AMD: Zen improves IPC by 40%+, AM4 FX CPU Announced

Fingers crossed...

I've got a 1090T which is long in the tooth. (My kids' machine...and my standby.) I've been getting the itch to replace the cpu/mobo w/an i5 or i7. (Leaning...strongly...towards an i7.)

Thats funny, Ive got the exact same thing. Had a 1090 for a while and ran it at 4.0 GHz forever. Moved it over to my little boy's rig when I moved up to the 8150 and then this 8350. Great little chip. Ill always hang onto it as a stand by.
 
I am waiting patiently for Zen, once its out, I will dump my 4790K and will never look back. The reason why I never get AMD right now is solely because Zen is not compatible with current boards. Pls give us a 5GHz CPU, I don't care about TDP.
 
i want amd to be just above haswell just high enough to tea bag intel...
 
I am waiting patiently for Zen, once its out, I will dump my 4790K and will never look back. The reason why I never get AMD right now is solely because Zen is not compatible with current boards. Pls give us a 5GHz CPU, I don't care about TDP.

That's pretty much the same situation for me. If Zen delivers, I will switch immediately. I'd still be on AMD if the 990FX platform wasn't so outdated. I do like most of Intel's products, but I absolutely hate their business practices. I try to only buy Intel stuff second hand so they don't get my money directly.
 
Fingers crossed...

I've got a 1090T which is long in the tooth. (My kids' machine...and my standby.) I've been getting the itch to replace the cpu/mobo w/an i5 or i7. (Leaning...strongly...towards an i7.)

If this new AMD cpu is competitive with the i7, then AMD my get my business.

(Other two rigs have an i7 and an FX8350. So, this one can be a toss-up between the two brands.)

I loved my 1090T, bought it when they first came out. But it was definitely not keeping up anymore so I replaced it with the i7 5820k. Night and day difference and my first Intel chip since the Pentium 3 lol.

I would love for AMD to generate some competition again, Intel is driving things forward but very slowly since they have no need to push hard right now to maintain their large lead.
 
Interesting what You guys are finding so great in those 1090T - I had overclocked 1055T and it was biggest piece of crap I've ever bough (well bulldozer was worse but i stayed safely away).

It was barely faster that core 2 Quad at similar speeds and when I moved to 2500k it was biggest most noticable cpu upgrade I've ever did.
 
Go second hand Xeon. Amazing the level of computing power you can get for low/mid-range prices.
 
I am hoping for a > $500 US 8 core / 16 threaded CPU. Meaning a processor that is AMD believes will compete with Intel's 8 core / 16 threaded processor not something that is the same performance level of 4 core / 8 threaded mainstream chip.

I'm looking for something that is decent enough to sink some money in. I'm not sure about why Jim Keller came to AMD but it is not to end up with a bad product if they wanted that then they could have just gotten Bulldozer better. So what is the big problem here for AMD to compete, if you look at gaming that will be "fixed" with DX12/Vulkan/Mantle if you look at something regarding Watts I'm not to sure if AMD will be able to do as well.

As long as the ballpark performance is close it is good enough if the next few generations of Zen can improve upon this.
 
I'm looking for something that is decent enough to sink some money in. I'm not sure about why Jim Keller came to AMD but it is not to end up with a bad product if they wanted that then they could have just gotten Bulldozer better.

Okay, dude, if you think you're better than Jim Keller, go put in an application. Otherwise you probably shouldn't be telling them how to design a CPU.
 
Okay, dude, if you think you're better than Jim Keller, go put in an application. Otherwise you probably shouldn't be telling them how to design a CPU.

It has nothing to do with me being better. Why don't you take a look at the last x86 cpu Jim Keller made. Just that the expectations for this one are way to high....
 
It was posted in another forum the interview with Keller which he states he was going to merge the best of what AMD had, bulldozer cores with puma. What he does say is it isnt a brand new IP but a furthering of the existing IP. He doesn't give great detail so we can't say there is no Keller magic, but I have to agree with Peter here and that this man KNOWS what he is doing.
 
I'm hoping that AMD will have at least two HT links in these new CPUs to allow multiprocessing.
 
I would say 40% would put IPC at around 70% to 80% of Haswell. Although IPC alone does not tell the whole story. We would have to see what frequency Zen will run at.

My math comes out to a 90% avg of the IPC of Haswell

Keller is quite good at designing CPUs, and Zen was good enough for AMD to completely change their 24 month roadmap to revolve around it. That's placing an awful lot of eggs in one basket folks.


Agreed on the base frequency is the most important missing detail.
 
Last edited:
I'm hoping that AMD will have at least two HT links in these new CPUs to allow multiprocessing.

I don't recall off the top of my head but I was under the impression that HT was dead in lieu of a new replacement..I know it was a few years ago I read about it, but with AMD (so sadly:() dead in the Big Data market, now would be a great time for them to switch to a new interconnect..

Don't get me wrong, I have nothing against HTT, and am a huge AMD fan..Its just dated and when I read your post a bell went off in my head..I would love it if were like the old days where mobile and desktop shared the same socket and you could use an expensive cpu in a 2P setup just by shorting a few traces with a pencil!! The good ole days!:cool:

If Zen turns out to clock well under water (I could care less about TDP as long as the MB is properly designed for it) then I will have one and this golden 3770K @ 5Ghz will be handed down to my Dad to replace his 20th Anniversary Pentium which is OC'd as well..
 
40% improvement in IPC should put it around Sandy/Ivy Bridge performance which is not great but good, especially considering that for all intents and purposes Sandy Bridge single core performance is enough.

If however that was true and 8c/16t CPU price was similar to 4c/8t i7 offerings (which is obviously much cheaper than 8c/16t i7's+expensive mobo) then I guess we could see big AMD comeback, people writing "AMD" in their sigs with pride :)

APUs with 4c/8t should have atleast PS4 level of GPU performance if not higher. Partly for marketing purposes and to keep safe distance from Intel iGPU offerings.

Oh, and as for ARM, why does it take them so damn long to release it? They have Opteron A1100 so it is a matter of making desktop version. By 2017 this 8c 2GHz A57 whing will be outdated. Maybe they will make something better by then... They should imho either release it now or give up.
 
40% improvement in IPC should put it around Sandy/Ivy Bridge performance which is not great but good, especially considering that for all intents and purposes Sandy Bridge single core performance is enough.

You make that claim, not knowing the final clocks (for a starter)?

lol
 
You make that claim, not knowing the final clocks (for a starter)?

He means at equal clockspeeds. We're talking about IPC.

That said, we don't really know the baseline for the 40% improvement since we haven't seen how well the Excavator cores perform yet.
 
He means at equal clockspeeds. We're talking about IPC.

That said, we don't really know the baseline for the 40% improvement since we haven't seen how well the Excavator cores perform yet.

Let me guess....this will be another hypetrain with little facts and a lot of wishfull thinking?
 
If it's anything like the 390X thread, it'll go on for about a year before we see any hardware! ;)

As for hype...well, I'll wait for the [H] review. If it's any good, instead of swapping my 1090T for an i7, I'll toss one of these bad boys in. Just hope it'll be out before Christmas.
 
The slides only mention a quad-core Zen, but they do mention putting a couple of these units together.

The "leaked" slides about the huge number of cores were proven to be fakes.
 
Let me guess....this will be another hypetrain with little facts and a lot of wishfull thinking?

All aboard! :)

I dunno, there was a lot of hype leading up to the Bulldozer launch but I don't recall AMD leaking any real info about it other than that it was a 'world record overclocker.' This doesn't smell quite as BS'ey to me. Unless the clock speeds are well below what Intel is doing it sounds like a nice upgrade. Keller's involvement is also reassuring.
 
I loved my 1090T, bought it when they first came out. But it was definitely not keeping up anymore so I replaced it with the i7 5820k. Night and day difference and my first Intel chip since the Pentium 3 lol.

I would love for AMD to generate some competition again, Intel is driving things forward but very slowly since they have no need to push hard right now to maintain their large lead.

ditto.
 
I would love for AMD to generate some competition again, Intel is driving things forward but very slowly since they have no need to push hard right now to maintain their large lead.

The way it is going for AMD is rather stupid, Intel maintains their lead in a segment that eventually will be less as smaller devices can do the same tasks. Intel has shown that it will do everything to maintain their position in the x86 market.

If AMD does not survive this for any reason Intel becomes a monopoly on desktop market in say 5 to 10 years, which will have other problems.

AMD already announced that their business will be spread now to ARM as well just to get out the strangle hold to generate money.

You can not demand or ask from AMD to beat Intel on cpu what you have to hope for is something that can come close.

What can happen is that if AMD is getting a good deal of ARM market they just leave x86 to Intel for simple reason as them having to answer for the monopoly status. This would be the only successful strategy for AMD to get rid of Intel in the short term (within 10 years).
AMD can not sustain growth even if they have a better product on the x86 market and many of the previous AMD management have failed to see this. If and when AMD gets enough money it prolly won't be from their x86 cores and to justify all the R&D cash to flood x86 platform would be another big mistake.
 
Last edited:
AMD have no place in mobile ARM space, ARM won't do much on server front and in desktop X86 won long time ago.

Good Zen performance is the only way AMD can survive.
 
AMD have no place in mobile ARM space, ARM won't do much on server front and in desktop X86 won long time ago.

Good Zen performance is the only way AMD can survive.

Like nvidia has done with the Tegra line. The only thing amd needs to do with K12 is beat nvidia's x1. Amd at this point has great gpu chips lackluster cpu arm chips are closer to gpu than x86... so amd K12 should be pretty beast now question is if they are aiming to make tablet SOC who is doing the communications bits separate samsung or Qualcomm modem? Or is amd going to make their own lte modem?
 
AMD have no place in mobile ARM space, ARM won't do much on server front and in desktop X86 won long time ago.

Good Zen performance is the only way AMD can survive.

Why won't ARM do much on the server front? Do you think a web app company cares if they run x86 or ARM? No they don't. Performance per dollar is all it matters.

Software compatibility is a non-issue. PHP, Java, Ruby, Python, etc,. work just fine on ARM.

If AMD can deliver a competitive server product, things will shift.
 
AMD have no place in mobile ARM space, ARM won't do much on server front and in desktop X86 won long time ago.

Good Zen performance is the only way AMD can survive.

What is mentioned is that ARM is the way forward for AMD to allow money to keep coming in. You are suggesting the same failed tactics that each and every person in previous AMD management has made. When AMD was faster it didn't result in a huge leap forward for the company. AMD takes on Intel it has to sell cpu to OEM and that is still not going anywhere.

Why would OEM take a big gamble on AMD cpu when they are getting what they need from Intel? Because the enthusiast market is not going to sustain AMD. (people who work for OEM have commented that no one ever gets fired for buying Intel cpu).

Like nvidia has done with the Tegra line. The only thing amd needs to do with K12 is beat nvidia's x1. Amd at this point has great gpu chips lackluster cpu arm chips are closer to gpu than x86... so amd K12 should be pretty beast now question is if they are aiming to make tablet SOC who is doing the communications bits separate samsung or Qualcomm modem? Or is amd going to make their own lte modem?

AMD ARM plans are not consumer oriented. That is why they dropped the compatible socket platform.
 
ARM is perfect symbol of AMD wunderwaffe tactics of previous managment who wanted to chase Intel through shortcuts so they were inventing more and more rediculus ways that were supposed to be a future of PC.

First moar coares in bulldozer
Then Apus
Then HSA
Then ARM cpus

At least Lisa Su seems to be realist and finally is doing some good work.
 
AMD ARM plans are not consumer oriented. That is why they dropped the compatible socket platform.

They dropped the server business and K12 is slated to be on an embedded platform. Draw better conclusions from that. They might use them in a server board of some sort but it seems to me they are targeting intel and Nvidia arm and that means tablets and phones. And that is a good idea but there are a lot of players in the arm field Qualcomm, Samsung are the two dominant ones with nvidia and a lot of Chinese brands trailing.
 
AMD have no place in mobile ARM space, ARM won't do much on server front and in desktop X86 won long time ago.

Good Zen performance is the only way AMD can survive.

Unfortunately according to AMD's own projections they're expecting the PC / desktop market to exhibit flat or negative growth in the near future.

So they're kind of building a CPU for a sector that isn't even going to grow much.

K12 could be interesting but it's coming way too late.
 
Unfortunately according to AMD's own projections they're expecting the PC / desktop market to exhibit flat or negative growth in the near future.

So they're kind of building a CPU for a sector that isn't even going to grow much.

K12 could be interesting but it's coming way too late.

I donno the release of each version of windows usually picks up pc sales...
 
I'd really love to a viable alternative socket for high end systems, and some competition for Intel to make them focus a bit more on the desktop or at least cut prices further than they are.

I want to step up to the X99 platform but it really irritates me how far behind it is versus the LGA1150 CPUs. If Zen is good, this might help there.
 
There is that theory but Windows 10 is going to be released a year before Zen is ready.

Plus as we've seen since Windows 7, Windows now runs on lesser hardware better than in the days of Vista.

Chances are if you slap Windows 10 on an old Vista machine from 2006 it will run even better.

SSDs are the only thing worth upgrading for the average Joe. I do a lot of SSD upgrades, not so much new laptops and desktops. Plenty of 2GHz+ dual core laptops with 4GB of ram in em that will live again.
 
ARM is perfect symbol of AMD wunderwaffe tactics of previous managment who wanted to chase Intel through shortcuts so they were inventing more and more rediculus ways that were supposed to be a future of PC.

First moar coares in bulldozer
Then Apus
Then HSA
Then ARM cpus

At least Lisa Su seems to be realist and finally is doing some good work.

ARM is a direct result of where it can deliver AMD more customers that you don't understand the dynamics of ARM market does not mean it is not there...
Also APU are pretty good they sold enough to keep updating the platform to FM2+ so not sure why you are listing this as a failure in the same line as you pretend know what you are talking about.
Intel making their graphics part in the cpu bigger each year as well so APU are doing something right.
HSA is not a trademark or business deal which makes them money but does allow hardware to function better as a result of these implementations, not to sure why you would list that..


They dropped the server business and K12 is slated to be on an embedded platform. Draw better conclusions from that. They might use them in a server board of some sort but it seems to me they are targeting intel and Nvidia arm and that means tablets and phones. And that is a good idea but there are a lot of players in the arm field Qualcomm, Samsung are the two dominant ones with nvidia and a lot of Chinese brands trailing.
They dropped SeaMicro that is something different ....
http://semiaccurate.com/2015/05/06/amd-puts-goodies-amid-vague-terms/
This is followed on by the long-awaited K12 ARM cores, one quote on the slides say that the new core is, “Highly-leveraged with x86 Core Development”. SemiAccurate takes this to mean that K12 will be something akin to Zen with a different decode structure to handle ARM’s ISA instead of x86. If this assumption is accurate, it would mean a very high performance ARM core, on par with current x86 offerings. That would change the game fairly radically if the two ISAs were on par with each other, mix and match indeed. The only problem we see is that K12 appears to have slipped a year to 2017, the more things change…
Unfortunately according to AMD's own projections they're expecting the PC / desktop market to exhibit flat or negative growth in the near future.

So they're kind of building a CPU for a sector that isn't even going to grow much.

K12 could be interesting but it's coming way too late.

It is a high performance part and current ARM cpu does not do that in the same scope as K12 does. Not even in 2017.
 
ARM is a direct result of where it can deliver AMD more customers that you don't understand the dynamics of ARM market does not mean it is not there...
Also APU are pretty good they sold enough to keep updating the platform to FM2+ so not sure why you are listing this as a failure in the same line as you pretend know what you are talking about.
Intel making their graphics part in the cpu bigger each year as well so APU are doing something right.
HSA is not a trademark or business deal which makes them money but does allow hardware to function better as a result of these implementations, not to sure why you would list that..

Have you checked AMD cpu market share in last 2 or 3 years ?

And I listed all of them because they are AMD marketing bullet points made with intention of fooling customers into buying cpu with completly inadequate performance for the time it was released with false hope than it will be great in the future.
 
Back
Top