• We should now be fully online following an overnight outage. Apologies for any inconvenience, we do not expect there to be any further issues.

ITT: We discuss future pricing and availability for AM3+ processors and mobos

Page 4 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

ShintaiDK

Lifer
Apr 22, 2012
20,378
146
106
It can have twice the perfs with the relevant application, that a 4C can do as well in lighltly threaded apps is not that relevant, what happens when the apps are updated.?.

i3 and FX4 will be stuck to their former perfs while the 8T CPUs will fly given their reserve.

How many years have we heard this? How many years are we to wait for something that may or may not happen?

There is no magic waiting, the FX8xxx wont get any faster, just constantly slower as time erodes them even further.
 

Abwx

Lifer
Apr 2, 2011
11,885
4,873
136
How many years have we heard this? How many years are we to wait for something that may or may not happen?

There is no magic waiting, the FX8xxx wont get any faster, just constantly slower as time erodes them even further.

You dont seem to follow the trend and numbers or you know them but are willfully ignoring them.

Do you realize that in october 2011 when the FX8150 was released we had apps that werent much Mthreaded, as a consequence the FX8150 was sold as a 2500K competitor, now go check the recent reviews that still use thoses CPUs and check where the 2500K is in respect of the 8150 currently, actualy the 2500K can no more compete with this CPU, perhaps you want some links..??.

That said if Mthread doesnt matter and wont matter more in the coming months/year then everytime someone ask if he should buy a 4770 or a 4690 just tell him that it s useless and that he should rely on an overclocked G3258 instead if he wants to go Intel....
 

ShintaiDK

Lifer
Apr 22, 2012
20,378
146
106
You dont seem to follow the trend and numbers or you know them but are willfully ignoring them.

Do you realize that in october 2011 when the FX8150 was released we had apps that werent much Mthreaded, as a consequence the FX8150 was sold as a 2500K competitor, now go check the recent reviews that still use thoses CPUs and check where the 2500K is in respect of the 8150 currently, actualy the 2500K can no more compete with this CPU, perhaps you want some links..??.

That said if Mthread doesnt matter and wont matter more in the coming months/year then everytime someone ask if he should buy a 4770 or a 4690 just tell him that it s useless and that he should rely on an overclocked G3258 instead if he wants to go Intel....

The 2500K still runs in circles around the FX8xxx in 99% of everything. Including new games.

And the 2500K still holds its value, same cant be said about the FX8150. Even BF4 looks terrible bad on the FX8150, something it should excel in. A complete flop of a CPU.

350x700px-LL-b0fc3bbb_http--www.gamegpu.ru-images-stories-Test_GPU-Action-Battlefield_4_Naval_Strike_-test-bf_4_proz.jpeg
 
Last edited:

Abwx

Lifer
Apr 2, 2011
11,885
4,873
136
The 2500K still runs in circles around the FX8xxx in 99% of everything. Including new games.

And the 2500K still holds its value, same cant be said about the FX8150. Even BF4 looks terrible bad on the FX8150, something it should excel in. A complete flop of a CPU.

350x700px-LL-b0fc3bbb_http--www.gamegpu.ru-images-stories-Test_GPU-Action-Battlefield_4_Naval_Strike_-test-bf_4_proz.jpeg

I was sure that you would provide a game graph, because games are about the single usage where the 2500k hold the comparison, (as if games were the only application of a 8 core CPU..) but let s look elsewhere, in apps that need cores and let s compare thoses two CPUs, and a lot of others, throughputs :

http://techreport.com/review/27018/intel-xeon-e5-2687w-v3-processor-reviewed/6

legacy-7zip-comp.gif



legacy-7zip-decomp.gif



legacy-qtbench.gif



legacy-tc-aes.gif


legacy-tc-twofish.gif


legacy2-pov-chess2.gif


Techreport were sympathetic to use the Povray test that was the less painfull for the 2500K, the other test used in this review is this one but there s no 8150 infortunately, or rather fortunately, isnt it..

pov-bench.gif





Fortunately there s some ICC compiled apps that help the 2500K "perform" well :

legacy-euler3d.gif


Another link where there s direct comparisons :

http://www.computerbase.de/2014-09/...ating-mit-anwendungen-und-spielen-1920-x-1080


You can see that in games the 2500K is not better and overall it s largely inferior to the 8150, it s time to stop believing in urban legends and look at real numbers instead.
 
Last edited:

ShintaiDK

Lifer
Apr 22, 2012
20,378
146
106
Another link where there s direct comparisons :

http://www.computerbase.de/2014-09/...ating-mit-anwendungen-und-spielen-1920-x-1080


You can see that in games the 2500K is not better and overall it s largely inferior to the 8150, it s time to stop believing in urban legends and look at real numbers.

You should check your link. The 2500K is still faster. But considering the place of the 8370E, I wouldnt take those numbers so serious.

If all you do is encoding movies, rendering and compress files. Sure the FX is faster. But most of us tend to do other things, specially for the main part of the time. And thats where the FX falls flat. And sales reflect this. The sales are appaling to put it mildly. Even the AMD said it was a failure.
 
Last edited:

cbn

Lifer
Mar 27, 2009
12,968
221
106
that said Beema has surely significantly better perf/watt than Kaveri, this latter is not at ease within a 19W TDP and should rather be used for the 25-35W range

Here are the specs for the Kaveri mobile APUs:

Kaveri-SKUs-640x265.jpg


Quad core @ 2.1 GHz/3.3 Ghz turbo with 384 Stream processors @ 533 Mhz looks like a really good spec for 17 watts.

So at this point I think we can say Kaveri clearly wins performance per watt against Beema A6-6310. This at a cost of increased die size for Kaveri though.

EDIT: In fact, here is passmark comparison between the processors:

http://www.cpubenchmark.net/cpu.php?cpu=AMD+FX-7500+APU (3419 cpu marks for FX-7500)

http://www.cpubenchmark.net/cpu.php?cpu=AMD+A6-6310+APU (2393 cpu marks for A6-6310)

However, for a basic desktop use case (where a lot of extra iGPU isn't needed), maybe a AM1/Beema (suitably souped up) could provide a decent enough level of performance. This at the cost of performance per watt, but since it is being used in a typical USFF desktop or AIO desktop I don't think this would be much of an issue (assuming TDP remains reasonable enough.)
 
Last edited:

Ramses

Platinum Member
Apr 26, 2000
2,871
4
81
I looked at the 2500 and it's middle of the road suit and tie siblings in August of 13, and bought an 8350 for less and it performed beautifully at anything I could find to do with it.
I'm looking at the posted and linked benchmarks above and I still don't see that it was a bad choice. I also notice the 2500k isn't available new from any of the usual suspects when I looked to see what it had been selling for now. I'm assuming it was still more than the buck eighty the 8350 was a year ago (and still is).

As an enthusiast, I don't just want fast, I want interesting.
None of the i3 and i5's were interesting, the i7's were, some of them, but
disproportionate money. Thus, AMD stuff. I'm yet to find the point where they fall flat.
I'm sure it's there, but I haven't seen it.
FX has had a nice long life and I expect it to continue right along selling for some time to come.
 

Abwx

Lifer
Apr 2, 2011
11,885
4,873
136
Here are the specs for the Kaveri mobile APUs:

Kaveri-SKUs-640x265.jpg


Quad core @ 2.1 GHz/3.3 Ghz turbo with 384 Stream processors @ 533 Mhz looks like a really good spec for 17 watts.

So at this point I think we can say Kaveri clearly wins performance per watt against Beema A6-6310. This at a cost of increased die size for Kaveri though.

However, for a basic desktop use case (where a lot of extra iGPU isn't needed), maybe a AM1/Beema (suitably souped up) could provide a decent enough level of performance. This at the cost of performance per watt, but since it is being used in a typical SFF desktop or AIO desktop I don't this would be much of an issue (assuming TDP remains reasonable enough.)

The 17W APU is a 2C, the lower power 4C is 19W, and no, Kaveri has not better perf/watt than Beema, the small cores are impossible to beat in this area, even mobile Haswell has lower perf/watt than Beema or Baytrail.

About the basic DT usage there s a point where i do not agree with most people here, that is the importance of GPUs.

Personaly i m not a gamer but i m not representative of the average user, about all the people i know do some gamings on their DT, few have very good set ups for such usage and they use what they have at disposal, more often than anything else they use an IGP, of course most people are ignorant that the GPU is what provide good perfs for games so they are relying on the urban lgend that the more powerfull the CPU the better it will do in anything including games, right, but only with a dGPU and that s what is completely canceled in all marketing motivated debates because this would point Intel s CPU s as being less good for the general public usages.
 

itsmydamnation

Diamond Member
Feb 6, 2011
3,079
3,915
136
So let me get this straight, enterprises will choose Intel apus because they are good enough but not AMDs...this train of though is very hard to follow.

yes, but not for this reason. Intel has several enterprise specific management features. Can you even get a TPM on an AMD MB?
 

Ramses

Platinum Member
Apr 26, 2000
2,871
4
81
I had an a8-6410 laptop for a bit last month.
Performed beautifully in any sort of basic/normal task.
With a 500gig ssd and 16gb ram on all of them, it felt..
Just like the i7-4510u I'm typing on now.
Just like my fx-9590 desktop.
Just like the fx-8350 desktop had before.

I just can't find much functional different that isn't task specific in any
of the newer decent chips. They are all plenty fast.
I am impressed with the performance given the low power consumption of the two
laptops vs this desktop. I'd have been happy to stay with the AMD stuff but the laptop
surrounding it was a cheap POS, so I ended up with the i7 HP.
 

cbn

Lifer
Mar 27, 2009
12,968
221
106
The 17W APU is a 2C, the lower power 4C is 19W

My mistake I should have written 19 watts since you were comparing Kaveri with that power level to Beema. See quote below from your post #75:

http://forums.anandtech.com/showpost.php?p=37029785&postcount=75

You are right that a bigger GPU would be area consuming, moreover given that it would need a dual channel controler, that said Beema has surely significantly better perf/watt than Kaveri, this latter is not at ease within a 19W TDP and should rather be used for the 25-35W range, this will undoubtly change with the Carrizos APUs, if the announced numbers are accurate this may well shift the OEMs attention to the Excavator variant rather than to the Beema replacement, ie, Carrizo-L.
 

cbn

Lifer
Mar 27, 2009
12,968
221
106
Kaveri has not better perf/watt than Beema, the small cores are impossible to beat in this area,

Did you not see the Passmark scores I posted in post #81?

EDIT: I guess you didn't since they were not quoted when you posted. Well, as you can see Kaveri does win in performance for cpu by quite a bit. Then, of course, there is a large iGPU specification difference between those 19 watt Kaveri and the 15 watt A6-6310 which doesn't look trivial either.
 
Last edited:

Abwx

Lifer
Apr 2, 2011
11,885
4,873
136
You should check your link. The 2500K is still faster. But considering the place of the 8370E, I wouldnt take those numbers so serious.

I was looking at the 2500k and the 8150 in computerbase.de charts, not the 8370E.

The charts says that the 2500K was a better choice in october 2011 ( you can check the october 2011 reviews) but with time the 8150 largely outmatched it.

If all you do is encoding movies, rendering and compress files. Sure the FX is faster. But most of us tend to do other things, specially for the main part of the time. And thats where the FX falls flat. And sales reflect this. The sales are appaling to put it mildly. Even the AMD said it was a failure.

They are talking commercialy speaking not perfs wise.

And yet you have a 4670 while you had a 2500K previously IIRC, so you re agreeing with me in some way, if all you needed was a CPU because :

most of us tend to do other things, specially for the main part of the time.
What are thoses things that mandate getting rid of the 2500k for a more potent 4C after only 2-3 years usage..??.

Actualy you just proved what i said above, that the 2500k was better on the short term while the 8150 was better for a mid/long term perspective, currently with your 4670 you have about the same throughput than a very lowly clocked FX8xxx.
 

ShintaiDK

Lifer
Apr 22, 2012
20,378
146
106
And yet you have a 4670 while you had a 2500K previously IIRC, so you re agreeing with me in some way, if all you needed was a CPU because :

What are thoses things that mandate getting rid of the 2500k for a more potent 4C after only 2-3 years usage..??.

Actualy you just proved what i said above, that the 2500k was better on the short term while the 8150 was better for a mid/long term perspective, currently with your 4670 you have about the same throughput than a very lowly clocked FX8xxx.

I didnt have a 2500K previously. I had a 3570K. And that one I still got because we needed a second gaming PC. And it still doesnt change the fact that the 2500K ages multiple times better than the FX8150.

However just because I may or may not replace it doesnt mean it age badly or not. We let reviews decide that. However when even the producing company says its a fail and you still champion it...
 
Last edited:

Abwx

Lifer
Apr 2, 2011
11,885
4,873
136
Did you not see the Passmark scores I posted in post #81?

EDIT: I guess you didn't since they were not quoted when you posted. Well, as you can see Kaveri does win in performance for cpu by quite a bit. Then, of course, there is a large iGPU specification difference between those 19 watt Kaveri and the 15 watt A6-6310 which doesn't look trivial either.

Numbers are indeed largely in favour of Kaveri, that said beware of benches without power comsumption numbers, i dont know this Passmark and if it s reliable, i ll check if there s some numbers for more precise comparison, Cinebench wich is used by Notebookcheck favour Beema too much, in integer tasks Kaveri is much better, this could explain the Passmark large difference between thoses APUs.

I benefit from the thread to point that notebookcheck did some reviews of Kaveri but they (the reviews) are terrible, they didnt even upodate the drivers and got an APU that was working at base frequency, 1.9GHz, in single thread mode, they posted an update but they didnt change the irrelevant scores.
 

cbn

Lifer
Mar 27, 2009
12,968
221
106
So does any have further comments about what I posted in #54?

So in Summary, here is what I would like to see happen:

1. Octocore AM3+ occupying lower price points. This will strengthen traditional desktop at the value level.

2. Kaveri and other big core APUs leaving traditional desktop and moving into mobile (Gaming laptops) or (to a lesser extent) very small desktops that use mobile derived chips.

3. Jaguar (Puma) based chips moving upscale into more and more entry level desktops that used to be occupied by the lower priced FM2/FM2+. This by making use of improving yields on 28nm to increase clocks and make use of quad core.

P.S. Regarding #3, As much as dislike the idea of SOCs in entry level desktops, that is pretty much the only option I see for AMD. (re: I just can't see Enterprise customers or the average user needing all the extra iGPU of Kaveri. But I am thinking if AMD increases clocks on the Jaguar quad cores the experience will be decent enough.)

Basically I would call the above 3 goals: S.T.O.P F.M. (as in stop FM2 and FM2+) :thumbsup:

System to openly promote FX motherboards

All joking aside on the acronym, I would seriously like to see a big reduction in FM series for the most part. I can't see much reason for it to exist. (Put those expensive to make Kaveris in Mobile application where they will be more appreciated for obvious reasons)
 

Abwx

Lifer
Apr 2, 2011
11,885
4,873
136
And it still doesnt change the fact that the 2500K ages multiple times better than the FX8150.
.

Lol, you call this a fact..?..

Then provide benches that say the contrary of those that i posted earlier, you are all talks without the slightest substance, i guess that relying to general hollow statement is all that is lefts when numbers are that indeniable.

In the waiting of your numbers i wont answer no more to non argumented posts.
 

ShintaiDK

Lifer
Apr 22, 2012
20,378
146
106
Lol, you call this a fact..?..

Then provide benches that say the contrary of those that i posted earlier, you are all talks without the slightest substance, i guess that relying to general hollow statement is all that is lefts when numbers are that indeniable.

In the waiting of your numbers i wont answer no more to non argumented posts.

Like what, some of these?

http--www.gamegpu.ru-images-stories-Test_GPU-RPG-dragon_age_inquisition-test-DragonAgeInquisition_proz_proz.jpg

http--www.gamegpu.ru-images-stories-Test_GPU-Action-Battlefield_4_China_Rising_-test-bf_4_proz.jpg

http--www.gamegpu.ru-images-stories-Test_GPU-Action-Far_Cry_4-nv-test-fc_proz.jpg

http--www.gamegpu.ru-images-stories-Test_GPU-Action-Metal_Gear_Solid_V_Ground_Zeroes_-test-mgs_proz.jpg

http--www.gamegpu.ru-images-stories-Test_GPU-Action-Watch_Dogs-test-proz_nvidia_ultra.jpg

http--www.gamegpu.ru-images-stories-Test_GPU-Action-Assassins_Creed_Unity-test-ac_proz.jpg
 

Maxima1

Diamond Member
Jan 15, 2013
3,549
761
146
The charts says that the 2500K was a better choice in october 2011 ( you can check the october 2011 reviews) but with time the 8150 largely outmatched it.
The IPC sucks. It also uses more watts than the Intel option. Because of Intel's advantage, they can offer far superior single-threaded performance and comparable multithreaded performance using less cores/threads.

I was sure that you would provide a game graph, because games are about the single usage where the 2500k hold the comparison, (as if games were the only application of a 8 core CPU..) but let s look elsewhere, in apps that need cores and let s compare thoses two CPUs, and a lot of others, throughputs :
In many apps, it has to be a lot better to even make much of a difference. Sort of like ISP's. People will take significantly less download speeds for a relatively small price reduction. On the contrary, for games, it can become unplayable or turn into a crappy experience. To be fair, this is more of an issue with mobile parts rather than the desktop.
 

Abwx

Lifer
Apr 2, 2011
11,885
4,873
136
Like what, some of these?

Of these games you should add...

What if i use 6 rendering benches in a row, would you consider it as a valuable point if i use six times the same kind of application to prove a point.?.

That s actualy what you are doing, to each different application i post you re answering with a game...

So we get your point, 6 games are equivalent to say :

Fritzchess
7 Zip or Winrar (but not both..)
3DS Max (or Povray but not both...)
GCC compiling
SHA
AES/Non AES encryption
X264 encoding

In all thoses apps the 2500K will be slaughtered, no wonder that you re patheticaly using the same bench ad nauseam, you know too well what the 2500K is worth compared to a 8150, heck, i posted enough graphs...
 

Ramses

Platinum Member
Apr 26, 2000
2,871
4
81
So does any have further comments about what I posted in #54?



Basically I would call the above 3 goals: S.T.O.P F.M. (as in stop FM2 and FM2+) :thumbsup:

System to openly promote FX motherboards

All joking aside on the acronym, I would seriously like to see a big reduction in FM series for the most part. I can't see much reason for it to exist. (Put those expensive to make Kaveris in Mobile application where they will be more appreciated for obvious reasons)

Watching FX prices and the marketing pushes form big retailers for them, and even some of them going in and out of stock, I suspect AMD is going to get every penny they can for them and that they are doing ok-enough right now. Then they will dry up. I also think the higher end examples will be a bit of a niche market chip, and they will always do well on ebay and such.
If they weren't selling or it was in there business plan to need to move more of them, they'd be cheaper, and they really aren't. I attribute that to good marketing and name recognition tagging along with the GPU business, same reason they are doing RAM and SSD's that aren't all that impressive performance wise that I can tell. I'm not sure what's up with the desktop FM stuff, spinoff of mobile that was cheap enough to put in a desktop to rake in a few bucks maybe? Got me. I suspect the mechanisations of big cpu business is over our collective heads, largely due to a lack of data to process.

I'd love to see the higher end FX chips get cheaper, and they will eventually until/unless they just pull the plug on them. I intend to keep on buying AMD when I can without creating a hardship for myself, because I believe the world is a better place with two companies supplying CPU's rather than one (intel), and AMD surely needs the money. If I could wish Cyrix back into business I would.

There were discussions not unlike this a year ago, and the FX line was old then. Here it is a year later, the ole FX is still hanging on, still performing good enough for everyone that isn't trying to game at 4k or something crazy.
I consider that a testament to it's quality, and in no small part to do with the responses in the "have you slowed down upgrading" thread since all this stuff is plenty fast for most folks.
I strongly suspect next year this time, there will still be FX CPU's circulating, hopefully a fair bit cheaper. As I tell my customers often though, why would I lower the price on something if it's already selling? Myself I can't think of anything I was doing last year, or am doing this year, or will be doing next year that my FX box does not blast through with ease. Long as that's the case they can keep popping them out I guess.

For all we know, AMD planned all of this from the start. They knew the single thread performance was bad. They knew how they compared to Intel parts. Maybe they knew upgrade cycles were slowing down as performance leveled off and this was a calculated chip? They had to have had an inkling how far Intel would progress in a year or three. Who knows.

I'd like to know when they broke even on the FX line. Or if they did.
Or how they can even calculate that with a CPU. The business aspect is moderately fascinating.
 

Abwx

Lifer
Apr 2, 2011
11,885
4,873
136
The IPC sucks. It also uses more watts than the Intel option. Because of Intel's advantage, they can offer far superior single-threaded performance and comparable multithreaded performance using less cores/threads.

In many apps, it has to be a lot better to even make much of a difference. Sort of like ISP's. People will take significantly less download speeds for a relatively small price reduction. On the contrary, for games, it can become unplayable or turn into a crappy experience. To be fair, this is more of an issue with mobile parts rather than the desktop.


Oh, the IPC argument again, but when Intel release a 10C or a 12C everyone is praising the exploit but no one seems to notice that the ST perf is worse than the one of about any FX, but hey, there s the core count, the same core count that make the 8150 explode the 2500K, you dont seems to have checked the graphs above, you can see that the 2500K has not the same MT perf than the 8150, not the same at all by as much as 50% in some instances..

Here the graph that validate AMD s technology, and it s aknowledged by Intel itself, from now i d like to hear the Intel supporters explicitely saying that the 10-12C from Intel sucks completely since they have lower ST perf than the FXs and that people should discard them.

68947.png




And about this review of course there s no FX in the Linux tests :

http://www.anandtech.com/show/8679/intel-haswellep-xeon-12-core-review-e5-2650l-v3-and-e5-2690-v3/2

Surely that they performs so miserably with this OS that Ian Cutress didnt want to bother AMD...
 

Abwx

Lifer
Apr 2, 2011
11,885
4,873
136
So does any have further comments about what I posted in #54?



Basically I would call the above 3 goals: S.T.O.P F.M. (as in stop FM2 and FM2+) :thumbsup:

System to openly promote FX motherboards

All joking aside on the acronym, I would seriously like to see a big reduction in FM series for the most part. I can't see much reason for it to exist. (Put those expensive to make Kaveris in Mobile application where they will be more appreciated for obvious reasons)


Ultimately we ll surely see AM3+ and FM2+ being fusioned, the exemple of Carrizo is telling, they will use a single infrastructure for two completely different APUs, with SoCs there will be probably no more FX like CPUs, at best we can expect a SoCeted 8 core K12 or Zen that will be a server part recycled for DT purposes.

Besides mobile doesnt mean higher ASPs, just look at Intel Btrails, they have a negative ASP while AMD s APUs are not that expensive, including Kaveri, if the market was sane the prices would be quite low, eventualy high ASPs are possible only because of Intel s quasi monopoly, without this monopoly it would be the same as in Android powered items, slim margins on low double digit prices.
 

jhu

Lifer
Oct 10, 1999
11,918
9
81
Oh, the IPC argument again, but when Intel release a 10C or a 12C everyone is praising the exploit but no one seems to notice that the ST perf is worse than the one of about any FX, but hey, there s the core count, the same core count that make the 8150 explode the 2500K, you dont seems to have checked the graphs above, you can see that the 2500K has not the same MT perf than the 8150, not the same at all by as much as 50% in some instances..
.

When Intel releases a 10 core or 12 core processor, they're called "Xeons" and go into datacenters and such. Performance/watt matters a lot in high density situations in a way that the Bulldozer/Piledriver Opterons can't match.
 

FX2000

Member
Jul 23, 2014
67
0
0
If you have $500 buy a console. $500 in a gaming PC is $400 once you inevitably put Windows in the equation. And an 8350 isn't an octa core. Its 8 modules, more like a quad. Same the 6300 is a tri-core. When Broadwell is out and then Skylake there is less and less reason to buy FX for a gaming box. Give it away for free and even then Intel's performance in the majority is still better.
How is it any less of a 8 core than say, an 5960X?
It's just 4 modules, with 2 cores in them. Sharing ALU's. Educate yourself and stop this garbage talk x). CMT>SMT