Originally posted by VrikaThat review was from the beginning of 2013. Back then I would have probably said that i3 is just fine for gaming.Technology evolves and for the past year we've had quite large number of games asking for more cores. If you're building a computer that should last a couple of years into the future, more cores than just 2 would be good investment.
Technology evolves much faster than the performance demand is when it comes to CPU . You can have n-number of cores but there is little utilization per application for them - most computing will be done in 2-4 threads and there is "nothing to put" into other threads.
As I said above, if you have the money, go for cheapest i5(under 200 USD it is a great opportunity) but i3 with hyperthreading will still do the job fine as well and saved money will do better spent on GPU or w/e else.
I have a i5 3570 3.4ghz and it runs pretty good and is middle of the road performance wise. I run a older 560ti gtx 2meg and 8g of ram with it and play Planetside 2 with no real problem on medium. In the larger battles I get 30fps which isn't too bad, but would like more for it is a fps game and time to kill matters a lot.
Played:SWG(pre NGE/CU sucked)Yep its true, anyone who quit SWG because of the NGE/CU missed out on a much better combat system. DCUO, Fallen Earth, STO, The Secret World. Battlefield series. Planetside 2. Still playing SWG.
4) It's true that only few games are able to properly use more than 4 threads, but we're talking about 2 cores vs. 6 cores. Your previous posts suggests that you think hyperthreading will somehow magically make the 2 cores to perform as fast as 4 cores would. That's not true, i3 still has only 2 cores and hyperthreading is just a way to utilize those 2 cores better.
4) Maybe you should look up what Hyperthreading actually is because it does precisely that magic - provides 2 logical processors per core = 4 threads, just like any Quad core. While not as powerful, it can process multithreaded operations like any multicore cpu.
I did not claim that the hyperthreading couldn't run 4 threads in 2 cores. The issue is, it does not double the processing speed of those cores. It might increase it 20% since running 2 threads at once means the processor can run other thread when it has to wait for something about the other thread. If hyperthreading increases the speed 20% compared to executing just a single thread, that's just 60% of processor's single-thread execution speed to each of the threads when both have a lot of stuff. Compared to having 4 cores and 100% of processor's single-thread execution speed for all of them.
Intel has massive advantage on single-thread execution speed, but I think no so massive that you should just take 2 of intel's cores against 6 of AMD's.
EDIT: But if one can afford i5, then I think that's better for gaming than anything that AMD can offer.
EDIT2: Look at the system requirements on Steam.
-Middle-Earth Shadow of Mordor: Processor: Intel Core i5-750, 2.67 GHz | AMD Phenom II X4 965, 3.4 GHz
-Far Cry 4: Processor: 2.6 GHz Intel® Core™ i5-750 or 3.2 GHz AMD Phenom™ II X4 955
-The Crew: Intel Core2 Quad Q9300 @ 2.5 GHz or AMD Athlon II X4 620 @ 2.6 GHz (or better)
Some of the new games are starting to ask for Intel i5 in their minimum requirements. That's why I think now it's the time to try to get more cores than i3 has to offer
Originally posted by VrikaI did not claim that the hyperthreading couldn't run 4 threads in 2 cores. The issue is, it does not double the processing speed of those cores. It might increase it 20% since running 2 threads at once means the processor can run other thread when it has to wait for something about the other thread. If hyperthreading increases the speed 20% compared to executing just a single thread, that's just 60% of processor's single-thread execution speed to each of the threads when both have a lot of stuff. Compared to having 4 cores and 100% of processor's single-thread execution speed for all of them.Intel has massive advantage on single-thread execution speed, but I think no so massive that you should just take 2 of intel's cores against 6 of AMD's.EDIT: But if one can afford i5, then I think that's better for gaming than anything that AMD can offer.EDIT2: Look at the system requirements on Steam.-Middle-Earth Shadow of Mordor: Processor: Intel Core i5-750, 2.67 GHz | AMD Phenom II X4 965, 3.4 GHz -Far Cry 4: Processor: 2.6 GHz Intel® Core i5-750 or 3.2 GHz AMD Phenom II X4 955-The Crew: Intel Core2 Quad Q9300 @ 2.5 GHz or AMD Athlon II X4 620 @ 2.6 GHz (or better) Some of the new games are starting to ask for Intel i5 in their minimum requirements. That's why I think now it's the time to try to get more cores than i3 has to offer
That is not how it works.
Usually, 1 or 2 threads puts way more load on core so whether other threads have much higher available computional power does not matter, they won't be utilized.
All what matters is whether you can run 4 threads for parallel processing, whether there are physical cores makes no difference.
That is why i3 beat FX-6300 in games - there is no use for more threads and i3 provides more punch per thread. In that case, more cores provide no advantage.
Look at the system specs you posted, new games asking for ancient CPU, that should put anyone into perspective how much CPU power is actually needed there.
Hyperthreading is nowhere near the same as twice the cores. The example I posted previously was with an i3 with 4 threads, and it still performed dismally next to an i5. This is not a one off case, it's very common with modern games that expect 4 cores.
i3: 36fps, i5: 60fps. In these instances an AMD FX 6300 will outperform an i3 considerably, as it does here.
Pretty much any gaming site will recommend against an i3 as two cores is not enough for gaming going forward. Up until 2 years ago it was more or less fine.
Originally posted by AthisarHyperthreading is nowhere near the same as twice the cores. The example I posted previously was with an i3 with 4 threads, and it still performed dismally next to an i5. This is not a one off case, it's very common with modern games that expect 4 cores.Take Crysis 3http://www.techspot.com/articles-info/642/bench/CPU_03.pngi3: 36fps, i5: 60fps. In these instances an AMD FX 6300 will outperform an i3 considerably, as it does here.Pretty much any gaming site will recommend against an i3 as two cores is not enough for gaming going forward. Up until 2 years ago it was more or less fine.
Same failed argument won't become any better if you repeat it.
As I pointed out to you already, the results of Crysis 3 benchmarks are rather due poor game optimization than anything else.
Anyone can check from plenty of other tests provided by very same site and see how i3 performs in other games to get a proper picture:
You intentionally picked the only benchmark there where i3 is losing breath. Enough said.
And last but not least, what you keep forgetting is the GPU being used for benchmarks. Unless you have a system with top end card like they have, you will hit GPU bottleneck MUCH earlier.
So yeah, if you are buying R290x, GTX980 or Titan, i3 might be a bad option. That is about all your point you are making here. Gratz.
Heck, even with top end GPU, the frame rate is still at very decent rates even on low end CPUs.
Hyperthreading is nowhere near the same as twice the cores. The example I posted previously was with an i3 with 4 threads, and it still performed dismally next to an i5. This is not a one off case, it's very common with modern games that expect 4 cores.
i3: 36fps, i5: 60fps. In these instances an AMD FX 6300 will outperform an i3 considerably, as it does here.
Pretty much any gaming site will recommend against an i3 as two cores is not enough for gaming going forward. Up until 2 years ago it was more or less fine.
Same failed argument won't become any better if you repeat it.
As I pointed out to you already, the results of Crysis 3 benchmarks are rather due poor game optimization than anything else.
Anyone can check from plenty of other tests provided by very same site and see how i3 performs in other games to get a proper picture:
You intentionally picked the only benchmark there where i3 is losing breath. Enough said.
And last but not least, what you keep forgetting is the GPU being used for benchmarks. Unless you have a system with top end card like they have, you will hit GPU bottleneck MUCH earlier.
So yeah, if you are buying R290x, GTX980 or Titan, i3 might be a bad option. That is about all your point you are making here. Gratz.
Heck, even with top end GPU, the frame rate is still at very decent rates even on low end CPUs.
An i3 is a bad option period. If you have to drop down to substandard parts to build a computer for gaming then you may want to get a different hobby.
Out of all the post I have ever read about building a gaming PC you sir are the only one I have ever seen argue that buying an i3 over an i5 is a good idea.
Congrats you win the "I don't know shit about building a PC award."
Originally posted by chums54An i3 is a bad option period. If you have to drop down to substandard parts to build a computer for gaming then you may want to get a different hobby. Out of all the post I have ever read about building a gaming PC you sir are the only one I have ever seen argue that buying an i3 over an i5 is a good idea.Congrats you win the "I don't know shit about building a PC award." Hyper threading lol
Sorry, you can't beat actual benchmarks. Feel free to check them out for yourself.
Hyperthreading is nowhere near the same as twice the cores. The example I posted previously was with an i3 with 4 threads, and it still performed dismally next to an i5. This is not a one off case, it's very common with modern games that expect 4 cores.
i3: 36fps, i5: 60fps. In these instances an AMD FX 6300 will outperform an i3 considerably, as it does here.
Pretty much any gaming site will recommend against an i3 as two cores is not enough for gaming going forward. Up until 2 years ago it was more or less fine.
Same failed argument won't become any better if you repeat it.
As I pointed out to you already, the results of Crysis 3 benchmarks are rather due poor game optimization than anything else.
Anyone can check from plenty of other tests provided by very same site and see how i3 performs in other games to get a proper picture:
You intentionally picked the only benchmark there where i3 is losing breath. Enough said.
And last but not least, what you keep forgetting is the GPU being used for benchmarks. Unless you have a system with top end card like they have, you will hit GPU bottleneck MUCH earlier.
So yeah, if you are buying R290x, GTX980 or Titan, i3 might be a bad option. That is about all your point you are making here. Gratz.
Heck, even with top end GPU, the frame rate is still at very decent rates even on low end CPUs.
So all the games that work fine on quad core processors but not on dual core processors are poorly optimised? And one of the biggest engines that exists is poorly optimised? If you want to believe that an i3 is powerful enough for every game out there then go ahead and think that.
Also it's fairly ludicrous to assume I went and found the only benchmark that supports my argument. I know that Crysis 3 is extremely well optimised for threading and is CPU intensive, so it's a good example. Just picking another from the URL you listed shows the same occurring,
Originally posted by AthisarAnd one of the biggest engines that exists is poorly optimised? If you want to believe that an i3 is powerful enough for every game out there then go ahead and think that. Also it's fairly ludicrous to assume I went and found the only benchmark that supports my argument. I know that Crysis 3 is extremely well optimised for threading and is CPU intensive, so it's a good example.
Originally posted by AthisarSo all the games that work fine on quad core processors but not on dual core processors are poorly optimised? ...Anyone can produce a benchmark from games that don't have high CPU requirements, but that proves nothing.
What "all other games" you talk about?
Just going down the list of benchmarks, all i3-3220(old Ivy bridge CPU):
Not suprisingly, CPU is stressed a lot by RTS but in that case, I guess very specific testing would be needed to find out what determines the performance. However, it will be still very fine for vast majority of games, even with top end gpu.
Not suprisingly, CPU is stressed a lot by RTS but there isn't enough large sample of benchmarked games to make any conclusion what determines the performance. However, it will be very fine for vast majority of games, even with top end gpu.
Taking an average of an arbitrary small selection of games doesn't show anything statistically relevant. As I've said, and as you already appear to know, games that don't have high CPU requirements will not struggle on an i3. That covers the vast majority of old console type games. The fact is that the new consoles use 8 cores, and game engines have been/are being adapted to rely on a higher number of cores, and this shows in the recommended requirements and the benchmarks. It may be fine for lots of games, but it's simply incorrect to say it makes no difference when it evidently can make a massive difference.
Originally posted by Athisar Taking an average of an arbitrary small selection of games doesn't show anything statistically relevant.
On the contrary, it provides statiscical relevance. There is nothing arbitrary about random sample of games.
You are just obtusely ignoring blatantly obvious truth presented in benchmarks above - games have in general fairly low CPU requirements. Only what sets aside are RTS, and those give a hard time to any CPU.
Taking an average of an arbitrary small selection of games doesn't show anything statistically relevant.
On the contrary, it provides statiscical relevance. There is nothing arbitrary about random sample of games.
You are just obtusely ignoring blatantly obvious truth presented in benchmarks above - games have in general fairly low CPU requirements. Only what sets aside are RTS, and those give a hard time to any CPU.
Consoles are entirely different machines.
Ok, there's no point arguing with you. I've made my points and backed them up, and I'll leave it there.
I went and browsed my usual go-to benchmark source: Anandtech. There weren't a lot of FX6300 vs i3 comparisons, but a few (Diablo 3, Rage, Skyrim, a couple others).
It looked like there were two cases: either it was close, and the two chips were within a few percentage of each other, or to was a a landslide case towards the FX6300 (those few cases that needed real cores).
Keep in mind if you are dropping to an H-series budget motherboard, you are also crippling your turbo frequency - and most benchmarks are going to show uncapped turbo performance (a Z-series
If you don't really save any money (or its just a few dollars), so sometimes perform 8% better, and sometimes perform 200% worse - that doesn't make a lot of sense to me.
I can't see any sane recommendation for a core i3 except for those people who are blind to Intel above all else in the name of brand loyalty. I don't think there is anything wrong with that as long as you realize it.
Two cores for the price of six, you say? What could possibly go wrong?
As I already told you: Nothing.
You spread false info and empty speculations.
1) i3 isn't substantially more expensive - there is about 10% difference in price. 2) Intel board is definitely not more expensive. 3) Yes, FX-6300 is definitely slower in per thread/core performance. 4) There aren't really any games that would sensibly utilize more than 4 threads and won't be for a long time. 5) Even if so, the bottleneck will occur on GPU long before than CPU - you won't notice the 100 USD difference in CPU once you get above certain, fairly low CPU power but you will damn make notice a difference in 100 USD spent on GPU. 6) FX-6300 is completely ill advice, it is a dead end platform.
As always with your posts - less theory, more practical application.
1) An FX-6300 is $110 at the moment on New Egg, $100 on Amazon, or $89 on NCIX. For comparison, the cheapest new Core i3 on those sites is $120 on New Egg, $112 on Amazon, or $110 on NCIX. And the cheapest Core i3 is not the fastest Core i3; there's more than one, and the better ones (higher clock speeds, more cache, etc.) cost more. The Core i3-4370 is at least $160, depending on where you buy it.
Is the $110 Core i3-4130 on NCIX faster in pure single-threaded performance than an FX-6300? Yeah, but not by that much--and not by nearly as much as a Core i3-4370. The FX-6300 can also be overclocked if you want to go that route; a Core i3 can't.
2) Decent quality LGA 1150 motherboards start around $100; Socket AM3+ around $70. If you're willing to go the cheap junk route, Socket AM3+ is still cheaper. The difference is because Intel charges more for their chipsets, and motherboard vendors pass along that cost to you.
3) In single-threaded performance, most Core i3 processors will be faster than an FX-6300. But per thread performance is a notion that doesn't even make sense. I'll come back to this later.
4) To the contrary, there are already a lot of games that use more than four threads. There's a huge difference between being able to put more than four threads to good use and needing more than four cores in order for the game to be running. A game could easily have a lot of threads that are idle most of the time. Being able to execute as many as want to go at once helps with performance consistency, even if it helps less with average frame rates.
5) This varies wildly by game, but if the CPU isn't the bottleneck, why not save money by going AMD?
6) If you're talking about upgrading to future CPUs that don't yet exist, basically all platforms are dead-end platforms. LGA 1366 never got an upgrade from the original CPUs for under $500. The only upgrade that ever came to LGA 1156 was the same CPU as before except 133 MHz faster. The only upgrade that came to LGA 1155 or the first LGA 2011 was a new generation that was slightly faster at stock speeds but couldn't overclock as far. The only upgrade that came to LGA 1150 was the same CPUs as before with higher stock clock speeds.
The days of an LGA 775 platform that could upgrade from the initial Pentium 4 to a later, vastly superior Core 2 Duo are long since over and probably never coming back, with plausibly a one-off exception if AMD's upcoming Zen architecture (due to arrive in 2016) fits some then-older socket that likely isn't even on the market yet.
-----
I don't think you know what hyperthreading is. Threads are not the same as cores, and having more threads resident on a core does help sometimes, but it's not the same thing as having more cores. For example, a GeForce GTX 580 can have up to 24,576 threads resident at a time, but that's not at all similar to having 24,576 shaders rather than the 512 that the card actually has.
The idea of hyperthreading is that a single CPU core can have two threads resident at a time rather than one. This allows a core to switch between those two threads very, very quickly, rather than having a considerable penalty for context switching. If one thread has nothing available to do for whatever reason, the other thread can execute some instructions instead. But if both threads are ready to do something at the same time, only one gets to go and the other has to sit and wait. If the two threads were on different cores, both can execute instructions whenever they're ready.
Intel claims that hyperthreading can improve performance by "up to" 30% in programs that would scale flawlessly to more cores. That's not double performance, or even anywhere remotely near double as you'd get with a second real core. Furthermore, if the first thread is already good at covering up latency so that it always has something ready to go, hyperthreading might not provide any benefit at all. I once did an experiment with something that trivially scaled to arbitrarily many cores and found that on my Core i7-860, eight threads was about 15% faster than four threads.
It might be possible to find some pathological corner case where hyperthreading does nearly double performance. But it's also possible--and probably easier--to find pathological corner cases where hyperthreading makes things run more slowly than it would without hyperthreading. Indeed, that was common in the early days of hyperthreading before WIndows got smarter about how it scheduled threads.
Ya'll need to stop giving him bad advice. Clearly what he needs is a Xeon E7-8890 v2. Only 15 cores is sufficient these days, anything less is just being irresponsible.
"The surest way to corrupt a youth is to instruct him to hold in higher esteem those who think alike than those who think differently."
That review was from the beginning of 2013. Back then I would have probably said that i3 is just fine for gaming.
Technology evolves and for the past year we've had quite large number of games asking for more cores. If you're building a computer that should last a couple of years into the future, more cores than just 2 would be good investment.
Technology evolves much faster than the performance demand is when it comes to CPU . You can have n-number of cores but there is little utilization per application for them - most computing will be done in 2-4 threads and there is "nothing to put" into other threads.
As I said above, if you have the money, go for cheapest i5(under 200 USD it is a great opportunity) but i3 with hyperthreading will still do the job fine as well and saved money will do better spent on GPU or w/e else.
I haven't looked at the internal workings of any game engine besides my own, but I can tell you that I use a lot more than 2-4 threads.
There is one thread that processes inputs. If you press a key or move the mouse, that thread timestamps the action, does some computation, and then passes data where it needs to go.
There is another thread that looks around to see if it needs to load anything else that isn't already loaded. That doesn't need to run continuously, of course. It does its check, then sleeps for 100 ms, checks again, and so forth. If it finds anything, it passes it along to a different thread.
There's another thread that just manages assets to load, largely to keep everything synchronized, primarily with a goal that the GPU never be asked to render assets that aren't already in video memory.
There's yet another thread whose only purpose is to take things that the GPU is going to be asked to draw and sort them into a more sensible order to reduce the number of OpenGL API calls necessary.
That's four threads right there, and those threads added together probably don't add up to 1% of the CPU work that a game does. Having them off in separate threads so that they can run when it makes sense makes things vastly easier to program. It probably also increases performance apart from a pathological level of optimizations that would make the code basically incomprehensible. How many threads will be active depends on a variety of factors--including the results of querying how many cores your system has available. But it will usually be in the double digits--with most of the threads idle most of the time,and sometimes all of the idle at once if you're waiting on the GPU.
That review was from the beginning of 2013. Back then I would have probably said that i3 is just fine for gaming.
Technology evolves and for the past year we've had quite large number of games asking for more cores. If you're building a computer that should last a couple of years into the future, more cores than just 2 would be good investment.
Technology evolves much faster than the performance demand is when it comes to CPU . You can have n-number of cores but there is little utilization per application for them - most computing will be done in 2-4 threads and there is "nothing to put" into other threads.
As I said above, if you have the money, go for cheapest i5(under 200 USD it is a great opportunity) but i3 with hyperthreading will still do the job fine as well and saved money will do better spent on GPU or w/e else.
I haven't looked at the internal workings of any game engine besides my own, but I can tell you that I use a lot more than 2-4 threads.
There is one thread that processes inputs. If you press a key or move the mouse, that thread timestamps the action, does some computation, and then passes data where it needs to go.
There is another thread that looks around to see if it needs to load anything else that isn't already loaded. That doesn't need to run continuously, of course. It does its check, then sleeps for 100 ms, checks again, and so forth. If it finds anything, it passes it along to a different thread.
There's another thread that just manages assets to load, largely to keep everything synchronized, primarily with a goal that the GPU never be asked to render assets that aren't already in video memory.
There's yet another thread whose only purpose is to take things that the GPU is going to be asked to draw and sort them into a more sensible order to reduce the number of OpenGL API calls necessary.
That's four threads right there, and those threads added together probably don't add up to 1% of the CPU work that a game does. Having them off in separate threads so that they can run when it makes sense makes things vastly easier to program. It probably also increases performance apart from a pathological level of optimizations that would make the code basically incomprehensible. How many threads will be active depends on a variety of factors--including the results of querying how many cores your system has available. But it will usually be in the double digits--with most of the threads idle most of the time,and sometimes all of the idle at once if you're waiting on the GPU.
Uhh, you need to talk to some programmers. Writing multithreaded code is far more difficult, thats why its taken this long to get to the point where you're starting to get programs that are multithreaded that aren't professional programs (i.e. cad/cam software, etc)
"The surest way to corrupt a youth is to instruct him to hold in higher esteem those who think alike than those who think differently."
All what matters is whether you can run 4 threads for parallel processing, whether there are physical cores makes no difference.
That is completely wrong. One physical core cannot have two threads execute instructions at the same time. If a core without hyperthreading wants to switch from executing one thread to another, there's a considerable penalty for doing so, as there is a period of time where it can't execute anything from either thread until the new thread is properly loaded. Hyperthreading makes it so that you can have two threads resident at once and switching between them is very, very fast. This also makes it sensible to switch between them more often, as sometimes you can say, the first thread has nothing to do, switch to the second thread and have it do something, then switch back to the first thread and have it do something else all faster than it would take just to switch threads once without hyperthreading.
So hyperthreading is a good thing. Don't get me wrong about that. But it's no substitute for having more cores. If Intel wanted to, they could readily make a single-core processor that can have 256 threads resident at a time. Such a chip would be badly beaten by a simple dual core processor outside of some very pathological workloads.
Taking an average of an arbitrary small selection of games doesn't show anything statistically relevant.
On the contrary, it provides statiscical relevance. There is nothing arbitrary about random sample of games.
You are just obtusely ignoring blatantly obvious truth presented in benchmarks above - games have in general fairly low CPU requirements. Only what sets aside are RTS, and those give a hard time to any CPU.
Consoles are entirely different machines.
Someone who would take frame rates from a bunch of games, average them, and claim that the average proves anything knows nothing about statistics.
If several games got 5 frames per second and one got 500, the average could easily be over 60. That doesn't mean that those games with 5 frames per second are playable. All it means is that the average is the wrong measure entirely.
Uhh, you need to talk to some programmers. Writing multithreaded code is far more difficult, thats why its taken this long to get to the point where you're starting to get programs that are multithreaded that aren't professional programs (i.e. cad/cam software, etc)
That depends very, very strongly on what you're trying to do. Some algorithms are intrinsically single-threaded and can't be parallelized. Some algorithms take a great amount of work to parallelize. Some are pretty easily parallelizable such that making it scale to many cores only adds a few percent to the work in coding it.
There is also an intrinsic amount of work it takes to write a parallel program, so for something simple enough that you can write a single-threaded version in ten minutes, a parallel version might well make it take two or three times as long. But that intrinsic amount of work becomes far less significant in larger projects.
Or to take an extreme example, if you've got a modern, highish-end GPU (basically, if you spent over $200 on it in the last few years and didn't buy something incredibly stupid), your GPU probably has several thousand threads going most of the time in most games, unless it's light enough on GPU usage that your video card is mostly idle while playing the game.
Most things in games are pretty easy to parallelize enough that no thread needs more than a small fraction of a core. The glaring exception is the thread that communicates with the GPU, and that is often the thing that creates a single-threaded bottleneck somewhere. But even that may well be going away soon, as both OpenGL and Mantle offer ways to greatly reduce the amount of work that thread has to do, and DirectX will join them next year.
Comments
Technology evolves much faster than the performance demand is when it comes to CPU . You can have n-number of cores but there is little utilization per application for them - most computing will be done in 2-4 threads and there is "nothing to put" into other threads.
As I said above, if you have the money, go for cheapest i5(under 200 USD it is a great opportunity) but i3 with hyperthreading will still do the job fine as well and saved money will do better spent on GPU or w/e else.
Played:SWG(pre NGE/CU sucked)Yep its true, anyone who quit SWG because of the NGE/CU missed out on a much better combat system. DCUO, Fallen Earth, STO, The Secret World. Battlefield series. Planetside 2. Still playing SWG.
I did not claim that the hyperthreading couldn't run 4 threads in 2 cores. The issue is, it does not double the processing speed of those cores. It might increase it 20% since running 2 threads at once means the processor can run other thread when it has to wait for something about the other thread. If hyperthreading increases the speed 20% compared to executing just a single thread, that's just 60% of processor's single-thread execution speed to each of the threads when both have a lot of stuff. Compared to having 4 cores and 100% of processor's single-thread execution speed for all of them.
Intel has massive advantage on single-thread execution speed, but I think no so massive that you should just take 2 of intel's cores against 6 of AMD's.
EDIT: But if one can afford i5, then I think that's better for gaming than anything that AMD can offer.
EDIT2: Look at the system requirements on Steam.
-Middle-Earth Shadow of Mordor: Processor: Intel Core i5-750, 2.67 GHz | AMD Phenom II X4 965, 3.4 GHz
-Far Cry 4: Processor: 2.6 GHz Intel® Core™ i5-750 or 3.2 GHz AMD Phenom™ II X4 955
-The Crew: Intel Core2 Quad Q9300 @ 2.5 GHz or AMD Athlon II X4 620 @ 2.6 GHz (or better)
Some of the new games are starting to ask for Intel i5 in their minimum requirements. That's why I think now it's the time to try to get more cores than i3 has to offer
That is not how it works.
Usually, 1 or 2 threads puts way more load on core so whether other threads have much higher available computional power does not matter, they won't be utilized.
All what matters is whether you can run 4 threads for parallel processing, whether there are physical cores makes no difference.
That is why i3 beat FX-6300 in games - there is no use for more threads and i3 provides more punch per thread. In that case, more cores provide no advantage.
Look at the system specs you posted, new games asking for ancient CPU, that should put anyone into perspective how much CPU power is actually needed there.
Hyperthreading is nowhere near the same as twice the cores. The example I posted previously was with an i3 with 4 threads, and it still performed dismally next to an i5. This is not a one off case, it's very common with modern games that expect 4 cores.
Take Crysis 3
http://www.techspot.com/articles-info/642/bench/CPU_03.png
i3: 36fps, i5: 60fps. In these instances an AMD FX 6300 will outperform an i3 considerably, as it does here.
Pretty much any gaming site will recommend against an i3 as two cores is not enough for gaming going forward. Up until 2 years ago it was more or less fine.
Same failed argument won't become any better if you repeat it.
As I pointed out to you already, the results of Crysis 3 benchmarks are rather due poor game optimization than anything else.
Anyone can check from plenty of other tests provided by very same site and see how i3 performs in other games to get a proper picture:
http://www.techspot.com/features/gaming/gaming-benchmarks/
Performance in 1 game does not make a rule.
You intentionally picked the only benchmark there where i3 is losing breath. Enough said.
And last but not least, what you keep forgetting is the GPU being used for benchmarks. Unless you have a system with top end card like they have, you will hit GPU bottleneck MUCH earlier.
So yeah, if you are buying R290x, GTX980 or Titan, i3 might be a bad option. That is about all your point you are making here. Gratz.
Heck, even with top end GPU, the frame rate is still at very decent rates even on low end CPUs.
An i3 is a bad option period. If you have to drop down to substandard parts to build a computer for gaming then you may want to get a different hobby.
Out of all the post I have ever read about building a gaming PC you sir are the only one I have ever seen argue that buying an i3 over an i5 is a good idea.
Congrats you win the "I don't know shit about building a PC award."
Hyper threading lol
Sorry, you can't beat actual benchmarks. Feel free to check them out for yourself.
So all the games that work fine on quad core processors but not on dual core processors are poorly optimised? And one of the biggest engines that exists is poorly optimised? If you want to believe that an i3 is powerful enough for every game out there then go ahead and think that.
Also it's fairly ludicrous to assume I went and found the only benchmark that supports my argument. I know that Crysis 3 is extremely well optimised for threading and is CPU intensive, so it's a good example. Just picking another from the URL you listed shows the same occurring,
http://www.techspot.com/articles-info/689/bench/CPU_02.png
Anyone can produce a benchmark from games that don't have high CPU requirements, but that proves nothing.
It is about Crysis, not i3.
http://www.dsogaming.com/pc-performance-analyses/crysis-3-pc-performance-analysis/
All I am saying that if you are on a budget and have to decide where to put 100 USD, rather spent in on GPU than i5.
What "all other games" you talk about?
Just going down the list of benchmarks, all i3-3220(old Ivy bridge CPU):
The Crew 2560x1600 - 45 fps
Far Cry 4 - 76 fps
Alien Isolation - 104 fps
Metro: Redux - 47 fps
WatchDogs: - 51 fps
Thief - 76 fps
Battlefield 4 - 95 fps
Batman Arkham Origins - 140 fps
Arma 3 - 45 fps
Splinter Cell Blacklist - 72 fps
Company of Heroes 2 - 29 fps
Bioshock Infinite - 63 fps
SimCity - 23 fps
In avg 66 fps.
Not suprisingly, CPU is stressed a lot by RTS but in that case, I guess very specific testing would be needed to find out what determines the performance. However, it will be still very fine for vast majority of games, even with top end gpu.
Taking an average of an arbitrary small selection of games doesn't show anything statistically relevant. As I've said, and as you already appear to know, games that don't have high CPU requirements will not struggle on an i3. That covers the vast majority of old console type games. The fact is that the new consoles use 8 cores, and game engines have been/are being adapted to rely on a higher number of cores, and this shows in the recommended requirements and the benchmarks. It may be fine for lots of games, but it's simply incorrect to say it makes no difference when it evidently can make a massive difference.
On the contrary, it provides statiscical relevance. There is nothing arbitrary about random sample of games.
You are just obtusely ignoring blatantly obvious truth presented in benchmarks above - games have in general fairly low CPU requirements. Only what sets aside are RTS, and those give a hard time to any CPU.
Consoles are entirely different machines.
Ok, there's no point arguing with you. I've made my points and backed them up, and I'll leave it there.
Yep, you can't argue actual benchmark results rebutting your points.
I went and browsed my usual go-to benchmark source: Anandtech. There weren't a lot of FX6300 vs i3 comparisons, but a few (Diablo 3, Rage, Skyrim, a couple others).
It looked like there were two cases: either it was close, and the two chips were within a few percentage of each other, or to was a a landslide case towards the FX6300 (those few cases that needed real cores).
Keep in mind if you are dropping to an H-series budget motherboard, you are also crippling your turbo frequency - and most benchmarks are going to show uncapped turbo performance (a Z-series
If you don't really save any money (or its just a few dollars), so sometimes perform 8% better, and sometimes perform 200% worse - that doesn't make a lot of sense to me.
I can't see any sane recommendation for a core i3 except for those people who are blind to Intel above all else in the name of brand loyalty. I don't think there is anything wrong with that as long as you realize it.
1) An FX-6300 is $110 at the moment on New Egg, $100 on Amazon, or $89 on NCIX. For comparison, the cheapest new Core i3 on those sites is $120 on New Egg, $112 on Amazon, or $110 on NCIX. And the cheapest Core i3 is not the fastest Core i3; there's more than one, and the better ones (higher clock speeds, more cache, etc.) cost more. The Core i3-4370 is at least $160, depending on where you buy it.
Is the $110 Core i3-4130 on NCIX faster in pure single-threaded performance than an FX-6300? Yeah, but not by that much--and not by nearly as much as a Core i3-4370. The FX-6300 can also be overclocked if you want to go that route; a Core i3 can't.
2) Decent quality LGA 1150 motherboards start around $100; Socket AM3+ around $70. If you're willing to go the cheap junk route, Socket AM3+ is still cheaper. The difference is because Intel charges more for their chipsets, and motherboard vendors pass along that cost to you.
3) In single-threaded performance, most Core i3 processors will be faster than an FX-6300. But per thread performance is a notion that doesn't even make sense. I'll come back to this later.
4) To the contrary, there are already a lot of games that use more than four threads. There's a huge difference between being able to put more than four threads to good use and needing more than four cores in order for the game to be running. A game could easily have a lot of threads that are idle most of the time. Being able to execute as many as want to go at once helps with performance consistency, even if it helps less with average frame rates.
5) This varies wildly by game, but if the CPU isn't the bottleneck, why not save money by going AMD?
6) If you're talking about upgrading to future CPUs that don't yet exist, basically all platforms are dead-end platforms. LGA 1366 never got an upgrade from the original CPUs for under $500. The only upgrade that ever came to LGA 1156 was the same CPU as before except 133 MHz faster. The only upgrade that came to LGA 1155 or the first LGA 2011 was a new generation that was slightly faster at stock speeds but couldn't overclock as far. The only upgrade that came to LGA 1150 was the same CPUs as before with higher stock clock speeds.
The days of an LGA 775 platform that could upgrade from the initial Pentium 4 to a later, vastly superior Core 2 Duo are long since over and probably never coming back, with plausibly a one-off exception if AMD's upcoming Zen architecture (due to arrive in 2016) fits some then-older socket that likely isn't even on the market yet.
-----
I don't think you know what hyperthreading is. Threads are not the same as cores, and having more threads resident on a core does help sometimes, but it's not the same thing as having more cores. For example, a GeForce GTX 580 can have up to 24,576 threads resident at a time, but that's not at all similar to having 24,576 shaders rather than the 512 that the card actually has.
The idea of hyperthreading is that a single CPU core can have two threads resident at a time rather than one. This allows a core to switch between those two threads very, very quickly, rather than having a considerable penalty for context switching. If one thread has nothing available to do for whatever reason, the other thread can execute some instructions instead. But if both threads are ready to do something at the same time, only one gets to go and the other has to sit and wait. If the two threads were on different cores, both can execute instructions whenever they're ready.
Intel claims that hyperthreading can improve performance by "up to" 30% in programs that would scale flawlessly to more cores. That's not double performance, or even anywhere remotely near double as you'd get with a second real core. Furthermore, if the first thread is already good at covering up latency so that it always has something ready to go, hyperthreading might not provide any benefit at all. I once did an experiment with something that trivially scaled to arbitrarily many cores and found that on my Core i7-860, eight threads was about 15% faster than four threads.
It might be possible to find some pathological corner case where hyperthreading does nearly double performance. But it's also possible--and probably easier--to find pathological corner cases where hyperthreading makes things run more slowly than it would without hyperthreading. Indeed, that was common in the early days of hyperthreading before WIndows got smarter about how it scheduled threads.
"The surest way to corrupt a youth is to instruct him to hold in higher esteem those who think alike than those who think differently."
- Friedrich Nietzsche
I haven't looked at the internal workings of any game engine besides my own, but I can tell you that I use a lot more than 2-4 threads.
There is one thread that processes inputs. If you press a key or move the mouse, that thread timestamps the action, does some computation, and then passes data where it needs to go.
There is another thread that looks around to see if it needs to load anything else that isn't already loaded. That doesn't need to run continuously, of course. It does its check, then sleeps for 100 ms, checks again, and so forth. If it finds anything, it passes it along to a different thread.
There's another thread that just manages assets to load, largely to keep everything synchronized, primarily with a goal that the GPU never be asked to render assets that aren't already in video memory.
There's yet another thread whose only purpose is to take things that the GPU is going to be asked to draw and sort them into a more sensible order to reduce the number of OpenGL API calls necessary.
That's four threads right there, and those threads added together probably don't add up to 1% of the CPU work that a game does. Having them off in separate threads so that they can run when it makes sense makes things vastly easier to program. It probably also increases performance apart from a pathological level of optimizations that would make the code basically incomprehensible. How many threads will be active depends on a variety of factors--including the results of querying how many cores your system has available. But it will usually be in the double digits--with most of the threads idle most of the time,and sometimes all of the idle at once if you're waiting on the GPU.
Uhh, you need to talk to some programmers. Writing multithreaded code is far more difficult, thats why its taken this long to get to the point where you're starting to get programs that are multithreaded that aren't professional programs (i.e. cad/cam software, etc)
"The surest way to corrupt a youth is to instruct him to hold in higher esteem those who think alike than those who think differently."
- Friedrich Nietzsche
That is completely wrong. One physical core cannot have two threads execute instructions at the same time. If a core without hyperthreading wants to switch from executing one thread to another, there's a considerable penalty for doing so, as there is a period of time where it can't execute anything from either thread until the new thread is properly loaded. Hyperthreading makes it so that you can have two threads resident at once and switching between them is very, very fast. This also makes it sensible to switch between them more often, as sometimes you can say, the first thread has nothing to do, switch to the second thread and have it do something, then switch back to the first thread and have it do something else all faster than it would take just to switch threads once without hyperthreading.
So hyperthreading is a good thing. Don't get me wrong about that. But it's no substitute for having more cores. If Intel wanted to, they could readily make a single-core processor that can have 256 threads resident at a time. Such a chip would be badly beaten by a simple dual core processor outside of some very pathological workloads.
Someone who would take frame rates from a bunch of games, average them, and claim that the average proves anything knows nothing about statistics.
If several games got 5 frames per second and one got 500, the average could easily be over 60. That doesn't mean that those games with 5 frames per second are playable. All it means is that the average is the wrong measure entirely.
I would personally go with either an I5 or if on a budget an AMD CPU.
Nothing wrong with an I3 necessarily. But if an I5 isnt an option then a quad or 6 core AMD would be a better choice for the money.
That depends very, very strongly on what you're trying to do. Some algorithms are intrinsically single-threaded and can't be parallelized. Some algorithms take a great amount of work to parallelize. Some are pretty easily parallelizable such that making it scale to many cores only adds a few percent to the work in coding it.
There is also an intrinsic amount of work it takes to write a parallel program, so for something simple enough that you can write a single-threaded version in ten minutes, a parallel version might well make it take two or three times as long. But that intrinsic amount of work becomes far less significant in larger projects.
Or to take an extreme example, if you've got a modern, highish-end GPU (basically, if you spent over $200 on it in the last few years and didn't buy something incredibly stupid), your GPU probably has several thousand threads going most of the time in most games, unless it's light enough on GPU usage that your video card is mostly idle while playing the game.
Most things in games are pretty easy to parallelize enough that no thread needs more than a small fraction of a core. The glaring exception is the thread that communicates with the GPU, and that is often the thing that creates a single-threaded bottleneck somewhere. But even that may well be going away soon, as both OpenGL and Mantle offer ways to greatly reduce the amount of work that thread has to do, and DirectX will join them next year.