The Best CPU for the Money: AMD FX vs. Intel Budget Shootout
- Thread starter Steve
- Start date
Where single core performance still matter, e.g. some games and application that are only aware of 2-4 threads, then the i3 is on par with an overclocked FX 8000. When games are aware of more cores or just are more heavily CPU optimized, same goes for apps, an 8 core FX will always jump ahead of an i3.
The best examples I can throw at you are Crysis 3, Battlefield 4 in both DirectX and Mantle and Thief in both DirectX and Mantle. Latest patch, latest drivers, of course.
In all these games, which are properly multithreaded, the 8 core FX takes its rightful place way ahead the fastest i3 and faster than an i5.
That’s a great bit of fan fiction there but here is the real story…
Obviously all games were patched to the latest version and the latest GPU drivers were used. All testing conducted using DirectX.
The overclocked FX-8320 makes up some nice distance here but it is not enough to beat the Core i3-4360.
The results are pretty close in Battlefield 4, you certainly aren't going to notice the difference between the FX-8320E and Core i3-4360 or the Core i5-4430 for that matter. Nonetheless the plucky little Core i3 did beat the FX processor.
No self respecting AMD fan boy should even acknowledge DX11 performance in Thief, Mantle all the way. For whatever reason the FX series gets hosed in Thief regardless of the version, it has been that way since day 1. Sadly Mantle support is only available in a few select games and won't help FX users who went with the green team.
Last edited:
GeforcerFX
Posts: 1,089 +547
That’s a great bit of fan fiction there but here is the real story…
Obviously all games were patched to the latest version and the latest GPU drivers were used. All testing conducted using DirectX.
The overclocked FX-8320 makes up some nice distance here but it is not enough to beat the Core i3-4360.
The results are pretty close in Battlefield 4, you certainly aren't going to notice the difference between the FX-8320E and Core i3-4360 or the Core i5-4430 for that matter. Nonetheless the plucky little Core i3 did beat the FX processor.
No self respecting AMD fan boy should even acknowledge DX11 performance in Thief, Mantle all the way. For whatever reason the FX series gets hosed in Thief regardless of the version, it has been that way since day 1. Sadly Mantle support is only available in a few select games and won't help FX users who went with the green team.
You should edit that into the article would add to the gaming benchmarks which I was hoping for more games benchmarked.
Honestly we picked the games we did because they pretty much tell the whole story. We could have benchmarked with a dozen of the latest AAA+ titles and the findings would have been very much the same. Most new AAA+ titles get benchmarked here at TechSpot with a range of AMD and Intel CPU’s so we have a pretty good idea of what is going on.
Having said all that you are right, we should add this data to the main review and will likely do so shortly.
GeforcerFX
Posts: 1,089 +547
Really the only time you will find AMD FX chips pushing the same FPS as Intel is when the GPU is the bottleneck, like we see on the High res part, most of the games show a difference of 2-3 FPS across the whole field, even if he had been testing with a a 4790k in the field as well it would prob being pulling no more then a 2-3 more FPS then the rest, but on the low res benchmarks the 4790k would be rocking all the chips since the CPU becomes the limiting factor not the GPU. Toms HW reviewed the 8370E with 3 different GPUs and this was pretty much what it showed on all the games they benched.
I just picked up a new 8320e for my parent's pc. The difference between that an their old Phenom 965be @3.4ghz (8gb ddr2@800) is amazing. I also put in a Gigabyte 990fxa-ud3 mobo and 8gb G-skill DDR3@1600 ram and a Radeon R9 280.
Overall the system is noticeably faster and they keep commenting that they can't believe the difference. The cpu is paired with an the old Phenom 2 copper heatsink (8320e one that comes with it is a cheap aluminum one) and I've only seen it go as high as 40C during the Windows 7 system score. It will never reach that temp for what they do normally.
I didn't do much testing as I myself own a fx-9370, but in my Mom's Farmville games it's night and day difference. You might think it's the video card making the difference, but it's mostly on the cpu. For some reason I have yet to diagnose the R9 280 rarely leaves the 300mhz power state...which in Farmville, it never leaves it. Also, I had the R9 280 in with the old cpu/mobo and it was still as slow at cold molasses.
Overall the system is noticeably faster and they keep commenting that they can't believe the difference. The cpu is paired with an the old Phenom 2 copper heatsink (8320e one that comes with it is a cheap aluminum one) and I've only seen it go as high as 40C during the Windows 7 system score. It will never reach that temp for what they do normally.
I didn't do much testing as I myself own a fx-9370, but in my Mom's Farmville games it's night and day difference. You might think it's the video card making the difference, but it's mostly on the cpu. For some reason I have yet to diagnose the R9 280 rarely leaves the 300mhz power state...which in Farmville, it never leaves it. Also, I had the R9 280 in with the old cpu/mobo and it was still as slow at cold molasses.
SuperVeloce
Posts: 133 +34
yeah
yeah, same here. The cpu temperatures might be lower, but that is because on haswell heat dissipation from die is the real problem. Amount of heat put from cpu into your computer case is larger on FX cpus.
And there is one more thing: max. allowed temperatures for intel are much higher
Jugoslav
Posts: 21 +1
Are you seriously now trying to suggest that FX processors run cooler than a Core i3-4360 or Core i5-4430. Even with a closed loop liquid cooler an FX-8320E runs no cooler than either the Core i3 or Core i5 processors using the little Intel box cooler.
Please refer to the orange charts in the review, they are displaying power consumption in watts.
Think of it like buying a heater for your bedroom, you want the one with the most watts. The FX-8320E is a way better heater than the Intel processors. Those Intel processors struggle to generate 100 watts of heat in most applications so a heavy Photoshop session just isn't going to keep you warm at night.
The AMD FX-8320E on the other hand pumps out almost 200 watts and if you need more heat just overclock it for a balmy 260 watts. If that's not enough heat thrown in a Radeon R9 290X or two, make sure you get the reference cooler though.
Jugoslav
Posts: 21 +1
Intel processors struggle to generate 100 watts of heat ...???
Since when is heat meausured in watts? I thought its meausured in celsius or fahrenheit.
More watts doesnt mean more heat. If power plant produces more watts it doesnt mean its hotter, its just means it produces more electricity.
cliffordcooley
Posts: 13,141 +6,441
It's not, but it is proportional. Since you made that statement, I'm guessing you don't understand how a heating element produces heat.
dividebyzero
Posts: 4,840 +1,271
AMD's processors have a small advantage regarding heat production based on the manufacturing process (SOI vs FinFET), and FinFET doesn't scale particularly well (input power vs heat generation) because of higher current leakage.
Having said that, within the terms of this review, Intel's i3 (and i5 for that matter) run cooler due in part to consuming less power. An AMD FX chip will consume around 10% more power than (assuming the chips workload consumes 80% of the 95W TDP for intensive workloads) the Intel chips (which generally consume around 82-88% of their rated TDP under the same conditions).
Even with SOI manufacturing taken into account, an i3/i5 will generate less heat than a comparable FX processor under the same workload.
G
Guest
Just wanted to chime in since I have been watching this thread. Jugoslav you aren't nearly as sharp as you think you are. You seem to lack basic comprehension.
The author never said heat was measure in watts. Your analogy is also f*!@#ing stupid, I am not going to point out why because its that stupid.
In fact your entire purpose for being here is stupid. What have you been trying to prove? The results you have been complaining about mean nothing. As an example who cares if the overclocked FX 8320E can score 1000pts or 1100pts in a given test, it still uses way too much power to get that result and isn't much faster than a Core i3 if it is faster at all.
Yes they are priced the same but out of the box the 8320E gets smoked by the Core i3 in most applications/games. You can't overclock the 8320E without extra cooling and at this point you are better off with a Core i5. There is no scenario where it makes sense to buy an 8320E over a Core i3 or Core i5, its that simple so just deal with it.
This has all been pointed out already though so are you just that thick?
Pretty much what I thought.
Don’t worry about it. Jugoslav is just a very passionate FX 8320E owner which is great. Granted trying so hard to discredit a review that pretty much states what all power users already knew wasn’t the best way of going about things.
Still it created discussion about the review and that is always a good thing. He even found a typo which was great
By all means let the discussion continue just don’t get personal.
Jugoslav
Posts: 21 +1
We are talking about processors, and not about heaters!? I mean WTF!!
does that mean Audi 2.0, will be hotter then Fiat 1.2... I dont think so.
How can u say its running cooler when max core temp in gaming is 45C, and as far as I know Intel idle temp is around 40C, and load temp is 68 and if you overclock it it goes above 90C. My cpu will never see 68C, not even in Prime95 test, so I dont know who is wrong here. And yes you are comparing watts with temperature.
WATT
Unit of Power ?
Symbol W
Named after James Watt
I care. We are talking about 529pts more in Pc Mark 8 Work bench and 291pts in Creative suite test on 4.5ghz cpu, here its tested on 4.6ghz, so the score should be even higher. That way 8320e would be just beneath i5, instead its beneath i3.
I know there are some variables, but yet score shouldnt be so different.
Here is the test where benchmarks are almost the same, and if you look at it closely, you will see that this test has a very different results.
http://www.legionhardware.com/articles_pagesprinter/msi_z87_xpower,1.html
Last edited:
cliffordcooley
Posts: 13,141 +6,441
That is what I thought, you don't see how they are related. Which is why you are making non-sense postings.
Put the same radiator in both cars and then comment again. You seem to be laboring under the delusion that an engine with more horse power doesn't require a larger radiator to dispense of the additional heat.
The same with CPU's when they pull more load, a larger cooler will be needed to remove the additional heat. Intel CPU's are more efficient, which is why they run cooler (contrary to what you are saying). They create less load (end result is wattage) in getting the same workload completed, which produces less heat (temperature as if you need me spelling this one out).
There is not one single device/gadget using electricity, that will not create heat (study the effect of moving electrons). I used the heating element because it is the easiest electric device to comprehend when it comes to how electric usage creates heat.
Jugoslav
Posts: 21 +1
Sorry my mistake I just remembered..... that my cpu is overclocked and undervolted, yes you can overclock and undervolt FX processors.
FX user can disable turbo, wich lowers vcore from 1.46 to 1.36V, then he can overclock the cores to 4ghz or more, and after all that he can undervolt the cpu. 8 core 4 ghz FX processor can run on 1.260V-1272V instead on 1.460V. 8 core 3.5ghz fx cpu can run on 1.164-1.188V. That can lower temperatures dramatically.
And if you add a 120mm fan on to the right side case panel, or 60mm fan on to VRM heatsink or on to the bottom of the motherboard that can further lower the temperatures. My Fx 8320 4ghz-all cores) in Aida64 stress test doesnt go over 44C on cores and socket temp doesnt go above 54C.
I really dont know why AMD didnt do all of that. We would avoid half load turbo, wich is max turbo but happens only on 4 cores, and it happens very rarley and we would also avoid full load turbo, wich happens on all of the cores, but only at 3.5ghz for 8320E processor, and happens most of the time. And if you dont enable HPC mode on stock processor, throttling happens wich backs down the cores to 2.9ghz ,that should also be removed from the cpu. If AMD processors wouldnt have all of that,
they would have much better performance, plus temperatures would be lower.
Once again Im very sorry I forgot that my cpu isnt stock.
FX user can disable turbo, wich lowers vcore from 1.46 to 1.36V, then he can overclock the cores to 4ghz or more, and after all that he can undervolt the cpu. 8 core 4 ghz FX processor can run on 1.260V-1272V instead on 1.460V. 8 core 3.5ghz fx cpu can run on 1.164-1.188V. That can lower temperatures dramatically.
And if you add a 120mm fan on to the right side case panel, or 60mm fan on to VRM heatsink or on to the bottom of the motherboard that can further lower the temperatures. My Fx 8320 4ghz-all cores) in Aida64 stress test doesnt go over 44C on cores and socket temp doesnt go above 54C.
I really dont know why AMD didnt do all of that. We would avoid half load turbo, wich is max turbo but happens only on 4 cores, and it happens very rarley and we would also avoid full load turbo, wich happens on all of the cores, but only at 3.5ghz for 8320E processor, and happens most of the time. And if you dont enable HPC mode on stock processor, throttling happens wich backs down the cores to 2.9ghz ,that should also be removed from the cpu. If AMD processors wouldnt have all of that,
they would have much better performance, plus temperatures would be lower.
Once again Im very sorry I forgot that my cpu isnt stock.
dividebyzero
Posts: 4,840 +1,271
Well, here is a Haswell i3 peaking at 61C under full load with a stock HSF.
Rubbish. The Intel processors used in this review are only overclockable via bus frequency. The OC record for the i5 tested here is 4.95% increase over stock on air based solely on upping the bus.
And yet, the review of the CPU with an AIO water cooler attached and using an open air test bench gives 64C using the same test at a frequency little removed from the one tested here.
I might also add that AMD's 970/990X/FX runs hotter than Intel's i3/i5 orientated motherboards (especially with overclocking factored in) . Check any review that logs voltage regulation, chipset, and southbridge temperatures. CPUs don't operate in a vacuum.
EDIT: I see you've amended some info. I will leave this post as it is since the information is still relevant in the context of the article.
cliffordcooley
Posts: 13,141 +6,441
I hope you don't think that FX is the only CPU's capable of that. I've over-clocked and under-volted both C2Q9400 and i7-2600K. I'm currently no longer doing so though.
Jugoslav
Posts: 21 +1
And yet, the review of the CPU with an AIO water cooler attached and using an open air test bench gives 64C using the same test at a frequency little removed from the one tested here.
64c cpu and 52c cores? I think thats pretty good for 4.7 ghz in Prime95 test
And here is the same processor overclocked on stock cooler. Its from the same site, how come you didnt see that.
http://www.guru3d.com/articles_pages/amd_fx_8320e_processor_review,18.html
Here are Other overclocked processors
And here is my FX 8320 4ghz on Scythe Katana 3 in Aida64 Stress test.
Max temp
Cores 43
CPU 52 ( labeled north bridge)
Last edited:
dividebyzero
Posts: 4,840 +1,271
I did see it, but whether it is the stock cooler or not wasn't mentioned. Just that the air cooling was achieved with a "simple heatpipe cooler" since the retail package wasn't shown in the product showcase. Whatever is received for review is what is photographed. Without knowing what cooler was used, I chose an example where the cooling is known.
In any event, the 61C recorded (on an open test bench) for the FX at 4.6 is still the same temp recorded by the i3 at stock offering better performance, and not having to worry about the runaway train power consumption.
Well, that was all pretty pointless, since 1. This review doesn't include any of those processors, and 2. There wasn't a comparison including an FX.
No one is doubting that SOI based processors run at lower temps than FinFET (for the exact reasons I outlined earlier), but it pretty pointless comparing an overclocked FX against an overclocked "K" Haswell when the article doesn't concern itself with the "K" SKUs, so you're basically talking a straw man argument....and of course, if you're holding up those CPUs as an example, shouldn't you also be holding up their relative performance achieved for that heat output?
Also bear in mind that the "K" processors are more likely to fall into different user scenarios not based solely upon budget - namely Crossfire/SLI, where AMD's FX's tend to fall flat on their collective faces compared to Intel's offerings.
In the end, you are trying to argue the superiority of an architecture laid down almost eight years ago, on an inferior process node, with inferior IPC, on a SOI process that is rapidly heading into history ...unless ET-SOI ever actually turns into a processor product
I did see it, but whether it is the stock cooler or not wasn't mentioned. Just that the air cooling was achieved with a "simple heatpipe cooler" since the retail package wasn't shown in the product showcase. Whatever is received for review is what is photographed. Without knowing what cooler was used, I chose an example where the cooling is known.
In any event, the 61C recorded (on an open test bench) for the FX at 4.6 is still the same temp recorded by the i3 at stock offering better performance, and not having to worry about the runaway train power consumption.
Well, that was all pretty pointless, since 1. This review doesn't include any of those processors, and 2. There wasn't a comparison including an FX.
No one is doubting that SOI based processors run at lower temps than FinFET (for the exact reasons I outlined earlier), but it pretty pointless comparing an overclocked FX against an overclocked "K" Haswell when the article doesn't concern itself with the "K" SKUs, so you're basically talking a straw man argument....and of course, if you're holding up those CPUs as an example, shouldn't you also be holding up their relative performance achieved for that heat output?
Also bear in mind that the "K" processors are more likely to fall into different user scenarios not based solely upon budget - namely Crossfire/SLI, where AMD's FX's tend to fall flat on their collective faces compared to Intel's offerings.
In the end, you are trying to argue the superiority of an architecture laid down almost eight years ago, on an inferior process node, with inferior IPC, on a SOI process that is rapidly heading into history ...unless ET-SOI ever actually turns into a processor product
Even with sufficient energy I couldn’t have said it better myself, as always nice info DBZ.
Peter Farkas
Posts: 654 +527
I still don't get why people are still buying AMD CPUs at all...
Similar threads
Latest posts
-
10 Things We Hate About Nvidia- Endymio replied
-
PlayStation 5 Pro will be bigger, faster, and better using the same CPU- poshflamingos replied
-
NASA's "next-gen" solar sail spacecraft is ready to launch for practical testing
- Cal Jeffrey replied
