Core i3 or i5

Toggle sidebar Toggle sidebar
M

mush08

Reputable
Apr 10, 2015
5
0
4,510
Hi, would like to know if there is any significant difference nowadays between the two since most games use hyper-threading will the 4 threads on a core i3 be similar to the four cores of the i5.
I know there are benefits of having actual cores than threads but just want to know how much of a difference is it since now apps and games support HT technology do they use the two threads like two CPUs.

I also heard that the 4th gen CPUs had heating problems is that only for overclocking or even for gaming on locked CPUs.

Is it worth paying double the price for an i5

Thanks
 


So a thread running a process is not similar to a core running a process, sorry im not sure how the two exactly differ.

Also does the integrated GPU in the 4th gen processors help when you have dedicated graphics card or is it only for the onboard VGA.

Any idea on the heating issues?
 
The difference between threads and cores with hyperthreading, an i3 can schedule 2 threads per core but still only has 2 cores doing the work. An i5 can process all 4 threads at the same time. Hyperthreading is basically a thread scheduler, if the core pauses while processing thread 1 it can jump to thread 2 but has to swap back and forth. That's why ht only offers a slight increase in multithreaded situations.

For gaming you'd want a dedicated graphics card. The igpu is basically a new version of onboard graphics but it's better than the old vga chips on motherboards. The igpu is plenty to do web browsing, office programs, light photo editing, youtube/dvd/bluray playback etc. Maybe even some light gaming if the games are lower resolution and visuals turned down some but probably not much for aaa titles like farcry4, crysis etc.

Heat issues were when intel moved from the soldered ihs (metal cpu plate) to one with a thermal paste inside between the ihs and physical chips. They improved it a bit with devil's canyon. So far on my chip I haven't had any heating issues at all. It's a gamble with binning and overclocking, I ran out of voltage headroom on my cpu. Trying to go for 4.7ghz required more voltage than I was comfortable with, around 1.35 to 1.36. Backing it off to 4.6ghz I can keep vcore at 1.28v. As far as temps, the ambient room temp of my pc room is around 70f/21-22c right now. It's idling at 26-27c. After 10 passes of prime95 v26.6 using small fft's, the highest core temp reached was 74c and mostly stayed around 72c. That's using an air cooler, not water cooling or anything so no real heat issues.
 


So its actually just one core doing all the work but with hyper-threading it can skip tasks that have to wait for other processes to complete and work on another task. So intel charge double the price for that :pfff: anyway better an i5 than i3.
I had a dual core and managed to play most games on min settings then after some games started to lag i thought of switching to a core2duo but decided to get a new board and ram and bought a g3240 and only noticed a slight performance increase still unable to play new games.
About the heating issue you mentioned that was solved after the devils canyon but i was planning to go for a locked cpu not the k series cause here it would be about $60 more.
I also saw that the i7 5820k doesn't have an igpu so if we were going for processors that have it and we have a dedicated vga we are paying extra for nothing.

 
I think the heat issues are relative. Maybe compared to sandy bridge the haswell have 'heat issues' but in the larger picture of things I don't really consider them that hot. Many cpu's hotter than haswell. Not sure how much if any performance the 5820k adds to gaming. Keep in mind it's a more expensive platform (x99) all the way around. Motherboards, ddr4 ram etc are all going to cost a bit more. Stock turbo is only 3.6 which is right around where the 4690k start and below where the 4790k starts in terms of core speed. It's a k series so it's overclockable but likely with a little less headroom due to having 6 physical cores on the same size die meaning heat will build up faster.

There are locked core i5's, if you're concerned about hyperthreading you can go with a xeon but it's a bit more than the i5's. Some people may see an igpu as a wasted expense, I see it more or less as included value. Even though I use a dedicated gpu, if my graphics card dies my machine isn't totally unusable. I can simply switch over to the graphics port on the motherboard and continue to use my pc even if it's not great for gaming. Most mainstream desktop cpu's have the igpu, the i3's, i5's, i7's (lga1150), not just a select few premium versions. The number of intel cpu's without integrated graphics are fewer, only some of the budget xeons don't have the igpu and the x99 extreme series 6-8 core cpu's. You could either look at it as paying extra for nothing or paying what you'd probably be paying anyway and getting a bonus.
 
An i3 has the throughput of about three practical cores (allowing for the two real cores and hyperthreading), the i5 has four practical cores (real cores), and the i7/xeon behaves as if it has about 5.5 practical cores (four real plus hyperthreading). This is based on general use benchmarks and assumes identical clock speeds. (3.5 Ghz)

I don't think that the price is double, in the USA at least, $110 for 4160, $180 for 4460. When I had to make that choice, for a machine where gaming was important, but not the only thing, I went with an i5.
 
Hyperthreading is a bit more than a thread scheduler. Some parts of the processor such as the registers are physically duplicated on the die, just not entire cores.

In terms of performance increase attributable to HT, it's highly variable. The best rule of thumb is 0-30% over not having it. Note that although i5s don't have HT, they do have Turbo Boost.

Comparing an i5 at double the cost of an i3? In the UK, that's pretty much an i3-4150 vs an i5-4690 which conveniently have the same base clock speeds (although the i5 does have Turbo Boost). Looking at the user benchmarks, you're talking about a 50-75% improvement of the i5 over the i3 where multicore stuff is concerned.

In short, on the one hand an i3 is very capable when it comes to general usage and games playing, on the other an i5 will offer a significant improvement, but not double. My general advice is to get an i5 if it can be afforded without skimping on other parts, but where there are budget constraints an i3 is certainly not a budget performer.
 
There is quite a misunderstanding within users at the moment about the i3 & i5 processors in regards to gaming.

Please see this recent article regarding gaming on i3 and i5 for some valuable insights: http://www.techspot.com/review/972-intel-core-i3-vs-i5-vs-i7/

In many cases, you will not be able to tell the difference between an i3 and an i7 - a lot depends on the rest of your equipment and specifically your monitor capability.
 
I've never heard of nor seen 50% performance from hyperthreading. I've seen 20-30% referenced as best case scenario though it hardly ever happens. Real world performance improvements from hyperthreading alone are usually around 10-12% on average or less. The reason I say hyperthreading alone, many times the i5 4690k and i7 4790k with ht are the two that get compared. Performance at first glance seems closer to 30% in some situations but what isn't addressed is the stock out of the box core speed differences which are partially to credit for the improvement in performance, not the ht on it's own. Once both cpu's are oc'd to the same frequency to eliminate the speed difference, the performance gap given from the extra 2mb cache and ht are much narrower and more indicative of real world ht benefits.

Just thought I'd post this since it's one of the first reviews I've seen on gtaV, a brand new game released that is more heavily threaded than many past games to get an idea of where the various cpu's perform. The i3 performs decent at 68fps while the i5 hits 81fps, just a couple fps behind a ht enabled i7 and 8 core haswell 5960x.

http://www.techspot.com/review/991-gta-5-pc-benchmarks/page6.html
 
I think it's a mistake to judge the value of hyperthreading in an i3 by looking at the improvement hyperthreading gives in an i7 over an i5.

Tech Buyers' Guru had a fair stab at evaluating the benefits of hyperthreading in gaming. They compared a Pentium (as equivalent to an i3 without HT as they could get), an i3, an i5 (ditto an i7 without HT) and an i7. HT always offered a distinct improvement on an i3 yet hardly any on an i7.

The fps improvement HT on the i3 offers does vary drastically, e.g. 5% on BF3 single player yet 50-75% in multiplayer; 25-30% in Hitman:Absolution; 25-50% in Tomb Raider. FarCry 3 showed the most distinctive gains - average fps went from 25 to 44 and minimum fps went from 16 to 35. That's the difference between being playable and unplayable.

So HT always gave at least some boost on an i3 and in some cases put it not far behind the i5, whereas using HT on the i7 frame rates barely changed and sometimes even dropped: Which is why I say it's wrong to talk down HT on an i3 by pointing to the lack of improvements it gives on an i7.

Same with the GTA V benchmarks above (though unfortunately a non-HT dual core can't be benchmarked on that game). One way of looking at it is that HT gives an i7 barely any improvement over an i5; another way is that HT allows a dual-core processor to run at six-sevenths the speed of a quad core. I'd call that a real-world improvement.
 
An i3 does have the most to benefit from ht, my comments were more or less that ht is unpredictable. It really varies from game to game and cpu to cpu despite being the same technology. The ht on an i3 is no different from that on an i7. It's not easy to find, at least in my searches but I did come across a few test sites that tested various cpu's with ht on and off.

In deus ex, the i3 3220 gained performance from ht at around 17%. The i7 3770k only gained around 7%. Fc3 was impressive, around 75% performance gain for the i3 while at the same time a game that showed ht performance boost for the i3 saw a 4% performance drop for the i7 with ht on. Hitman absolution also showed a decent performance improvement for the i3 + ht but again a negative effect (slight) with ht enabled on the i7. Bf3 the i3 saw a 12% improvement with ht on, the i7 once again saw a slight hit with ht enabled. It very erratic and unpredictable, ht in general as well as performance boosts on the i3. 50% performance increases aren't the norm but the exception.

http://techbuyersguru.com/CPUgaming.php

In 3dsmax, ht hurt render performance. The i3 actually rendered faster without ht enabled. Comparing versions 2011, 2012 and 2013, the results were similar except for version 2012 which saw way worse performance with ht enabled. 2011 and 2013 scaled better. Ht in and of itself is no guarantee, either in games or other applications. It's a toss of the coin if it's going to help or hurt performance and by how much is a complete crapshoot from one program to another.

http://www.extremetech.com/computing/133121-maximized-performance-comparing-the-effects-of-hyper-threading-software-updates/2
 
Throughput, the word I used, is not the same as performance.

Any Intel processor running one thread is going to 'scale' mostly at it's clock speed, with allowances for the various cache's and bus speeds and widths.
The same for two threads, where a G3258 at 4.8Ghz is essentially the same as a 4.8Ghz 4770K throughput.

As the number of threads increases, the processor become 'saturated' and has to start scheduling and prioritizing things.

If a game uses a lot of threads, it will do better on a multi-core processor, and Hyperthreading simulates additional cores, but is limited by the actual cores.

Most games use few enough threads that an i5 can handle the load with its four cores, most of the time, which is why an i7 shows very similar performance.

Benchmarks need to be taken with a grain of salt because the benchmarkers are running only the game in question, with little to no other processes beyond OS and monitoring. I know that when I am gaming, I have an e-mail feed, and various other processes going on; I do not shut everything else down to game and so do not get the theoretical performance.
 
Your actual comment was
"hyperthreading is basically a thread scheduler...That's why ht only offers a slight increase in multithreaded situations.
Click to expand...
which is frankly wrong in a thread titled 'i3 or i5'.

Clearly not for an i3, since it the i3+HT keeps outperforming the dual core. All you're doing is saying 'HT isn't much good on an i7 so it's not much good on an i3' which is flying in the face of the benchmarks.

So clearly the effects of HT aren't the same - that's why it's pointless talking about what HT does for an i3 by comparing i7s and i5s.


It's not an i3. They're using an HP Z620 which runs on dual Xeon CPUs - those are eight core processors.

Here are some 3dsmax benchmarks where you an see a 3.5 GHz i3 compared to a 3.3 GHz Pentium - in the 2013 version it renders more than twice as fast. Is that all down to the extra 200 MHz and 1 MB cache, or is hyperthreading helping?
 
In vegas pro, the i3 rendered 26% faster than the 3258. Not all of that is hyperthreading considering the i3 has a 300mhz faster base clock stock to stock, 25% more cache and 25.6 gb/s memory bandwidth vs the 21.3gb/s of the pentium. More than just ht is at play there so it's still not really an apples to apples comparison.

Obviously the 2013 version of 3dsmax had some odd behavior with the pentium. You can see how things scaled with 2012 yet only the pentium suffered badly during 2013. It's happened with other apps and other cpu's where odd unexpected anomalies have happened.

Since I was corrected earlier on ht, just thought I'd share.
"Hyper-threading works by duplicating certain sections of the processor—those that store the architectural state—but not duplicating the main execution resources. This allows a hyper-threading processor to appear as the usual "physical" processor and an extra "logical" processor to the host operating system (HTT-unaware operating systems see two "physical" processors), allowing the operating system to schedule two threads or processes simultaneously and appropriately."
http://en.wikipedia.org/wiki/Hyper-threading

"Because they still share so many resources, only one thread will technically execute at once. The advantage of adding the HT logic is that if a thread is executing and stalls for any reason, the other thread can be switched in very fast while the cause of the stall in the first thread is addressed."
http://www.overclock.net/a/hyperthreading-explained

So basically, in simple terms without digging too deep into details, yes - hyperthreading is in essence thread scheduling. Despite sharing other resources, the fact remains that two threads are not executed in parallel since the physical cores can only execute one at a time. It can switch back and forth, but at the heart of it it's still processing one _then_ the other. Not true simultaneous execution.
 
You'd get a lot more out of this, and save yourself some time, if you read what was already written. I'll just requote myself rather than explain it all again.

In terms of performance increase attributable to HT, it's highly variable. The best rule of thumb is 0-30% over not having it...HT always gave at least some boost on an i3 and in some cases put it not far behind the i5, whereas using HT on the i7 frame rates barely changed and sometimes even dropped.
Click to expand...

In short, on the one hand an i3 is very capable when it comes to general usage and games playing, on the other an i5 will offer a significant improvement, but not double. My general advice is to get an i5 if it can be afforded without skimping on other parts, but where there are budget constraints an i3 is certainly not a budget performer.
Click to expand...

Hyperthreading is a bit more than a thread scheduler. Some parts of the processor such as the registers are physically duplicated on the die, just not entire cores.
Click to expand...
 
I read it the first time you posted it. I also read the links I provided and was just taking the opportunity to clarify that some of the things I said previously were correct despite being argued.

No one said the i3 is a bad cpu, certainly a budget option. Depending on the game though the weakness of being dual core vs quad core is beginning to show. Bf4 at 1080, the i3 has min fps of 95-98 where the i5's are between 135-142fps min. On a 60hz monitor not a major deal but could pose a problem for those running on 120hz monitors. Frame times are also around 25% better for the quad core.

Sleeping dogs and tomb raider showed little difference, metro last light however the i3 drops to 38-39fps while the quad core only drops to 50-52fps while average framerates on the i3 are 30% slower than the i5. Starcraft saw similar with low min fps drops and higher frametimes from the i3 compared to the i5.

http://www.hardwarepal.com/best-cpu-gaming-9-processors-8-games-tested/

In non gaming tasks, ht performance becomes even weaker in many benchmarks.
http://www.anandtech.com/bench/product/1197?vs=1261

The i3 does show a lot of improvement over the g3258 which I would say is to be expected. The i3 4360 is $130, the g3258 is $65 - putting the i3 over double the cost of the pentium anniversary. For less than the price difference between the 3258 and i3 someone could easily have the i5 4590.
 
That's good that you realise that now, because originally you said that "ht only offers a slight increase in multithreaded situations". Anyway, on pricing...


Leaving aside that $130 is not over double $65, you've picked out one of the more expensive i3s to compare to one of the cheaper ($180) i5s, which also isn't the i5 in those Anandtech benchmarks but a slower one.

The i5 4690K in the benchmarks is $240, $110 over the given i3-4360. Moreover an i3-4160 can be had for $100 - $35 more than the Pentium, $80 less than the i5-4590 and $120 less than the price of the i5-4690K in those benchmarks, whilst being only 100 MHz (3.6 vs 3.7 GHz) slower than the 4360 - in short, talking about 4360 pricing is pointless since nobody with sense would buy it over the 4160.

And in those benchmarks, the i5 never gives double the performance of the i3. Sometimes it comes close, most of the time it's maybe 25-50% better.


The OP (since answering him/her is what it's all about) asked whether there was a significant difference between an i3 with HT and an i5. They knew that physical cores are better than logical but wanted to know by how much. They wanted to know if an i5 is worth paying double for.

There's usually a significant difference in benchmark performance, but never double and in a lot of games it's less than twenty percent. Whether it's worth paying double for an i5, ultimately and as always, comes down to what an individual will be using their system for and what they desire: e.g. whether 95 fps minimum in BF4 @ 1080p means weakness in a processor is a matter of opinion - some might be very happy with that.

 
It would be nice if new games continue to scale well like gtaV does. As it is, just a few months ago the 3258 was considered 'good enough' in the majority of games for a nice little inexpensive overclocking cpu. Not even a year later people aren't recommending them anymore and they're quickly losing ground in gaming performance aside from what are progressively becoming 'older' games. I imagine the i3 has a bit more life in it than the 3258, the cut off between 2 threads and 4 is a steep one given current gaming needs. 4-5yr old k series i5's are still quite capable and continue to push on even with current games and gpu's. They're more expensive yes but with a longer viable system life to go with. I'm sure the i3 falls somewhere in the middle.

Each game has varying requirements, so trying to find a blanket cpu at the lower end of the budget is a bit more difficult. Just like deciding what software for editing someone's going to be using or likely to be using, same goes with games. Each person's needs are different. Some prefer to go with budget setups and upgrade more frequently, others who pay out more at the onset can typically go longer on their system before needing to upgrade. $100 extra for a system that should last a good 3-4 years or more isn't that big of a price premium, especially given the cost of gpu's and those are usually upgraded more often.

Like you said, depends what else they do with their pc. I chose an i5 but I use my system for more than gaming so it makes a substantial difference. Anymore a 'decent' gpu for most games is an easy $300. Some will upgrade every year, others every other year. Spending $600 more or less in 2-3yrs on gpu's, $100 difference across 4yrs for a cpu which will push higher fps throughout that time isn't that bad of a premium. If the op knows for the next couple of years they won't be using a 120hz monitor the i3 may be fine. If in a year they have the funds, wish to move up to a 120hz monitor they may find the i3 struggling during some titles.
 
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom