Haswell i3-4150 vs FX 8320/e for budget gaming rig?

[Abwx]

I didn't think you believed in your own BS enough to give me a time frame for when FX will over take i5's either.

Thanks for playing.

There s a 4670k in the graphs above....in case you didnt notice, and the "i" under scrutinity in this thread has 3 following, not a 5...

I hope it will serve to debunk people who are making the ridiculous statement that an i3 is a valuable opponent for a FX, people can see what some fanatics are up to, that is advising to buy at same price a CPU that has half the perfs because it is....almost equal, but inferior, in games than the FX8xxx.

On hardwre.fr games suite the i5 has 25% advantage over the FX, down from 42% last year, how much in a year or two.?.

And still, hardware.fr were very sympathetic to remove BF4 for the current suite and replace it with a single threaded game that help the i5 getting a better score, they use also ARMA 3, wich is lowly threaded, and none of the recently updated games set apart Crysis 3, otherwise the i5 would had trouble getting more than 10-15% advantage in their games benches, the i3 is already below the FX...

 

[2is]

Shifting goal posts much? Below is what YOU said

On future titles i5s stand no chance against a FX8, if anything it s the integer throughput that define the max theorical perf, to get an idea of where the things will converge look at multithreaded integer based banchmarks.

To that I asked you a very simple question. How many years do you want before FX8 overtakes i5?

This is at least the third time and all you do is try and talk your way out of it. If you really believed your own nonsense, you'd have provided an answer to that very simple question.

So how long do you want? A year? or two?

 

[Abwx]

Shifting goal posts much? Below is what YOU said

To that I asked you a very simple question. How many years do you want before FX8 overtakes i5?

This is at least the third time and all you do is try and talk your way out of it. If you really believed your own nonsense, you'd have provided an answer to that very simple question.

This is in the doing actualy, i wouldnt give more than 2 years to the i5 to be also irrelevant in games compared to the FX, there are already games where the i5 has no substancial advantage, in the threaded games Hardware.fr did notice that the code is often mediocre with bizarre scalings and that they are greatly improvable, let s check a few of theses games :

And now let s check the games where the FX is not as good due to either low thread counts or still unoptimised code, notice that Hyperthreading has sometimes being removed as it s counterproductive, an i3 would be stuck with two cores and no HT in such cases, that s the reason some highly threaded Intel CPUs can do badly...

Now that i m looking at the graphs i m more inclined to think that two years is exagerated, look like it will be less time, that said, tell us where the much hyped i3 would be placed in thoses charts, would be curious if i get yet another "opinion" that it would be better than a FX8 in games.

 

[2is]

Got it, so in 2015 then? FX8 > i5 is what you're predicting? My argument with you in this thread has always been about i5 and never the i3. I don't shift goal posts, I'll leave that to you and the rest of the people preaching the same thing for the last 8 years. That isn't an opinion either.

 

[AtenRa]

Low budget systems will first be GPU limited and then CPU limited in the vast majority of todays new games at 1080p. With Mantle and DX-12 on the way, plus 4K becoming mainstream the next couple of years, CPU performance in gaming is becoming less important every year.

Both FX and Core i3 will be plenty for the majority of new games today. I would recommend the FX for every game using Frostbite 3 and Cryengine as both have been shown to scale up to 8 threads and both use Mantle.

Also, have a look what happens when the game is optimized for todays CPUs.

http://gamegpu.ru/action-/-fps-/-tps/metro-last-light-redux-test-gpu.html

Metro Last Light

Metro Last Light Redux

 

[Abwx]

Got it, so in 2015 then? FX8 > i5 is what you're predicting? My argument with you in this thread has always been about i5 and never the i3. I don't shift goal posts, I'll leave that to you and the rest of the people preaching the same thing for the last 8 years. That isn't an opinion either.

I said that the i5 had 42% advantage last year and 25% currently, it s not a theorical thing it s a fact, the FX is gaining while the i5 is downgraded progressively.

Now, if you look at my post history you ll notice that it is only recently that i ended posting in threads where people are asking for advices, a few times ago i didnt bother much to debate such issues, thing is that a lot of people here are deliberatly giving bad advices because of brand loyalism if not fanatism, let s summarize this thread :

A member ask for advice about an FX8 or an i3 and automaticaly a few cluless are advising for the i3 with the wrong claim that it s better than the FX8, once i question this choice the goalpost is moved to the i5 with everybody advising to buy a more expensive MB than expected under the guise that it can be upgraded to the i5 and that this latter is better in games and that there s an upgrade path, but which upgrade path exactly.?? An i5.?.But by the time he upgrade, even to a Broadwell i5, the FX will be largely at i5 level in games, so that s just a downgrade in respect of what he will get comparatively to an FX now.

But then let look at the money, a more expensive MB and an i3 that cost about 130$, that is, 130$ now and 200$ later while an FX cost 140$ and will get him a better result than an i3 now and an i5 later in games, for other apps it s a no contest , now and in the future...

As said some must be really fanatics to give such bad advices without even checking the numbers, it is all urban legends and hearsays recycled ad nauseam, as ridiculous as using Cinebench in single thread to check which CPU is the faster, that is, using a soft in a way that will never be used in any universe, only a fool would do renderings in ST, but nevermind, it s the single core score that matters for such "specialists"..

 

[BSim500]

if tweaked the power value can be reduced by 15-17%% down to 55W/8C.

Another one of your favorite "undervolted AMD vs stock Intel" apples and orange comparisons? Let's do the opposite - if "tweaked" (ie, undervolted) you can get can a 55w i3 down to 35w whilst the stock FX is 95w... (See what you did there?)

On hardwre.fr games suite the i5 has 25% advantage over the FX, down from 42% last year, how much in a year or two.?.

Yet again I'll repeat what was explained to you earlier - 12% of that relative 17% "growth" in performance FX-8350 vs FX-8100 against an i5 is down to higher clock speeds (4.0GHz vs 3.6GHz). And the increased gap between FX-8xxx vs FX-4xxx scores and i7 vs i5 vs i3 vs Pent are the result of a jump to "next gen gaming engines" as a result of the better threaded consoles which occurs once every 7-8 years. Are 16x core PS5 consoles going to be released next year, then 32x core PS6 in 2016? No (and even if they were you wouldn't see remotely the same figures for each doubling of cores due to Amdah's Law - people have actually benchmarked games on 12C/24T / 15C/30T Xeon's you know). Are clock speeds going to increase 12% per year? Also no (for both brands). All you're doing is taking a once in a decade console architectural shift and trying to comically "pretend" it's some annual linear 20% threading efficiency growth which is so delusional it's not worth commenting on any longer. :thumbsdown:

An FX-8320E isn't a bad chip compared to the i3, but your claims of "An FX-8320 will destroy an i5 in 2015 - this is the year" are the same baseless broken worn-out record today that they were when started up in general in 2006 by people who have difficulty grasping game code doesn't scale up like x264 codecs and "theoretical throughput" obsessions are meaningless because of that. Even the "best case" benchmarks here by myself and AtenRa show the 100% core increase of 8-cores AMD's getting only roughly 25-35% more than 4-core AMD's - and again, a chunk of that is +400Mhz base clock difference. So that's "25-30% better threading growth in total", not "25-30% better threading growth per year". Using your "maths" by 2020, FX-8320's will enjoy 100% load on 14 of their 8 cores.

 

[NTMBK]

Go for the i3. More modern motherboards (with support for M.2 SSDs), better single threaded performance, and in 3 years' time you can get a second hand i5 or i7 for the board and make it last a few years longer.

 

[beginner99]

Just save another $100 and then get an I5. How old is OP? Christmas is near. Put parts or money for parts on wishlist.

 

[Abwx]

Another one of your favorite "undervolted AMD vs stock Intel" apples and orange comparisons? Let's do the opposite - if "tweaked" (ie, undervolted) you can get can a 55w i3 down to 35w whilst the stock FX is 95w... (See what you did there?)

Why didnt you quote my post completely.?.

At stock an 8320E is 63W with all cores loaded, a 8370 is 65W, that s what the measurements show, if you had read accurately you would had seen that i gave the 33W figure for the i3.

These are real numbers for both the i3 and the FX, no need to come with your made up estimations based on about nothing, when i post thoses TDP figures it s out from real measurements, not urban legends.

Yet again I'll repeat what was explained to you earlier - 12% of that relative 17% "growth" in performance FX-8350 vs FX-8100 against an i5 is down to higher clock speeds (4.0GHz vs 3.6GHz).

Lol, the number is about the FX8350 tested in games last year and the same FX8350 tested in games this year, so much for your frequency growth, anyway keep on doing points out of misquotings and half quotes, this will surely help your credibility on the long run...

 

[BSim500]

Why didnt you quote my post completely.?.

At stock an 8320E is 63W with all cores loaded, a 8370 is 65W, that s what the measurements show, if you had read accurately you would had seen that i gave the 33W figure for the i3.

Because you're still doing what you did before - cherry pick the lowest *measured* figure "63w at 50% CPU load" you can find then compare it to the worst case "55w official TDP at 100% load" Intel figure based on nothing more than an already debunked "theoretical throughput" figures that supposedly "justify" the "50% AMD load" by 'matching' a Haswell i3 to an AMD FX-4xxx, but actually look positively silly in reality:-

On what planet is a 36min / 43avg FX-4300 ("half an FX-8320") "the same throughput" as an i3's 57min / 69fps when the Intel is almost 60% faster both min & avg? It's examples like this that show your "theoretical throughput" figures are totally bogus and divorced from reality, even when your whole post is quoted "in context". :thumbsdown:

Edit : "8 core AMD "E" chips use the same sub 65w load power as an i3".

Please - just stop it...
http://images.anandtech.com/graphs/graph8427/67026.png
http://hothardware.com/articleimages/Item2225/power.png
http://img.hexus.net/v2/cpu/amd/FX/8370e/Power2.png

 

[Gikaseixas]

OP, instead of picking the FX or i3 why don't you add a bit more money and get a proper gaming cpu? I would start with a i5.
The FX is not bad, I have one and it plays everything but nothing compares to a fast (ipc) haswell core in games. If you're buying now, get an i5.
The i3 while decent too will most likely struggle in future games. We're seeing some bench's that clearly show this.

 

[Flapdrol1337]

If I look at test results the i3 is probably a better pick than the i5 for price/performance, still my gut says get an i5

Wouldn't consider an 8 core FX, sixcore + overclock, maybe, but it just performs awful in games that want per-tread perfomance, like arma 3, total war, mechwarrior online, planetside 2, etc.

 

[Abwx]

Because you're still doing what you did before - cherry pick the lowest *measured* figure "63w at 50% CPU load" you can find then compare it to the worst case "55w official TDP at 100% load"

You keep creating misquotes and misleadings, the 63W are at stock with all 8 cores loaded at 100%, is it clear enough that you are inventing made up numbers out of my real numbers and voluntarly misquoting me, it is lies, do you even realize it, or is it willfully.?.

Here again thoses numbers, at 100% LOAD ON ALL CORES AND STOCK SETTINGS :

FX8320E 63W

FX8370E 65W

i3 4130 33W

Rather than living on the past try to get the new picture, AMD did a great job with thoses E series, the two models above can be overclocked at 4.0 and TDP is 85-87W in this case, with all cores loaded if ever you didnt understand...

For whom is interested in the technical sides i already pointed that theses chips are the Richland versions of the FX even if AMD didnt communicate about this, the voltages are the same as well as the frequency headrooms, despite a 32nm node they managed to get the 3.3 frequency stable at 1.116V , that s 1.5% more than an i5 4670K and quite remarkable but still not up to Intel s 22nm wich allow 3.5 at 1.0V in the 4770K, do the maths, the FX has to dissipate 21% more because of the process due to higher voltage, but they compensate by 10% thanks to SOI substrat that allow to have lower switching losses than a same sized bulk node.

 

[BSim500]

Here again thoses numbers, at 100% LOAD ON ALL CORES AND STOCK SETTINGS :

FX8320E 63W

FX8370E 65W

http://hothardware.com/articleimages/Item2225/power.png
http://img.hexus.net/v2/cpu/amd/FX/8370e/Power2.png
http://images.anandtech.com/graphs/graph8427/67026.png
http://www.kitguru.net/wp-content/uploads/2014/09/power-draw1.png
http://www.guru3d.com/articles_pages/amd_fx_8370_and_8370e_processor_review,7.html
http://www.vortez.net/articles_pages/amd_fx_8370e_review,5.html
http://www.phoronix.com/scan.php?page=article&item=amd_fx_2sep&num=8

So where's your linked "source" for the 65w max load figures then, given every other major review site is getting up to double and averaging 90-120w?...

"63w on 8 fully loaded cores" doesn't even pass the "common sense smell test" given the chips still use 32nm (and the move from 32nm Sandy to 22nm Ivy Bridge resulted in a drop from 95w to 77w even for quad cores)...

 

[Abwx]

If I look at test results the i3 is probably a better pick than the i5 for price/performance, still my gut says get an i5

Wouldn't consider an 8 core FX, sixcore + overclock, maybe, but it just performs awful in games that want per-tread perfomance, like arma 3, total war, mechwarrior online, planetside 2, etc.

In Total War, 1080p ultra + 780ti, an i3 at 3.5 does 31.2 FPS, the FX 8350 27.4 FPS, huge advantage for the i3, isnt it.?.

In ARMA 3 it s 28.3 and 20.6

In Watchdogs it s 37.4 and 60.3

In Crysis 3 it s 23.2 and 33.6

So i guess that neither Watchdog nor Crysis are to be accounted for, only the games where the i3 is better are relevant.

Notice that in all thoses scores the i3 is maxed out while the FX has a lot of margin left, also one has to realize that these scores are made on minimal systems where all other apps are shut down, to get thoses FPS with an i3 you ll have to shut off every other soft, set apart the game of course, the slightest usage by any other apps and thoses scores will collapse, it wont be the case with the FX.

 

[Flapdrol1337]

In games it's never going to be near full load avx powerconsumption, unfortunately. I don't think it's that relevant.

 

[BSim500]

In games it's never going to be near full load avx powerconsumption

I agree but he did say "65w at 100% load on all cores", where is where I'm calling BS.

 

[Abwx]

http://hothardware.com/articleimages/Item2225/power.png
http://img.hexus.net/v2/cpu/amd/FX/8370e/Power2.png
http://images.anandtech.com/graphs/graph8427/67026.png
http://www.kitguru.net/wp-content/uploads/2014/09/power-draw1.png
http://www.guru3d.com/articles_pages/amd_fx_8370_and_8370e_processor_review,7.html

So where's your linked "source" for the 65w max load figures then, given every other major review site is getting roughly double?...

My source is here :

http://www.hardware.fr/focus/99/amd-fx-8370e-fx-8-coeurs-95-watts-test.html

Measurements at the main and at the CPU 12V rail level before the VRMs wich can be estimated at 90% efficency for all plateforms.

It s much more professional than Anand s phony measurements, because what they do not mention is that Intel CPUs are rapidly throttling under IBT while AMDs do not.

Even using Prime 95, wich use AVX and consume more than IBT, Hardware.fr didnt manage to get thoses figures, at most they got 80W TDP with power virus, notice that the 33W figure i gave for the i3 wouldnt be valid in such a test, you should had also posted the figures for this CPU but i guess that only AMD has to be trashed and that you wouldnt contradict me if i give a low number for an Intel CPU...

To get back to the numbers i posted theses are extracted under Fritzbchess wich is a little more power hungry that Cinebench.

Besides i looked at Guru 3d review, they measure 99W power delta at the main, accounting for the PSU losses this yield 90W at the 12V rail level, extract the VRMs 10% losses and we re at 81W assuming that all the delta is consumed by the CPU, and it is under Prime 95 wich is more power hungry than IBT, so there s really a lot of insight in the graph you posted, that is, how did Anandtech get thoses figures while HFR got about the same result as Guru3D with P95.??.

 

[Flapdrol1337]

In Total War, 1080p ultra + 780ti, an i3 at 3.5 does 31.2 FPS, the FX 8350 27.4 FPS, huge advantage for the i3, isnt it.?.

In ARMA 3 it s 28.3 and 20.6

In Watchdogs it s 37.4 and 60.3

In Crysis 3 it s 23.2 and 33.6

So i guess that neither Watchdog nor Crysis are to be accounted for, only the games where the i3 is better are relevant.

Notice that in all thoses scores the i3 is maxed out while the FX has a lot of margin left, also one has to realize that these scores are made on minimal systems where all other apps are shut down, to get thoses FPS with an i3 you ll have to shut off every other soft, set apart the game of course, the slightest usage by any other apps and thoses scores will collapse, it wont be the case with the FX.

Well, the 8350 is i5 money, and an i5 will perfom similar or much better than a 8350 depending on the game. i5 might be cheaper because you can safely install it on a $40 motherboard.

And yes, I do think the per-thread performance is more important than the total multithreaded capability, the games that need the per-thread performance will run awfully on an FX on all settings, while you can still play the multithreaded heavy games well on the i3's when you lower settings.

 

[NTMBK]

It s much more professional than Anand s phony measurements, because what they do not mention is that Intel CPUs are rapidly throttling under IBT while AMDs do not.

The reason that the Intel CPUs are throttling on IBT is because of the ridiculously good AVX throughput of Haswell. The clock drops a bit, but it is still way out in front of Ivy Bridge, Piledriver etc for SIMD.

 

[Abwx]

Well, the 8350 is i5 money, and an i5 will perfom similar or much better than a 8350 depending on the game. i5 might be cheaper because you can safely install it on a $40 motherboard.

And yes, I do think the per-thread performance is more important than the total multithreaded capability, the games that need the per-thread performance will run awfully on an FX on all settings, while you can still play the multithreaded heavy games well on the i3's when you lower settings.

40$ MBs are H81 based and a lot of people here insisted on the upgradability with a Broadwell as a good reason to go i3, the OP did choose the i3 mainly based on this perception.

As for the i3 it may be decent in games benched but what about in a real system where there are some apps that ask for a few ressources from time to time, let say an app that is downlading an update or anything else.?.

Anyway if you re interested in cores and HT scaling, with unfortunately only a handfull games tested, there s a comparison at HFR, they use a FX and a 4770K that is used to simulate an i3 and an i5 at 3.5, this is where i extracted the numbers i posted above :

http://www.hardware.fr/focus/101/perfs-avec-2-4-6-8-coeurs-4-jeux-loupe.html

 

[NTMBK]

40$ MBs are H81 based and a lot of people here insisted on the upgradability with a Broadwell as a good reason to go i3, the OP did choose the i3 mainly based on this perception.

Even without Broadwell support (which I honestly think isn't going to be that big a deal), he can still upgrade to a Haswell quad core and get a big improvement. Whereas with AM3+ there is nowhere he can upgrade to.

 

[Enigmoid]

My source is here :

http://www.hardware.fr/focus/99/amd-fx-8370e-fx-8-coeurs-95-watts-test.html

Measurements at the main and at the CPU 12V rail level before the VRMs wich can be estimated at 90% efficency for all plateforms.

It s much more professional than Anand s phony measurements, because what they do not mention is that Intel CPUs are rapidly throttling under IBT while AMDs do not.

Even using Prime 95, wich use AVX and consume more than IBT, Hardware.fr didnt manage to get thoses figures, at most they got 80W TDP with power virus, notice that the 33W figure i gave for the i3 wouldnt be valid in such a test, you should had also posted the figures for this CPU but i guess that only AMD has to be trashed and that you wouldnt contradict me if i give a low number for an Intel CPU...

To get back to the numbers i posted theses are extracted under Fritzbchess wich is a little more power hungry that Cinebench.

Besides i looked at Guru 3d review, they measure 99W power delta at the main, accounting for the PSU losses this yield 90W at the 12V rail level, extract the VRMs 10% losses and we re at 81W assuming that all the delta is consumed by the CPU, and it is under Prime 95 wich is more power hungry than IBT, so there s really a lot of insight in the graph you posted, that is, how did Anandtech get thoses figures while HFR got about the same result as Guru3D with P95.??.

Fritz is a complete BS benchmark and you know it. Comparing Conroe and HW there is virtually no gain in performance.

Cherry picking power numbers (No one cares about the 12V rail, they care about the power the CPU is pulling from the PSU). A crappy platform does not make a processor power efficient (original atom), nor can it be ignored.

Their results make absolute no sense. All FX cpus idle at the same power (no they do not). Haswell idles >2x higher than FX? No it does not.

Logic is not: Hardware.fr gets X while sites A-M get Y. All other sites must be wrong, because HWFR must be right. But rather the opposite.

 

[AtenRa]

On what planet is a 36min / 43avg FX-4300 ("half an FX-8320") "the same throughput" as an i3's 57min / 69fps when the Intel is almost 60% faster both min & avg? It's examples like this that show your "theoretical throughput" figures are totally bogus and divorced from reality, even when your whole post is quoted "in context". :thumbsdown:

FX4300 = 34/45
Core i3 4330 = 33/40

FX4300 = 91/111
Core i3 4330 = 102/111

Not to mention all those tests are run without any AntiAliasing. When you have a $300-400 or higher GPU and you enable 2x MSAA(no need to be higher) you become GPU limited in the vast majority of new AAA games at 1080p.
When you have a sub $300 GPU you are GPU limited even without enabling any AA.

As i have said before, low budget systems are GPU limited most of the time.
And for those that promote the upgrade path 3 years down the line, at that time he can upgrade to a new motherboard + $130 CPU that will be better than spending $200-250 for an older Core i7.