In general desktop work, the flagship with i7 runs noticably faster than the i5 with Vista. However, Vista with 8GB of RAM and SSD HD, runs at a very respectable speed and is not that far behind the i7. I could certainly comfortably live with it if necessary.

On CPU Benchmark and a few other test sites, the i7-960 Bloomfield was rated significantly higher than the i5-2400 Sandybridge, even though the actual core speed for the i5-2400 is slightly faster than i7-960 in the Intel specs. I suspect the difference is due the i7 having hyper-threading with 2 threads in each core. This is a picture of the results in CPU Benchmark.


35aromx.png

However, it's at this point I need information from someone who is more knowledgable about CPUs than I am. As I understand it, hyper-threading doubles the band width, which in effect allows double the amount of data to be processed, without actually increasing the native speed of the core.

This would mean that even though a single silicon core in an i5 is slightly faster, the i7 effectively has twice as many "cores", (hyper-threads), so it's processing twice as much data, even though it's at a marginally slower rate.

The closest analogy I can come up with, and correct me please if I'm wrong, is the comparison of two electric wires in parallel running at a slighty slower power rating than a single wire. Each single slower wire would deliver say only 80% of the Coulombs, or Farads if you prefer, than the faster single wire delivers. However, the total Coulombs delivered in any given time span by both wires in parallel is 80 + 80 = 160%. Which in computer language means 60% more data processed by the 2 hyper-threads in a single core than that processed by the slightly faster single silicon core.

Any input would be gratefully received.
 
I understand the same as you re hyperthreading;
the parallel wires analogy is a fair one (or maybe two
machines each working just a tad slower than a single
one). The more "parallelity" the better, if all else is equal :)
 
Thanks DocJohn. I have just been in touch with Intel live chat, and they pretty much confirmed what I said and you've described as "parallelity" ... good call! I'm about to check the Intel Benchmarks for both processors, but I don't think they rate them side by side except in terms of general specs. This is a copy of the transcript from live chat, with Intel techo speaking:

"The amount of cores available, same as the amount of threads will always influence. The cores are the units that read and execute program instructions such as add, move data, and branch; but the multiple cores can run multiple instructions at the same time, increasing overall speed for programs that require parallel computing (requiring the use of multiple cores). Therefore, Hyper threading will always also influence on the performance."

Update: Have since checked side by side specs of both CPUs and overall i5 Sandybridge is faster, but i7 Bloomfield had 25.6 bandwidth compared to 21 in i5.

The bandwidth % difference is roughly about the performance difference for data transfers that I have measured in comparing i5 and i7 on my two machines. This includes copying data packages of 10GB onto my hard drive from a USB3 external storage hard drive, and from a 2nd data storage partition on my main HD.

Cheers PB :D
 
Life was much simpler when you just read the clock frequency.....
I'd certainly have to agree with that statement.

The other reason I wanted to check this out was because I was told by a senior member in another forum that I was misleading people reading my post by saying my i7 was faster in processing data than my i5. And that all the benchmarks showed i5-2400 to be faster than i7-960. Yet every benchmark site I visited showed i7 as considerably better in performance; which backed up my own personal use of both CPUs. So I was curious to find the truth. I was also told by the same person that the number of cores had nothing to do with with the speed of a computer's perfomance. The Intel tech chat info shows that also to be untrue.

I don't personally have sufficient technical background in computers to know the fine detail of the working of CPUs, but I do have a background of tertiary level studies in electronics. Unfortunately at 71 years of age, I don't have the motivation to start studying up on stuff again. So I'll just content myself to read reviews! LOL!

Cheers PB :wave:
 
The other reason I wanted to check this out was because I was told by a senior member in another forum that I was misleading people reading my post by saying my i7 was faster in processing data than my i5. And that all the benchmarks showed i5-2400 to be faster than i7-960. Yet every benchmark site I visited showed i7 as considerably better in performance; which backed up my own personal use of both CPUs. So I was curious to find the truth. I was also told by the same person that the number of cores had nothing to do with with the speed of a computer's perfomance. The Intel tech chat info shows that also to be untrue.
Well, a CPU with a faster clock, (assuming the same number of operations per clock cycle), will obviously process data faster than a slower clocked unit.

The end result is software dependent. Only if "X" software is capable of utilizing all the cores and threads present, will a CPU with slower cores but more paths be faster.

Bench marking is typically done in many different ways, and like all other methods of generating statistics, it can be made to say want you want it to say.

Accordingly, if you run benchmarking software that favors many threads, CPUs with less threads / cores will suffer. That's pretty much a given, but not necessarily what you should expect, or even need, in your day to day computing. That is still predicated on a case by case basis.

Who's the fastest, and who has the most "bandwidth", is great for pissing contests, but an enormous waste if you're just surfing the web.
 
Well, a CPU with a faster clock, (assuming the same number of operations per clock cycle), will obviously process data faster than a slower clocked unit.

The end result is software dependent. Only if "X" software is capable of utilizing all the cores and threads present, will a CPU with slower cores but more paths be faster.

Bench marking is typically done in many different ways, and like all other methods of generating statistics, it can be made to say want you want it to say.

Accordingly, if you run benchmarking software that favors many threads, CPUs with less threads / cores will suffer. That's pretty much a given, but not necessarily what you should expect, or even need, in your day to day computing. That is still predicated on a case by case basis.

Who's the fastest, and who has the most "bandwidth", is great for pissing contests, but an enormous waste if you're just surfing the web.
That's interesting. The two main CPU benchmark web sites that showed i7960 well ahead of i52400 use actual user data supplied to their web site by running their software tester program. This program has about 7 or 8 different parameters tested. Then it compiles a score for your particular CPU based on those parameters. The end result you see on the benchmark chart which I've shown a picture of, is the average of the scores of literally thousands of tests done on actual PCs and the data collected by CPU Benchmark. Results are updated every day.

I tested actual time taken to transfer 10GB of data from an external USB3 hard drive to desktop; and from a 2nd storage partiton on the hard drive to desktop. And the i7 was significantly faster than the i5. The only problem is the CPUs have different sockets so can't go on the same mobo. Which means that even though the OS is the same, with bare bones installation and no other programs loaded, the hardware is not.

In those tests for CPU Benchmark and Futuremark, the i52400 doesn't have as many slots for RAM as the i7960, so most users are probably running i7 with a lot more RAM.

I guess there's really no definitive answer. From a practical point of view for my own personal use though, using the appropriate board for each CPU, and 12GB RAM versus 8GB, the faster machine by about 30% is the i7960. And if I'm using software that uses hyper-threading I imagine it would be faster stil?
 
Don't you think the 960 should be faster? It's a premium product running triple chalnnel RAM. If it wasn't faster, or couldn't do more work per unit of time, it would be monumentally stupid to buy one. Don't ya think?

I'm missing what you're trying to prove.

If Windows has detected all 8 cores/threads it should be able to use them while copying. Fire up the task manager and watch the pretty green lines bounce up and down. See what's really going on.

That doesn't say you're going to install DVD Shrink, a 6+ year old program, and have the same results.

You can't burn a DVD faster with a 960, than you can with an old E8400, (or less). The DVD drive won't let you.

So everything is task related.
 
Don't you think the 960 should be faster? It's a premium product running triple chalnnel RAM. If it wasn't faster, or couldn't do more work per unit of time, it would be monumentally stupid to buy one. Don't ya think?

If Windows has detected all 8 cores/threads it should be able to use them while copying.

I'm missing what you're trying to prove.
Yes I always thought the i7 would be a lot faster than the i5. However, as mentioned above, a post on another forum by a senior member stated that the i5-2400 was ahead of the i7-960.

This made me wonder if I had been "monumentally stupid" about a year ago when I paid $328 for the I7-960 for my flagship & $186 for the i5-2400 for my backup machine.

Re performance:

  • Task manager showed all 8 threads running, with between 3 to 7 threads being active depending on the program being run.
  • On the Windows 7 Performance Test for i7-960, out of a possible score of 1 - 7.9 the overall performance was 7.6 and the CPU was 7.6. In the same test, i5-2400 scored 5.4 for CPU.
  • 5 different benchmark sites which use scores from many thousands of actual PCs, and rate 7 - 8 different parameters, all put the i7 well above the i5. In fact I couldn't find one site that didn't.
That has restored my confidence. Thanks for your input.
 
You will only really notice the difference while video editing or rendering 3D scene's (or any number of other multi threaded programs). In day to day tasks you will hardly notice it... Even in gaming it shouldn't be too far behind.
Also those sb's love to be overclocked. So the matter of being "monumentally stupid" all depends if you actually need the extra power.
 
You will only really notice the difference while video editing or rendering 3D scene's (or any number of other multi threaded programs). In day to day tasks you will hardly notice it... Even in gaming it shouldn't be too far behind.
Also those sb's love to be overclocked. So the matter of being "monumentally stupid" all depends if you actually need the extra power.
Thanks for the input Dantrag.

I have noticed on the performance tab in task manager that hyperthreading kicks in, shown by more than 4 threads being used, when simply surfing the net, or playing videos on VLC Media Player, and quite a few other programs.

Even so, it probably is a bit of over-kill in terms of what the main PC is used for. However, I want as fast a machine as I can get for everyday use, without getting into the $1,000 range for a CPU. As was pointed out by Cap'n Cranky, speed of performance can be rate limited by certain applications, like MS Office 2010 Word, for example. It's slow no matter what's opening it. And ditto for optical drives.

I guess my preferences have been biased toward hyperthreading since the change in speed seen when single solid cores first became hyperthreading, which if my memory serves me right, was about 10 years ago. Then came dual core, followed by core 2 duo with 4 hypertheads, and accompanying increased performance with each new model.

So my thinking was, if 4 solid cores are good, then 4 solid cores with 8 hyperthreads should be better. But then that has to be balanced with a later version architecture in the cores of i5.

Having said that, with SATA3 SSD hard drives, and USB3 external HD storage, speeds can be quite blisering. I have all my personal data on a 2nd storage partiton on my main HD on a Corsair 128GB SATA3 SSD. It takes exactly 27 seconds to drag/drop 3.07GB of music files from the storage partition to the GUI. I've also got an Acronis image of Win7 with all programs loaded, (but no personal data), in the storage partition.

In the event of a total crash of the OS, it takes about 7 minutes using Acronis to restore the pristine image of Win7 which occupies 28.3GB. And about another 2 or 3 minutes to drag/drop all my personal data into the GUI, which includes Favorites, Contacts, email accounts imported, etc ... leaving the originals intact on the storage partition. Then add to that whatever time it takes to get updates for Windows and 3rd party programs.

From go to whoa that means the fully loaded OS can be restored in about 14 - 15 minutes. And that beats the hell out of doing a clean install.
 
  • On the Windows 7 Performance Test for i7-960, out of a possible score of 1 - 7.9 the overall performance was 7.4 and the CPU was 7.6. In the same test, i5-2400 scored 5.4 for CPU.
Sorry to butt in but how did you only get a CPU rating of 5,4 for the i5?
I have a Pentium G840 and my overall rating is 5,4 and my CPU rating is 6,7.
 
Sorry to butt in but how did you only get a CPU rating of 5,4 for the i5?

I have a Pentium G840 and my overall rating is 5,4 and my CPU rating is 6,7.
Because of the different sockets, i5 is running on DH67CL mobo. 5.4 was the performance score that was shown on the Performance meter in Computer > Properties. It stayed the same after doing refresh of the rates. More than that I can't tell you.

What I'd like to know is how you got a 6.7?
 
Here are some more scores for reference. I checked a few machines I have for WEI of the Processor (Calculations per second).
  1. Core2 Q9400 = 7.2 (I'm surprised this is higher than 7.1)
  2. i3-2100 = 7.1
  3. i5-2400 = 7.1
  4. i7-2600K = 7.6
 
Hi, sorry i missed that you were running the i5 on vista and the window OS's rate differently.
You should put 7 on there then test it but it would have been awesome if a Pentium could have got a higher rating than a i5,for me at least.
 
Here are some more scores for reference. I checked a few machines I have for WEI of the Processor (Calculations per second).
  1. Core2 Q9400 = 7.2 (I'm surprised this is higher than 7.1)
  2. i3-2100 = 7.1
  3. i5-2400 = 7.1
  4. i7-2600K = 7.6

BMfan said:
Hi, sorry i missed that you were running the i5 on vista and the window OS's rate differently. You should put 7 on there then test it but it would have been awesome if a Pentium could have got a higher rating than a i5,for me at least.
Oooopps! Yeah, was running the tests for i5 on Vista Ultimate 32bit. Loaded Win7 64bit on Intel DH67CL mobo and got totally different result. Both CPUs were run on SATA3 SSD Corsair HDs. Flagship has 12GB RAM & backup PC has only 8GB.

These are the results for Win7 64bit run on each of the Intel mobos DH67CL socket 1155 & DX58SO2 socket 1366. Obviously they can't be run on the same mobo due to different sockets.

CPU ----------------------------- i5: 7.4 - i7: 7.6
RAM ----------------------------- i5: 7.6 - i7: 7.6
Graphics ----------------------- i5: 5.2 - i7: 7.4
Gaming Graphics ------------ i5: 4.6 - i7: 7.4
Hard Disk ---------------------- i5: 7.8 - i7: 7.8

Not a lot in it for CPU speed. Graphics and gaming results would be expected as DH67CL mobo is using onboard graphics, and while still full high def, is not the same as a single Gigabyte GTX 550Ti graphics card, which is what I'm using on the DX58SO2 mobo ... I'm not using SLI bridge to 2 cards.

Also checked Futuremark Benchmark results here. i7-960 was equal to i7-965 Extreme @ 7550 and i5-2400 came in at 7030. As mentioned above, I couldn't find one benchmark web site where i5-2400 beat i7-960, using actual tests from thousands to millions of users who ran the web site software tester on their PCs, with up to 15 parameters tested, and results averaged.

[CENTER]
14l0g9u.jpg
[/CENTER]

In fairness to both CPUs, in gerneral usage a lot of users with i7-960 would be gamers and using SLI bridge connectors to two graphic cards, plus 3 channels of RAM supported by LGA1366 socket as opposed to two in the LGA1155 used by i5-2400.

Probably not al lot in it for general use unless the programs are being run on hyper-threading ... but then again, a lot of current programs are. This is backed up by the performance monitor results on Task Manager, which shows anywhere between 5 - 7 threads running on a lot of programs I've tested it on. Still haven't hit the 8 thread usage yet ...any suggestions on programs to try?
 
Back