AMD Zen Rumours Point to Earlier Than Expected Release

alxlwson said:
So, where would +40% IPC put Zen in relation to current Intel offerings?
Click to expand...
It depends how that 40% IPC increase is generated. AMD stated that IPC increased by up to 40% on a core basis, but that includes SMT now. A Pentium 4 with HyperThreading (Intel's marketing name for SMT) had up to a 30%-40% IPC increase with multi-threaded code, but that was unusual. Mostly it was only up to ~10-15% with multi-threaded code.

I think in an ideal situation, with a perfect and consistent 40% IPC increase in both single-threaded and multi-threaded code could bring Zen close to Haswell on a per core basis.

Given that Zen hasn't even taped out yet (AMD says it takes about 12 months from tape out to release, a schedule AMD has never once met on a new design), and that Haswell is coming up on 3 years old soon, I don't think even the rosiest projections for Zen are looking good at this point. Matching the performance of what will be a 4 year old processor when Zen is finally released isn't exactly brag-worthy, and Zen is pretty unlikely to meet that level of performance.
 
alxlwson said:
So, where would +40% IPC put Zen in relation to current Intel offerings?
Click to expand...

IPC would put it somewhere between ivy-bridge and Haswell.

I expect the clockspeed will decrease from the 4.7 GHz FX. How much it will decrease is unknown.
 
Last edited:
drescherjm said:
IPC would put it somewhere between ivy-bridge and Haswell.

I expect the clockspeed will decrease from the 4.7 GHz FX. How much it will decrease is unknown.
Click to expand...

That's very bad news, if we are to believe "leaked" slides. Surely that uber processor guy did better than that. I can't imagine AMD doing that. That makes zero sense. We should all take that with a grain of salt.
 
Given that Zen hasn't even taped out yet
Click to expand...

Zen (in this case Summit Ridge) DID tape out already. A while ago, actually. The whole "Zen has yet to tape out" thing is a misnomer, the guy was actually talking about the server variant when he said it hadn't taped yet. That's why every time you see AMD mention Zen launching in 2016 they always say HEDT or high-end whatever somewhere close by. Summit Ridge launches on AM4 and then the APU and server variants come shortly later.

That's very bad news, if we are to believe "leaked" slides. Surely that uber processor guy did better than that. I can't imagine AMD doing that. That makes zero sense. We should all take that with a grain of salt.
Click to expand...

What are you referring to here? The clockspeed or the IPC? It being between Ivy and Haswell would be very good considering it's a clean-sheet design. Intel has been iterating their current architecture FOR AGES at this point. I mean it still traces its lineage back to Pentium 3. Sure if you look at the building blocks you wouldn't be able to say they look even remotely similar but the fact is they've been modifying that uarch for over a decade at this point. AMD is starting from scratch basically with much less spending money than Intel has.

In terms of clockspeed I'm not sure why anyone would expect this design to hit 4.7ghz or anywhere close to that on stock. This uarch doesn't have pipelines as long as 15h and wasn't designed with that sort of thing in mind. For the octocores I'd expect 3.3ghz or somewhere in that range for stock clocks. That's realistic since they said the top TDP they're targeting is 95W for AM4. Don't expect miracles on Samsung/GloFo 14LPP.
 
NaroonGTX said:
Zen (in this case Summit Ridge) DID tape out already. A while ago, actually. The whole "Zen has yet to tape out" thing is a misnomer, the guy was actually talking about the server variant when he said it hadn't taped yet. That's why every time you see AMD mention Zen launching in 2016 they always say HEDT or high-end whatever somewhere close by. Summit Ridge launches on AM4 and then the APU and server variants come shortly later.



What are you referring to here? The clockspeed or the IPC? It being between Ivy and Haswell would be very good considering it's a clean-sheet design. Intel has been iterating their current architecture FOR AGES at this point. I mean it still traces its lineage back to Pentium 3. Sure if you look at the building blocks you wouldn't be able to say they look even remotely similar but the fact is they've been modifying that uarch for over a decade at this point. AMD is starting from scratch basically with much less spending money than Intel has.

In terms of clockspeed I'm not sure why anyone would expect this design to hit 4.7ghz or anywhere close to that on stock. This uarch doesn't have pipelines as long as 15h and wasn't designed with that sort of thing in mind. For the octocores I'd expect 3.3ghz or somewhere in that range for stock clocks. That's realistic since they said the top TDP they're targeting is 95W for AM4. Don't expect miracles on Samsung/GloFo 14LPP.
Click to expand...

Was talking about IPC. It's a big deal... it doesn't seem they can afford to not be competitive with the Intel generation that will be out with launch. They need to be on par, otherwise they won't even get money from the holdouts that have been waiting for these new parts. I really don't see releasing a part (clean sheet or not) that is four-five years behind your competition is going lure new customers or even keep existing ones.
What's the current IPC deficit between the current i7 and the FX9xxx?

I'm not really concerned with power consumption.
A 220w part running full bore 24/7/365 uses ~ 1927kWh *9.43c= $181 a year, if I did my math correctly. Now, seeing as very rarely is a desktop part ever ping 100%, or 24/7/365, a more logical appropriation is to cut that down to being taxed maybe 25% of the year. That is probably still very high. So $181x.25= ~$45 a year, or $3.75 a month.

Hot or not, just needs to be as fast as whatever Intel has on the market, or shortly thereafter.
 
Last edited:
Darakian said:
It's also unknown how many cores will be packed in.
Click to expand...

I'd say that it's safe to assume that the top Zens will have eight cores (16 threads). It would be around the same size (or even smaller) as quad core skylakes because they will not have GPU.
 
I'd say that it's safe to assume that the top Zens will have eight cores (16 threads).
Click to expand...

I expect 4 core / 8 threaded processors and 6 core / 12 threaded processors on the 95W AM4. These will compete directly with mainstream i5s and i7s.

And 8 core / 16 threaded and above on the 200W socket G competing with Intel's Enthusiast platform.
 
Last edited:
Was talking about IPC. It's a big deal... it doesn't seem they can afford to not be competitive with the Intel generation that will be out with launch. They need to be on par, otherwise they won't even get money from the holdouts that have been waiting for these new parts. I really don't see releasing a part (clean sheet or not) that is four-five years behind your competition is going lure new customers or even keep existing ones.
Click to expand...

Your expectations are not realistic. And even if Zen does launch in the ballpark between Ivy and Haswell, how is that not good? Broadwell barely had any noticeable increase over Haswell, and ditto for Skylake. Intel's last biggest jump in IPC and/or perf/core was between Westmere and Sandy Bridge, they haven't had anything significant since then and it's understandable why (protip: it has nothing to do with "not having any competition" like so many ill-informed always mention in the community.)

AMD has already secured numerous design wins with OEM's and other clients for Zen-based chips. If Zen was a dud, why would that happen? Obviously those OEM's and clients saw Zen and what it could offer and liked it enough to use it in their upcoming solutions. The idea that if it doesn't achieve parity with Skylake, something nobody at AMD could've possibly foreseen FOUR YEARS AGO when the design first began, would mark it as a failure is absurd. That is only a sentiment an unrealistic enthusiast could have.

For OEM's and enterprise and whatnot clients, they would choose the AMD parts either due to perf/watt or unique features or just outright cheaper price. Same with consumers. It's not rocket science really. AMD had to start somewhere and securing design wins with Zen for the serverspace, where they stand to gain MUCH MUCH more revenue from than something as niche as HEDT or even consumer DIY desktop was only the logical step to take.

I expect 4 core / 8 threaded processors and 6 core / 12 threaded processors on the 95W AM4. These will compete directly with mainstream i5s and i7s.

And 8 core / 16 threaded and above on the 200W socket G competing with Intel's Enthusiast platform.
Click to expand...

There hasn't been any official word on a "Socket G" (nice job making something up again, Fuad) yet from AMD. Though it's very likely there will be a successor to Socket G34, it hasn't materialized yet. Anyway it has already been established that AM4 will be getting a 95W 8C/16T Zen CPU. Numerous PR slides from AMD have shown "Up to 8 CPU" for AM4 and similar rhetoric on them before.
 
Guesstimating the IPC of Zen based on the 40% statement is meaningless unless someone can demonstrate what 40% over Excavator would actually be. The Haswell guesstimate is based around 40% over Steamroller's IPC.

drescherjm said:
I expect 4 core / 8 threaded processors and 6 core / 12 threaded processors on the 95W AM4. These will compete directly with mainstream i5s and i7s.

And 8 core / 16 threaded and above on the 200W socket G competing with Intel's Enthusiast platform.
Click to expand...

That's a whole lot of wasted space since the Zen doesn't have an iGPU taking up half of it unlike Skylake. The 8 core Zen is suppose to be released this year on AM4. There's suppose to be a 16 core APU and a 32 core server CPU in the works.
 
Last edited:
I have not seen any benches or figures given from users in regards to Excavator, unfortunately. Only some geekbench results. FM2+ is getting some Excavator Athlons soon but they will be locked-multiplier and most likely only released in limited markets (like China.) Bristol Ridge will launch on AM4 in a few months and will contain locked/unlocked APU's and Athlon variants so I guess we can get some decent projections then? The figure I've seen from AMD was that XV offered 5~15% higher IPC than Steamroller does.
 
alxlwson said:
So, where would +40% IPC put Zen in relation to current Intel offerings?
Click to expand...

It is a very floaty figure ;) . What it is based on is that the CMT part is gone which was a headache to get decent IPC on.

The figures you get to hear from AMD are just that they are done for purposes which will make people think that the chip is going to be "good". Sometimes it is for investors or other purposes. The proof lies in the early samples and or first release candidate.

Supposedly you need 6 to 9 months for a new revision of the cpu, by now it should be clear to AMD where they are standing...
 
NaroonGTX said:
Your expectations are not realistic. And even if Zen does launch in the ballpark between Ivy and Haswell, how is that not good? Broadwell barely had any noticeable increase over Haswell, and ditto for Skylake. Intel's last biggest jump in IPC and/or perf/core was between Westmere and Sandy Bridge, they haven't had anything significant since then and it's understandable why (protip: it has nothing to do with "not having any competition" like so many ill-informed always mention in the community.)

AMD has already secured numerous design wins with OEM's and other clients for Zen-based chips. If Zen was a dud, why would that happen? Obviously those OEM's and clients saw Zen and what it could offer and liked it enough to use it in their upcoming solutions. The idea that if it doesn't achieve parity with Skylake, something nobody at AMD could've possibly foreseen FOUR YEARS AGO when the design first began, would mark it as a failure is absurd. That is only a sentiment an unrealistic enthusiast could have.

For OEM's and enterprise and whatnot clients, they would choose the AMD parts either due to perf/watt or unique features or just outright cheaper price. Same with consumers. It's not rocket science really. AMD had to start somewhere and securing design wins with Zen for the serverspace, where they stand to gain MUCH MUCH more revenue from than something as niche as HEDT or even consumer DIY desktop was only the logical step to take.
Click to expand...

Of course Zen has OEM wins, just like FX and Athlons had OEM wins. Dell and co. don't want to be totally devoid of AMD based designs.
 
I still think AMD should switch to factors of 3 for their products as a differentiator to Intel and Nvidia.

CPUs - 3 core - 6 core - 9 core - 12 core

GPUs - 1GB - 3GB - 6GB - 9GB

They go '1 higher/louder!'
 
daglesj said:
I still think AMD should switch to factors of 3 for their products as a differentiator to Intel and Nvidia.

CPUs - 3 core - 6 core - 9 core - 12 core

GPUs - 1GB - 3GB - 6GB - 9GB

They go '1 higher/louder!'
Click to expand...

Well they have done three core processors in the not so distant past.
Gpu ram is a different story. They have to take whatever offerings are available from oem, then adjust quantity to fit for bus width.
 
polonyc2 said:
Zen is a good name for it...all AMD fans need to meditate and pray that it doesn't suck
Click to expand...

I'm not an AMD or Intel fan (I just buy what is best) but I really hope AMD can pull off a miracle with Zen. If they don't that's probably the end of them, which means 5-10% performance jumps per year from Intel for the rest of eternity. If no one buys the GPU/ATI division we could even be left with a monopoly in both the CPU and GPU markets, which would be horrible for PC enthusiasts.
 
Last edited:
alxlwson said:
Was talking about IPC. It's a big deal... it doesn't seem they can afford to not be competitive with the Intel generation that will be out with launch.

Hot or not, just needs to be as fast as whatever Intel has on the market, or shortly thereafter.
Click to expand...

AMD hasn't released a new CPU faster than Intel's equivalent offering since the core 2 duo. Thinking they need to release the fastest CPU on the market to stay competitive is crazy IMHO.

If they can release a quad core cpu with an IPC that puts them in the ball park of a haswell and price them similar to their current CPUs that would be a huge deal. (that would be ~$100 for i5 4690 performance)
 
Kardonxt said:
AMD hasn't released a new CPU faster than Intel's equivalent offering since the core 2 duo. Thinking they need to release the fastest CPU on the market to stay competitive is crazy IMHO.

If they can release a quad core cpu with an IPC that puts them in the ball park of a haswell and price them similar to their current CPUs that would be a huge deal. (that would be ~$100 for i5 4690 performance)
Click to expand...


I'm not saying they need to do it to stay competitive. I'm saying they need to do it to stay alive and return to profit. There needs to be a draw for people to get back to AMD. If they can't release a blazing CPU, they're going to end up like they are now. Burning money with shit sales. The only way I can see that being the case is to have something around current Intel performance, and it just be stupid cheap. Perhaps moving an enormous amount of volume at a low profit margin could return them to profit.
If they don't have a wicked offering, Intel is going to destroy them in an advertising campain. Big box employees are gonna laugh at the sound of AMD to customers after they get finished with the brainwashing of the Intel Retail Edge program.
 
alxlwson said:
I'm not saying they need to do it to stay competitive. I'm saying they need to do it to stay alive and return to profit. There needs to be a draw for people to get back to AMD. If they can't release a blazing CPU, they're going to end up like they are now. Burning money with shit sales. The only way I can see that being the case is to have something around current Intel performance, and it just be stupid cheap. Perhaps moving an enormous amount of volume at a low profit margin could return them to profit.
If they don't have a wicked offering, Intel is going to destroy them in an advertising campain. Big box employees are gonna laugh at the sound of AMD to customers after they get finished with the brainwashing of the Intel Retail Edge program.
Click to expand...

What they need to stay alive is a platform that OEMs will take up. One that will give OEMs a reason to give AMD their money.

Something cheap, and that can easily be supplied.
 
Kardonxt said:
If they can release a quad core cpu with an IPC that puts them in the ball park of a haswell and price them similar to their current CPUs that would be a huge deal. (that would be ~$100 for i5 4690 performance)
Click to expand...
It's unlikely that AMD would price a new CPU that low unless it had major performance disadvantages. The reason AMD sells ~$100 quad core CPUs is because they're roughly comparable in CPU performance (generously, and iGPU performance hasn't a factor in CPU pricing at all) to dual core Intel models which sell in that price range -- or it's unsold stock that would otherwise be written off due to no demand.

The density of TSMC/Samsung 14nm/16nm FF processes isn't much better than 20nm planar processes, so die sizes might be a bit large on Zen models. That has usually been a recipe for disaster since AMD generally has larger CPU die sizes, and sells those chips for lower prices. There would be no reason for AMD to sell a quad core Zen for that low if it didn't absolutely need to.
 
alxlwson said:
I'm not saying they need to do it to stay competitive. I'm saying they need to do it to stay alive and return to profit. There needs to be a draw for people to get back to AMD. If they can't release a blazing CPU, they're going to end up like they are now. Burning money with shit sales. The only way I can see that being the case is to have something around current Intel performance, and it just be stupid cheap. Perhaps moving an enormous amount of volume at a low profit margin could return them to profit.
If they don't have a wicked offering, Intel is going to destroy them in an advertising campain. Big box employees are gonna laugh at the sound of AMD to customers after they get finished with the brainwashing of the Intel Retail Edge program.
Click to expand...


The reason why AMD is in a rut right now is because they have been cutting into their margins by selling larger CPU's (die size) at lower prices. They are setting their MSRP based on comparative performance. If they were to do that when they have much better performance that won't get them out of the hole they are in, it would be better for them to increase their CPU prices and maintain the performance per cost ratio.
 
They had an architecture AMD couldn't capitalize on, since Intel does not release any information about yields it is rather silly posting about die size since you can have lower die size but if the yields are bad......
 
alxlwson said:
I'm not saying they need to do it to stay competitive. I'm saying they need to do it to stay alive and return to profit. There needs to be a draw for people to get back to AMD. If they can't release a blazing CPU, they're going to end up like they are now. Burning money with shit sales. The only way I can see that being the case is to have something around current Intel performance, and it just be stupid cheap. Perhaps moving an enormous amount of volume at a low profit margin could return them to profit.
If they don't have a wicked offering, Intel is going to destroy them in an advertising campain. Big box employees are gonna laugh at the sound of AMD to customers after they get finished with the brainwashing of the Intel Retail Edge program.
Click to expand...

JMccovery said:
What they need to stay alive is a platform that OEMs will take up. One that will give OEMs a reason to give AMD their money.

Something cheap, and that can easily be supplied.
Click to expand...

You guys are both right. The need an awesome CPU that will entice the OEM's. What people keep forgetting is the requirements are completely different for that market segment.

They need a chip that's decently fast on par with current Intel offerings, not ones from years past. This has nothing to do with the actual performance, even though us enthusiasts know full well it's not needed. This has everything to do with marketing and how ill-informed average joe consumer really is. Average joe is going to make his decisions based on the idiot teenager working Best Buy or Computer Shopper magazine, they don't peruse decent online forums to get the real answers. If they do happen to even look at a numbers, they only care if x is bigger than y. They don't care what x and y actually mean. Not only this, the recent trend for past several years is for OEM's to put the AMD chip's in the lowest end computers they offer. When average joe has a bad time with the low end computer, he doesn't think it's his fault, he blames the logos on the box. Sometimes that means they switch OEM's (sometimes to Intel only ones) or sometimes they blame that AMD sticker. Granted, this last bit is only a minor source of the problem, but it is there. Getting AMD chips into high end boxes that are built well can only help spread word of mouth advertising about their product.

The second issue, and arguably more important to win over the OEM's is to not only get the performance nearly on par with Intel, they need to do so with performance per watt. This issue is again for reasons us enthusiasts don't relate to. OEM's have very little profit margin on the boxes they do sell. If they can get away with using an ounce of copper less in a heatsink, they will. If they can order 1 million 300 watt power supplies and forgo ordering 400 watt power supplies since all boxes have same power requirements, they will. If they can make the motherboard with less power circuitry and save pennies a motherboard, they will. These little cost cutting things add up big when you are moving millions of units.

The final issue is volume. If and when their chips become popular again with the OEM's, they have to be able to supply enough in quantity at prices per unit competitive with Intel.
 
Last edited:
Even if its a good chip the OEMs wont buy it in any great number.

You'll have 54 Intel boxes and the one AMD box in the store.

Kind of like how it is now. Performance means nothing. The CPU market at the budget end is a mess right now. Confusing enough for enthusiasts with Atom chips masquerading as Pentium. ULV i3 chips and normal i3 chips, Celerons that could be Atoms or maybe Pentiums, crappy AMD E1 chips etc. etc.

I now just tell folks to dig out a cheap i5 equipped machine and be done with it. They keep them for 5-6 years anyway so no ill store.
 
Pieter3dnow said:
They had an architecture AMD couldn't capitalize on, since Intel does not release any information about yields it is rather silly posting about die size since you can have lower die size but if the yields are bad......
Click to expand...


Intel has been on 14nm for close to 2 year now? They don't have any problems with yields on 14nm. Doesn't take very long to get acceptable yields once you have functional chips of a certain size being produced on it.... I don't think Intel would have produced any chips on a fab process they felt that they would be getting bad yields anyways. Just like they were forced to push back 10nm they ended up, doing what they had to do on 14nm with Kaby Lake.
 
razor1 said:
Intel has been on 14nm for close to 2 year now? They don't have any problems with yields on 14nm. Doesn't take very long to get acceptable yields once you have functional chips of a certain size being produced on it.... I don't think Intel would have produced any chips on a fab process they felt that they would be getting bad yields anyways. Just like they were forced to push back 10nm they ended up, doing what they had to do on 14nm with Kaby Lake.
Click to expand...

Actually they did have issues on 14nm to begin with. Even the performance was lack luster.
 
razor1 said:
Intel has been on 14nm for close to 2 year now? They don't have any problems with yields on 14nm. Doesn't take very long to get acceptable yields once you have functional chips of a certain size being produced on it.... I don't think Intel would have produced any chips on a fab process they felt that they would be getting bad yields anyways. Just like they were forced to push back 10nm they ended up, doing what they had to do on 14nm with Kaby Lake.
Click to expand...

You dont remember the P3 1000 (or was it the 1133) released just to try and match AMD speed and performance days do you? I am also just giving you shit as I dont think they would release chip that had problems with yields now days.
 
razor1 said:
That was quickly fixed wasn't it? Within 1 quarter, 4 months.
Click to expand...

Probably, doesn't seem to have been an issue in the last while. Seemed to be the issue for the first few months, but like most news we generally only ever hear the bad, things go well... SILENCE.
 
1/2 bent said:
You dont remember the P3 1000 (or was it the 1133) released just to try and match AMD speed and performance days do you? I am also just giving you shit as I dont think they would release chip that had problems with yields now days.
Click to expand...


Oh yeah I remember that, but Intel has such a large lead on nodes and process now that it doesn't matter for them to rush things out on new nodes anymore. They are close to a generation ahead with node processes, that's like a 2 year lead maybe a bit more.
 
pxc said:
The density of TSMC/Samsung 14nm/16nm FF processes isn't much better than 20nm planar processes, so die sizes might be a bit large on Zen models. That has usually been a recipe for disaster since AMD generally has larger CPU die sizes, and sells those chips for lower prices. There would be no reason for AMD to sell a quad core Zen for that low if it didn't absolutely need to.
Click to expand...

It's not a 2-node size jump like the straight up numbers would suggest, but the TSMC/Samsung processes are not *too* much lower density than Intel's 14nm process. In other words, not a huge disparity there.
 
Derfnofred said:
It's not a 2-node size jump like the straight up numbers would suggest, but the TSMC/Samsung processes are not *too* much lower density than Intel's 14nm process. In other words, not a huge disparity there.
Click to expand...
I'm not sure where you're getting your info, but there is a significant difference: https://www.semiwiki.com/forum/content/3759-intel-versus-tsmc-14nm-processes.html That's virtually as much as there was between Intel's 14nm process and TSMC's or Samsung's 20nm processes. This is no secret. TSMC and Samsung have both noted the lack of scaling at the 14/16nm node years ago.

And I didn't even mention the other deficits that TSMC and Samsung have, like transistor performance. :p
 
Same source, different perspective, evidently, haha. And I clearly misremembered the delta as smaller than it is--more on the range of 10% than 20% (whoops, mea culpa). :)

Agreed on the fact that in all aspects Intel is winning on fab, no doubt about it.
 
Well we also have to keep in mind Intel has full control of their processes and also they are much more capable at a similar node vs. other companies at the same same node. Just look at the latest Phi, its a much larger chip than any other fab can create.

So AMD does have its work cut out for them even though Intel has been slowing increasing performance gen to gen. I'm hopefully wishing they will catch up, but in reality, I have a knot in my stomach.
 
Last edited:
I have been thinking about Zen and the reported 40% IPC increase and some of the reactions across many forums on its release and what they expect. Honestly 40% IPC increase above my 8350 would be nice and impressive, yet they are speaking of excavator which is 5-10% above Piledriver making it more so. Anyway even just above mine I think it will be a win, no matter where Intel stands at that time. Seriously this gives AMD users like myself a great upgrade from what we have. Not to mention there are a lot of users still using Sandybridge that may be old AMD users that want to come back. The unified socket is nice too.

Anyway if the 40% holds I think it will be met well. However if it does I wouldn't expect prices to cling to current lineups as most hope for. They will likely price close to Intels similarly performing chips.
 
I have heard all kinds of great AMD IPC rumors in the past since ... well a long time ago. Many of those rumors directly from AMD employees. I have seen none of those come to fruition. I would like nothing more than to see AMD bang out some 40% IPC jumps on desktop apps, but I am not holding my breath.
 
Even if AMD does meet the 40% or greater IPC jump, there's still gonna be people bitching about it for contrived reasons as usual.
 
NaroonGTX said:
Even if AMD does meet the 40% or greater IPC jump, there's still gonna be people bitching about it for contrived reasons as usual.
Click to expand...

Just like how one makes a "If they do" argument then you get the rash number who state they wont or that only fanboys believe it, completely missing the point of the original argument.
 
NaroonGTX said:
Even if AMD does meet the 40% or greater IPC jump, there's still gonna be people bitching about it for contrived reasons as usual.
Click to expand...

I expect there will be a lot of complaining about the price from people who do not understand why the bulldozer cores are priced the way they are today. Part of this is I expect to achieve a 40% improvement in IPC and to allow for HT I expect a single Zen core will have about as many transistors as an entire 2 core bulldozer module. There will also be complaining that Zen does not blow Intel out of the water and people complaining about the reduced clock speeds.

For me I am hopeful that Zen will produce a legitimate upgrade from my i7 970 in both single threaded and multithreaded performance + the ability to overclock and have ECC. I am also interested in very interested NUC type devices that perform better than 1 to 2GHz ARM devices but do not cost $300 to $500US.
 
Last edited:
-Strelok- said:
Only the most dedicated fanboys really believe anything AMD says anymore. Until they deliver, I don't care what their marketing team tries to tell me.
Click to expand...

Except that the one who designed this CPU knows what he is doing. Therefore, the 40% increase would not be unusual especially considering the IPC right now is only a little better than something that was produced 5 years ago.

I like the 6700k I upgraded too but, I am finding that, in fact, it was not a real difference over the 8350 in day to day use. (Also, I game at 4k so I am not seeing any real significant difference there either if any at all.) The only reason I did not upgrade to another AMD cpu is because I would have had to gone the 9590 route and a new mainboard. I just did not want to invest in a platform anymore that I would be replacing in a years time anyways, one way or the other.
 
You must log in or register to reply here.
Back
Top