• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

frequency vs fsb vs cache vs cores vs onboard memory controller?

p3ndldqny2

p3ndldqny2

p3ndldqny2

Associate
Joined
24 Oct 2009
Posts
138
Can someone either provide me with a link to a good tutorial, or just say it on here but how does one judge all these varying factors with regards to a CPU's performance?

It appears now we cant look at frequency, so i look at fsb and cache size, but then ive heard people say ignore the fsb etc

:confused:

To me, as a programmer, more cores doesnt really interest me because i know from a programming viewpoint its very hard to try utillise all cores.
 
Basically what you need to do is decide what it is your want your CPU to be good at and then research that. Or post here and we can discuss it.

If you want to work out the value of Pii down to the 32 mil mark then a 980X is king. But if you want to render a media file then the top of the hill AMD hexcore is king. You see where I am going here. Horses for courses.

By and large currently Intel holds the performance crown currently.
 
By and large FSB doesn't make that much odds anymore. Back in the days of 66 vs 100 vs 133, yeah... now it's arguable if it even exists :)

Cache is, to the best of my knowledge, great for number crunching. Photoshop, video rendering, etc, I think do much better with a large cache. Arguably compile times might be better, but unless you're working on something utterly gigantic, any modern CPU will chew through it just a few seconds anyway.

Cores... well, obviously, side by side running ^^ Depending on what you're programming that may or may not follow naturally - I've been lucky enough to be given a project that can pretty much multi thread indefinitely, so I'm spending my time working out the best ways to make my program scalable from dual to quad to N :) Such work does exist, it's just a question of what your job usually requires as to whether you'll ever need to.

But I agree with 1Day above - start by deciding what you want to do with your CPU. Half the population of this forum can discuss AMD vs Intel vs dual vs quad for days :)
 
eddiew said:
By and large FSB doesn't make that much odds anymore. Back in the days of 66 vs 100 vs 133, yeah... now it's arguable if it even exists :)
Click to expand...


FSB is how you get your multiplier limited CPU to higher frequencies, it is the clock frequency on which the CPU, HT link, north bridge and memory depend.
 
eddiew said:
Cache is, to the best of my knowledge, great for number crunching. Photoshop, video rendering, etc, I think do much better with a large cache. Arguably compile times might be better, but unless you're working on something utterly gigantic, any modern CPU will chew through it just a few seconds anyway.
Click to expand...

I dont play games, but i design real-time systems which do a lot of calculations per second/millisecond.

So at the moment im writing a real-time OS detection system, i would say calculations are done every second, but if i could make the period half a second it would be even better.

Therefore, multicore can be useful, but the ability to sequentially calculate many instructions per second is very useful.

I guess that would mean i number-crunch more?
 
i think cache can make a difference in the right scenario (then again, so does almost everything else). Taking the phenom II and athlon II. Both pretty much the same die except the phenom has a nice huge slab of shared L3 cache and the athlon does not. superpi 1m times between the two are almost the same (assuming same clockspeed and core count). super pi 32m and the phenom takes a significant leap ahead. Evidently the athlons cache is not large enough to hold 32 million digits in one go XD.

however, an extra core or two is probably going to be much better in the long run than a few more megs of cache.
 
thingemajib said:
i think cache can make a difference in the right scenario (then again, so does almost everything else). Taking the phenom II and athlon II. Both pretty much the same die except the phenom has a nice huge slab of shared L3 cache and the athlon does not. superpi 1m times between the two are almost the same (assuming same clockspeed and core count). super pi 32m and the phenom takes a significant leap ahead. Evidently the athlons cache is not large enough to hold 32 million digits in one go XD.

however, an extra core or two is probably going to be much better in the long run than a few more megs of cache.
Click to expand...

Do you know where i can get superpi results of various processors, like a graphical comparison?

I found this website:
http://www.cpubenchmark.net/high_end_cpus.html

but it seems to be a little too general, SuperPi results would be great
 
You can go to HWBOT and have a look at a couple of thousand CPU comparison and Super Pi.

But based on what CPU you have been using I can say with certainty that the i7 range of CPU's will offer a huge boost in work output.

However the next generation Sandybridge is due out very soon and will be the next main Intel product line for a while. So I would wait a few more months and go that route. The on-board cache makes all the difference.

I will go over to HWBOT and have a look for a link that shows you the data you want to see.


HWBOT


And go have a look at AnandTech too.
 
Last edited:
p3ndldqny2 said:
Do you know where i can get superpi results of various processors, like a graphical comparison?

I found this website:
http://www.cpubenchmark.net/high_end_cpus.html

but it seems to be a little too general, SuperPi results would be great
Click to expand...

cant find my exact example (i have seen it somewhere...), but i found similar:
http://www.pcstats.com/articleview.cfm?articleid=2489&page=7

compare the athlon ii 635 (3.1ghz quad) with the phenom ii 955 (3.2ghz quad). It does 1M, 8M and 16M. Notice how the phenom pulls away more and more in each test. Annoyingly that site hasnt reviewed the athlon 640, but like i said, its close enough for me to make my point.
 
You're never going to evaluate CPUs (and the platforms they work on) with tech specs like clock freq, cache, fsb, etc etc. Best is to look at benchmarks for applications similar to yours. With such a unique app as what you're coding, even that could be of limited use and you'd be better testing for yourself.
 
thingemajib said:
cant find my exact example (i have seen it somewhere...), but i found similar:
http://www.pcstats.com/articleview.cfm?articleid=2489&page=7

compare the athlon ii 635 (3.1ghz quad) with the phenom ii 955 (3.2ghz quad). It does 1M, 8M and 16M. Notice how the phenom pulls away more and more in each test. Annoyingly that site hasnt reviewed the athlon 640, but like i said, its close enough for me to make my point.
Click to expand...

You couldnt possibly imagine how much i'd love to get my hands on one of those Xeons with 24MB of L3 :p !!!!!!!

(Thanks for the link)
 
1Day said:
You can go to HWBOT and have a look at a couple of thousand CPU comparison and Super Pi.

But based on what CPU you have been using I can say with certainty that the i7 range of CPU's will offer a huge boost in work output.

However the next generation Sandybridge is due out very soon and will be the next main Intel product line for a while. So I would wait a few more months and go that route. The on-board cache makes all the difference.

I will go over to HWBOT and have a look for a link that shows you the data you want to see.


HWBOT


And go have a look at AnandTech too.
Click to expand...

Thanks for this, i shall have a good look!
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom