Sigh! Zen sounds like many of their other, not yet available, products: 40% better performance, power power draw, yadda yadda yadda. But when the time comes, they can't come close to what they promised.
So this all seems very optimistic to me. How many times do companies that are having severe problems make announcements that this is the new stuff that will change everything? All too often. How many times are they right? Not that often.
Sigh! How I hate to generalized comments who didn't even read the article before posting.
They aren't talking about 40% better performance, they talk about 40%+ more IPC, which is very specific and only part of what constitutes the actual performance of a CPU. That means they are breaking away from the module design they introduced with Bulldozer and focus more on beefier cores. They didn't even talk about the power draw yet.
If Zen comes with 8C+HT in 2016 it will compete with Broadwell-E on Socket 2011, which is extremely expensive and could very well have about the same IPC, a big step forward from the current situation. It is pretty safe to assume that even Cannonlake on Socket 1151 will only come with 4C+HT.
This isn't very optimistic at all. They target an IPC performance level which is around Intel's Haswell generation.
So please, stop posting such uninformed comments, or at least try to read and understand the article before hand. Thank you!
Depending on the actual performance of the CPU the costs will vary. If it is performing on the same level as Broadwell-E then you better bet that it will cost as much as Broadwell-E. If it performs better, it will cost more. This is also of course if the power levels also fit the bill and we have yet to see how well the FinFET they are using performs in that arena.
That is how it has and always will work.
I am optimistic yet, as I think melgross was saying, hesitant to trust anything from AMDs marketing. Hell I don't even 100% trust Intel's marketing or any other company as they always use best case scenarios and not real world scenarios.
SO we can hope this will turn it around for AMD because I don't think they can last much longer running as they have been.
That is just plain and simple a wrong statement. Just take a brief look at AMDs history and you will find that products on the same level as their competitors usually are cheaper. So if and 8C+HT Zen performs on the same level an an Broadwell-E 8C+HT, it surely will be cheaper. With more performance, it might be on a similar price level.
The last time AMD held the CPU performance crown they had CPUs that cost more than Intel. That was during the Pentium 4 and Athlon 64 days. FX series CPUs cost about $1K+ and were the best you could get.
However, when Intel released the first Core 2 Quad at $500 bucks it beat the FX series that was still way higher than it.
And that is historical fact. You can try to deny it all you want but when AMD has a better performing part, they can demand a higher price. If you look at their GPUs it shows as well. Whenever their GPUs were the top performers they were equal to or higher than NVidia yet when they were lower performing they were cheaper. When the HD7970GHz came out, it launched at $549 bucks but could only be purchased for about $600. And it justified the price as it pounded the G600 series. However, if we look at the HD6970 it launched at $369 (cheaper than the GTX480 even) because overall it performed a bit worse than the GTX580 at $500 bucks.
That is how the market works and people tend to deny it but it is and it will happen. I know so many people who want to think that AMD is a company that cares about them. Well they do but more that you buy their products and they make money, just like NVidia, Intel and most every other company out there.
That said, I am excited for the R9 390X and I have been buying Radeons since the 9700Pro but that wont stop me from seeing the truth that there is a chance that it will cost more than the equivalent or lower performing NVidia/Intel product.
Exactly! I've said this so many times but all the AMD fans who think that company is some benevolent entity that steals from the rich to give to the poor.
All you hear repeated on the internet is how we need AMD to not just thrive, but survive, because competition is good but the fact is, the last time AMD thrived, they priced people like me out of the market for their CPUs! Athlon 64s STARTED at $400 ($450 after mark-ups) so I had no problems at all switching to Intel when their $330 E6600 blew everything AMD had out of the water for a fraction of the price.
Same thing with the 7970, as you mentioned. AMD sure was competitive then and used the opportunity to piggy-back on last-gen 40nm pricing. AMD fans said it was fully deserved, press didn't seem to mind either, but the 7970's performance was nothing special which allowed Nvidia to jump their 2nd fastest Kepler chip to their flagship SKU, and create a new ultra premium part called Titan.
Competition drives prices down. That's the main point I guess. Intel still charges $1K for their top tier Core and $2.5k for their top Xeon. If AMD delivers, you can say goodbye to that pricing, or at least better performance increase the more you pay, not the silly 20-30% for 3x the price.
Yeah once again, no guarantee of that, indeed you were just given 2 examples of competition driving prices up. What if AMD pulls off a miracle and actually beats Intel performance, and instead of staying the course of bargain basement pricing, looks to be rewarded for their efforts? They could slot right in lock and step with Intel's pricing and who loses? AMD users who aren't accustomed to paying more than say $200 for their CPUs.
First, please attempt to extract the MASSIVE nvidia carrot visibly protruding from your butt... and secondly please abandon all attempts to appear impartial in your commentary. It just makes you seem even more idiotic...
"you were just given 2 examples of competition driving prices up"
That isn't what those examples showed, they showed only that inter-generational rifts result in new products that are faster and cheaper than the older products.
In addition, since performance continued to increase and that performance became cheaper, they show that competition LOWERED prices, which was best for the consumer.
Companies must ALWAYS charge the most they can get - the most the market will tolerate.
Competition is a king! In short range, if AMD manages to make a desent product, the price will be somewhat high, because AMD cash flow at this moment is far from stellar. But in long run all companies also care for market share, and that is easier with relatively cheaper product. So it will force Intel to think they cost, and that means AMD also have to think their prizing. Same thing in GPU world. In long run there will be market share competition and that means prize competition. In short range, the need to balance the financial accounts is very important to the company, because if they don't they are dead company sooner or later. You can not do business, without income. AMD has been in tough situation for a long time and that does not help them in the prize competition, because they don't have money reserves for bloody prize competition. But if they manage to get their finance account to balance and even get some profit, they can try to gain bigger market share and that will means better prizes to us!
"The last time AMD held the CPU performance crown they had CPUs that cost more than Intel. That was during the Pentium 4 and Athlon 64 days" That was only valid for the very highest end. In the mainstream (where most people and sales are) at a similar performance point AMD was always cheaper. Just as an example Q1 2005: Athlon X2 4200+ was outperforming a Pentium D 840 both being at ~$540. At that time, given the same money, you could easily build a better rig going for AMD. That changed in time but AMD in it's entire history was rarely the one with higher cost parts at any given performance point.
Again, not accurate. If you wanted an Athlon 64 you could not find one for less than $400. Only Socket A based Athlon. It was a shocking role reversal for AMD and their fans since they went from no CPUs over $400 to none under 400 for 64bit. It's like they gave their fans the middle finger once they got the lead!
AMD did have equal to or higher pricing during those days. When the Core 2 series hit though, AMDs parts were still higher than Intels parts. Then again at the same time you could hve bought a Pentium D 805 for $150 bucks, gotten a good air cooler and overclocked it to levels where it beat the top end Pentium EE and Athlon 64 FX for about 1/5th of the price.
One reason why was that they didn't have enough product for demand because they actually closed down their last American based FAB just before launching the Athlon 64 line and their retail channels were low. That caused some to price higher than MSRP, much like how the HD7970GHz was running for about $1K if I remember correctly because of the bitcoin craze that caused them to be over valued and short on supply.
That said, my main point is that if AMD has a great performing part we will see prices equal to or higher than Intel. That is how it works. They are not going to have a better performing part and give it away, especially when they have mostly bled money the past 5+ years.
Same goes for every company. They are all out to make a profit, that is after all the idea behind capitalism.
I never said it was faster than Athlon 64. I said that when AMD has a performance advantage, it will cost more or will be equal in price. The FX-60 launched at $1031 (was probably $1100 retail). At the same time the top end Intel was the Pentium EE 955 which was $999. The FX-60 was a better CPU.
The FX-62 was higher priced than that when it and AM2 launched in May of 2006, around $1400, and yet when Core 2 Duo launched in July the E6700 was only $530 (at least half the price) and it outperformed it by quite a bit.
You can find all of this information in reviews from Anandtech via Google easily.
It isn't a AMD sucks, more a AMD is a business out to make money.
If Zen performs equal to or better than Intel it will be priced equal to or higher. That is how a business runs.
From what I can see - this company has quite a fan-base.
If they price them all out again, it won't be forgotten.
Me, I buy the highest spec price CPU available, regardless of who makes it. But sometimes just not on initial launch, saving myself *a little* cash in the process.
If AMD makes the fastest chip, that is what I will buy.
Assuming same clockspeeds, which is a big if and a risk AMD took in releasing that graph. They just have to be careful because if its 40% higher IPC and 25% lower clocks, that's not going to cut it, people are going to be bigtime disappointed because all they'll remember are the claims of 40% higher IPC.
There's that, if Zen has a shorter pipeline I'd expect lower factory clocks. But the other side of it being Bulldozer never hitting the target clocks it was meant for anyways, unless you wanted that 220W CPU.
If they want to stuff 8 cores into 95W power envelope, they would likely have lower single-thread performance than Intel's offering.
Thing is, for most tasks (except pre-DX12 gaming), single-threaded performance is overrated, as the computer is usually responsive enough.
Then, 8 Zen cores with lower 25% lower clock speed compared to 4 Skylake cores (even if they have 15% better IPC than Haswell), would show its teeth on workloads, where performance matters. (assuming AMD and especially their foundry partners deliver)
2x perf/watt increase for GCN GPU tech moving from 28nm to finfet sounds good on the surface, but that would only just be slightly surpassing Maxwell's efficiency on 28nm.
Any links to support such bold claims? Last time I've checked there was 300w vs 370w difference (total system performance). And CPU was hardlly about 100w in that picture.
Can't say there's too much that's interesting on there other than Zen in 2016. Q2 2015 Graphics parts will go a long way in determining if AMD makes it to 2016. If they can make it to Zen and their FinFet graphics parts and they're good, maybe they'll pull out of this funk.
But the odds are certainly stacked against them at this point.
My biggest question for them in 2015 is, what are they possibly going to fab (and sell) at GloFo this year to satisfy that $1Bn take or pay wafer commitment?
So anything worth buying will come after 2016? I was looking forward to R9 390 series to see if it will be as good as GTX 970 but if AMD is aiming to match Maxwell's efficiency in 2016, R9 390 would either be hotter or slower.
The 390 being hotter than Maxwell isn't that big a deal. There's a whole continuum in there, from "just barely hotter " to "OMGWTFBBQGTX480". Only the later would be worthy of concern in a performance-oriented desktop.
Does AMD really need to have an "Analysts day" talk that gives away their roadmap and plans to their competitor Intel?
They really need to learn from Apple and the A7 and just do everything in secret and only announce it when it's done. Apple's competitors are still stunned and trying to react from that announcement.
AMD needs to spend less time giving away their roadmap, and more time just building the product.
Their "roadmap" should just be product releases. It would help if they just disappeared for 2 years to work on these products. Come back when its ready.
They're all watching Star Trek in Liquid VR and speaking "Vulkan" to each other, so expect 1-2% speed optis for 3D, and some forlorn nostalgia for mantle. After all that they sit and contemplate their belly buttons, to focus, as zen is coming.
I must say, I like the overall strategy they laid out. I am very very cautiously optimistic.
1) I am glad they said they are staying out of mobile for now. Until K12, they won’t really have anything extra to offer, and unless K12 scales well down to 2W, they won’t have it either. In that market, you either differentiate (Qualcomm – modems and custom cores), or have low costs using standard ARM architecture (MediaTek, Samsung – that also leverages its manufacturing edge).
To support their view, just look at nVidia (that just announced it leaves the market) and Intel (that 100% subsidizes its products, just to get foothold – the failure went so far, that they have hidden mobile segment in their segment reporting)
2) Releasing Zen as performance part first is a good idea, because that is market where (i) there are best margins, and (ii) they still have a lot of credibility in that market or at the very least they cooperate with partners, that could help them to some design wins. Therefore, if they manage to offer parts, that are competitive with Intel’s upper-midrange, it might be quite a win for them.
My prediction is, that if they stuff 8 cores in 95W envelope, they will likely have lower single-threaded performance than Intel’s best, and multi-threaded may be better. Given how overrated single-threaded performance is (except for pre-DX12/Mantle games), it might be reasonable trade-off.
At the same time, those who remember K7 vs. PIII times will recall, how all of sudden was Intel able to squeeze quite a lot of performance from their chips (though it was due to 180nm upgrade vs 250nm).
The key question here are yields and actual quality of the manufacturing process. If, and that is a big if, it works well, AMD finally won’t have such a huge process disadvantage now. But AMD just has to take this bet, the alternative is really going bust this time.
3) AMD’s APUs always had better performance than Intel’s, but let’s be honest, they never really had punch to play modern games and higher settings. Now with HBM and hopefully more efficient architecture overall, they may be able to release APUs, that will be truly gaming-capable. This may secure them some design wins, assuming the CPU part is not resource hog (we will see Carizzo).
A number of thoughts: 1. Glad to see the f**kwits on the AMD board finally realise Dirk Meyer was right. Racing into low margin areas was foolish in the extreme, they burned a load of cash and time on it and its not sure if the company can recover.
2. 40% IPC on the BD series isn't great. That would leave it commensurate with Intel's 2011-2012 level IPC.
3. Will Zen actually be buyable in calendar year 2016? Or is it going to slip to "sometime" in 2017?
4. The HBM concept is great, for APU, GPU and even HPC dedicated CPUs, unfortunately due to (1) they don't have the resources to execute Zen and HBM concurrently.
5. GF also needs to deliver, while not GF at the time, it was the 65 nm process that screwed Barcelona, the TLC was a side-issue. Moving to 45nm went a long way to getting performance parity with Intel's best at the time. Unfortunately by then, the damage was done.
6. If Intel's process transitions are slowing up, then that does improve GF et al's chances of catching up.
7. I hope they make it. The market needs them to make it. [I've an old PII x6 I'd like to replace, ideally with Zen - I always go for the underdog!!]
Don't buy AMD CPU again. I don't know if you have a comparable Intel platform to compare with your PII x6 or not but I do. I had a C2Q Q6600 and a PII X4 925. I got the Q6600 to run at 3.0GHz at reduced voltage and X4 925 at 3.2GHz at stock voltage both on $100ish Gigabyte mATX board. The Intel platform is more responsive and overall gives me less problem because Intel RAID is more stable. I have since replaced the Q6600 with 3770K and will eventually replace the X4 925 with an Intel.
Funnily enough, I did have a Q6600, still have it indeed, but routing memory through the FSB is a killer for what I need to do from time to time. In this, even a 65nm Barcelona was faster than Harpertown never mind Clovertown!
Uh that's a worthless comparison. What does RAID have to do with it?
Here's one - I have a Phenom II X4 965 machine and a Core i5 4670k machine and I don't notice any difference with general responsiveness in usage, applications and of course gaming (which is mostly GPU bound for me).
It wasn't called out in the article, but anyone notice "Disruptive Memory Bandwidth" under "Next Generation Opteron" (a different column from ARM)? That sounds like HBM on an x86 AMD CPU in 2016-17.
This really was barely an article, just a recycling of marketing materials. Whether AMD can ever survive let alone return to profitability is very much open to question.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
47 Comments
Back to Article
melgross - Wednesday, May 6, 2015 - link
Sigh! Zen sounds like many of their other, not yet available, products: 40% better performance, power power draw, yadda yadda yadda. But when the time comes, they can't come close to what they promised.So this all seems very optimistic to me. How many times do companies that are having severe problems make announcements that this is the new stuff that will change everything? All too often. How many times are they right? Not that often.
w0mbat - Wednesday, May 6, 2015 - link
Sigh! How I hate to generalized comments who didn't even read the article before posting.They aren't talking about 40% better performance, they talk about 40%+ more IPC, which is very specific and only part of what constitutes the actual performance of a CPU. That means they are breaking away from the module design they introduced with Bulldozer and focus more on beefier cores. They didn't even talk about the power draw yet.
If Zen comes with 8C+HT in 2016 it will compete with Broadwell-E on Socket 2011, which is extremely expensive and could very well have about the same IPC, a big step forward from the current situation. It is pretty safe to assume that even Cannonlake on Socket 1151 will only come with 4C+HT.
This isn't very optimistic at all. They target an IPC performance level which is around Intel's Haswell generation.
So please, stop posting such uninformed comments, or at least try to read and understand the article before hand. Thank you!
jimmy$mitty - Wednesday, May 6, 2015 - link
Depending on the actual performance of the CPU the costs will vary. If it is performing on the same level as Broadwell-E then you better bet that it will cost as much as Broadwell-E. If it performs better, it will cost more. This is also of course if the power levels also fit the bill and we have yet to see how well the FinFET they are using performs in that arena.That is how it has and always will work.
I am optimistic yet, as I think melgross was saying, hesitant to trust anything from AMDs marketing. Hell I don't even 100% trust Intel's marketing or any other company as they always use best case scenarios and not real world scenarios.
SO we can hope this will turn it around for AMD because I don't think they can last much longer running as they have been.
w0mbat - Wednesday, May 6, 2015 - link
> That is how it has and always will work.That is just plain and simple a wrong statement. Just take a brief look at AMDs history and you will find that products on the same level as their competitors usually are cheaper. So if and 8C+HT Zen performs on the same level an an Broadwell-E 8C+HT, it surely will be cheaper. With more performance, it might be on a similar price level.
jimmy$mitty - Wednesday, May 6, 2015 - link
The last time AMD held the CPU performance crown they had CPUs that cost more than Intel. That was during the Pentium 4 and Athlon 64 days. FX series CPUs cost about $1K+ and were the best you could get.However, when Intel released the first Core 2 Quad at $500 bucks it beat the FX series that was still way higher than it.
And that is historical fact. You can try to deny it all you want but when AMD has a better performing part, they can demand a higher price. If you look at their GPUs it shows as well. Whenever their GPUs were the top performers they were equal to or higher than NVidia yet when they were lower performing they were cheaper. When the HD7970GHz came out, it launched at $549 bucks but could only be purchased for about $600. And it justified the price as it pounded the G600 series. However, if we look at the HD6970 it launched at $369 (cheaper than the GTX480 even) because overall it performed a bit worse than the GTX580 at $500 bucks.
That is how the market works and people tend to deny it but it is and it will happen. I know so many people who want to think that AMD is a company that cares about them. Well they do but more that you buy their products and they make money, just like NVidia, Intel and most every other company out there.
That said, I am excited for the R9 390X and I have been buying Radeons since the 9700Pro but that wont stop me from seeing the truth that there is a chance that it will cost more than the equivalent or lower performing NVidia/Intel product.
chizow - Wednesday, May 6, 2015 - link
Exactly! I've said this so many times but all the AMD fans who think that company is some benevolent entity that steals from the rich to give to the poor.All you hear repeated on the internet is how we need AMD to not just thrive, but survive, because competition is good but the fact is, the last time AMD thrived, they priced people like me out of the market for their CPUs! Athlon 64s STARTED at $400 ($450 after mark-ups) so I had no problems at all switching to Intel when their $330 E6600 blew everything AMD had out of the water for a fraction of the price.
Same thing with the 7970, as you mentioned. AMD sure was competitive then and used the opportunity to piggy-back on last-gen 40nm pricing. AMD fans said it was fully deserved, press didn't seem to mind either, but the 7970's performance was nothing special which allowed Nvidia to jump their 2nd fastest Kepler chip to their flagship SKU, and create a new ultra premium part called Titan.
lilmoe - Wednesday, May 6, 2015 - link
Competition drives prices down. That's the main point I guess. Intel still charges $1K for their top tier Core and $2.5k for their top Xeon. If AMD delivers, you can say goodbye to that pricing, or at least better performance increase the more you pay, not the silly 20-30% for 3x the price.chizow - Wednesday, May 6, 2015 - link
Yeah once again, no guarantee of that, indeed you were just given 2 examples of competition driving prices up. What if AMD pulls off a miracle and actually beats Intel performance, and instead of staying the course of bargain basement pricing, looks to be rewarded for their efforts? They could slot right in lock and step with Intel's pricing and who loses? AMD users who aren't accustomed to paying more than say $200 for their CPUs.Ranger101 - Thursday, May 7, 2015 - link
A little advice chizow...First, please attempt to extract the MASSIVE nvidia carrot visibly protruding from your butt...
and secondly please abandon all attempts to appear impartial in your commentary. It just makes
you seem even more idiotic...
Thanks in advance.
chizow - Thursday, May 7, 2015 - link
AMD butthurt defense ranger on duty I see!looncraz - Friday, May 8, 2015 - link
"you were just given 2 examples of competition driving prices up"That isn't what those examples showed, they showed only that inter-generational rifts result in new products that are faster and cheaper than the older products.
In addition, since performance continued to increase and that performance became cheaper, they show that competition LOWERED prices, which was best for the consumer.
Companies must ALWAYS charge the most they can get - the most the market will tolerate.
haukionkannel - Thursday, May 7, 2015 - link
Competition is a king! In short range, if AMD manages to make a desent product, the price will be somewhat high, because AMD cash flow at this moment is far from stellar. But in long run all companies also care for market share, and that is easier with relatively cheaper product. So it will force Intel to think they cost, and that means AMD also have to think their prizing. Same thing in GPU world.In long run there will be market share competition and that means prize competition. In short range, the need to balance the financial accounts is very important to the company, because if they don't they are dead company sooner or later. You can not do business, without income.
AMD has been in tough situation for a long time and that does not help them in the prize competition, because they don't have money reserves for bloody prize competition.
But if they manage to get their finance account to balance and even get some profit, they can try to gain bigger market share and that will means better prizes to us!
costeakai - Thursday, May 14, 2015 - link
it depends on what they would do with the profit: eat it, or reinvest it...close - Thursday, May 7, 2015 - link
"The last time AMD held the CPU performance crown they had CPUs that cost more than Intel. That was during the Pentium 4 and Athlon 64 days"That was only valid for the very highest end. In the mainstream (where most people and sales are) at a similar performance point AMD was always cheaper. Just as an example Q1 2005: Athlon X2 4200+ was outperforming a Pentium D 840 both being at ~$540. At that time, given the same money, you could easily build a better rig going for AMD. That changed in time but AMD in it's entire history was rarely the one with higher cost parts at any given performance point.
chizow - Thursday, May 7, 2015 - link
Again, not accurate. If you wanted an Athlon 64 you could not find one for less than $400. Only Socket A based Athlon. It was a shocking role reversal for AMD and their fans since they went from no CPUs over $400 to none under 400 for 64bit. It's like they gave their fans the middle finger once they got the lead!jimmy$mitty - Thursday, May 7, 2015 - link
AMD did have equal to or higher pricing during those days. When the Core 2 series hit though, AMDs parts were still higher than Intels parts. Then again at the same time you could hve bought a Pentium D 805 for $150 bucks, gotten a good air cooler and overclocked it to levels where it beat the top end Pentium EE and Athlon 64 FX for about 1/5th of the price.One reason why was that they didn't have enough product for demand because they actually closed down their last American based FAB just before launching the Athlon 64 line and their retail channels were low. That caused some to price higher than MSRP, much like how the HD7970GHz was running for about $1K if I remember correctly because of the bitcoin craze that caused them to be over valued and short on supply.
That said, my main point is that if AMD has a great performing part we will see prices equal to or higher than Intel. That is how it works. They are not going to have a better performing part and give it away, especially when they have mostly bled money the past 5+ years.
Same goes for every company. They are all out to make a profit, that is after all the idea behind capitalism.
medi03 - Thursday, May 7, 2015 - link
BS.Prescott was sloer than Athlon 64, consumed more power and also cost more.
jimmy$mitty - Friday, May 8, 2015 - link
I never said it was faster than Athlon 64. I said that when AMD has a performance advantage, it will cost more or will be equal in price. The FX-60 launched at $1031 (was probably $1100 retail). At the same time the top end Intel was the Pentium EE 955 which was $999. The FX-60 was a better CPU.The FX-62 was higher priced than that when it and AM2 launched in May of 2006, around $1400, and yet when Core 2 Duo launched in July the E6700 was only $530 (at least half the price) and it outperformed it by quite a bit.
You can find all of this information in reviews from Anandtech via Google easily.
It isn't a AMD sucks, more a AMD is a business out to make money.
If Zen performs equal to or better than Intel it will be priced equal to or higher. That is how a business runs.
Notmyusualid - Wednesday, May 6, 2015 - link
From what I can see - this company has quite a fan-base.If they price them all out again, it won't be forgotten.
Me, I buy the highest spec price CPU available, regardless of who makes it. But sometimes just not on initial launch, saving myself *a little* cash in the process.
If AMD makes the fastest chip, that is what I will buy.
Thermogenic - Thursday, May 7, 2015 - link
Of course, Haswell will be two generations old by then. AMD has a rough road ahead of them, but I wish them the best. Competition is always good.tipoo - Wednesday, May 6, 2015 - link
40% IPC over Excavator, so 1.68x Piledriver in total? That's Haswell territory, if not above it...Hm.chizow - Wednesday, May 6, 2015 - link
Assuming same clockspeeds, which is a big if and a risk AMD took in releasing that graph. They just have to be careful because if its 40% higher IPC and 25% lower clocks, that's not going to cut it, people are going to be bigtime disappointed because all they'll remember are the claims of 40% higher IPC.tipoo - Wednesday, May 6, 2015 - link
There's that, if Zen has a shorter pipeline I'd expect lower factory clocks. But the other side of it being Bulldozer never hitting the target clocks it was meant for anyways, unless you wanted that 220W CPU.ppi - Thursday, May 7, 2015 - link
If they want to stuff 8 cores into 95W power envelope, they would likely have lower single-thread performance than Intel's offering.Thing is, for most tasks (except pre-DX12 gaming), single-threaded performance is overrated, as the computer is usually responsive enough.
Then, 8 Zen cores with lower 25% lower clock speed compared to 4 Skylake cores (even if they have 15% better IPC than Haswell), would show its teeth on workloads, where performance matters. (assuming AMD and especially their foundry partners deliver)
tviceman - Wednesday, May 6, 2015 - link
2x perf/watt increase for GCN GPU tech moving from 28nm to finfet sounds good on the surface, but that would only just be slightly surpassing Maxwell's efficiency on 28nm.medi03 - Thursday, May 7, 2015 - link
Any links to support such bold claims? Last time I've checked there was 300w vs 370w difference (total system performance). And CPU was hardlly about 100w in that picture.chizow - Wednesday, May 6, 2015 - link
Can't say there's too much that's interesting on there other than Zen in 2016. Q2 2015 Graphics parts will go a long way in determining if AMD makes it to 2016. If they can make it to Zen and their FinFet graphics parts and they're good, maybe they'll pull out of this funk.But the odds are certainly stacked against them at this point.
My biggest question for them in 2015 is, what are they possibly going to fab (and sell) at GloFo this year to satisfy that $1Bn take or pay wafer commitment?
FlushedBubblyJock - Thursday, May 7, 2015 - link
They can make some ps5 and xbone "metal" and some AMD spider ramJumangi - Wednesday, May 6, 2015 - link
Blah blah...were AMD and we will be be releasing some really cool stuff...eventually....just wait guys...please?So sad how far AMD has fallen in recent years...
jwcalla - Wednesday, May 6, 2015 - link
Why do I have a sinking feeling that these things will only be competitive against two-generation old Intel tech?Peichen - Wednesday, May 6, 2015 - link
So anything worth buying will come after 2016? I was looking forward to R9 390 series to see if it will be as good as GTX 970 but if AMD is aiming to match Maxwell's efficiency in 2016, R9 390 would either be hotter or slower.Black Obsidian - Thursday, May 7, 2015 - link
The 390 being hotter than Maxwell isn't that big a deal. There's a whole continuum in there, from "just barely hotter " to "OMGWTFBBQGTX480". Only the later would be worthy of concern in a performance-oriented desktop.medi03 - Thursday, May 7, 2015 - link
Even R9 290x beats GTX 970vFunct - Thursday, May 7, 2015 - link
Does AMD really need to have an "Analysts day" talk that gives away their roadmap and plans to their competitor Intel?They really need to learn from Apple and the A7 and just do everything in secret and only announce it when it's done. Apple's competitors are still stunned and trying to react from that announcement.
AMD needs to spend less time giving away their roadmap, and more time just building the product.
Their "roadmap" should just be product releases. It would help if they just disappeared for 2 years to work on these products. Come back when its ready.
FlushedBubblyJock - Thursday, May 7, 2015 - link
They were all working on Mantle. mow that's dead, so now they're all working on Liquid VR.They all have their 3D goggles on all day and are "optimizing" between pron.
FlushedBubblyJock - Thursday, May 7, 2015 - link
They're all watching Star Trek in Liquid VR and speaking "Vulkan" to each other, so expect 1-2% speed optis for 3D, and some forlorn nostalgia for mantle.After all that they sit and contemplate their belly buttons, to focus, as zen is coming.
costeakai - Thursday, May 14, 2015 - link
happi - Thursday, May 7, 2015 - link
Company in AMD's situation cannot really afford to do that. Or to be more precise, its management, unless they want to be kicked out.See Intel, we also know about Skylake in this level of detail quite a bit ahead.
ppi - Thursday, May 7, 2015 - link
I must say, I like the overall strategy they laid out. I am very very cautiously optimistic.1) I am glad they said they are staying out of mobile for now. Until K12, they won’t really have anything extra to offer, and unless K12 scales well down to 2W, they won’t have it either. In that market, you either differentiate (Qualcomm – modems and custom cores), or have low costs using standard ARM architecture (MediaTek, Samsung – that also leverages its manufacturing edge).
To support their view, just look at nVidia (that just announced it leaves the market) and Intel (that 100% subsidizes its products, just to get foothold – the failure went so far, that they have hidden mobile segment in their segment reporting)
2) Releasing Zen as performance part first is a good idea, because that is market where (i) there are best margins, and (ii) they still have a lot of credibility in that market or at the very least they cooperate with partners, that could help them to some design wins. Therefore, if they manage to offer parts, that are competitive with Intel’s upper-midrange, it might be quite a win for them.
My prediction is, that if they stuff 8 cores in 95W envelope, they will likely have lower single-threaded performance than Intel’s best, and multi-threaded may be better. Given how overrated single-threaded performance is (except for pre-DX12/Mantle games), it might be reasonable trade-off.
At the same time, those who remember K7 vs. PIII times will recall, how all of sudden was Intel able to squeeze quite a lot of performance from their chips (though it was due to 180nm upgrade vs 250nm).
The key question here are yields and actual quality of the manufacturing process. If, and that is a big if, it works well, AMD finally won’t have such a huge process disadvantage now. But AMD just has to take this bet, the alternative is really going bust this time.
3) AMD’s APUs always had better performance than Intel’s, but let’s be honest, they never really had punch to play modern games and higher settings. Now with HBM and hopefully more efficient architecture overall, they may be able to release APUs, that will be truly gaming-capable. This may secure them some design wins, assuming the CPU part is not resource hog (we will see Carizzo).
jabber - Thursday, May 7, 2015 - link
I'll wait for disappointment day...oh sorry, I mean launch day!Atari2600 - Thursday, May 7, 2015 - link
A number of thoughts:1. Glad to see the f**kwits on the AMD board finally realise Dirk Meyer was right. Racing into low margin areas was foolish in the extreme, they burned a load of cash and time on it and its not sure if the company can recover.
2. 40% IPC on the BD series isn't great. That would leave it commensurate with Intel's 2011-2012 level IPC.
3. Will Zen actually be buyable in calendar year 2016? Or is it going to slip to "sometime" in 2017?
4. The HBM concept is great, for APU, GPU and even HPC dedicated CPUs, unfortunately due to (1) they don't have the resources to execute Zen and HBM concurrently.
5. GF also needs to deliver, while not GF at the time, it was the 65 nm process that screwed Barcelona, the TLC was a side-issue. Moving to 45nm went a long way to getting performance parity with Intel's best at the time. Unfortunately by then, the damage was done.
6. If Intel's process transitions are slowing up, then that does improve GF et al's chances of catching up.
7. I hope they make it. The market needs them to make it. [I've an old PII x6 I'd like to replace, ideally with Zen - I always go for the underdog!!]
Peichen - Thursday, May 7, 2015 - link
Don't buy AMD CPU again. I don't know if you have a comparable Intel platform to compare with your PII x6 or not but I do. I had a C2Q Q6600 and a PII X4 925. I got the Q6600 to run at 3.0GHz at reduced voltage and X4 925 at 3.2GHz at stock voltage both on $100ish Gigabyte mATX board. The Intel platform is more responsive and overall gives me less problem because Intel RAID is more stable. I have since replaced the Q6600 with 3770K and will eventually replace the X4 925 with an Intel.Atari2600 - Friday, May 8, 2015 - link
Funnily enough, I did have a Q6600, still have it indeed, but routing memory through the FSB is a killer for what I need to do from time to time. In this, even a 65nm Barcelona was faster than Harpertown never mind Clovertown!mikato - Monday, May 11, 2015 - link
Uh that's a worthless comparison. What does RAID have to do with it?Here's one - I have a Phenom II X4 965 machine and a Core i5 4670k machine and I don't notice any difference with general responsiveness in usage, applications and of course gaming (which is mostly GPU bound for me).
SydneyBlue120d - Thursday, May 7, 2015 - link
I wonder if a custom version of Zen could be used for the next generation Nintendo platform...stephenbrooks - Tuesday, May 19, 2015 - link
It wasn't called out in the article, but anyone notice "Disruptive Memory Bandwidth" under "Next Generation Opteron" (a different column from ARM)? That sounds like HBM on an x86 AMD CPU in 2016-17.beck2050 - Tuesday, May 26, 2015 - link
This really was barely an article, just a recycling of marketing materials. Whether AMD can ever survive let alone return to profitability is very much open to question.