Is there any reason to use FX CPUs right now?

[TurtleCrusher]

Hardware.fr's test was with 1x GTX 780Ti, GameGPU's was done with 2x 780 Ti's in SLI. So the fact the faster Intel CPU's got up to 20% higher fps on GameGPU (but the FX-8350 was within 7% of itself) was more about removing a mild GPU bottleneck on hardware.fr's test than his usual made-up "cherry picked scenes" excuses. As you can see by this chart, a single 780Ti dips down to 49fps (and average 56fps) where in 780TI SLI the min fps is 67fps and 85avg.

What's more hardware.fr included the amusing review note : "The framerate Watch Dogs for its part measured at a 20s racing in a pretty busy part of the game. We found scenes with framerate from 10 to 20% lower but the automatic backup system prevented us to use them in a very reproducible, as the changing environment." (whatever that means...)

So given all that, it's not really surprising that CPU's that can do up to 100fps are being held back on the hardware.fr's test. Which presumably is why he loves hardware.fr so much - if you can artificially cap the fastest Intel CPU's by 15-20% with a GPU bottleneck, then you can "narrow the gap" in CPU comparisons...) :sneaky:

Since when are we spending $900 on two video cards to pair with an FX-83xx that can be had for $115?

 

[BSim500]

Since when are we spending $900 on two video cards to pair with an FX-83xx that can be had for $115?

"We" aren't. But if you're going to test CPU's on the metric of CPU gaming performance, then it makes sense to eliminate GPU bottlenecks. Hence why every major review site tends to test CPU's on top-end cards and not 7770's at 3840x2160...

 

[Abwx]

What's more hardware.fr included the amusing review note : "The framerate Watch Dogs for its part measured at a 20s racing in a pretty busy part of the game. We found scenes with framerate from 10 to 20% lower but the automatic backup system prevented us to use them in a very reproducible, as the changing environment." (whatever that means...)

So given all that, it's not really surprising that CPU's that can do up to 100fps are being held back on the hardware.fr's test. Which presumably is why he loves hardware.fr so much - if you can artificially cap the fastest Intel CPU's by 15-20% with a GPU bottleneck, then you can "narrow the gap" in CPU comparisons...) :sneaky:

So according to you only AMD CPUs would endure a 10-20% lower framerates in said scenes, you dont realize that the gap would still be the same in %age..??..

I guess that it s not easy to deny real numbers, it s even impossible, also you dint adress the fact that the FX8350 has higher framerates with a single card that Game.GPU with a SLI set up...

And there s no GPU bottleneck, otherwise the 4790 wouldnt perform 14% better with about the same frequency difference than a 4770k, that is, perfect scaling, but hey, the GPU is bottlenecked for any intel CPU that is below a 4790, you see there s a simple explanation that fit some people bias...

Seriously, you dont even catch such basic statistics and you want to discuss such matters..??..

 

[BSim500]

So according to you only AMD CPUs would endure a 10-20% lower framerates in said scenes, you dont realize that the gap would still be the same in %age..??

Hang on, I thought I was on your extra special "ignore list"? LOL.

It wouldn't necessarily be the same in percentage since AMD & Intel's don't increase core loading the same way in heavy scenes. As usual in many games, if a particular scene spikes say a main physics thread load up far more than other threads, it will saturate one AMD core much earlier than one Intel one. 70% load / 30% remaining of an AMD core has less overhead to deal with a major single thread spike in "heavy" areas of a game / level than 70% load / 30% remaining of an Intel core due to the massive disparity in IPC, and since games don't scale up anywhere near 100% by piling on MOAR CORES, then it often won't scale the same way.

Reality check : Even in well threaded Watch Dogs, an FX-8350 is barely 20% faster than an FX-4300 despite having +100% the parallel horsepower plus a 5% clock speed advantage on top (ie, adjusted for clock speed, a +100% increase in cores results in a +15% in speed on the same architecture). And that's one of the better threaded games of 2014 where an FX-9590 doesn't get trounced by an i3 despite a +1.5GHz clock speed advantage and 6x extra cores.

Going through the list of games I posted, even ignoring Intel, there's often as little as 3% FX8350 vs FX4300 difference. The largest 8 vs 4 core AMD gap I've seen is AC: Unity with around 36% (49 vs 36fps). Meanwhile the i5's are regularly up to +70-100% faster than the same FX4300 (72 vs 36). If you genuinely cannot see what pulling literally double the fps per core (which are scaling in parallel on AMD's by only +0-36% per +100% core count with most being around 10-20%, ie, still well below the lowest blue "Ahmdal's Law" line even in the best threaded games) means as far as your comical predictions of "obsolescence for i5's purely because they don't have 8 cores", then I'm really not sure what else to say...

also you dint adress the fact that the FX8350 has higher framerates with a single card that Game.GPU with a SLI set up...

Maybe it has something to do with the fact hardware.fr ran their tests without any AA vs GameGPU's 4x MSAA? Another apples vs oranges "fair comparison" of yours. (It's also why some of us don't see hardware.fr as some infallible God-like unquestionable 'authority' on benchmarking...)

Seriously, you dont even catch such basic statistics and you want to discuss such matters..??..

I think you need to calm down and stop being so permanently angry when talking to people. You may find you'll get a lot more positive responses and less mocking / sarcastic replies with a less 'perma-rage' style of posting...

 

[AtenRa]

Haswell i5 - $180
Motherboard - $45+
Cooler - free with CPU

Total: $225


FX-8320 - $150
Overclocking board - $100+
Hyper212 Evo cooler - $35

Total: $285


It's also possible to go with a power supply that is approximately 100-200w smaller in an i5 system, saving another $20-40.

Cheapest Haswell i5-4430 cost $185 at newegg. This one is only at 3GHz base and 3.2GHz Turbo.

Next one the Core i5 4460 cost $190 at newegg.

With $45 H81 motherboard total cost goes to $230-235

As for the FX,

Asrock 970 PRO3 R2.0 = $55

It supports all FX83xx and the new FX83xxE. You can OC ANY FX83xx to 4.4GHz with this board easily. It has 6x SATA-6 ports, it supports CF and it has 4x Memory Dimms.

Also, every 125W TDP FX cpu like FX8320/FX6350 comes with a heat-pipe copper cooler that it is adequate to OC the 8-Core FX to 4.4GHz (Turbo off, under-voltage)

FX8320 = $150
FX6350 = $126

Asrock 970 Pro3 R2.0 = $55

Total = $181(FX6350) or $205(FX8320)

Now, if you dont want to OC and you are also care a lot about power consumption then you can go for the Core i5 4460. If you want to play with your PC(OC) and you also like to have the ability to CF then you go with the FX and a faster GPU like the R9 290.

Simple as that.

 

[TurtleCrusher]

I'd save $10 by going with the

$115 FX-8310
$25 Arctic Freezer Pro 7 Rev 2

The HSF isn't audible in my case even when ramped up and temps don't go past 56C when doing prime95. While I have the stock HSF on the shelf I'm sure it would've been fine. It feels about as heavy as the heatsink listed above.

Browsing through TigerDirect, they have a barebones kit of the 8310, cooler, MB, 8GB (2x4)ram, 1TB HDD, decent case and an 80% efficient (claimed) 650w PSU for $304 after rebate. There's deals to be had for system builders.

 

[Abwx]

Hang on, I thought I was on your extra special "ignore list"?

Let say that i m inclined to give some benefit of the doubt, and to be frank i have no one on ignore list in any forum, that s just not an habit of mine...

It wouldn't necessarily be the same in percentage since AMD & Intel's don't increase core loading the same way in heavy scenes. As usual in many games, if a particular scene spikes say a main physics thread load up far more than other threads, it will saturate one AMD core much earlier than one Intel one. 70% load / 30% remaining of an AMD core has less overhead to deal with a major single thread spike in "heavy" areas of a game / level than 70% load / 30% remaining of an Intel core due to the massive disparity in IPC, and since games don't scale up anywhere near 100% by piling on MOAR CORES, then it often won't scale the same way.

AMD cores are not that more stressed than Intel s ones, the overhead is the overhead, if an AMD is at 70% the Intel will also be at this rate, the difference will be in performance, wich can be seen on the graph, if the remaining 30% are used then the FPS will simply increase by 30% in both cases, the IPC is already accounted, why should it be accounted a second time?..

IF the intel does 70 FPS and the AMD 55 at both 70% then at full throughput the FPS will be increased by 21 and 16.5 FPS respectively.

Reality check : Even in well threaded Watch Dogs, an FX-8350 is barely 20% faster than an FX-4300 despite having +100% the parallel horsepower plus a 5% clock speed advantage on top (ie, adjusted for clock speed, a +100% increase in cores results in a +15% in speed on the same architecture). And that's one of the better threaded games of 2014 where an FX-9590 doesn't get trounced by an i3 despite a +1.5GHz clock speed advantage and 6x extra cores.


Going through the list of games I posted, even ignoring Intel, there's often as little as 3% FX8350 vs FX4300 difference. The largest 8 vs 4 core AMD gap I've seen is AC: Unity with around 36% (49 vs 36fps). Meanwhile the i5's are regularly up to +70-100% faster than the same FX4300 (72 vs 36). If you genuinely cannot see what pulling literally double the fps per core (which are scaling in parallel on AMD's by only +0-36% per +100% core count with most being around 10-20%, ie, still well below the lowest blue "Ahmdal's Law" line even in the best threaded games) means as far as your comical predictions of "obsolescence for i5's purely because they don't have 8 cores", then I'm really not sure what else to say...

Watchdog is scaling quite oddly with core count, it s like it use 6 cores on AMD but need 8 cores to do efficently so, hardware.fr also did some tests on this aspect, the difference between a FX4 and FX8 is 30%, you can see below the scaling thing in this game for common CPUs, the link provide more exemples :

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

Maybe it has something to do with the fact hardware.fr ran their tests without any AA vs GameGPU's 4x MSAA? Another apples vs oranges "fair comparison" of yours. (It's also why some of us don't see hardware.fr as some infallible God-like unquestionable 'authority' on benchmarking...)

Without AA in WD but on ultra settings, this allow to remove some GPU bottleneck since the cards perfs would collapse with AA and the games would then be GPU limited, did you think about this..?

I think you need to calm down and stop being so permanently angry when talking to people. You may find you'll get a lot more positive responses and less mocking / sarcastic replies with a less 'perma-rage' style of posting...

Although my style may sound harsh i generaly, if not always, discuss things based on numbers with possibly no ad hominem, as such your post would find a much better usage if it was a self recommended advice...

 

[escrow4]

I'd save $10 by going with the

$115 FX-8310
$25 Arctic Freezer Pro 7 Rev 2

The HSF isn't audible in my case even when ramped up and temps don't go past 56C when doing prime95. While I have the stock HSF on the shelf I'm sure it would've been fine. It feels about as heavy as the heatsink listed above.

Browsing through TigerDirect, they have a barebones kit of the 8310, cooler, MB, 8GB (2x4)ram, 1TB HDD, decent case and an 80% efficient (claimed) 650w PSU for $304 after rebate. There's deals to be had for system builders.

Except the mobo is a 780G model from 2008, the PSU is a firecracker and the case is barely adequate. And in prices outside America the 4460 and 8350 are both $245 here in Australia which makes even less sense to favour FX. And with Broadwell and Skylake + Z170 chipset coming I can't see why anyone would get a bottom of the barrel FX combo.

 

[BSim500]

if an AMD is at 70% the Intel will also be at this rate

A 4GHz Haswell core at 70% usage / 30% idle is not the same as an FX at 70% usage / 30% idle due to massive +50-70% differences in IPC. The 2.8GHz 70% will be doing more at the same load AND the 1.2GHz 30% idle has greater overhead before core saturation takes place in the event of 1 main thread spiking any single core (eg, demanding 1.4Ghz more AMD cycles which may only take 800-900MHz more on the Intel). It's not "either/or", it's both. Swap "AMD" for P4 and the same still applies if it makes you feel any better.

Without AA in WD but on ultra settings, this allow to remove some GPU bottleneck since the cards perfs would collapse with AA and the games would then be GPU limited, did you think about this..?

So your idea of a fair comparison is AA vs no AA, Ultra vs High, single vs SLI, etc, and you just "hope" they cancel each other out? The more you obsessively cling to hardware.fr even with stupid disparities like that, the more people just "switch off" and find decent benchmarks on other sites.

Although my style may sound harsh i generaly, if not always, discuss things based on numbers with possibly no ad hominem, as such your post would find a much better usage if it was a self recommended advice...

You're joking right? Have a good evening.

 

[TurtleCrusher]

Except the mobo is a 780G model from 2008, the PSU is a firecracker and the case is barely adequate. And in prices outside America the 4460 and 8350 are both $245 here in Australia which makes even less sense to favour FX. And with Broadwell and Skylake + Z170 chipset coming I can't see why anyone would get a bottom of the barrel FX combo.

Prices are different in Austrailia, no doubt. That why I'm on a US site discussing US prices and deals. Who would've known! Prices in AU are a product of Geoblocking, so in your specific case going on the Whirlpool forums might be better if you talk prices.

Is the mobo going to be a problem for someone who would pair that system with a similar class GPU like the 750ti or R7 260? How about a single SSD or HDD? The motherboard is just fine. Most of you act like those who would pair an FX will do so with a Triple-Crossfire 290X configuration, 8 SSDs in raid with needed room for all internal watercooling loops with configurable VRM heatsinks that have WC plugs already attached, a dual-KillerNIC and an enthusiast grade soundcard built in.

GET REAL. The majority of people would pair that with a 750ti and a 1TB HDD and call it good. That's still more powerful than the average gaming system.

The PSU is more of a 550w but is definitely not one of Ultra's "cheap" PSUs. It'll work A-OK.

Here in the US the price difference is huge between FX and an i5/i7.

 

[Abwx]

A 4GHz Haswell core at 70% usage / 30% idle is not the same as an FX at 70% usage / 30% idle due to massive +50-70% differences in IPC. The 2.8GHz 70% will be doing more at the same load AND the 1.2GHz 30% idle has greater overhead before core saturation takes place in the event of 1 main thread spiking any single core (eg, demanding 1.4Ghz more AMD cycles which may only take 800-900MHz more on the Intel). It's not "either/or", it's both. Swap "AMD" for P4 and the same still applies if it makes you feel any better.

You didnt understand my previous post, using the 30% remaining throughput allow the 4770K to get 21 more FPS while the FX will get only 16.5, you understand that if the i7 start from a higher FPS it will need more throughput delta in absolute value to get 30% more FPS than the FX to get his 30% more FPS, in both cases the difference is 30%..

As said you re accounting twice the IPC difference...

So your idea of a fair comparison is AA vs no AA, Ultra vs High, single vs SLI, etc, and you just "hope" they cancel each other out? The more you obsessively cling to hardware.fr even with stupid disparities like that, the more people just "switch off" and find decent benchmarks on other sites.

The idea is to not bottleneck the GPU, with AA the differences between the Intels and the AMDs would be much lower in Hardware.fr review, the GPU would be FPS limited and you would be the first to protest that the CPUs differences are not highlighted and buried by a limited GPU, that s why i relied on Hardware.fr reviews while it s you that brought Game.gpu.ru who use a SLI to counter the lower FPS due to using AA, hence they are introducing another variable, variable SLI scaling, in tests that are already hard to sort out..

You're joking right? Have a good evening.

I m dead serious, and you just proved my point since you had to dig in a 2011 post of mine, although i must congratulate you for the choice of the member i was answering to, i let others decide if this was really a troll post considering our respective post histories....

 

[Abwx]

Except the mobo is a 780G model from 2008, the PSU is a firecracker and the case is barely adequate. And in prices outside America the 4460 and 8350 are both $245 here in Australia which makes even less sense to favour FX. And with Broadwell and Skylake + Z170 chipset coming I can't see why anyone would get a bottom of the barrel FX combo.

173 and 150€ in Europe for the 4460 and 8350 respectively, a 8320 is at 130 and a 8320E at 132€, the 4460 is below all thoses CPUs in applications and not that extraordinary better in games, it just lack the base and turbo frequencies of a 4670K to be significantly better in this latter case, it will best the FX8350 on Arma III and TotalWar R2, be on par on watchdog and below in Crysis, for anything other than games it s a no contest in favour of the FXs, comparatively it s not as future proof as the FXs...

 

[mrmt]

Except the mobo is a 780G model from 2008, the PSU is a firecracker and the case is barely adequate. And in prices outside America the 4460 and 8350 are both $245 here in Australia which makes even less sense to favour FX. And with Broadwell and Skylake + Z170 chipset coming I can't see why anyone would get a bottom of the barrel FX combo.

AMD is in the middle of an inventory crisis right now. Expect a lot of these bottom of the barrel combos in the next two quarters.

 

[frozentundra123456]

173 and 150€ in Europe for the 4460 and 8350 respectively, a 8320 is at 130 and a 8320E at 132€, the 4460 is below all thoses CPUs in applications and not that extraordinary better in games, it just lack the base and turbo frequencies of a 4670K to be significantly better in this latter case, it will best the FX8350 on Arma III and TotalWar R2, be on par on watchdog and below in Crysis, for anything other than games it s a no contest in favour of the FXs, comparatively it s not as future proof as the FXs...

applications will only favor FX in heavily threaded applications. To say it will be faster in "anything" other than games is a gross distortion of the facts.

Edit: BTW, how can something be "future proof" when it isnt even "present proof". Best you can hope for is that it will partially catch up.

 

[mrmt]

Edit: BTW, how can something be "future proof" when it isnt even "present proof". Best you can hope for is that it will partially catch up.

Were AMD current designs future proof they wouldn't be building something from the stratch, but rather doubling down their bet on the R&D path. That they elected to not do so in their future designs says everything about how future proof the thing will be.

 

[Enigmoid]

Watchdog is scaling quite oddly with core count, it s like it use 6 cores on AMD but need 8 cores to do efficently so, hardware.fr also did some tests on this aspect, the difference between a FX4 and FX8 is 30%, you can see below the scaling thing in this game for common CPUs, the link provide more exemples :

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

The 8 core FX chips are removing part of the module penalty. 4->8 core scaling is quite a bit less than what is shown.

 

[Abwx]

applications will only favor FX in heavily threaded applications. To say it will be faster in "anything" other than games is a gross distortion of the facts.

Edit: BTW, how can something be "future proof" when it isnt even "present proof". Best you can hope for is that it will partially catch up.

Right that there are some benches like Sunspider where the i5 will do better, but is 200ms rather than 300ms that relevant, so not only the perf difference although substancial wouldnt be noticeable but it isnt even representative of real world perf with say Firefox that use 4 cores for sure, it would be like estimating a CPU perf with Cinema 4D with only a single thread test of this soft.

So the FX is already present proof unless one use irrealistic and meaningless tests.

Personaly i pointed softs used by Hardware.fr, thoses are rendering, encoding, photo editing, archiving, chesses and compiling, now could you point me where the FXs would be lacking and if it would be noticeable, let s discuss in practice, what are thoses softs of yours that say that it s neither present proof or eventualy not future proof, i gave my references now take some risks and produces your own numbers that fuel your point, or lack of, at this point i dont know...

 

[Abwx]

The 8 core FX chips are removing part of the module penalty. 4->8 core scaling is quite a bit less than what is shown.

Not sure when looking at other Mthreaded game, actualy i dont know how things are managed, Hardware.fr said that there are some curious results, indeed.

They start with 4 threads for the FX in this game, perfect scaling from 4 to 6 cores, yet there are forcibly modules whose two cores are used when going to 6 from 4 threads :

Edit : Hardware.fr didnt use a single module for a single core, 2C test use one module, 4C test use two modules.

 

[frozentundra123456]

Right that there are some benches like Sunspider where the i5 will do better, but is 200ms rather than 300ms that relevant, so not only the perf difference although substancial wouldnt be noticeable but it isnt even representative of real world perf with say Firefox that use 4 cores for sure, it would be like estimating a CPU perf with Cinema 4D with only a single thread test of this soft.

So the FX is already present proof unless one use irrealistic and meaningless tests.

Personaly i pointed softs used by Hardware.fr, thoses are rendering, encoding, photo editing, archiving, chesses and compiling, now could you point me where the FXs would be lacking and if it would be noticeable, let s discuss in practice, what are thoses softs of yours that say that it s neither present proof or eventualy not future proof, i gave my references now take some risks and produces your own numbers that fuel your point, or lack of, at this point i dont know...

anand cpu bench

I count seven benchmarks there that a soon to be 3 generation old lower level i5 wins, not counting gaming, which it won every game tested. I am sure a Haswell 4460 would be considerably faster.

Of course, I am sure you will say those are all menaingless benchmarks, but most or all of those multithreaded benchmarks you love so much are just as irrelevant to a lot of users. Matter of fact, the only cpu demanding thing I do is gaming, so they are irrelevant to me.

I dont do any editing or encoding, but if I did, I could set up the job to run overnite, so FX being faster does not matter, at least if one applies the reasoning in your post. Gaming matters more to me because I do it in real time. So if one stretches "logic" far enough, I guess you can find as reason to disregard any benchmark you wish.

 

[TurtleCrusher]

anand cpu bench

I count seven benchmarks there that a soon to be 3 generation old lower level i5 wins, not counting gaming, which it won every game tested. I am sure a Haswell 4460 would be considerably faster.

Of course, I am sure you will say those are all menaingless benchmarks, but most or all of those multithreaded benchmarks you love so much are just as irrelevant to a lot of users. Matter of fact, the only cpu demanding thing I do is gaming, so they are irrelevant to me.

I dont do any editing or encoding, but if I did, I could set up the job to run overnite, so FX being faster does not matter, at least if one applies the reasoning in your post. Gaming matters more to me because I do it in real time. So if one stretches "logic" far enough, I guess you can find as reason to disregard any benchmark you wish.

All you do is gaming? Surely you have an i3 then, right?

 

[Abwx]

anand cpu bench

I count seven benchmarks there that a soon to be 3 generation old lower level i5 wins, not counting gaming, which it won every game tested. I am sure a Haswell 4460 would be considerably faster.

Of course, I am sure you will say those are all menaingless benchmarks, but most or all of those multithreaded benchmarks you love so much are just as irrelevant to a lot of users. Matter of fact, the only cpu demanding thing I do is gaming, so they are irrelevant to me.

I dont do any editing or encoding, but if I did, I could set up the job to run overnite, so FX being faster does not matter, at least if one applies the reasoning in your post. Gaming matters more to me because I do it in real time. So if one stretches "logic" far enough, I guess you can find as reason to disregard any benchmark you wish.

Ok, so you evacuated the problem by stating that only games matters for you when it comes to raw throughput, so be it if it s your choice but IIRC your CPU is a 2500k and your GFX is a 7790, you know what that means.?.

That the CPU will be bottlenecked well before it reach a significant part of its throughput, with your GFX about any CPU would be relevant and would yield the same scores, as a hint here Hardware.fr review of the 7850K, they use a 7750 and R5 250 with an i3 and a Kaveri, this latter in both single and dual graphic tests, sure that there are few games but still, the trend is clear enough :

http://www.hardware.fr/articles/913-10/gpu-dual-graphics-jeux.html

 

[frozentundra123456]

All you do is gaming? Surely you have an i3 then, right?

No.

 

[frozentundra123456]

Ok, so you evacuated the problem by stating that only games matters for you when it comes to raw throughput, so be it if it s your choice but IIRC your CPU is a 2500k and your GFX is a 7790, you know what that means.?.

That the CPU will be bottlenecked well before it reach a significant part of its throughput, with your GFX about any CPU would be relevant and would yield the same scores, as a hint here Hardware.fr review of the 7850K, they use a 7750 and R5 250 with an i3 and a Kaveri, this latter in both single and dual graphic tests, sure that there are few games but still, the trend is clear enough :

http://www.hardware.fr/articles/913-10/gpu-dual-graphics-jeux.html

I have a i5 2320 and HD7770. Point is , I was just trying to show how absurd it is to cherry pick benchmarks and say any benchmark that does not fit ones preconceived ideas is somehow invalid.

 

[soccerballtux]

Yes, similar to the Cinebench performance numbers while using compiled binaries instead of precompiled binaries. The results are what they are even in the most ideal circumstances (custom compiled, CPU specific binaries): FX 8350 has to use 8 threads and higher frequencies to get about the same performance that a Haswell Core i5 does at lower frequencies with 4 threads.

oh. those results never were clear to me.
-On the 8350s how many threads were being used (particularly the samples/s column confused me, but looking closer now I understand)
-if the Turbo was functioning 100% of the time (back then I didn't know what the stock non-turbo 8350 freq was)
-why you didn't use core vs. module on the single/multi threaded FX parts

also, I'm not aware what the most recent Intel gen is, and consider parity (more or less) with the i7 2600 to be more than sufficient to consider AMD FX a win and worth the money at the <$150 pricepoint.

but I see what you say about the single threaded performance now :/

 

[escrow4]

173 and 150€ in Europe for the 4460 and 8350 respectively, a 8320 is at 130 and a 8320E at 132€, the 4460 is below all thoses CPUs in applications and not that extraordinary better in games, it just lack the base and turbo frequencies of a 4670K to be significantly better in this latter case, it will best the FX8350 on Arma III and TotalWar R2, be on par on watchdog and below in Crysis, for anything other than games it s a no contest in favour of the FXs, comparatively it s not as future proof as the FXs...

Magical futureproof FX . . . . then Skylake will hit around Christmas and put it in the cemetery once and for all.

You are always go on about future proofing but right now FX is just outclassed. It will just be poorer and poorer as Intel continues with tick-tock.