speculation zen processor

AndreRio

[H]ard|Gawd
Joined
Nov 23, 2011
Messages
1,240
I want to know more about the zen cpu. how much will it cost care to speculate? and, performance? more like or better than a i7-6900k, or more like a i7-6800k/5820k?
 

flatty

Limp Gawd
Joined
Feb 2, 2016
Messages
178
Im curious too :)) experience saying AMD will not recover the gap, but who knows what pleasant surprise :)
 

Darakian

Supreme [H]ardness
Joined
Apr 12, 2004
Messages
4,698
It's worth considering some historical perspective. When AMD was able to bring the pain to intel with the athlon 64 intel had been investing into a failed and marketing driven architecture. Intel isn't on an expensive and futile journey right now. The ace that AMD might have is in HSA, but that seems a few years out at least.

My guess is that zen will be a conservative core after the bulldozer gamble. It will be competent and well priced, but not terribly exciting.
 

DrDoU

2[H]4U
Joined
Jun 4, 2007
Messages
2,703
It's worth considering some historical perspective. When AMD was able to bring the pain to intel with the athlon 64 intel had been investing into a failed and marketing driven architecture. Intel isn't on an expensive and futile journey right now. The ace that AMD might have is in HSA, but that seems a few years out at least.

My guess is that zen will be a conservative core after the bulldozer gamble. It will be competent and well priced, but not terribly exciting.


What do you mean by conservative core?
 

Darakian

Supreme [H]ardness
Joined
Apr 12, 2004
Messages
4,698
What do you mean by conservative core?
What I mean by conservative is "not taking undo risk". See #3 here conservative

Without knowing all the millions of decisions going in to making a new cpu architecture; I expect AMD will take the "safe" option at each or at least at most.
 

pxc

Extremely [H]
Joined
Oct 22, 2000
Messages
33,064
Adding SMT isn't a conservative design choice. ;) The complexity needed to add that to a core is very risky in design terms. SMT is very likely responsible for most of the "40% improvement per clock" that AMD often touts.

I think, as an alternative, you have to consider that Zen is the result of a very long design cycle. Whether that's good or bad is really the question. Certainly, there should have been 5 years that AMD put into a new design after the underwhelming Bulldozer completion.
 

Dan_D

Extremely [H]
Joined
Feb 9, 2002
Messages
59,578
It's worth considering some historical perspective. When AMD was able to bring the pain to intel with the athlon 64 intel had been investing into a failed and marketing driven architecture. Intel isn't on an expensive and futile journey right now. The ace that AMD might have is in HSA, but that seems a few years out at least.

My guess is that zen will be a conservative core after the bulldozer gamble. It will be competent and well priced, but not terribly exciting.

People need to understand historical events more clearly. The reason why AMD was able to bring a competitive product to market are due to several things lining up perfectly for AMD. AMD acquired NexGen Systems and got access to their very competitive NX686 architecture which was in development at the time. This was the basis for K6 and sold well for AMD. This was important after the very lackluster K5 processors less than stellar performance and sales. While K6 was no Pentium Pro killer, it was a good upgrade for socket 7 systems and offered good performance for less money. Also, Digital Equipment Corporation was acquired by Compaq, and essentially shut down. The developers of the DEC Alpha were hired by AMD to work on K7. Some of the advancements in that platform were incorporated into it as a result. Lastly, Intel bet on clock speed with Netburst and ultimately it wasn't a great gamble for them as they were unable to ramp clock speeds to the levels they originally intended. Intel believed they could achieve 5GHz clocks and couldn't do anything reliable past 3.73GHz. Even their 3.8GHz Prescott CPU was short lived.

AMD has only been competitive on a performance level with Intel 5 years out of the last 20+. Also, look at AMD's history further and you'll see one bad management decision after another. They've often sold off one or more profitable divisions in order to keep it's CPU division afloat. AMD's made several promises over the years concerning performance, generated a lot of hype and then backed off at the last minute. AMD's promise of Zen being 40% faster than Excavator may very well be true, but in the past such statements indicate the performance levels are best case scenario and only cover one or two benchmarks at best. IPC performance improvements stated for previous architectures essentially followed this format. Even if Zen is truly that much faster than Excavator, that only puts AMD on par with Haswell at best. Intel's got Broadwell, Skylake, Broadwell-E, and they'll have Kaby Lake by the time Zen launches. Depending on when Zen actually hits, it may even have to face off with Cannonlake. It will eventually during it's life cycle, but the timing is everything with regard to sales.

Unless Zen has a 20% or better clock speed advantage over Intel's more recent architectures, I don't think Zen will compete the way K7 and K8 did.
 

Mackintire

2[H]4U
Joined
Jun 28, 2004
Messages
2,932
Clock for clock going from Haswell to Broadwell isn't huge in terms of overall performance in games.

I've commented on this before, AMD has NOTHING that can trump what Intel has on the desktop.

ZEN integrated APUs on the laptops..... yeah AMD might be able to turn that market upside down.

My reasoning is: Intel's GPUs suck, native quad core laptops don 't really exist below the $1k price point.
 

Dan_D

Extremely [H]
Joined
Feb 9, 2002
Messages
59,578
Clock for clock going from Haswell to Broadwell isn't huge in terms of overall performance in games.

I've commented on this before, AMD has NOTHING that can trump what Intel has on the desktop.

ZEN integrated APUs on the laptops..... yeah AMD might be able to turn that market upside down.

My reasoning is: Intel's GPUs suck, native quad core laptops don 't really exist below the $1k price point.

There is a difference between a product being financially successful and dominating the competition. Intel dominated financially but didn't hold the performance crown during the K7/K8 days. This is because the two are unrelated.
 

JustReason

razor1 is my Lover
Joined
Oct 31, 2015
Messages
2,483
Adding SMT isn't a conservative design choice. ;) The complexity needed to add that to a core is very risky in design terms. SMT is very likely responsible for most of the "40% improvement per clock" that AMD often touts.

I think, as an alternative, you have to consider that Zen is the result of a very long design cycle. Whether that's good or bad is really the question. Certainly, there should have been 5 years that AMD put into a new design after the underwhelming Bulldozer completion.
I think they thought that with bulldozer some software changes were coming, namely a DX12 like API. Hence why they released processors with higher core counts. Look at some of these DX12 benches where the AMD CPUs are performing quite well, not better than Intel by any stretch of the imagination, but a far better showing than in DX9-11. Oh should have mentioned that a DX12 like API WAS in the works when bulldozer and GCN were being built/dreamed/engineered (saw posts in 2009 when it was killed, the new API).
 

pxc

Extremely [H]
Joined
Oct 22, 2000
Messages
33,064
I don't think gaming was the main reason for AMD releasing 8 core processors. That was simply to stay in the game vs quad core Intel processors, for selling outside of the CPU bargain basement ghetto. DX12 has little or nothing to do with that choice IMO. AMD could have certainly been convinced that higher core count processors would be better utilized in the future, and that wasn't a bad bet. However, it became a necessity to compete with 2x more cores than the competition used, and had little advantage given the rapid shift to mobile where AMD's module-based processor power consumption was uncompetitive. The limitation of that, and the primary reason BD was a disappointment, was that AMD failed to fix the prior problems that held back K10: poor IPC and poor single-threaded performance.
 

ebduncan

[H]ard|Gawd
Joined
Feb 1, 2008
Messages
2,008
Amd Zen probably won't be the fastest chip on the block. However i'd expect Amd to close the gap quite a bit, considering this is the closest they have been to Intel on process in a while. Zen's performance will rely solely on the process node its on, and what kind of clock speeds they can ultimately achieve on the 14nm finfet process. Good news is the 14nm globalfoundries/samsung node is known to have a 0.8 volt supply voltage, making it a higher performance node in general compared to intel's at 0.75 supply voltage. The bad news Intel's process is generally regarded as being superior, as it offers better transistor density. Lucky for AMD that intel has a rather large gap in the desktop market between the skylake 6700k and now the Broadwell-E I7-6800k. So I'd expect AMD to hit between those markets.
 

tangoseal

[H]F Junkie
Joined
Dec 18, 2010
Messages
9,330
I personally think that AMD is going for broke. They need something that will reinvigorate their brand. Something that will bring choice back the market and they want the choice to be AMD for a while.

This processor has not been nearly as hyped as past so called FX (current versions) chips have been. I think the new processor will be delivered with a "Walk the Walk, not talk the talk" approach. If not AMD is going to get punished badly by the market.

I am realistically bored of having an Intel 3930K as it is not much slower than the latest 6 core that was just released. Such a shame. The AMD in my box would be a refreshing feeling if it can deliver. And I think I am far from the only to think and feel this way at this point in time.

And honestly I miss the feeling I had when I the AMD Phenom II 965 Black Edition. It was fast and really felt different even though functionally it is the same as any processor as far as end result.
 

TheBlueChanell

Supreme [H]ardness
Joined
Apr 15, 2005
Messages
4,660
People need to understand historical events more clearly. The reason why AMD was able to bring a competitive product to market are due to several things lining up perfectly for AMD. AMD acquired NexGen Systems and got access to their very competitive NX686 architecture which was in development at the time. This was the basis for K6 and sold well for AMD. This was important after the very lackluster K5 processors less than stellar performance and sales. While K6 was no Pentium Pro killer, it was a good upgrade for socket 7 systems and offered good performance for less money. Also, Digital Equipment Corporation was acquired by Compaq, and essentially shut down. The developers of the DEC Alpha were hired by AMD to work on K7. Some of the advancements in that platform were incorporated into it as a result. Lastly, Intel bet on clock speed with Netburst and ultimately it wasn't a great gamble for them as they were unable to ramp clock speeds to the levels they originally intended. Intel believed they could achieve 5GHz clocks and couldn't do anything reliable past 3.73GHz. Even their 3.8GHz Prescott CPU was short lived.

AMD has only been competitive on a performance level with Intel 5 years out of the last 20+. Also, look at AMD's history further and you'll see one bad management decision after another. They've often sold off one or more profitable divisions in order to keep it's CPU division afloat. AMD's made several promises over the years concerning performance, generated a lot of hype and then backed off at the last minute. AMD's promise of Zen being 40% faster than Excavator may very well be true, but in the past such statements indicate the performance levels are best case scenario and only cover one or two benchmarks at best. IPC performance improvements stated for previous architectures essentially followed this format. Even if Zen is truly that much faster than Excavator, that only puts AMD on par with Haswell at best. Intel's got Broadwell, Skylake, Broadwell-E, and they'll have Kaby Lake by the time Zen launches. Depending on when Zen actually hits, it may even have to face off with Cannonlake. It will eventually during it's life cycle, but the timing is everything with regard to sales.

Unless Zen has a 20% or better clock speed advantage over Intel's more recent architectures, I don't think Zen will compete the way K7 and K8 did.

AMD does have a couple of good things going for them this time around.

A) If Zen does indeed launch late Q3/early Q4 of this year it will still be competing with Skylake / Kabylake and it'll be the first time since the 65nm node / pre-glofo that there will be a process parity between the two. AMD has basically been a process node behind since the original Phenom's.

B) To add to that the recent GloFo/Samsung team-up means that for the foreseeable AMD will have access to quality fabs that have the cash to keep pushing forward with both node shrinks and R&D into Silicon replacements / supplements.

C) Jim Keller was brought on board and given full reign/control in designing a competitive high-end x86 CPU as well as steering AMD's CPU boat in the right direction post-Zen before he left for Tesla. He has a pretty damn good track record in both the x86 and the ARM departments and has been sought after or worked for some of the largest tech companies around

D) It's basically make or break for them at this point, if Zen is Stars/Bulldozer all over again they are pretty much done for and they know it. I believe them when they say that early ES / Lab results have met or exceeded expectations. If this was not the case and Zen was a pos I believe we'd of seen some moves preparing for spin-offs, re-structuring, executive shake-ups etc.

E) A majority of the current benches/info we have on the Excavator core used in the current / 7th Gen APU's has been fairly positive from both the IPC increase perspective and the platform's overall efficiency and performance relative to both 6th Gen APUs and Skylake CPU's in the same TDP envelope. This bodes well for Zen taking into account that the IPC increase metrics AMD are claiming are using this core as their baseline.

F) Zen doesn't need to compete on a 1:1 or even exceed in IPC compared to Intel. They need to be close enough and do so while using similar or less power. If they can come within 85-95% of Intel's current high-end and bring a new platform in AM4 that competes in features with x99/z170 for less money they will be sitting pretty, imo.

It should be really interesting to see what actually happens once Summit Ridge / Zen launches and I'm hoping AMD will give me a good reason to upgrade. I am looking to move my workstation to a 6-8 core i7 setup but will wait and see what kind of performance profile Zen offers. If anything it could at least nudge Intel into lowering the prices of their HEDT CPU's and Motherboards.

As I stated above, there are a lot of factors working in AMD's favor this time around and I'm really hoping they emerge post-Zen as a much leaner/meaner and competitive company. They are known to fuck things up and their track record especially on the CPU side isn't always the greatest but only time will tell.
 
Last edited:

Pieter3dnow

Supreme [H]ardness
Joined
Jul 29, 2009
Messages
6,784
Amd Zen probably won't be the fastest chip on the block. However i'd expect Amd to close the gap quite a bit, considering this is the closest they have been to Intel on process in a while. Zen's performance will rely solely on the process node its on, and what kind of clock speeds they can ultimately achieve on the 14nm finfet process. Good news is the 14nm globalfoundries/samsung node is known to have a 0.8 volt supply voltage, making it a higher performance node in general compared to intel's at 0.75 supply voltage. The bad news Intel's process is generally regarded as being superior, as it offers better transistor density. Lucky for AMD that intel has a rather large gap in the desktop market between the skylake 6700k and now the Broadwell-E I7-6800k. So I'd expect AMD to hit between those markets.

It is not even about the gap it is about the follow through as long as AMD can build on Zen as a very solid foundation then the rest might pick up as we move along. Does AMD need the performance crown from Intel , not really AMD needs to be able to sell Zen and chip away some money so they can funnel it back into R&D and when the time comes have another improvement ready.

If you want to dissect this check how Intel fell asleep while ARM ran away with the mobile products. If anything the major changes have been around power saving. Which would bode well for AMD future endeavours since that is the road they took with Jim Keller ...
 
Last edited:

Pieter3dnow

Supreme [H]ardness
Joined
Jul 29, 2009
Messages
6,784
Not to sure if Charlie "knows" more then we do but he felt that he should share this :
What specs were revealed about Zen? 40% more IPC, presumably over Caririzo, 8 cores with 16 threads, and all for the new AM4 platform. The performance number should put Zen on par with Haswell, not bleeding edge but more than in the game, and the core count is borderline pointless for consumers but chopped dies will probably be really solid. That said it is obviously a ground-up new architecture and the first threaded core AMD is doing. Expect sea changes when the numbers come out. AMD also said the very different Zen server part will sample to a few OEMs in Q3 with mass samples in Q4, that should give you a nice window for launch.
 
Joined
Dec 19, 2010
Messages
545
If this is a speculation thread, let's say AMD's Zen shits on everything, would the immediate profit drive them to be able to rollout their architectural updates / lithography changes faster? Or are they hurting for cash and this is a "last chance" so to speak to grab a mainstream or HEDT market?

Also, lets look at another possibility, if AMD's Zen sucks major donkey dick and flops for some reason, would they really be facing an absorption from some other large tech company? Also, who do you guys think would acquire AMD? I really wouldnt want Intel to be the only major player.

If this sounds retarded I apologize.
 

drescherjm

[H]F Junkie
Joined
Nov 19, 2008
Messages
14,927
If zen the 4 core / 8 threaded Zen does not perform as well as an ivy bridge i5 (both at stock frequencies) I expect AMD to go through bankruptcy proceedings in the next 2 or so years. The debt repayments are too large for another failure.
 

Pieter3dnow

Supreme [H]ardness
Joined
Jul 29, 2009
Messages
6,784
If this is a speculation thread, let's say AMD's Zen shits on everything, would the immediate profit drive them to be able to rollout their architectural updates / lithography changes faster? Or are they hurting for cash and this is a "last chance" so to speak to grab a mainstream or HEDT market?
Also, lets look at another possibility, if AMD's Zen sucks major donkey dick and flops for some reason, would they really be facing an absorption from some other large tech company? Also, who do you guys think would acquire AMD? I really wouldnt want Intel to be the only major player.
If this sounds retarded I apologize.

You are not the only person stuck in this mind set. AMD x86 license is not transferable so if AMD gets sold the x86 part is as good as gone, there is a change Intel will negotiate but you never know how this will play out.

If Zen succeeds then AMD would have little to no need to rush changes. The rushing of changes would only shorten the life spawn of Zen. If they applied all of the changes before a (example) 5 year period then during the latter part of those last changes they would need funds available to R&D Zen successor to coincide.
If zen the 4 core / 8 threaded Zen does not perform as well as an ivy bridge i5 (both at stock frequencies) I expect AMD to go through bankruptcy proceedings in the next 2 or so years. The debt repayments are too large for another failure.

You meant something else then what you wrote down because as it known now there will be no 4 core 8 threads part for a good while ;) .
 

drescherjm

[H]F Junkie
Joined
Nov 19, 2008
Messages
14,927
You meant something else then what you wrote down because as it known now there will be no 4 core 8 threads part for a good while

No I meant the 4 core / 8 threaded Zen CPU on the AM4 platform in 2017. Although the availability of the 8 core / 16 threaded CPU may also be in 2017 as well with the rest of Zen.
 

Mackintire

2[H]4U
Joined
Jun 28, 2004
Messages
2,932
Keep in mind that Fabrication Space must be purchased BEFORE you have the demand. Even if AMD sells ZEN like hotcakes loaded with crack. They will not be able to increase sales or market share faster than X. Demand may be high, but their ability to product is directly related to how much risk they want to open themselves up to.
 

JustReason

razor1 is my Lover
Joined
Oct 31, 2015
Messages
2,483
If AMD sells then Intel MUST negotiate the X86 "in good faith", a part of the agreement mandated by the court. Now if AMD flat out dies with no sale (not sure if that is even possible) then not sure how the courts mandated that.
 

Trimlock

[H]F Junkie
Joined
Sep 23, 2005
Messages
15,228
AMD came out with their current architecture hoping it would compete with Intel on the big data server market. Having lots of of cores with shared resources was a good idea but Intel trumped them too easily with their big dies and extreme power efficiency (important to server farms).

AMD still has a great server chip but they are out classes in both energy use and performance, the price difference is easily made up in cost to run for these businesses so it was a hard sell for AMD.

They did manage decent CPUs for laptops but they couldn't get them in packages with the OEMs that made them attractive. They got shoehorned into shitty laptops that no one wanted.
 

TheBlueChanell

Supreme [H]ardness
Joined
Apr 15, 2005
Messages
4,660
If this is a speculation thread, let's say AMD's Zen shits on everything, would the immediate profit drive them to be able to rollout their architectural updates / lithography changes faster? Or are they hurting for cash and this is a "last chance" so to speak to grab a mainstream or HEDT market?

Also, lets look at another possibility, if AMD's Zen sucks major donkey dick and flops for some reason, would they really be facing an absorption from some other large tech company? Also, who do you guys think would acquire AMD? I really wouldnt want Intel to be the only major player.

If this sounds retarded I apologize.

I'm not sure if anything has changed but AMD's licensing agreement with Intel for x86 stipulates that if AMD is acquired / changes hands the license becomes moot and the new ownership are not legally allowed to manufacture x86 CPU's. This could of changed in the last 5-6 years and it's also possible that some legal negotiations/shenanigans could be pulled to nullify this either due to government intervention as it would basically leave Intel as the sole player in the x86 space or some wrangling of the patents/IP AMD currently holds and cross-licenses to Intel and other industry players.

I'd assume as an absolute worse case the CPU / EESC units would be shuttered and all of it's highly valuable IP/patents would be sold off or they would be purchased specifically for said patents. It's GPU/RTG arm could be spun-off as a separate and newly independent entity/player or it could be gobbled up by someone like Samsung, Apple, Intel or even someone mildly obscure like Microsoft.

Let's hope none of the above happens though, that Zen ends up competitive and AMD's ends 2016 / kicks of 2017 with a bang and as a much more focused, competent and aggressive player/competitor. They really do have a unique product portfolio and the ability to create designs/chips that no-one else in the industry can. If they can muster up a competitive CPU core to complement their excellent GPU designs we should start to see some crazy APU's and applications of their tech mid-2017 and out.
 

Lunas

[H]F Junkie
Joined
Jul 22, 2001
Messages
9,990
i would say zen will be somewhere around 6th gen intel perf but not 7th. i dont see amd smt doing anything but closing the gap.


I'm not sure if anything has changed but AMD's licensing agreement with Intel for x86 stipulates that if AMD is acquired / changes hands the license becomes moot and the new ownership are not legally allowed to manufacture x86 CPU's. This could of changed in the last 5-6 years and it's also possible that some legal negotiations/shenanigans could be pulled to nullify this either due to government intervention as it would basically leave Intel as the sole player in the x86 space or some wrangling of the patents/IP AMD currently holds and cross-licenses to Intel and other industry players.

.

the license stipulates a new agreement will be made. so basically if amd gets bought by say samsung intel and samsung have to go to the negotiation table and hammer out fair compensation for renewal.

AMD could enter into partnerships though.
 
Last edited:

TheBlueChanell

Supreme [H]ardness
Joined
Apr 15, 2005
Messages
4,660
AMD came out with their current architecture hoping it would compete with Intel on the big data server market. Having lots of of cores with shared resources was a good idea but Intel trumped them too easily with their big dies and extreme power efficiency (important to server farms).

AMD still has a great server chip but they are out classes in both energy use and performance, the price difference is easily made up in cost to run for these businesses so it was a hard sell for AMD.

They did manage decent CPUs for laptops but they couldn't get them in packages with the OEMs that made them attractive. They got shoehorned into shitty laptops that no one wanted.

I would of agreed with this 3-5 years ago maybe but now for 90% of projects/server deployments it simply doesn't make sense to use something like an Opty 6300 setup over any of the Xeon E3/E5 v1-2+. The CPU's themselves generate loads more heat, compete dismally from a PPW perspective and both the C32/G34 chipsets themselves guzzle up power due to legacy design (NB/SB), outdated process (65nm-45mm don't remember off-hand) and lack any modern connectivity (and even EFI/GPT in some cases).

If your needs called for lots of cores in a smaller footprint then they made sense, could offer a great bang for your buck and also made excellent hypervisor hosts but even those advantages have been kind of reneged by Intel the last few years. It's hard to even find Opteron based models from both Dell and HP, the last time I deployed one into production was probably late 2011-ish. If Zen can compete closely with Intel in IPC and PP/w they should be able to put a ding in Intel's marketshare with their new Opty's as they will most likely end up much cheaper and offer more cores/$. If they come in within 85-90% of Intel's performance they will fly off the shelves, imo.
 

TheBlueChanell

Supreme [H]ardness
Joined
Apr 15, 2005
Messages
4,660
i would say zen will be somewhere around 6th gen intel perf but not 7th. i dont see amd smt doing anything but closing the gap.




the license stipulates a new agreement will be made. so basically if amd gets bought by say samsung intel and samsung have to go to the negotiation table and hammer out fair compensation for renewal.

AMD could enter into partnerships though.

That is good news should our worst case scenario end up being how everything pans out. Part of me wonders though if any potential buyer would have any interest in resuming/restarting x86 production / R&D. If Zen is a pile of shit that would essentially mean said buyer would most likely have to / want to start the next design again over from scratch, I'm not entirely certain it'd be worth it for a multitude of reasons. The landscape of the market / industry is also completely different than it was just 5-7 years ago and it's entirely possible that Intel could remain the sole provider of x86 CPU's without having to worry about any kind of anti-trust/monopoly headaches. ARM designs are starting to compete/potentially overtake Intel in the low power/mobile/lower-end space and sooner rather than later we will start seeing them more and more in the datacenter. That would pretty much leave Intel with just the HPC/SMB/App segments all to themselves and ARM making inroads everywhere else.
 

Pieter3dnow

Supreme [H]ardness
Joined
Jul 29, 2009
Messages
6,784
No I meant the 4 core / 8 threaded Zen CPU on the AM4 platform in 2017. Although the availability of the 8 core / 16 threaded CPU may also be in 2017 as well with the rest of Zen.
Well good luck waiting for those :) Zen APU is prolly first part that will use those 4 core 8 threads:) .
Keep in mind that Fabrication Space must be purchased BEFORE you have the demand. Even if AMD sells ZEN like hotcakes loaded with crack. They will not be able to increase sales or market share faster than X. Demand may be high, but their ability to product is directly related to how much risk they want to open themselves up to.
You could also link that to the amount of motherboards produced ;) What is risk ? They prolly know at which point there is market saturation ...
 

Trimlock

[H]F Junkie
Joined
Sep 23, 2005
Messages
15,228
I would of agreed with this 3-5 years ago maybe but now for 90% of projects/server deployments it simply doesn't make sense to use something like an Opty 6300 setup over any of the Xeon E3/E5 v1-2+. The CPU's themselves generate loads more heat, compete dismally from a PPW perspective and both the C32/G34 chipsets themselves guzzle up power due to legacy design (NB/SB), outdated process (65nm-45mm don't remember off-hand) and lack any modern connectivity (and even EFI/GPT in some cases).

If your needs called for lots of cores in a smaller footprint then they made sense, could offer a great bang for your buck and also made excellent hypervisor hosts but even those advantages have been kind of reneged by Intel the last few years. It's hard to even find Opteron based models from both Dell and HP, the last time I deployed one into production was probably late 2011-ish. If Zen can compete closely with Intel in IPC and PP/w they should be able to put a ding in Intel's marketshare with their new Opty's as they will most likely end up much cheaper and offer more cores/$. If they come in within 85-90% of Intel's performance they will fly off the shelves, imo.
Yea these days it would make no sense at all to build an AMD based server, OEMs know that and even AMD has given up.
 

drescherjm

[H]F Junkie
Joined
Nov 19, 2008
Messages
14,927
Yea these days it would make no sense at all to build an AMD based server, OEMs know that and even AMD has given up.

I fully expect Zen to change that. The 14/16 nm processes with higher IPC and lower clocks will certainly be more efficient. I wish there was more information about the server/high end workstation platform. I am interested in that one instead of AM4 to upgrade my 6 core / 12 threaded windows workstation because I can use the extra cores (my primary usage of the box is medical imaging research).
 

cageymaru

Fully [H]
Joined
Apr 10, 2003
Messages
20,785

Lunas

[H]F Junkie
Joined
Jul 22, 2001
Messages
9,990
Intel and AMD have delayed their new chips to January 2017 due to too much inventory of undesirable stock. Guess neither company can dump their existing stock of flavorless CPUs onto consumers as we know that new stuff is coming. This makes no sense to me but whatever. Just relaying the news. Don't shoot the messenger.
Intel, AMD new CPU platforms target January 2017 launch

Where I saw it at first.
Report claims Intel, AMD will both delay next-gen processors to early 2017 | ExtremeTech
It makes no sense for AMD to delay the zen cpu they are releasing is supposed to sit as the replacement for the 8350 which at this point is 2 years old.

The only thing they have in recent bin is low to mid range apu and cpu and the new zen chips to be released first are not for that end of the market...
 
Last edited:

cageymaru

Fully [H]
Joined
Apr 10, 2003
Messages
20,785
yes because delays will help clear the stock of old shitty processors that have been on shelves for years... AMD aside from apu and crippled apu has nothing that has not been on market for over 3 years. It makes no sense for AMD to delay when the one they have ready to go is supposed to be the desktop mainstream high end where the only thing new in the old stock is low-mid end apu and cpu that have been out for 4 years...

Once again don't shoot the messenger but....
Amazon Best Sellers: Best Computer CPU Processors

Motherboards.
Amazon Best Sellers: Best Computer Motherboards
 
Last edited:

Trimlock

[H]F Junkie
Joined
Sep 23, 2005
Messages
15,228
ExtremeTech has been known for reporting on terrible rumors before, I hope this is false I'm looking forward to Zen.
 

ebduncan

[H]ard|Gawd
Joined
Feb 1, 2008
Messages
2,008
Intel and AMD have delayed their new chips to January 2017 due to too much inventory of undesirable stock. Guess neither company can dump their existing stock of flavorless CPUs onto consumers as we know that new stuff is coming. This makes no sense to me but whatever. Just relaying the news. Don't shoot the messenger.
Intel, AMD new CPU platforms target January 2017 launch

Where I saw it at first.
Report claims Intel, AMD will both delay next-gen processors to early 2017 | ExtremeTech

AMD have slides that state q3/q4 2016 release. They also keep saying they are on track, every chance they get, so I would expect Zen before 2017 at least reviews of it.
 

Pieter3dnow

Supreme [H]ardness
Joined
Jul 29, 2009
Messages
6,784
I'm hoping they are on track also. I really want something new.

Looking at your signature yeah that would change a lot with Zen. Supposedly we should be really lucky if this cpu would even reach 4 ghz. Let's hope that the AM4 platform is up to scratch and allows us some higher rated DDR4 memory to be used. In a way were lucky with the timing on the whole thing (cpu ram motherboards).

I'm with ebduncan can't see the delay coming for Zen unless motherboards are the issue ...

SemiAccurate Forums - View Single Post - Official AMD Zen uarchitecture thread

for what it is worth ;)
 
Last edited:

Dan_D

Extremely [H]
Joined
Feb 9, 2002
Messages
59,578
F) Zen doesn't need to compete on a 1:1 or even exceed in IPC compared to Intel. They need to be close enough and do so while using similar or less power. If they can come within 85-95% of Intel's current high-end and bring a new platform in AM4 that competes in features with x99/z170 for less money they will be sitting pretty, imo.
Agreed. I've said something similar in other Zen threads. The platform's features are often overlooked, but if AMD can offer the same features or better as X99 at Z170 prices it would be a home run for them provided the CPU comes close enough performance / price wise with a reasonable power envelope. The main difference in my opinion is that I think the power envelope doesn't matter too much so long as the performance is 85-95% of what Intel's offerings are and it's priced accordingly. For the desktop, the power envelope is far less important aside from it's relationship to heat and overclocking.

It should be really interesting to see what actually happens once Summit Ridge / Zen launches and I'm hoping AMD will give me a good reason to upgrade. I am looking to move my workstation to a 6-8 core i7 setup but will wait and see what kind of performance profile Zen offers. If anything it could at least nudge Intel into lowering the prices of their HEDT CPU's and Motherboards..
Price adjustments are very likely assuming Zen is good and at least competes well on a price / performance ratio. Price changes on motherboards aren't going to happen. I can't get into precise details as numbers change, and motherboard manufacturers aren't entirely forthcoming with the information concerning build costs. However, I do know that many of the Z170 and X99 motherboards are produced with thinner profit margins that you might imagine. The Intel chipset itself is a significant portion of a motherboards overall cost. A PLX chip for example is roughly $75-$85 alone. An Alpine Ridge controller or ASMedia 1142 USB 3.1 controller adds additional cost. For Alpine Ridge adding in Thunderbolt 3 certification drives the cost up even further. The Z170X Gaming G1 from GIGABYTE for example isn't something that could ever be sold for sub-$300 price points for example. It costs $400+ because it has to.

While I don't know what Zen's chipsets will be like, I can say that the 990FX wasn't half what the equivalent Intel chipset at the time was priced at. You couple that with less restrictive VRD specifications and fewer guidelines from AMD and more corners can be cut on an AMD processor compatible motherboard than could ever be cut on the Intel side. This is pretty much how it's always been and it's one of the many reasons why AMD based motherboards can offer the same feature set at half the price. Many people think it's purely a fact of price gouging and it simply isn't true. At least, not past the chipset itself. Intel controls that and only they know how much those truly cost to produce. The price of the chipset for the motherboard manufacturers simply isn't cheap.

As I stated above, there are a lot of factors working in AMD's favor this time around and I'm really hoping they emerge post-Zen as a much leaner/meaner and competitive company. They are known to fuck things up and their track record especially on the CPU side isn't always the greatest but only time will tell.
I agree with many of your points, but we just don't know how well AMD is truly going to do with Zen. AMD's track record isn't too good as you say, but more than that it often cannibalizes its profitable divisions to maintain it's CPU business despite the fact that their track record of success is piss poor financially speaking. AMD can get a cash infusion off selling it's graphic division one time. It's like being on a sinking boat where the water pump is the only thing keeping it from going under and then selling that pump before they get to shore or repair the hole. That's how AMD thinks unfortunately.

So I'm less confident than you seem to be, but we'll see what happens. If nothing else it will be interesting when Zen comes out at the very least.
 

Pieter3dnow

Supreme [H]ardness
Joined
Jul 29, 2009
Messages
6,784
While I don't know what Zen's chipsets will be like, I can say that the 990FX wasn't half what the equivalent Intel chipset at the time was priced at. You couple that with less restrictive VRD specifications and fewer guidelines from AMD and more corners can be cut on an AMD processor compatible motherboard than could ever be cut on the Intel side. This is pretty much how it's always been and it's one of the many reasons why AMD based motherboards can offer the same feature set at half the price. Many people think it's purely a fact of price gouging and it simply isn't true. At least, not past the chipset itself. Intel controls that and only they know how much those truly cost to produce. The price of the chipset for the motherboard manufacturers simply isn't cheap.

I agree with many of your points, but we just don't know how well AMD is truly going to do with Zen. AMD's track record isn't too good as you say, but more than that it often cannibalizes its profitable divisions to maintain it's CPU business despite the fact that their track record of success is piss poor financially speaking. AMD can get a cash infusion off selling it's graphic division one time. It's like being on a sinking boat where the water pump is the only thing keeping it from going under and then selling that pump before they get to shore or repair the hole. That's how AMD thinks unfortunately.

The problem with Intel chipset and motherboards and only having one source means you can ask anything and keep prices artificially high the sole reason for Intel leaving Socket 7. We all remember how those Intel Sata ports were broken on one of the Intel chipsets :) . Intel can burden costs towards chipset development and no one can refute it. From the company that cut cost on solder and case lids....

When you see clear blame from AMD side of things why did Intel need to do all of those nasty things as well. AMD management would do themselves in without any help from Intel ?

I can tell you one thing that AMD did right that was hiring Jim Keller and that was something which happened to late but still there not giving up ....
 
Top