Should AMD release an updated 8 core FX (improved cores, smaller process)?

[superstition]

So actually there is no real reason to have an 8-core Excavator at 28nm and that is why AMD hasnt made one.

There are reasons to offer something more recent than 2012 tech in the market. The question is whether or not the company has the means.

 

[VirtualLarry]

probably just not enough people with AM3+ boards out there anymore.

While I still have an AM3+ board or two, I agree, Newegg's selection of AM3+ boards (aside from a few prominent high-end "enthusiast" models) are drying up, seemingly fast.

 

[tential]

Ask yourself how many they even sell and can sell.

AMD already answered with no, there isn't any ROI in it.

A new processor for the sake of a new processor with minor improvements in amds position would kill brand value.... The brand struggles already among the vast majority of consumers.. That'd be the last nail in an already nail ridden coffin for amd pr wise especially on gaming forums like neogaf where amd is still the "hot loud, omg never buy it only nvidia+Intel for life!"

 

[jpiniero]

AMD barely has enough money to release Zen as it is...

 

[ElFenix]

AMD would be better served in getting an EX core APU into AM4 (or whatever zen's platform will be called) with a proper die shrink down to 14/16FF along with whatever new graphics AMD has planned for that node. 2017 is a long ways off.

 

[erunion]

Is that the node or AMD's choice of transisitors (low-leakage vs. high)?

Or is this because only 32nm used SOI?
:

Its the node, but its not limited to GF. No fab offers a 4+ghz process, outside of Intel.

AMD doesnt have access to higher clocking process than 32nm, and won't anytime soon. This explains AMDs need for a new arch, Zen. Zen will grow performance through IPC increase, because clocks wont increase and might even(probably will) decrease at 14nm FF.

 

[Phynaz]

Its the node, but its not limited to GF. No fab offers a 4+ghz process, outside of Intel.

Technically GloFo does, as they now own the fab that manufactures POWER for IBM.

 

[dark zero]

Its the node, but its not limited to GF. No fab offers a 4+ghz process, outside of Intel.

AMD doesnt have access to higher clocking process than 32nm, and won't anytime soon. This explains AMDs need for a new arch, Zen. Zen will grow performance through IPC increase, because clocks wont increase and might even(probably will) decrease at 14nm FF.

IBM wants to talk one or two things to.you about that...

 

[erunion]

Technically GloFo does, as they now own the fab that manufactures POWER for IBM.

Right, IBM 22nm is an outlier. I dont recall what clocks its shipping at, though.

 

[dark zero]

Right, IBM 20nm is an outlier. I dont recall what clocks its shipping at, though.

IBM could deliver up to 3.5 Ghz without problems... More than that is an Fx at 125 watts.

 

[NTMBK]

AMD don't have money or time to waste. Focus on finishing Zen.

 

[SPBHM]

a minor update would probably have a minor or no impact on sales at this point, because it wouldn't solve the worst characteristics that are a keeping consumers away from the FX line,

as for 6 cores or more fm2+, I think AMD would not want to use resources for that because they want the same die to be used for laptops, and over 4 cores there is no market for laptops? they are better using the space for their IGPs I guess

but yes, it think it would be nice if suddenly they dropped something improved for Am3+ for example or Athlon X6 for FM2+, but I don't see it making sense at this point... they need more than a minor update, and they need a new platform with it, maybe Zen will deliver.

 

[Boze]

I think it'd probably be best for AMD to focus on Zen & AM4 motherboards at this point. It'd be nice to spend $100 - $200 to drop a faster processor in place of the FX-6300 that's powering one of my home servers, but AMD needs to focus on the big picture if they're going to mount any kind of defense against Intel.

 

[jpiniero]

I don't even think you will see socket Carrizo either. AMD's only real new product in 2016 might be Arctic Islands.

 

[videogames101]

Right, IBM 22nm is an outlier. I dont recall what clocks its shipping at, though.

IBM shipped 32nm systemZ chips with 5.5 GHz clocks mate, I don't have 22nm numbers but I'm betting its comparable.

EDIT: 5GHz on 22nm (apparently using pulse-clocked latches, which is neat - didn't AMD use these awhile back?)

Source: http://www.researchgate.net/publica..._System_z_Microprocessor_ISSCC_2015_slidedeck

Keep in mind anyone could be shipping 5GHz chips on modern processes if they were willing to take a large enough IPC hit.

 

[Zstream]

I'll take a 16/8 core 1.5ghz chip, but I'm probably one of the few. Looking to do real multitasking on a htpc server.

 

[Soulkeeper]

I had hopes that they'd release a non-gpu 28nm FX (4 modules) on the fm2 platform. I'd be running that now if it existed.

 

[alexruiz]

I would surely take one... or a few.

 

[superstition]

IBM shipped 32nm systemZ chips with 5.5 GHz clocks mate, I don't have 22nm numbers but I'm betting its comparable.

EDIT: 5GHz on 22nm (apparently using pulse-clocked latches, which is neat - didn't AMD use these awhile back?)

Source: http://www.researchgate.net/publica..._System_z_Microprocessor_ISSCC_2015_slidedeck

Keep in mind anyone could be shipping 5GHz chips on modern processes if they were willing to take a large enough IPC hit.

5.x GHz at 22nm sounds promising.

 

[Arachnotronic]

I don't even think you will see socket Carrizo either. AMD's only real new product in 2016 might be Arctic Islands.

There will be socketed Carrizo in 2016.

 

[h4rm0ny]

Why would anyone buy an updated FX when its performance is still horrible behind?

I would. I'm currently waiting to see how a few things play out before spending the large amount of money necessary to upgrade from where I am (an FX-8350). A compatible but upgraded 8 core using the technologies refined for Carrizo would make a moderately priced nice little bump in performance in the interim and let me draw out the expensive full upgrade cycle a little longer. Long enough to see how Zen turns out, in fact.

But then since I joined these forums I have noticed there cannot be an AMD thread without posts from yourself appearing to attack them in any way you can. I don't understand why you do that, but regardless, the answer to your question is that yes, there are those of us who would like it. And they're also still selling FX-8350's and their ilk, so it's not like the only buyers would be those upgrading from an existing and close FX- chip.

EDIT: But specifically to the OP's question, I don't know if they should release one. Costs for producing it have got to be a lot less than doing a real new chip but AMD funds are extremely tight, I imagine. Only they could really know if it would be cost-effective to do so, but I imagine it probably wouldn't be. Several of the things that make Carizzo great are really specific to lower-power laptops. Scale it up higher-end and its relative gain over previous generations gets smaller. They should probably focus everything on Zen at this point.

 

[sm625]

There is zero chance of them making enough money off it to cover the cost. So no, they absolutely should not and will not do something like this. Unless somebody comes to them with a large chunk of money and says we'll buy 50000 pieces if you can reduce the TDP by 25% while providing the same performance at whatever specific application they need. AND also assuming that AMD engineers were able to tell the sales guys that this could actually be done.

 

[Shehriazad]

Nah.


Gen 1 and 2 are too old to still get any decent gains and even more cores won't help.

Kaveri base would be better by a good margin and 4 CPU modules would be fine for sure...but they would have to put a lot of work into that...waste of money.

I mean sure...they'd have like +10% singlethreaded and like what was it +15% multithreaded performance on that...likely even more with a good L3 cache...but it's too much effort. Sure, it would probably have a +20% total improvement with a decent L3 cache on there (including better IPC) compared to 2nd gen FX...but it's still socket FM2+.


Lastly there is Carrizo...it was designed with HDL in mind and only exists as mobile chip (correct me if I'm wrong)...that one is just an utter waste of time for the high end sector.


Plus...WHY would you want to make better FX chips right now? Athlon 860K and FX 8XXX fulfill ALL the needs sufficiently right now.

Need a super cheap quad core that you can play with? 860K on FM2+ is marvellous for the price.
Need 8 threads but don't have a lot of money and possibly want to play the newest games with it? FX 8XXX will do just fine....since it'll still be able to deliver 60+ fps on the vast majority of new games. (At least if they're optimized enough to actually use more than 2 threads...which ARE the vast majority of new releases nowadays).

And if you need 120+ fps in all situations or want to play strategy games in 4K...then anything AMD could offer right now is totally not what you would want to buy, anyway...in that case your budget will have to be a lot higher, anyway. And I'm not paying 300$ for some updated FX chip. I'd rather wait for Zen or buy something Intel...because say what you want, but Intel just is too good in the $250-$300 space. Unless they suddenly managed to make a 4 module Kaveri based chip that has a baseclock of 4.5GHZ and a boost of 4.7Ghz with no more than 95w TDP.
That's the only chip I'd deem worthwile in that area right now...and it would be a novelty chip with very low availability on a platform almost no one bought.

 

[ShintaiDK]

There is zero chance of them making enough money off it to cover the cost. So no, they absolutely should not and will not do something like this. Unless somebody comes to them with a large chunk of money and says we'll buy 50000 pieces if you can reduce the TDP by 25% while providing the same performance at whatever specific application they need. AND also assuming that AMD engineers were able to tell the sales guys that this could actually be done.

They need to sell a million+. Not 50000

 

[LTC8K6]

There is zero chance of them making enough money off it to cover the cost. So no, they absolutely should not and will not do something like this. Unless somebody comes to them with a large chunk of money and says we'll buy 50000 pieces if you can reduce the TDP by 25% while providing the same performance at whatever specific application they need. AND also assuming that AMD engineers were able to tell the sales guys that this could actually be done.

If it could be done, it should be done already.

No need to wait for a request and a big order.