Future APU's and the end of FX

NaroonGTX

Gawd
Joined
Jun 3, 2013
Messages
710
Rg9fKas.png


This map shows that there is no FX planned for the future, not even in 2015. So it seems that the reports of FX-9000 being the end of the FX line were true. I guess this sucks for people who liked hexa- and octo-cores, but it's possible that we'll see hexacore APU's in the future on FM2+.

Interesting to note that the 65W max TDP for Carrizo is confirmed, as is Carrizo having Excavator-based x86 CPU cores and still being on FM2+, so FM2+ isn't dead after Kaveri as some have speculated.

Elsewhere I have seen mentions of a future APU codenamed "Basilisk", though no info is revealed on it yet. I can only speculate that it will succeed Carrizo, quite possibly being Excavator-based but offering DDR4 support. This would probably be the opening for the FM3 platform.
 
Where is this "end of FX" again ?

From what I see, the FX continues right through 2015..

Misleading title is misleading..
 
no, what is most interesting is that this graphic anticipates AMD sticking with the FM2+ and DDR3 to the end of 2015!
 
Where is this "end of FX" again ?

From what I see, the FX continues right through 2015..

Misleading title is misleading..

Yeah, FX continues... with nothing more than Vishera, no actual new processors will hit AM3+, so the platform itself has hit a dead-end. There's nothing misleading about it.
 
Yeah, FX continues... with nothing more than Vishera, no actual new processors will hit AM3+, so the platform itself has hit a dead-end. There's nothing misleading about it.

I kind of don't care, how long as AM3+ been out? compare that to the average intel chipset. It's not so bad.

I just want them to release an 8 core / 4 module excavator chip, they should have more space by then (hopefully, please someone let them build on something smaller than 28nm)
 
I kind of don't care, how long as AM3+ been out? compare that to the average intel chipset. It's not so bad.

I just want them to release an 8 core / 4 module excavator chip, they should have more space by then (hopefully, please someone let them build on something smaller than 28nm)

With all the slides released so far since the recent APU13, this is what I'm expecting:

  • Vishera FX (Piledriver) continues on AM3+ until end of 2014.
  • Kaveri APU (Steamroller) releases in 2014. -> FM2+
  • Carizzo APU (Excavator) releases in 2015. -> FM2+
  • Excavator-based APU refresh releases in 2016 -> FM3, DDR4
I'm going to assume that when high density libraries are first used with Excavator modules, we'll probably see a return to 4 modules (8 core) processors, but with L3 cache. If you don't recall, high density libraries was announced by AMD some time ago last year I think. You can read about it here on Anandtech.

My thinking, with that, AMD will be able to squeeze a lot more in a smaller space-- GPU units, CPU units, L3 cache-- all on 20nm process. Intel may be at 14nm at this point when this happens for AMD.

That's what I'm going to expect some time after 2015.
 
This discussion is a bit old ;) . If you checked prices on FM2+ mainboards you know where this is going.

AMD can't make money on AM3+ products in the way they do with their APU. What we see here is that AMD positioned in the budget region less then $90 for a mainboard with a cheap cpu that don't have to cost over $120-$140.

And yes there will be no true FX cpu line appearing on the FM2+.
 
Last edited:
3m/6c APU would be fine by me, hopefully they release such a beast.
 
I don't really care I'm not likely to be changing sockets for years. Quite happy with my current FX setup
In 4/5 years time I'll update to a DDR4 12 core something if it's an APU then fine.

I do make the odd build so I might look at the APU's for there.

The problem with the APU concept is if you are using a graphics card it's fairly pointless. Yes it might save some buyers on buying a GPU (that's great) but it's really not a solution for some people.

And those who don't need GPU performance much, and more CPU bang per buck..it's not ideal there either. So I have mixed feelings on this path myself
 
I don't really care I'm not likely to be changing sockets for years. Quite happy with my current FX setup
In 4/5 years time I'll update to a DDR4 12 core something if it's an APU then fine.

I do make the odd build so I might look at the APU's for there.

The problem with the APU concept is if you are using a graphics card it's fairly pointless. Yes it might save some buyers on buying a GPU (that's great) but it's really not a solution for some people.

And those who don't need GPU performance much, and more CPU bang per buck..it's not ideal there either. So I have mixed feelings on this path myself

You have to think of the GPU on the APU as another smaller processor. I like to stream games. If I had one of the current Intel processors, I could use QuickSync to transcode my video from 1080p to 720p in real time on the little GPU that is worthless normally. If the Intel chip's GPU can handle it as pathetic as they are, imagine what an AMD version could handle! Not knocking Intel but it's common knowledge that the GPU on their chips is severely lacking power.

In the future I could see Windows allocating background processes to it. Or maybe you set a program to run on the GPU. A high powered APU with an equally powered GPU on the die unlocks lots of possibilities.
 
AMD have already said they will release 8 core APUs in 2014, they said they will make an FM2+ version of the Jaguar APU.
 
AMD have already said they will release 8 core APUs in 2014, they said they will make an FM2+ version of the Jaguar APU.

That wouldn't make sense because the Jaguar cores are mobile CPU cores, and are an entirely different architecture from Steamroller modules. A 4 module (8 core) Jaguar-based APU on socket FM2+ would actually be a step backwards compared to Kaveri-based APU in terms of performance.

Given there won't be Steamroller-based FX processors in 2014, the only way forward is making a Steamroller-based 4 module (8 core) APU, but we probably won't see that in 2014. We might see a 4 module (8 core) APU based on Excavator modules in 2015 or later, however.
 
I wouldnt mind if the new Kaveri only ever sees a quad-core version if it (with overclocking) could compete with an i5 4670. If it can't, then I might have to go blue team on my next upgrade...
 
While I agree we've seen the last of the FX line, I wonder if they'll come out with a refresh. Prices on 8300's are dropping pretty quick like 8320's at $100 AMD 8350's at $170. I guess they could be letting the 9000's just take over and ride out AM3+ til '15 but I wonder if they'll shine up the Piledriver 8 cores and put out a couple new ones and put them out.
 
Rather disappointing if that roadmap is true.

Not really AMD is just limiting the amount of products it wants to compete on. They can't get the same power usage and ipc on the AM3+ platform. And an APU is not as bad as it sounds but the requirement is that you need to have HSA(HQ+mantle) for games to perform.

The drawback is (which is disappointing) fewer cores.
 
Also remember that AMD changed its roadmap nearly 10 times in 2013, nothing is certain.
 
Info on Kaveri won't be definite until CES 2014, which is a few days before the initial Kaveri launch itself on Jan. 14th. But the AMD of today is really secretive and doesn't like to talk about future products publicly anymore, except only in the most vaguest of terms.
 
Can I just say that if the title of this thread is true, then is unfortunate for some. However most will be content with the APU IF, and I emphasize the word IF as the only real way to sucess is to show the world how amazing APU. Mantle, AND HSA are in tackling real-world tasks and software that was designed around those three key ideas. If popular software devs utilize the potential that HSA, Mantle, and APUs offer then I think by 2015 the majority of users will not miss FX anymore.

However realistically, all these future of FX doomed talk are followed by the words, rumors and assumptions. No where in any of the roadmaps have we seen what they intend to release in 2016. They might kill off FX, they might keep it. Who knows
 
azn6dgi.png


No mention of GPU boost clocks, unfortunately.

In regards to the phrasing "end of FX", right now it is the end. We have absolutely no information about ANYTHING that AMD has planned past 2015. All we even know about 2015 itself is that Carrizo will release in that timeframe, and of course more GPU at some point. So for now, FX as a concept is dead.
 
So the moment when AMD manages to secure future games will have good multithreaded support due to putting 8 cores in consoles they stop updating 8 core cpu lines.

No wonder the company is joke.
 
Well, i think that one should expect some kind of "refresh" with some models for FX coming out, but other than that, it all points towards in deed the "death" of FX and AMD concentrating on APUs...

I was afraid it was going to happen...

http://hardforum.com/showpost.php?p=1040251669&postcount=136

On the bright side, i can now pat myself on the back and say "good job" for buying the FX-6300 instead of wasting time waiting for Steamroller...

Now, i will wait for FX to go EOL and see if i can grab an octacore maybe for peanuts (like when they were selling high clock Phenom II for 65 euros). I still have a hope that they may release a 95W octacore, but i won't hold my breath.

But, at any case, i will be staying on AM3+ platform for a long time. Plenty of power for non gamer. But it is sad to see AMD effectively "retreating" from the high-end battlefield. Intel has won there. I understand that it is more profitable for AMD though, since the APUs are beating the Intel graphics and you can make cheap platforms for average Joes...
 
So the moment when AMD manages to secure future games will have good multithreaded support due to putting 8 cores in consoles they stop updating 8 core cpu lines.

No wonder the company is joke.

You have a point there, but at the end, how much of the market share are the gamers that use PC for gaming? The APUs are more versatile (i could use one!), cheap and good-all around. Probably they will start adding cores to the APUs in the future and catch up on performance with the FX line...

Besides, for the time being even a hexacore is enough for gaming. By the time more cores are used, AMD probably thinks that HSA and Mantle will take over the need for new FX...

At the end, even amongst gamers, not all are there counting FPS all the time. Some just want to have the game "playable" and after a point, more CPU power is overkill.
 
So the moment when AMD manages to secure future games will have good multithreaded support due to putting 8 cores in consoles they stop updating 8 core cpu lines.

No wonder the company is joke.

Haha nice observation.
 
So the moment when AMD manages to secure future games will have good multithreaded support due to putting 8 cores in consoles they stop updating 8 core cpu lines.

No wonder the company is joke.

Maybe you didn't hear it but they tested BF4 and Mantle on a FX8350 8 cores 2.ghz (yes 2) and The game is still GPU bound.

But I'm guessing you don't care either way. Mantle solves the CPU bottleneck.

Well, i think that one should expect some kind of "refresh" with some models for FX coming out, but other than that, it all points towards in deed the "death" of FX and AMD concentrating on APUs...

I was afraid it was going to happen...

http://hardforum.com/showpost.php?p=1040251669&postcount=136

On the bright side, i can now pat myself on the back and say "good job" for buying the FX-6300 instead of wasting time waiting for Steamroller...

Now, i will wait for FX to go EOL and see if i can grab an octacore maybe for peanuts (like when they were selling high clock Phenom II for 65 euros). I still have a hope that they may release a 95W octacore, but i won't hold my breath.

But, at any case, i will be staying on AM3+ platform for a long time. Plenty of power for non gamer. But it is sad to see AMD effectively "retreating" from the high-end battlefield. Intel has won there. I understand that it is more profitable for AMD though, since the APUs are beating the Intel graphics and you can make cheap platforms for average Joes...

The sad part is that mainboards are still expensive so AM3+ upgrade for AM3 users is not happening then Kaveri and FM2+ should be very affordable and with Mantle and a decent graphics card (R9 290/280) still be able to play games.
 
This is why I asked in the other thread: Are an L3 cache still important to gaming nowadays?

If not and the performance difference is negligible, it may be about time to move on to Socket FMx (FM2+, FM3) for the foreseeable future in terms of AMD processors and gaming.

On the downside:
You lose 6 and 8 core models, and a large shared L3 cache.​
On the upside:
You gain a GPU for compute possibly via Mantle, and HSA/hUMA implementations (if the software takes advantage of them). You also gain native PCI-Express 3.0 and a newer, more modern chipset compared to the aging 700-series/800-series based 900-series and HyperTransport.​
I'm still going to hold out hope that with high density libraries, just maybe we'll see at least a 6 core (3 module) APU based on Excavator on maybe 20nm. (Intel will be at 14nm or 10nm at this point.)

What I'm interested in is the performance. Hopefully the single-threaded performance has met up with Sandy Bridge/Ivy Bridge from Intel, and Excavator hopefully will match or beat Haswell when that one is out. Multi-threaded performance has been very good for AMD and it should be a big improvement for Kaveri APUs over Richland.
 
Is that Fruehe quote from here? :D

The proof is in the pudding..I'd like to say I'm not foaming at the mouth, but I am.
 
I wouldn't be too miffed about losing the L3 cache. The problem with the L3 cache on Bulldozer and Piledriver is that the latency on it was too slow for it to be effective in consumer (non-server) workloads, which is why we saw A8's/A10 quads having similar/the same performance as FX-4xxx chips. AMD said they'd isolated the causes of the bad latencies, but said it wasn't a high priority, and that's most likely because no APU's have L3 cache and none of the future ones (at least Kaveri and Carrizo) will either.

As for ST perf, I think Kaveri will be on par with or slightly faster than Nehalem, and Carrizo will be at SB levels. We'll see in a few weeks, though.
 
I am willing the bet that we still see a steamroller or excavator based 2.0ghz 8 core APU.
 
I am willing the bet that we still see a steamroller or excavator based 2.0ghz 8 core APU.

Im not totally giving up on us not seeing anything new on the FX front. These roadmaps change all the time and this one is showing Piledriver carrying on til 2015! That just doesnt seem right to me at all. Its already over a year old and theyre gonna run it for another 2 years?! I just cant see that happening.

Im sure FX is a very low priority for AMD but with them just basically leaving that spot blank on their roadmaps, I kinda think they will bring something out in 2014 for the desktop market.
 
Im not totally giving up on us not seeing anything new on the FX front. These roadmaps change all the time and this one is showing Piledriver carrying on til 2015! That just doesnt seem right to me at all. Its already over a year old and theyre gonna run it for another 2 years?! I just cant see that happening.

Im sure FX is a very low priority for AMD but with them just basically leaving that spot blank on their roadmaps, I kinda think they will bring something out in 2014 for the desktop market.

Since all R&D is already sunken cost on Pilediver i can see it as something realistic as long as AMD can sell them for more than producing them costed.

And it's not like another 10% IPC from Intel makes big diffrence to them.
 
Can I just say that if the title of this thread is true, then is unfortunate for some. However most will be content with the APU IF, and I emphasize the word IF as the only real way to sucess is to show the world how amazing APU. Mantle, AND HSA are in tackling real-world tasks and software that was designed around those three key ideas. If popular software devs utilize the potential that HSA, Mantle, and APUs offer then I think by 2015 the majority of users will not miss FX anymore.

However realistically, all these future of FX doomed talk are followed by the words, rumors and assumptions. No where in any of the roadmaps have we seen what they intend to release in 2016. They might kill off FX, they might keep it. Who knows

IMHO the biggest loss is for home server type folks. The Bulldozer/Piledriver FX chips were/are fantastic home server chips, especially since they support ECC and IOMMU out of the box. When my server comes due for an upgrade, this might force me into buying actual server hardware which is going to be a lot more pricey.
 
Maybe you didn't hear it but they tested BF4 and Mantle on a FX8350 8 cores 2.ghz (yes 2) and The game is still GPU bound.

But I'm guessing you don't care either way. Mantle solves the CPU bottleneck.



The sad part is that mainboards are still expensive so AM3+ upgrade for AM3 users is not happening then Kaveri and FM2+ should be very affordable and with Mantle and a decent graphics card (R9 290/280) still be able to play games.



About mantle games being gpu bound on 8 core 8350s downclocked to 2GHz... there is still the fact that those games were able to utilize 8 cores for workloads.

Kaveri cuts the core count in HALF !!!!!!!!

So the question is are mantle games with four 4 GHz Kaveri cores still gpu bottlenecked vs an 8350 with all 8 cores working in a crippled state?

If so, then it won't matter as much that it's only 4 cores, but can we just assume that?
 
About mantle games being gpu bound on 8 core 8350s downclocked to 2GHz... there is still the fact that those games were able to utilize 8 cores for workloads.

Kaveri cuts the core count in HALF !!!!!!!!

So the question is are mantle games with four 4 GHz Kaveri cores still gpu bottlenecked vs an 8350 with all 8 cores working in a crippled state?

If so, then it won't matter as much that it's only 4 cores, but can we just assume that?

all depends on the architecture.. a good example at least for AMD processors was the jump from phenom I to phenom II.. the phenom II X2's were out performing even the quad core phenom I's clock for clock. so just because you halve the amount of cores doesn't mean you automatically lose the same amount of performance. it's just a wait and see game at this point.
 
What, no 5 GHz 200+ W part? I would have been certain to see such a successful processor venture in their plans...

Now to be serious, it's nice to see TDP's topping out at stated 65W with the Carrizo offering, if true.
 
What, no 5 GHz 200+ W part? I would have been certain to see such a successful processor venture in their plans...

Now to be serious, it's nice to see TDP's topping out at stated 65W with the Carrizo offering, if true.

And, 95W for even the highest end 7850K, a mere 5W less than the Richland, but it's an improvement nonetheless. A mobile Kaveri APU might even show more improvements in TDP.
 
And, 95W for even the highest end 7850K, a mere 5W less than the Richland, but it's an improvement nonetheless. A mobile Kaveri APU might even show more improvements in TDP.

Given the much improved IGP, it's actually kind of remarkable.
 
I guess when it comes down to it AMD's FX series isn't that profitable for them. APU is quite popular and I have caught myself looking at FM2+ stuff which usually ends with me slapping myself back into reality. If the APU stuff that's coming out now was out in June when I built my PC I could have saved some good money. I bought an FX 8350 and an HD 7790. That's $350 bucks right there. I am hoping Mantle is as good as the wizards who praise it's magic say.

All that said, this graphic could be completely false.
 
And, 95W for even the highest end 7850K, a mere 5W less than the Richland, but it's an improvement nonetheless. A mobile Kaveri APU might even show more improvements in TDP.

Given the much improved IGP, it's actually kind of remarkable.

I agree with you both. An improvement in TDP is a welcome sight, especially given the IGP capability to come.
 
Back
Top