Is there any reason to use FX CPUs right now?

[MiddleOfTheRoad]

MiddleOfTheRoad, I largely agree, but for a home server I'd probably use a cat core or Atom-based solution, depending on what it's serving.

EDIT: In a situation where you're paying for electricity and working with constant heavy loads, an i7 will end up cheaper overall, and as fast or faster, while in a situation where you're never working with large loads, both 1150 and AM3 will be poor choices in terms of ecnomy. An FX fits the niche where you'll have very occasional spikey loads that are well threaded. There is the added upfront cost of a video card though, which is not included with FX chips...

AM3+ motherboard usually come with a motherboard-integrated Radeon 3000 or GeForce 7025. While nothing to call home about performance-wise, they are perfectly fine for use in a home server.

I measured the power draw of my friend's recently purchased FX 8320e versus my brand new i7 4790K -- and his system pulls less power under load according to my kill-a-watt meter.... Although my machine is a beast with lots of lights and touchscreen for fan controls, etc.

You definitely nailed the usage for a home server, though -- my home server is indeed an AM1 machine. An i7 or FX is probably overkill unless you are a power user.

 

[frozentundra123456]

Someone remind him of the topic.

I shoved you fx performance under mantle in comparison to intel cpus.

970 vs 290 is for the whole other forum. Its quite expected to see nv have better framerates in game based on their technology (modded civ5)

OK, lets look at mantle. In a gaming evolved game, under the api that is supposed to magically transform FX, it is *still* 15 percent slower. In addition, very few games use mantle and it will be several years before DX12 becomes widespread. So why would one want to endure subpar performance for several years in most games, only to move up to less subpar performance at best when DX12 becomes common, all the while using more power?

 

[frozentundra123456]

Surfing doesnt require a powerfull CPU, those tests are inflated for the purpose, a task that would take 0.1s instead of 0.15 is repeated a lot of times such that it will yield 10s and 15s but that s certainly not representative of the real experience, do you download 100 differents pages before starting to read the first one..?.

Also most people game on integrated GPUs and in this respect AMD s APUs are vastly superior to any Intel APU, a 80$ APU will trounce an i7 in this register, so in short Intel is better where it doesnt matter at all since the performance difference is not perceptible, that is, when browsing, and are lagging where it matters, gaming.

Would you pretend that a i3, or even i5, would give a better experience in a familial PC than a Kaveri.?

Possibly, but for this you ll have to spend even more money to buy a dGPU, otherwise the Intel APUs will be less good.

Nice shift of the goalposts to apus.

 

[Abwx]

i7 is significantly better than the i5 perf/Watt wise, it s not due to HT but mainly to a much better binning that allow 20% lower comsumption than the i5 when disabling HT, the i5 is challenged by the AMD FX E series when it comes to efficency, in integer MT tasks the FX8370E has often better perf/watt than a i5 4670K, what will set the difference is the plateform, with a 970 chipset the 8370E is competing without problem against a i5, this latter will be better in Single thread perf/Watt but not as good in well MThreaded workloads, particularly on integer code.

Nice shift of the goalposts to apus.

That s not a goal post shift, i think that it s important to state that AMD APUs are better at games than Intel s when using the IGP, as it makes no doubt that some people willfully fuel the confusion that Intel could be better with either a dGFX or an IGP.

 

[Enigmoid]

i7 is significantly better than the i5 perf/Watt wise, it s not due to HT but mainly to a much better binning that allow 20% lower comsumption than the i5 when disabling HT, the i5 is challenged by the AMD FX E series when it comes to efficency, in integer MT tasks the FX8370E has often better perf/watt than a i5 4670K, what will set the difference is the plateform, with a 970 chipset the 8370E is competing without problem against a i5, this latter will be better in Single thread perf/Watt but not as good in well MThreaded workloads, particularly on integer code.



That s not a goal post shift, i think that it s important to state that AMD APUs are better at games than Intel s when using the IGP, as it makes no doubt that some people willfully fuel the confusion that Intel could be better with either a dGFX or an IGP.

8370E is not more efficient than the 4670K.

 

[jpiniero]

You know it's funny; I was looking at the 8320E review and the 1100T is much more competitive with it than I would have thought even with the power consumption difference.

The other factor of course is that the 900 series chipset draws much more power than any of Intel's recent stuff.

 

[BSim500]

the i5 is challenged by the AMD FX E series when it comes to efficency, in integer MT tasks the FX8370E has often better perf/watt than a i5 4670K

Not this delusional "FX's draw less power than i5's" inverse-reality cr*p again... Even your own beloved "hardware.fr" with their extra-special "alternative measurements" and "30w i5's if you measure them in the 'right' way" don't show what you're claiming...
http://img.hexus.net/v2/cpu/amd/FX/8370e/Power2.png
http://images.anandtech.com/graphs/graph8427/67026.png
http://hothardware.com/articleimages/Item2225/power.png
http://www.phoronix.com/scan.php?page=article&item=amd_fx_2sep&num=8
http://www.vortez.net/articles_pages/amd_fx_8370e_review,5.html
http://www.guru3d.com/articles_pages/amd_fx_8370_and_8370e_processor_review,7.html
http://images.anandtech.com/graphs/graph8427/67026.png

"But everyone on the planet except hardware.fr is wrong because I say so!" - Abwx

OK, so let's find some of their charts, one power consumption the other efficiency:-
http://www.hardware.fr/getgraphimg.php?id=60&n=1
http://www.hardware.fr/getgraphimg.php?id=62&n=1

There simply is no lucid debate to be had here... :thumbsdown:

 

[SlowSpyder]

Not this delusional "FX's draw less power than i5's" inverse-reality cr*p again... Even your own beloved "hardware.fr" with their extra-special "alternative measurements" and "30w i5's if you measure them in the 'right' way" don't show what you're claiming...
http://img.hexus.net/v2/cpu/amd/FX/8370e/Power2.png
http://images.anandtech.com/graphs/graph8427/67026.png
http://hothardware.com/articleimages/Item2225/power.png
http://www.phoronix.com/scan.php?page=article&item=amd_fx_2sep&num=8
http://www.vortez.net/articles_pages/amd_fx_8370e_review,5.html
http://www.guru3d.com/articles_pages/amd_fx_8370_and_8370e_processor_review,7.html
http://images.anandtech.com/graphs/graph8427/67026.png

"But everyone on the planet except hardware.fr is wrong because I say so!" - Abwx

OK, so let's find some of their charts, one power consumption the other efficiency:-
http://www.hardware.fr/getgraphimg.php?id=60&n=1
http://www.hardware.fr/getgraphimg.php?id=62&n=1

There simply is no lucid debate to be had here... :thumbsdown:

He said in integer multi-threaded tasks. We'd need to see the performance and power use to know if that is true or not. Just posting power numbers doesn't prove anything one way or another. I have no idea if he is correct or not, but just pointing out better overall power consumption isn't the argument being made.

 

[BSim500]

He said in integer multi-threaded tasks. We'd need to see the performance and power use to know if that is true or not. Just posing power numbers doesn't prove anything one way or another. I have no idea if he is correct or not, but just pointing out better overall power consumption isn't the argument being made.

"For consumption test we try to use software that is for all architectures fairly representative of what we get in applications in terms of performance and consumption. Our choice is currently focused on Fritz Chess Benchmark..."

"The Fritz chess benchmark puts a pointed emphasis on threading, specifically reflecting the integer performance of these CPUs."

It's already heavily biased in his favor and he's still wrong no matter how often he repeats his absurd claims...

 

[AtenRa]

Not this delusional "FX's draw less power than i5's"

He said perf/watt

in integer MT tasks the FX8370E has often better perf/watt than a i5 4670K,

 

[SlowSpyder]

"For consumption test we try to use software that is for all architectures fairly representative of what we get in applications in terms of performance and consumption. Our choice is currently focused on Fritz Chess Benchmark..."

"The Fritz chess benchmark puts a pointed emphasis on threading, specifically reflecting the integer performance of these CPUs."

It's already heavily biased in his favor and he's still wrong no matter how often he repeats his absurd claims...

Which link is that from, how do they score? I'd be surprised if the 32nm FX can keep up in performance per watt, but I'd be curious how close it can get in a multithreaded test. I know even with undervolting I wasn't too close, if I remember correctly, in Cinebench. But, I only had one person to compare against and only tried one clockspeed. Anyway, I'm genuinely curious to see the scores vs. power used.

 

[Abwx]

8370E is not more efficient than the 4670K.

There s tasks where it has intrinsical better perf/Watt, what about if the FX8370E consume 20% more than a 4670K and get you 25-35% better perf.?.

Integer throughput of the FX is a strong point and a consequence of the server orientated design.

 

[Erenhardt]

OK, lets look at mantle. In a gaming evolved game, under the api that is supposed to magically transform FX, it is *still* 15 percent slower. In addition, very few games use mantle and it will be several years before DX12 becomes widespread. So why would one want to endure subpar performance for several years in most games, only to move up to less subpar performance at best when DX12 becomes common, all the while using more power?

CIV BE got mantle from GE, but the games is still the same old civ5 that was bought by nvidia. So your 15% have nothing to do with CPU, but with GPUs. Come back when thread title finally goes through your skull.

 

[AtenRa]

Which link is that from, how do they score? I'd be surprised if the 32nm FX can keep up in performance per watt, but I'd be curious how close it can get in a multithreaded test. I know even with undervolting I wasn't too close, if I remember correctly, in Cinebench. But, I only had one person to compare against and only tried one clockspeed. Anyway, I'm genuinely curious to see the scores vs. power used.

Cinebench is the best case scenario for Intel CPUs.
It is also the worst case scenario for AMD CPUs

You can try x264 that i believe is integer, Pov-Ray i believe is more about float

 

[Abwx]

He said in integer multi-threaded tasks. We'd need to see the performance and power use to know if that is true or not. Just posting power numbers doesn't prove anything one way or another. I have no idea if he is correct or not, but just pointing out better overall power consumption isn't the argument being made.

The graph above are saying it all, although since some people point Hardware.fr as biaised, here their bias :

First they changed the 1150 board when the FX8370E was released as a mean to display better perf/watt for the 4670K, they did use an Asus MB that is the most efficient of the market as a mean to get 5% better perf/Watt at the plateform level but for sure their 4670K numbers are not represenative of the average 1150 MB.

Second they changed their AM3+ MB to use a more power hungry Sabertooth even with a 8370E, certainly not that representative since most will use a 970 chipset.

Third they used 1600MHZ DDR3 for the AM3+ instead of the officialy supported 1866, they said that it was due to the 1150 not supporting more than 1600, as if it was AMD s fault that Intel doesnt officialy support 1866 RAM, yet when they tested the Haswell E it didnt seem to be a problem that this latter had much better RAM bandwith, they used the officaly max supported frequency.

Finaly Hardware.fr is owned by LDLC, a online IT store, and the reviewer, Marc Prieur, officialy advised buying the 4670K in this store PCs offerings, wich he changed to the 4690K the very day the FX E was released, no wonder why his Intel set up has been given a lot of attention, not counting that his power supply efficency seems to increase from 88% to 94% when switching from AMD to Intel..

So here all Hardware.fr bias, that said they are much less biaised than most sites and that s why their reviews can be assumed as being in the top 10% best reviews.

 

[SlowSpyder]

I believe you are right, Abwx. In those tests the FX 8370E is faster than an i5 by a healthy margin. The FX 8370E uses more power, but I bet the FX 8370E does have better performance per watt in those multi-threaded tests. If nothing else they are probably quite close, which isn't bad for the FX built on an older process and being an older design.

 

[Abwx]

"For consumption test we try to use software that is for all architectures fairly representative of what we get in applications in terms of performance and consumption. Our choice is currently focused on Fritz Chess Benchmark..."

"The Fritz chess benchmark puts a pointed emphasis on threading, specifically reflecting the integer performance of these CPUs."

It's already heavily biased in his favor and he's still wrong no matter how often he repeats his absurd claims...

It s you who is biaised because you are cluless about Fritz, actualy it s very favourable to Intel when comparing the i5 to an FX since the former did get 8% better IPC in this soft than Ivybridge i5s, it s a very good IPC improvement given that increasing IPC in Integer code is much harder than on FP.

What else have you to say that will show that you dont know the numbers and are talking out of clulessness..?.

Yet i provided several time this link :

http://www.hardware.fr/articles/897-25/gains-moyennes-cpu.html

 

[Abwx]

I believe you are right, Abwx. In those tests the FX 8370E is faster than an i5 by a healthy margin. The FX 8370E uses more power, but I bet the FX 8370E does have better performance per watt in those multi-threaded tests. If nothing else they are probably quite close, which isn't bad for the FX built on an older process and being an older design.

Thank you, much appreciated.


All numbers point the FX as a very good chip that was designed with future proofing in mind, indeed if AMD chips were not competitive there wouldnt be such truck loads of viral marketing flooding all AMD related threads.


The power figures at the 12V CPU rail level are 72W for the 8370E and 60W for the 4670K with power hungry multithreaded softs, actualy Fritz use more power than Cinebench for instance.

That said let see how Fritz "favour" AMD compared to other chess games....

So using the chess game that is most favourable to Intel is assumed as being a bias that favour AMD, typical bad faith of some people when faced by numbers that do not suit their preference...

 

[TheELF]

Surfing doesnt require a powerfull CPU, those tests are inflated for the purpose, a task that would take 0.1s instead of 0.15 is repeated a lot of times such that it will yield 10s and 15s but that s certainly not representative of the real experience, do you download 100 differents pages before starting to read the first one..?.

You wanted more examples for things the old I5 is better,you choose to ignore them all.
Yes you don't need a very strong CPU for the interwebs so why should people get expensive and power hungry FX?

Almost nobody renders or converts videos all day long or does anything the
FX are good in,they surf ,watch videos and play games.

The one thing that people are starting to do a lot and was kind of slow on the intel,convert videos for media players,smartphones or even real time streaming ,intel addressed by integrating quick sync into the IGPU.

All in all you can do everything a normal person does better with an intel,even if you have to use a discrete card for games.

Yes an AMD will also do all these thing quite good enough, still why spend your money on a CPU that is very good at doing stuff you don't use a lot
if you can spend the same amount of money on a CPU that is very good at doing stuff you actually do use a lot.

 

[Abwx]

You wanted more examples for things the old I5 is better,you choose to ignore them all.
Yes you don't need a very strong CPU for the interwebs so why should people get expensive and power hungry FX?

In that case no need of i5 and i7s as well, or is that applicable only to AMD.?.

And the FX is not that power hungry, in Intel case the eventual power savings are wasted anyway in the CPU price...

Almost nobody renders or converts videos all day long or does anything the
FX are good in,they surf ,watch videos and play games.

As already pointed for another member dont try to pretend that you re a whole forum, terms like "nobody" are typical of a lack of argument, what you re forgeting is that a FX will do your encoding/conversion while you re surfing or playing games, try to do that with a i5, or worse, with a i3...

The one thing that people are starting to do a lot and was kind of slow on the intel,convert videos for media players,smartphones or even real time streaming ,intel addressed by integrating quick sync into the IGPU.

Quicksync will give a lower quality than CPU encoding, you didnt knew about it or do you want the HFR link where they test all accelerated encodings.?..

All in all you can do everything a normal person does better with an intel,even if you have to use a discrete card for games.

According to your own requirements, youtube, surfing and such tasks there will be no perceptible difference between an Intel and an AMD but for sure there will be a difference if you dont use a dGFX with the Intel CPU, yet more $ wasted and your power efficency relative advantage bursted in flames, litteraly...

Yes an AMD will also do all these thing quite good enough, still why spend your money on a CPU that is very good at doing stuff you don't use a lot
if you can spend the same amount of money on a CPU that is very good at doing stuff you actually do use a lot.

I explained that you can do everything at the same time with an FX, that s the big difference with overhyped i3s and i5, indeed it require a lot of misleading to get someone buy an i3 at same price as a FX83xx, hence why AT CPU forum become AT Gaming forum once there s comparison done by Intel supporters.

But even that gaming mantra is becoming useless, Computerbase.de show the i3 4330 as 3% better than the FX8350 in 1080p games while a 4690K has 17% lead, but that s assuming that your PC do nothing else than serving the game otherwise the FX will be better than both thoses CPUs if you happen to use background tasks while gaming, and there will be some inherently.


http://www.computerbase.de/2014-09/...2/#diagramm-abschliessendes-performancerating

 

[Enigmoid]

Not sure why people care about ATX12 power usage when its the wall socket usage that determine the total power of the computer.

 

[jhu]

Not sure why people care about ATX12 power usage when its the wall socket usage that determine the total power of the computer.

Just one person.

 

[Abwx]

Not sure why people care about ATX12 power usage when its the wall socket usage that determine the total power of the computer.

At the main and with averagely priced MBs it will be 105-110W with a 4670K and 125-130W with a FX8370E, if not less for the latter since using a Sabertooth get you 137W.

Anyway the power comsumption argument is moot, as said the tiny saving is more than cashed by Intel thanks to inflated pricings.

Just one person.

I provided numbers for the main, so your post is just what it is, ad hominem, i guess that you re transposing the disagreement we got when you claimed that the FX couldnt possibly be better or at least as good than Haswells in encoding and rendering, wich i proved you to be wrong.

 

[nemesismk2]

well my answer to "Is there any reason to use FX CPUs right now?" yes i have a reason i own 3 am3+ motherboards from Gigabyte and also have amd fx 4300, amd fx 8350 and amd athlon ii x2 270. Ow that is three reasons.

 

[BSim500]

Not sure why people care about ATX12 power usage when its the wall socket usage that determine the total power of the computer.

The real reason being it tries to hide the AM3 platform inefficiency. Eg, with 1 thread Intel uses 30w (12v) but 63w (240v) a 110% difference. AMD however, uses 35w (12v) but 96w (240v) a 174% difference. Under full load, the FX "appears" to draw only 13% more (12v only), yet in actual reality, draws 37% more (240v). The fake "12v" figures also "appear" to show the FX idling at only half that of the i5 Haswell. In practice (240v), it idles at 41% more (which is where 95% of computers spend 95% of their time). Hence the endless stream of "12v only hardware.fr" (and other power consumption) cherry-picking... :sneaky:

My utility company certainly doesn't "subtract 10-20%" off my electricity bill for my appliances PSU losses either. Shame because I'd love an energy supplier like Abwx who appears to get a 47-63% discount from being able to run his 96-137w FX chip at a mere 35-72w without requiring a PSU or motherboard...