[TechSpot] i3 vs i5 vs i7

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

toyota

Lifer
Apr 15, 2001
12,957
1
0
@toyota

Yeah, the Crysis bench looks improbable to me as well. Had this been the case, we all would have had i3's. Maybe the bench session wasn't heavy on cpu.
yeah there are some parts of the game that are not super cpu heavy but it still makes no sense for the 4790 to fall behind. worse case scenario should have them getting the same framerate. after all the 4790 is clocked higher w turbo and has more cache too so even if HT did not help it should have at lease been tied.
 
Last edited:

Magic Carpet

Diamond Member
Oct 2, 2011
3,477
233
106
Human error?

The guys were having a party, a few beers spilled over the keyboard, got the results mixed. #$#$ happens, you know.

Whatever i3's clock, it's way below 4790K, so no way the FPS should be lower, I agree.
 
Last edited:

SimianR

Senior member
Mar 10, 2011
609
16
81
Goes to show you that for strictly gaming, you can get away with an i3 and a fairly high end GPU and still do quite well.
 

toyota

Lifer
Apr 15, 2001
12,957
1
0
Yup, wish they had included the G3220
the game would not be playable in spots with that cpu and running very high settings. if anything in the background needed even 2 or 3% of that cpu the game would hitch as its basically pegged the whole time with the game.
 

SteveGrabowski

Diamond Member
Oct 20, 2014
8,957
7,666
136
With the lower driver overhead for Nvidia drivers vs. AMD (according to the video), wouldn't a GTX 970 paired with an i3 match an R9 290 paired with an i5 for the most part?

Anyway, the cheapest R9 280 on Newegg is $190 - same as the cheapest GTX 960. There is a very good Sapphire Vapor-X R9 280X for $210. That's the card to get if you have the CPU to support it fully. If you don't you might as well get the GTX 960 and save the power and heat - performance isn't going to be any worse.

Microcenter has an XFX R9 280 for $165 after rebate which can be ordered online, and newegg has a Sapphire R9 280 for $171 after rebate. Nah man, an i3 + GTX 970 won't touch an i5 + R9 290. An i3 does a little better in gaming benchmarks than does an FX-8350, which is still pretty far below even a stock i5-2500k, much less an i5-4440. Eurogamer has only seen this with i3, not i5.
 
Last edited:

SteveGrabowski

Diamond Member
Oct 20, 2014
8,957
7,666
136
@toyota

Yeah, the Crysis bench looks improbable to me as well. Had this been the case, we all would have had i3's. Maybe the bench session wasn't heavy on cpu.

Yeah, Crysis 3 chews up Core i3s in every other benchmark I have ever seen. It's so well parallelized even AMD CPUs can really hang.

proz.jpg
 

SteveGrabowski

Diamond Member
Oct 20, 2014
8,957
7,666
136
yeah there are some parts of the game that are not super cpu heavy but it still makes no sense for the 4790 to fall behind. worse case scenario should have them getting the same framerate. after all the 4790 is clocked higher w turbo and has more cache too so even if HT did not help it should have at lease been tied.

Maybe they did their Crysis 3 benchmark in the tutorial. :biggrin:
 

toyota

Lifer
Apr 15, 2001
12,957
1
0
Yeah, Crysis 3 chews up Core i3s in every other benchmark I have ever seen. It's so well parallelized even AMD CPUs can really hang.

http://gamegpu.ru/images/stories/Test_GPU/Action/Crysis 3/test/proz.jpg
thats before the 1.3 patch that added better HT support. every cpu with HT including i7 got a nice boost.


pic upload



and Crysis 3 runs a lot better with Haswell so thats why even the i3 does good in this game. still the 4790 should not ever be slower in Crysis 3 so something is up with the techspot results.
 
Last edited:

SteveGrabowski

Diamond Member
Oct 20, 2014
8,957
7,666
136
thats before the 1.3 patch that added better HT support. every cpu with HT including i7 got a nice boost.


pic upload



and Crysis 3 runs a lot better with Haswell so thats why even the i3 does good in this game. still the 4790 should not ever be slower in Crysis 3 so something is up with the techspot results.

Interesting, thanks for posting that. Where did you find that benchmark?
 

crashtech

Lifer
Jan 4, 2013
10,695
2,294
146
the game would not be playable in spots with that cpu and running very high settings. if anything in the background needed even 2 or 3% of that cpu the game would hitch as its basically pegged the whole time with the game.
The point wouldn't be to prove anything about the G3220, but to be able to see how much of an advantage HT confers to a lowly i3.

@SteveGrabowski re the old Crysis 3 graph, Haswell i3s do much better than would be inferred by the Sandy Bridge i3 result.
 
Last edited:

toyota

Lifer
Apr 15, 2001
12,957
1
0
The point wouldn't be to prove anything about the G3220, but to be able to see how much of an advantage HT confers to a lowly i3.

@SteveGrabowski re the old Crysis 3 graph, Haswell i3s do much better than would be inferred by the Sandy Bridge i3 result.
no it would serve no point at all because on higher settings the CPU would just be tripping on itself in many areas of this game. the average framerate may look okay overall but it will stutter like crazy in spots. I've already tried this running just two cores of my CPU which would smoke the crap out of that G3220. plus I still have the rest of my CPU that was taking care of any background tasks.
 
Last edited:

crashtech

Lifer
Jan 4, 2013
10,695
2,294
146
no it would serve no point at all because on higher settings the CPU would just be tripping on itself in many areas of this game. the average framerate may look okay overall but it will stutter like crazy in spots. I've already tried this running just two cores of my CPU which would smoke the crap out of that G3220. plus I still have the rest of my CPU that was taking care of any background tasks.
I know you don't need to see dual core results because you are a member of the quad core police, but whether you think so or not, there would be value in illustrating for others how badly a 2C/2T CPU performs vs those with higher thread capabilities.
 

toyota

Lifer
Apr 15, 2001
12,957
1
0
I know you don't need to see dual core results because you are a member of the quad core police, but whether you think so or not, there would be value in illustrating for others how badly a 2C/2T CPU performs vs those with higher thread capabilities.
why don't you pay more attention to what I'm trying to tell you instead of making ignorant comments? I told you I tried to run it on just two cores of my CPU. it hitched and stuttered in many places and would not be playable in many areas of the game. two cores in my CPU will destroy that Pentium and again I still have CPU left for background tasks. so just looking at the average frame rate would not tell you the whole story because it would not show the hitching. do you really not understand that simple concept? :rolleyes:
 

crashtech

Lifer
Jan 4, 2013
10,695
2,294
146
All I am trying to say is it would be helpful for OTHERS to see actual benchmark results, even if they did take your word as the gospel, it's nice to have third party confirmation and a pretty graph to prove it.

Also, I did already state that these tests need to show minimums, so back at you as far as paying attention. ;)
 

toyota

Lifer
Apr 15, 2001
12,957
1
0
U
All I am trying to say is it would be helpful for OTHERS to see actual benchmark results, even if they did take your word as the gospel, it's nice to have third party confirmation and a pretty graph to prove it.

Also, I did already state that these tests need to show minimums, so back at you as far as paying attention. ;)
it was not necessarily the minimum that was causing the hitching. I never saw it drop below 30 or 40 frames per second most of time but it was quick fast drop that made the hitching. that's what I'm trying to get through your thick skull that it's not going to be reflected in just the frames per second numbers. this isn't something I'm pulling out of my rear as I tested the game myself on some different levels. and it would only be worse with that Pentium CPU as it is much slower than two of my cores and has no more resources left for any background task.
 
Last edited:

crashtech

Lifer
Jan 4, 2013
10,695
2,294
146
If it can't be quantified by showing low minimums or a nasty looking frame time variance graph, it's gonna be a hard sell to say the part won't work. Also, different people have different tolerances for what is perceived as stuttering in-game.
 

TheELF

Diamond Member
Dec 22, 2012
4,027
753
126
U it was not necessarily the minimum that was causing the hitching. I never saw it drop below 30 or 40 frames per second most of time but it was quick fast drop that made the hitching.

Run dxtory(or riva tuner) and limit fps to 30 or 40fps and see how smooth it will play,and how much of the processor will remain free for background tasks.

It's not rocket science but still, its not by chance that computer users used to be called geeks and nerds,you have to know what you're doing.
 

Blue_Max

Diamond Member
Jul 7, 2011
4,223
153
106
Goes to show you that for strictly gaming, you can get away with an i3 and a fairly high end GPU and still do quite well.

Considering the $125 CDN difference between i3 and i5, and $250 CDN difference between i3 and i7, it makes sense to get the i3 and and better video card than i5+mediocre card or i7+wimpy card.

I can totally appreciate the whole "budget" thing. ;)
 

toyota

Lifer
Apr 15, 2001
12,957
1
0
Run dxtory(or riva tuner) and limit fps to 30 or 40fps and see how smooth it will play,and how much of the processor will remain free for background tasks.

It's not rocket science but still, its not by chance that computer users used to be called geeks and nerds,you have to know what you're doing.
it's also not rocket science to know that what I was doing would represent exactly how nearly every single person with a Pentium CPU would be trying to play the game. I was just doing some real world testing and nothing more but if it makes you feel somehow superior to think that I would be too damn incompetent to figure out a way to try and actually make it playable on my own system if I had to then tell yourself that.
 

TheELF

Diamond Member
Dec 22, 2012
4,027
753
126
it's also not rocket science to know that what I was doing would represent exactly how nearly every single person with a Pentium CPU would be trying to play the game. I was just doing some real world testing and nothing more but if it makes you feel somehow superior to think that I would be too damn incompetent to figure out a way to try and actually make it playable on my own system if I had to then tell yourself that.

So if more benchmarks would include 2c/2t with maybe a note telling them to vsync or limit their frames if they have stutter than a lot more people would have a much better experience when gaming,what's your problem with that happening?
 

Blue_Max

Diamond Member
Jul 7, 2011
4,223
153
106
How's this for 2C/2T vs 2C/4T benchmarks?

http://www.anandtech.com/show/8232/...ary-edition-review-the-intel-pentium-g3258-ae

A bit old now, but basically the OC 2/2 holds its own pretty well on single GPU. Falls behind pretty substantially in some cases on dual GPU setup. But, uh...how many people are going to cheap out on the CPU and then run multi-GPU?

It was beaten every time by the i3, even when overclocked to the moon.

Glad I re-read the article you linked... I'll be sure to spend the extra $60C to get an i3 vs G3258. Thanks!
 

crashtech

Lifer
Jan 4, 2013
10,695
2,294
146
How's this for 2C/2T vs 2C/4T benchmarks?

http://www.anandtech.com/show/8232/...ary-edition-review-the-intel-pentium-g3258-ae

A bit old now, but basically the OC 2/2 holds its own pretty well on single GPU. Falls behind pretty substantially in some cases on dual GPU setup. But, uh...how many people are going to cheap out on the CPU and then run multi-GPU?
Can't understand why the SLI setup. Nobody is going to run a SLI rig with a dual core. And what is with the 4330 beating the 4360 in BF4? That's not right:

65188.png


If not for that glaring inconsistency, I would say this graph tells the story re minimums on 2T CPUs.
 
Apr 20, 2008
10,067
990
126
Can't understand why the SLI setup. Nobody is going to run a SLI rig with a dual core. And what is with the 4330 beating the 4360 in BF4? That's not right:

65188.png


If not for that glaring inconsistency, I would say this graph tells the story re minimums on 2T CPUs.

It's also single player, which almost any toaster can play. Going up to multiplayer (what most people play on BF3/4) the dual cores would lag behind immensely.
 

crashtech

Lifer
Jan 4, 2013
10,695
2,294
146
It's also single player, which almost any toaster can play. Going up to multiplayer (what most people play on BF3/4) the dual cores would lag behind immensely.
Anecdotally, I see my nephew playing BF4 MP with a 4.0GHz G3258 fairly well as long as the connection is good and the server is not too full. Playing with 10-15 players is not a real problem for him, my unofficial FRAPS observations has him falling into the low 40s, but then again he also has a GTX770 and plays at 1600x900, which might be helping out some.