FX 8370 Review

[Enigmoid]

Guess that you have trouble getting out of urban legends, in applications the 8370E is as good as 4670K, and it will be significantly better within the expected usage time, granted the 4670K has quite a lead in games but as times goes by dont expect it to perform much better contrary to the FX.

<SNIP>

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

I agree that the 8370E is a good chip but its not priced cheap enough to make up for it. Its $40 less making the build cost similar over the lifetime of the chip.

Performance in applications is neck and neck with the 4670K. However, why buy a K chip unless overclocking, the 4670 non K is $188 making it cheaper. Or the K chip can overclock.

http://www.ncixus.com/products/?usa...=BX80646I54690&manufacture=Intel&promoid=1342

Application performance is neck and neck and that is using perfectly threaded applications. Even future games thus are unlikely to exceed the performance of the 4670k, at best matching it on average. With the 4670 non K sitting at $188 (might be a sale or something) the 8370E needs a bit of a price cut.

 

[Abwx]

I agree that the 8370E is a good chip but its not priced cheap enough to make up for it. Its $40 less making the build cost similar over the lifetime of the chip.

Performance in applications is neck and neck with the 4670K. However, why buy a K chip unless overclocking, the 4670 non K is $188 making it cheaper. Or the K chip can overclock.

http://www.ncixus.com/products/?usa...=BX80646I54690&manufacture=Intel&promoid=1342

Application performance is neck and neck and that is using perfectly threaded applications. Even future games thus are unlikely to exceed the performance of the 4670k, at best matching it on average. With the 4670 non K sitting at $188 (might be a sale or something) the 8370E needs a bit of a price cut.

Depend of the applications, in winrar and winzip wich are integer tasks the 4670K is just slaughtered, 4670k has 6% lead in Fritz but the 8370E is better on Houdini and Stockfish (also chess games), the 4670K get some boost in X264 when AVX2 is implemented but nothing overhelming.

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

What the 8370E would need is a better chipset that has X99 like comsumption, that is below 7W, this would cut the power comsuption by 25W given that USB3 and SATA3 added controlers would be rendered useless , the power footprints would be reduced to 112W , wich is 5-10W more than 1150 plateforms using a 4670K, hence allowing the 8370E to have better perfs/watt even at the plateform level in the applications where it excel, too bad AMD was totaly blind on this matter, but could they guess that Haswell would got 15-20% worse perf/watt than IB and that the FX would be so competitive perfs/watt wise with such low power variants..??.

 

[Ramses]

6% lol.. Anything under 20% makes me want to take a nap.

 

[Enigmoid]

Depend of the applications, in winrar and winzip wich are integer tasks the 4670K is just slaughtered, 4670k has 6% lead in Fritz but the 8370E is better on Houdini and Stockfish (also chess games), the 4670K get some boost in X264 when AVX2 is implemented but nothing overhelming.

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

What the 8370E would need is a better chipset that has X99 like comsumption, that is below 7W, this would cut the power comsuption by 25W given that USB3 and SATA3 added controlers would be rendered useless , the power footprints would be reduced to 112W , wich is 5-10W more than 1150 plateforms using a 4670K, hence allowing the 8370E to have better perfs/watt even at the plateform level in the applications where it excel, too bad AMD was totaly blind on this matter, but could they guess that Haswell would got 15-20% worse perf/watt than IB and that the FX would be so competitive perfs/watt wise with such low power variants..??.

I agree with the chipset.

Winrar and 7-zip are low ipc applications often waiting on RAM, in this case HT benefits greatly.

I don't trust hardwarefr power measurements. The haswell vs. IVB vs. SB numbers don't make sense (IVB is virtually the same as SB when every other site reports it being substantially lower). You can't use fritz as a measure of efficiency.

Haswell is absolutely not 15-20% worse perf/W than IVB. On mobile its anywhere from equal to draw though performance has gone up more than the 35 -> 37W or 45 -> 47W jump. On servers the efficiency is pretty much the same. On the desktop it uses on average about 10% more power but then its about 10% faster.

 

[Abwx]

I don't trust hardwarefr power measurements. The haswell vs. IVB vs. SB numbers don't make sense (IVB is virtually the same as SB when every other site reports it being substantially lower). You can't use fritz as a measure of efficiency.

Haswell is absolutely not 15-20% worse perf/W than IVB. On mobile its anywhere from equal to draw though performance has gone up more than the 35 -> 37W or 45 -> 47W jump. On servers the efficiency is pretty much the same. On the desktop it uses on average about 10% more power but then its about 10% faster.

And yet they made great efforts to show Intel in good light, not that they are biaised but they are very carefull to not overstimate thoses CPUs power comsumption and they tend to use the most efficient Intel MBs.

They measure power at the 12V rail level wich is much more accurate than using the main deltas and the numbers are what they are, pleasant or unpleasant but they are far more relevant than AT for instance.

In this page you have the comsumption under Fritz (as a clue Prime 95 use only 24% more power than this soft) and the scores/watt on the two graphs at the bottom of the page :
http://www.hardware.fr/articles/913-9/cpu-consommation-overclocking.html

The 2500K has same efficency as the 4670K; the 3570K has 20% better efficency, the i3 IB has stellar efficency compared to its HW sibling..

 

[Enigmoid]

And yet they made great efforts to show Intel in good light, not that they are biaised but they are very carefull to not overstimate thoses CPUs power comsumption and they tend to use the most efficient Intel MBs.

They measure power at the 12V rail level wich is much more accurate than using the main deltas and the numbers are what they are, pleasant or unpleasant but they are far more relevant than AT for instance.

In this page you have the comsumption under Fritz (as a clue Prime 95 use only 24% more power than this soft) and the scores/watt on the two graphs at the bottom of the page :

The score/watt :

http://www.hardware.fr/articles/913-9/cpu-consommation-overclocking.html

The 2500K has same efficency as the 4670K; the 3570K has 20% better efficency, the i3 IB has stellar efficency compared to its HW sibling..

And I said before fritz is a terrible benchmark.

 

[Abwx]

And I said before fritz is a terrible benchmark.

Yes, it is mercyless...

Btw, what is a good bench since this one doesnt suit you..??.

Edit : i checked the numbers in CB 11.5 , using the power delta and the score (x100 to get a clear reading) we get :

2600K 10.1

3770K 13.7

4770K 12

IB has 14% better perf/watt than the 4770K and CB11.5 use ICC.

http://www.computerbase.de/2014-09/amd-fx-8370e-im-test/2/#diagramm-cinebench-r115

http://www.computerbase.de/2014-09/...mm-leistungsaufnahme-volllast-cinebench-x-cpu

 

[SlowSpyder]

Exaggerate much. Of course the extra power consumption is not a life changing experience. But for gaming, you get better performance and lower power use with an i5. The only advantage of the FX is lower initial cost, which is mitigated if not totally eliminated over the life of the platform by its higher power use. So I have to agree with Shintai on this one. Unless you use a lot of the productivity apps the FX excels at, it makes no sense to buy the FX. It never fails to amaze me the straw man arguments and logical fallacies introduced into these threads to attempt to justify the power use of the FX.

FX's aren't as efficient as Intel CPU's. But I believe the cost impact that has on your average computer user / gamer is nothing in real life.

 

[Enigmoid]

Yes, it is mercyless...

Btw, what is a good bench since this one doesnt suit you..??.

Edit : i checked the numbers in CB 11.5 , using the power delta and the score (x100 to get a clear reading) we get :

2600K 10.1

3770K 13.7

4770K 12

IB has 14% better perf/watt than the 4770K and CB11.5 use ICC.

http://www.computerbase.de/2014-09/amd-fx-8370e-im-test/2/#diagramm-cinebench-r115

http://www.computerbase.de/2014-09/...mm-leistungsaufnahme-volllast-cinebench-x-cpu

Yep and you see that SB != HW.

 

[Abwx]

Yep and you see that SB != HW.

In a bench that is Intel optimised, isnt it, but you know that my point was that IB has better perf/watt than HW by 15-20% according to the numbers i posted a few posts earlier, so far CB 11.5 seems to agree.

 

[Ramses]

FX's aren't as efficient as Intel CPU's. But I believe the cost impact that has on your average computer user / gamer is nothing in real life.

My electric supplier has some really good usage and data tracking stuff online, the way they, and I assume most, places tier and change the going rate throughout the day/week/season makes it difficult to figure it out, but rough math says my 9590, two 280x's, two monitors, stereo, nas, modem, router, UPS a lamp and laptop and whatever else happens to be plugged in here, is about $20 a month for 15 hours a day, two of which are gaming. And that was padding it. Jack crap, basically. I expected it to be higher but it explains why I don't really see it on my bill. Be worth thinking about if one was deploying dozens of them or something but for home use, imo, yawn. There's a nuke plant across the hills a few miles, I'd be more concerned if it was using a bunch of water as we are short of that here lol.

 

[frozentundra123456]

FX's aren't as efficient as Intel CPU's. But I believe the cost impact that has on your average computer user / gamer is nothing in real life.

That may be true, but the same argument then must be applied to the higher initial cost of the intel cpu. You cant just say the cost is insignificant in one case (extra power consumption of the FX) and then use a similarly small savings to justify buying the FX because it is cheaper. This kind of logic just makes my head spin.

 

[AtenRa]

I agree that the 8370E is a good chip but its not priced cheap enough to make up for it. Its $40 less making the build cost similar over the lifetime of the chip.

Performance in applications is neck and neck with the 4670K. However, why buy a K chip unless overclocking, the 4670 non K is $188 making it cheaper. Or the K chip can overclock.

http://www.ncixus.com/products/?usa...=BX80646I54690&manufacture=Intel&promoid=1342

Application performance is neck and neck and that is using perfectly threaded applications. Even future games thus are unlikely to exceed the performance of the 4670k, at best matching it on average. With the 4670 non K sitting at $188 (might be a sale or something) the 8370E needs a bit of a price cut.

You also have the FX8320E at $149,99, that is still a 95W TDP and it can also be overclocked to FX8370E level or higher.

The only problem i ever had with the FX line was the luck of Micro ATX format motherboards. For some reason only Full-ATX motherboards with the latest 9xx series chipset were ever made.

Also it is a loss for everyone AMD didnt have a 20nm HP process in 2014. A 20nm 8-Core Excavator could have 25% less power, 10-20% higher IPC and 10-20% higher Throughput at the same frequency of 4GHz at almost half the die size. They could make it FM2+ compatible and directly compete even against 4C 8T Haswell.
I believe 5GHz would be reached at 125W TDP easily even using 20nm bulk.

 

[Ramses]

People and money are frequently not logical. Look hard, you'll see it all over. It's an unspoken principle of capitalism. You think on the surface that computer stuff, since we benchmark and measure and try to be all scientific about it, would be a pretty hard and cut and dry thing, but it's not. It's just as emotional and illogical as anything else people get "in to". Lot of parallels to the hobby car market. Tons.

That said, between the slightly cheaper and the emotional appeal from AMD's kickin marketing dept, and a life long desire to take the road less traveled and not to run into 25 people with the same shit I have between home and the grocery store, and the fact that they are ALL plenty fast for anything but certain games(and even then AMD does OK), and you get people deciding to buy AMD.
Simple. There are other reasons, and other cases where folks just don't know any better, that's why we have benchmarks and review sites. I promise you if I was being hampered in the slightest I'd buy an i5 or i7 or whatever it took to not be hampered. But I'm happy as a clam with this box after buying two of almost everything in it short of the 16gig of cas7(happy overkill) over the last year finding parts that suited me.

It's not too much of a stretch if you hop POV's for a bit. I have lots of irrefutable reasons I keep buying AMD stuff, none of them are money or performance/benchmark related, the money is functionally irrelevant and the performance is plenty high on everything but the absolute bargain basement stuff. I replaced a 955BE with a 1090T, and it was faster but really not a ton faster, but I dug it enough to buy an 8350 since it'd be neat to see 8 cores in task manger and I figured it'd give me a bit more multitasking breathing room(and it did), and a year later as soon as the price hit the magic sub-250 I bought a 9590. Purely for the hell of it again, the 8350 was more than enough, but I dig the vibe of the 9590, so hey. I notice it keeps selling out too, so I suspect I'm not all by my lonesome.
Nobody with a clue is surprised that Intel stuff is faster, or more efficient, or more expensive, or that they sell more of em, it's Intel for effs sake. The Honda Accord of the CPU world. I'm not in love with em myself, if I needed a Civic or an Accord I'd buy one. Till then I'll stick with old Ford van's and 80's VW and Mercedes diesels. That analogy makes no sense if you don't know car junk probly.

With the number of people in the world today one can build a business on people buying things just for the hell of it, emotionally, to be different, because it's $2 cheaper, or any number of other illogical reasons. It don't have to be the best, it just needs to appeal to certain people, and enough to keep you afloat.

I sell to em every day.

I haven't been as satisfied on all levels with a computer as I am with this one since maybe a pair of P3's on a 440GX board with U320.
Or my long lost beloved Pentium Pros..
Computing is supposed to be FUN.

 

[AtenRa]

I haven't been as satisfied on all levels with a computer as I am with this one since maybe a pair of P3's on a 440GX board with U320.
Or my long lost beloved Pentium Pros..
Computing is supposed to be FUN.

Yeap, i recently had that moment tinkering with A10-7700K and A10-7850K. Playing around with those APUs trying to see how much they can perform was more interesting and fun than anything i did with PCs the last year or so.
Intel locked people out of OC when everyone could tinker and play even with the lowest Celeron a few years back. Look how exited people are with unlocked Intel Pentiums. Most people didnt even need to buy the anniversary Pentium, they mostly bought it just to play with it.
AMD is not the fastest this days but it sure is more fun to play around with and APU or FX processor than just simple buy a locked Intel CPU.

 

[Ramses]

Yeah, believe me I was taping pins on slot 1 cpu's and crap back years ago, peltiers on cellys, dual s370 Tualatin's, that stuff was fun. I got no beef with intel, other than not developing the Tualatin and introducing that POS p4 HT that I still have one of running in the shop for music and netflix. I go where my heart leads me whenever I possibly can, so many times in life one does not have that luxury. Hell when AMD stuff was faster and cheaper, I was running Intel lol.
Best isn't always best.
And you can't benchmark that.

 

[AtenRa]

Hell when AMD stuff was faster and cheaper, I was running Intel lol.
Best isn't always best.
And you can't benchmark that.

Hahahaha it was more fun to play with Pentium 4 2.4GHz and OC to 3.7GHz than play with AMD Athlon and get 200-300MHz OC top(air cooled). It was more fun to reach and overcome 4GHz water cooled on the Pentium 4, i dont believe anyone OCing at the time cared about power consumption. Power consumption was not even in the vocabulary at that time. You needed more power ?? you bought a bigger PSU :biggrin:
You wouldn't be concern with efficiency, you only cared to play with the damn CPU and OC the highest you could.

 

[Enigmoid]

In a bench that is Intel optimised, isnt it, but you know that my point was that IB has better perf/watt than HW by 15-20% according to the numbers i posted a few posts earlier, so far CB 11.5 seems to agree.

What does 'intel optimized' mean here? The comparison is between intel processors so any 'optimizations' will benefit the processors equally. And you are forgetting that only half of what you said is right, HW is more efficient than SB.

 

[SeanJ76]

AMD is 10 years behind Intel Processors.........if not more.....just in case you guys have not seen the huge bottleneck AMD cpus cause to gpus
8350@4.8ghz(7970) vs SB 2500k @stock 3.3ghz(660ti)
https://www.youtube.com/watch?v=HX6N9uJtzPA

 

[frozentundra123456]

You can pick benchmarks, and interpret them however, but bottom line is that Haswell gives longer battery life in mobile devices than SB or IVB, which was the intent. It also has the integrated voltage regulator, which will add some power consumption but reduce chipset power consumption.

 

[DrMrLordX]

So, sort-of back to the topic at hand.

We've seen the 8370E reviewed, and we even have a few data points showing that it can overclock pretty well on air/water compared to, say, the 8350. Looks like at least some of what TheStilt said about leakage is, in fact, true. Perhaps all of it.

So, where's the 8320E in all this? Has anyone seen any 8320E overclocks yet?

 

[AtenRa]

AMD is 10 years behind Intel Processors.........if not more.....just in case you guys have not seen the huge bottleneck AMD cpus cause to gpus
8350@4.8ghz(7970) vs SB 2500k @stock 3.3ghz(660ti)
https://www.youtube.com/watch?v=HX6N9uJtzPA

https://www.youtube.com/watch?v=LFcMRtq1zGo

https://www.youtube.com/watch?v=ch0mR4Evu8I

 

[jhu]

AMD is 10 years behind Intel Processors.........if not more.....just in case you guys have not seen the huge bottleneck AMD cpus cause to gpus
8350@4.8ghz(7970) vs SB 2500k @stock 3.3ghz(660ti)
https://www.youtube.com/watch?v=HX6N9uJtzPA

10 years? Clearly someone (you) is smoking crack. 10 years ago Intel was still making Pentium 4 for the consumer market. By all accounts, Bulldozer/Piledriver/Steamroller is significantly faster than Pentium 4 even if they only match Phenom II (single threaded) or Sandy Bridge (multi-threaded) at best.

 

[Ramses]

Power consumption was not even in the vocabulary at that time. You needed more power ?? you bought a bigger PSU :biggrin:

Kinda made me chuckle that everyone got all environmentally conscious too, I was out of the loop for a few years and came back and thought I was missing something when everyone was talking about power consumption
Just what's en-vogue I guess...

 

[Abwx]

What does 'intel optimized' mean here? The comparison is between intel processors so any 'optimizations' will benefit the processors equally.

And you are forgetting that only half of what you said is right, HW is more efficient than SB.

OK, 50% right because of IB/HW but actualy also half right on SB/HW since there are softs where the former does better than the latter, that sum up to 75%