End Of The Line For AMD FX Processors. (UPDATE)

[Andrew48]

but..but..my fx 8320...

[szewei]

@Dylan522p

Lol I mean an a10 apu with more cores = a12 I don't mean arm cores

Sent from my HTC Butterfly using Tapatalk

[DudeWazap]

I'm hoping that they are still secretly working on a new FX chip and a month before release they just tell everyone. It would be awesome and take everyone by surprise.

[7850OC]

I'm hoping that they are still secretly working on a new FX chip and a month before release they just tell everyone. It would be awesome and take everyone by surprise.

it be awesome that they bring to FM2 a new FX or a Phenom III x12 

[NutzInTheHead]

I gave up on AMD CPUs after my Phenom 2. My Athlon 64 back in 2004 was the best processor I ever owned. Loved it.

[Snickerzz]

What If they release phenom iii? O:

[TheProfosist]

Source? I know Skylake will support DDR4 but never heard anything about Broadwell doing so. I may be wrong though.

Once again, you're underestimating technology. I don't see anything wrong with someone owning a 3770K, 8350, GTX 680 or any other pieces of reasonably high end hardware from 2012 (Assuming we're in 2014). Honestly, not everyone has the money to upgrade every single cycle.

Im not seeimg why you would want to upgrade every single cycle? The only reason I upgraded from a 3820 to a 4930k was because I originally wanted a 6 core but didnt have the money and my board supported it. Heck the 2nd gen high end core 2 quads are just now really showing their age. Also for about a month a was stuck on a Phenom II x4 965 with a 460 and everything was fine and dandy. If I had to I could have continued ro use it and the two 1280x1024 monitors I had setup on it.

[TheProfosist]

I'm hoping that they are still secretly working on a new FX chip and a month before release they just tell everyone. It would be awesome and take everyone by surprise.

I dont think their investors would appreciate that...

[Humbug]

The scary part about this roadmap is not the lack of new fx CPUs.

Rather It's fact that the upcoming APUs even as far forward as 2015 are still projected to be lower performance parts than piledriver fx CPUs. The roadmap shows that AMD is going to continue to sell an architecture released in 2012 all the way into 2015, and not just sell it but it will continue to be their premium performance part. Just ridiculous...

[LAwLz]

My post isn't biased, well not according to the people that have been commenting on it.

Ahahaha you can't be serious. It is ridiculously biased. Just look at the "what about games" section. You cherry picked those two benchmarks like crazy. Sadly it seems like the post where I pointed that out in has been removed (am I being censored for telling the truth in your thread, or maybe I am getting it mixed up with some other thread where people cherry picked benchmarks?). There were like 20 gaming benchmarks on that site that showed Intel winning, and two that showed AMD winning, and you ONLY showed the two where AMD wins. You even admitted that you cherry picked them to "illustrate either a weakness or a strong point of the architecture" (funny thing that you pretty much only cherry picked stuff to illustrate the strong point of the AMD architecture). It's called cherry picking, or "fallacy of incomplete evidence". Not telling the whole truth is as bad as lying, and that's exactly what you are doing in that post.

 

 

And I do believe you are referring to the microOP cache that cuts the misprediction penalty by 3 cycles.

I think you are thinking of the same thing but that's not all it does. With the µOP cache the fetch unit can sometimes completely be powered off if it detects that the macro-ops it is going to fetch is already in the cache. That saves a ton of power and also cuts down the pipeline by several stages (which increases performance in integer calculations among other things).

 

 

AMD's & Intel's microOPs are quite different, they're not comparable in fact : http://www.realworldtech.com/bulldozer/5/ .

Thank you for agreeing with me. My point was that you are ignoring a ton of stuff in your other post and basically only goes "Intel are good at floatpoints and AMD are good at integers" which in some cases might be true but it is far more complex than that. Your link just further proves that there are more to CPUs that what your post makes it out to be.

 

 

The main reason Intel CPUs consume less power is because of the smaller die size as a result of the smaller manufacturing process that is a fact.

Certainly cutting the branch misprediction penalty does improve efficiency by improving performance but does not result in a lower power consumption on its own.

1) It's not just die size that makes Intel CPUs far less power hungry. You are completely ignoring a ton of things such as race-to-idle, leakage and the fact that AMD CPUs runs at a much higher frequency (and yes, higher frequency does mean higher power consumption). Die size is not the reason why Intel CPUs uses less power. Want proof? OK, let's compare the i7-8320 vs the FX-8350. Both have a die size of about 300mm² (294 and 315 respectively), both uses 32nm transistors and the amount of transistors are fairly similar as well (1.27B and 1.2B respectively). Neither one has a GPU that is powered down in these tests by the way, so there is no die are wasted.

So as you can see they are very, very similar in all of those regards.

i7-3820 = 184 watts

FX-8350 = 239 watts

A 30% difference even though they are very similar.

 

Anandtech's tests shows a similar story (please note that it's not the same version of x264 being used, and one is for the first pass and the other one is the second pass. Please note that the i5-2500K uses less power in the old version so it's pretty safe to say that if he tested the 3820 in the new test it would use even less power).

i7-3820 = 158.6 watts

FX-8350 = 195.2 watts

 

Die size is not an indicator of how much power something uses.

 

Yes, improving branch prediction does improve power consumption, as well as performance. Not having to flush the cache and start over means less wasted cycles, and that directly translates to less power used and higher performance.

 

 

You also need to realize that intel's AVX (& AVX2) instruction set is a flaoting point instruction set, and in fact the upcoming AVX-512 instruction set is going to be used in Intel's Knight's Landing architecture a sucessor to the current line of many core Xeon Phi, essentially a GPU.

So Intel is actually off-loading its floating point work to the GPU which is exactly what AMD has already done with Kaveri.

The APU and heterogeneous architectures are the future, AMD is leading it and Intel is saliently admitting it.

Not sure if that's just something you added to defend AMD, or if that was suppose to be a counter argument to my "you also have to take into consideration different instructions sets". Knight's Landing is NOT a GPU. Yes they will use AVX-512 for Knight's Landing. So what's your point? AVX-512 is not just about float points (it can also be used for integer operations) and it's not specific to GPUs (it is actually a CPU introduction set which will come to Xeon processors in the future). So I don't really get why you brought it up at all.

 

Again, I've already said that APUs might be the future, but they are not the present. In the perfect world we would be able to hardware accelerate almost all tasks, but we have a very long way to go on the software side. Even if we were already there, you could still get higher performance from having a really good dedicated CPU and a really good dedicated GPU, compared to having a gimped CPU and a gimped GPU on the same chip.

 

By the way, have you seen Iris and Iris Pro?? They perform better than the highest end AMD APU (even on the graphics side) so yeah, AMD is leading in terms of APU but basically only because Intel isn't really trying. When Intel gets serious they can outperform AMD even in iGPU performance.

[Mooshi]

I don't see why people are disappointed. Many of Intel's highest end chips are crappy "APUs" with the cpu part being glorious and the graphics part being complete shit. If AMD improves the CPU performance of their APUs in their future to be better than their best FX chips, things are going to be great. Stop being all doom and gloom. It's going to be a rocky road, but I refuse to believe AMD is just laying down, they probably have something in mind they are saving for later.

[Fetzie]

Many sources says the broadwell platform will have ddr4.It should be true because the problem is that broadwell is not compatible with Haswell nor Haswell refresh.Thus by needing to redesign things,they may as well put ddr4 in which is by then more affordable.

 

What I had heard, is that Haswell-E will have DDR4, and Skylake would have DDR4. Broadwell, being the Haswell refresh, would not support DDR4.

[fojacko]

Well considering they have significantly improved performance in single core applications. As long as they can keep the multi-core performance even slightly similar to the fx series (hey if they're limited to 4 cores, they might even improve it slightly). They're still in the game.

[nicholaswillems]

Guess Intel will be giving us even smaller increases in CPU now because they'll basically own the market now...

 

Even smaller? Then like 0,5% increases max?

[Raxxdude]

I think that all AMD needs to do is to make an APU as large or slightly larger than the X79 platform CPU's from intel. This will give them the space they need on the die to create an amazing CPU GPU combo.

[7850OC]

It might be the end for FX CPUs, but I highly doubt AMD is done with high performance desktop CPUs. Besides, if I remember correctly, they said they were done with that in like 2011 or something when Bulldozer was a disappointment, yet they continued and gave us Piledriver FX. 

 

AMD, I believe in you. Gief cheap 12 core CPU please. I like cores. :3

i also want a dual cpu motherboard support (for my 24 cores) 

[Giggitygoebbels]

What I had heard, is that Haswell-E will have DDR4, and Skylake would have DDR4. Broadwell, being the Haswell refresh, would not support DDR4.

Haswell refresh is not broadwell,and broadwell is incompatible with haswell.

[Humbug]

I don't see why people are disappointed. Many of Intel's highest end chips are crappy "APUs" with the cpu part being glorious and the graphics part being complete shit. If AMD improves the CPU performance of their APUs in their future to be better than their best FX chips, things are going to be great. Stop being all doom and gloom.

the problem is that the roadmap shows APUs continuing to be low performance parts. According to the roadmap even in 2015 the new APUs will be slower than current fx piledriver- which is very sad.

[Kobz360]

Im not seeimg why you would want to upgrade every single cycle? The only reason I upgraded from a 3820 to a 4930k was because I originally wanted a 6 core but didnt have the money and my board supported it. Heck the 2nd gen high end core 2 quads are just now really showing their age. Also for about a month a was stuck on a Phenom II x4 965 with a 460 and everything was fine and dandy. If I had to I could have continued ro use it and the two 1280x1024 monitors I had setup on it.

exactly!

[Avcn42]

AMD has not had a new CPU since 2012

Wrong. FX 9590 is from 2013

[Osmium]

Ha I knew it!

 

And now I will go cry myself to sleep.

[maximilian4o]

WHYYYY ???

 

[TheProfosist]

Wrong. FX 9590 is from 2013

Its not really "new"

[javelance]

This doesn't necessarily mean the end of eight core processors just the end of fx cpus.  It is still possible that AMD will make 8 core apus.

[TheProfosist]

This doesn't necessarily mean the end of eight core processors just the end of fx cpus.  It is still possible that AMD will make 8 core apus.

i just imagined 8 even weaker lower clocked cores is that bad?