AMD CPUs, SoC Rumors and Speculations Temp. thread 2

[juanrga]

Considering PD also sampled at this stage at 2.4, I am not concerned that a ceiling will arise...however...what I hear is optimism that they are trying to hit 3.6 base. What I am hearing is that they expect to be within 300 MHz of that target assuming current trends continue.

I expected 4.0 GHz in 2014. That was what I was told was their goal, at that time. Things change, and expectations change, and at this time, the current information I have says they expect to be close to 3.6 + turbo on the top SKU in HEDT. Server parts will be lower...in fact, I would be amazed if they hit 3.6 on a 16c/32t part...I would expect server chips to be much closer to 2.8-3.1. Even if you review the older opterons, lots of those were still relatively low clocks on that uarch with higher core counts.

Maybe they expect ~3.6 on the top 4-core option? Not even Intel has 3.6 GHz on their 8-core CPUs.

http://www.intel.com/buy/us/en/product/components/intel-core-i7-6900k-32ghz-20mb-smart-cache-box-515529

They do hit 3.2

But not in 14LPP on a 95W socket.

 

[juanrga]

The cinebench single thread scores for Zen are 50% higher than PD.

If we take AMD at face value and assume that the 40% number is IPC over carrizo. That puts the PD process about another 20% behind that. So, we are looking at 60% difference on uarch alone. Now, we consider that the IPC gains are 60%, and yet we have a 50% straight improvement in cinebench single thread...?

10% Clock speed deficit is the reason. This puts Zen's flagship part at 4.0*0.9 = ~3.6 GHz

Math is math.

Since we have confirmed that AMD only has 2.4 ghz Zen silicon, I don't see how there could have been a Cinebench score for a 3.6ghz Zen processor. Based on what we know this processor doesn't exist.

 

[-Fran-]

There's just one thing that doesn't fit the picture in my head for the clocks thing: engy samples low clocked.

What sort of engy samples did they give? Opteron? Did they gave the very first, un-tweaked versions? Was it a consumer Engy sample (as in, not fully validated)?

I think engy samples, while providing a good baseline for a clock-per-clock analysis, they do not represent final silicon in the slightest. And in that topic, has anyone found or seen any clock-per-clock metric leaked for Zen vs PD (or any CPU, really xD)?

Cheers!

 

[Cazalan]

There's just one thing that doesn't fit the picture in my head for the clocks thing: engy samples low clocked.

What sort of engy samples did they give? Opteron? Did they gave the very first, un-tweaked versions? Was it a consumer Engy sample (as in, not fully validated)?

I think engy samples, while providing a good baseline for a clock-per-clock analysis, they do not represent final silicon in the slightest. And in that topic, has anyone found or seen any clock-per-clock metric leaked for Zen vs PD (or any CPU, really xD)?

Cheers!

The engineering samples just need to be good enough to validate their motherboards and start testing for platform compatibility issues. Clock speeds aren't super important initially as there is much to do already. There are so many I/O that need validating. SATA/USB/PCIe to name a few. And of course the BIOS firmware that needs validating.

Prior to launch they will of course need higher clocked samples to make sure the motherboards can handle the power delivery and thermal requirements. Although that is probably not as big of a concern these days as the power estimation tools have come a long way there. The key players have extensive experience in selecting proper regulators and multi-phase layouts for a range of motherboards (cheap to ultra gamer extreme).

 

[sleeper357]

The cinebench single thread scores for Zen are 50% higher than PD.

If we take AMD at face value and assume that the 40% number is IPC over carrizo. That puts the PD process about another 20% behind that. So, we are looking at 60% difference on uarch alone. Now, we consider that the IPC gains are 60%, and yet we have a 50% straight improvement in cinebench single thread...?

10% Clock speed deficit is the reason. This puts Zen's flagship part at 4.0*0.9 = ~3.6 GHz

Math is math.

Since we have confirmed that AMD only has 2.4 ghz Zen silicon, I don't see how there could have been a Cinebench score for a 3.6ghz Zen processor. Based on what we know this processor doesn't exist.

could be the same as carrizo being confirmed only as a 2.5ghz part and later releasing a 3.5ghz chip?

 

[Cazalan]

The sample may run at 2.4 on air but who knows what it runs at on a high end peltier cooler or LN2 setup. If they want to characterize higher clocks they have a multitude of options.

 

[8350rocks]

The cinebench single thread scores for Zen are 50% higher than PD.

If we take AMD at face value and assume that the 40% number is IPC over carrizo. That puts the PD process about another 20% behind that. So, we are looking at 60% difference on uarch alone. Now, we consider that the IPC gains are 60%, and yet we have a 50% straight improvement in cinebench single thread...?

10% Clock speed deficit is the reason. This puts Zen's flagship part at 4.0*0.9 = ~3.6 GHz

Math is math.

Since we have confirmed that AMD only has 2.4 ghz Zen silicon, I don't see how there could have been a Cinebench score for a 3.6ghz Zen processor. Based on what we know this processor doesn't exist.

comparing clock to clock ES, that processor does exist....

 

[8350rocks]

The sample may run at 2.4 on air but who knows what it runs at on a high end peltier cooler or LN2 setup. If they want to characterize higher clocks they have a multitude of options.

Precisely this...

 

[jimmysmitty]

The cinebench single thread scores for Zen are 50% higher than PD.

If we take AMD at face value and assume that the 40% number is IPC over carrizo. That puts the PD process about another 20% behind that. So, we are looking at 60% difference on uarch alone. Now, we consider that the IPC gains are 60%, and yet we have a 50% straight improvement in cinebench single thread...?

10% Clock speed deficit is the reason. This puts Zen's flagship part at 4.0*0.9 = ~3.6 GHz

Math is math.

Since we have confirmed that AMD only has 2.4 ghz Zen silicon, I don't see how there could have been a Cinebench score for a 3.6ghz Zen processor. Based on what we know this processor doesn't exist.

could be the same as carrizo being confirmed only as a 2.5ghz part and later releasing a 3.5ghz chip?

Doubtful. Not the same process node and I doubt they will get more clock speed on the same node.

The sample may run at 2.4 on air but who knows what it runs at on a high end peltier cooler or LN2 setup. If they want to characterize higher clocks they have a multitude of options.

Neither of those cooling solutions mean anything though. Water cooling, maybe. Peltier is not too far off but has never been used in a normal setup and LN2 is just not indicative of anything since it is not viable and is why when people got all uppity about 8GHz Phenoms I didn't care, nor did I care about 8GHz Pentium 4s, because it is not a viable solution for anyone.

Air is the one that matters the most since that is what the majority uses so whatever the top clock speed on air is is what will determine the base and turbo clocks. Water is the second most but will not determine anything more than maximum enthusiast overclocks.

If Zen is clocked at 2.4GHz it is not the end of the world unless they are on par with Intel per clock with too old of a generation of CPU, especially since Intels 14nm is hitting base clocks of 4GHz.

 

[juanrga]

My original doubt about that 50% higher Cinebench score isn't solved. I am now reading two different (and contradictory) replies. One of them suggests that the sample was overclocked at 3.6GHz using extreme cooling. The other reply says that, on the contrary, the FX-8350 was underclocked to 2.4GHz to match the frequency of the sample.

Which is the valid reply?

 

[juanrga]

The cinebench single thread scores for Zen are 50% higher than PD.

If we take AMD at face value and assume that the 40% number is IPC over carrizo. That puts the PD process about another 20% behind that. So, we are looking at 60% difference on uarch alone. Now, we consider that the IPC gains are 60%, and yet we have a 50% straight improvement in cinebench single thread...?

10% Clock speed deficit is the reason. This puts Zen's flagship part at 4.0*0.9 = ~3.6 GHz

Math is math.

Since we have confirmed that AMD only has 2.4 ghz Zen silicon, I don't see how there could have been a Cinebench score for a 3.6ghz Zen processor. Based on what we know this processor doesn't exist.

could be the same as carrizo being confirmed only as a 2.5ghz part and later releasing a 3.5ghz chip?

I was asking something different. I was asking how a CB score could be measured at 3.6GHz when it was confirmed that the Zen silicon samples are clocked at 2.4GHz.

 

[salgado18]

If Zen is clocked at 2.4GHz it is not the end of the world unless they are on par with Intel per clock with too old of a generation of CPU, especially since Intels 14nm is hitting base clocks of 4GHz.

I think it is better for this discussion if we use the clocks of the 8-core Intel cpu against 8-core Zen, which is 3.2GHz base and 3.7GHz turbo (i7-6900k). We can't mix 8-cores from AMD against 4-cores from Intel.

 

[-Fran-]

I think it is better for this discussion if we use the clocks of the 8-core Intel cpu against 8-core Zen, which is 3.2GHz base and 3.7GHz turbo (i7-6900k). We can't mix 8-cores from AMD against 4-cores from Intel.

Why not?

I think we should mix them all and compare them all, haha.

Cheers!

 

[8350rocks]

I heard something interesting today...apparently, on the 14LPP node, I am hearing they pick up approximately 40% more clock speed for the same power draw over the 28nm node on GPUs.

Obviously...that is not a direct comparison to the 32nm PD-SOI node...and clocks on GPUs are typically much lower than CPUs; however...that does bode well for the prospect of clock speed across the board.

 

[8350rocks]

My original doubt about that 50% higher Cinebench score isn't solved. I am now reading two different (and contradictory) replies. One of them suggests that the sample was overclocked at 3.6GHz using extreme cooling. The other reply says that, on the contrary, the FX-8350 was underclocked to 2.4GHz to match the frequency of the sample.

Which is the valid reply?

I was not given specifics of the benchmark...or how the comparison was obtained.

However...both are possible options...they could have used any cooling solution to reach a specific clockspeed, they could have also underclocked as well...they could have even done both.

Both are valid options...and if you are comparing clock for clock, the manner in which they made the clocks equivalent is unimportant.

 

[-Fran-]

I heard something interesting today...apparently, on the 14LPP node, I am hearing they pick up approximately 40% more clock speed for the same power draw over the 28nm node on GPUs.

Obviously...that is not a direct comparison to the 32nm PD-SOI node...and clocks on GPUs are typically much lower than CPUs; however...that does bode well for the prospect of clock speed across the board.

If you're making a parallel between the GPU process and CPU process, then I think it's not entirely valid. The 16nm FinFET process TSMC and GloFo are using is aimed to sub-2Ghz speeds. Zen should be way above that speed. Saying it in another way, using the 1080 as example, it can go up to 2.2Ghz easily, but if you think about that speed for Zen, it doesn't sound as good 😛

Cheers!

 

[8350rocks]

I heard something interesting today...apparently, on the 14LPP node, I am hearing they pick up approximately 40% more clock speed for the same power draw over the 28nm node on GPUs.

Obviously...that is not a direct comparison to the 32nm PD-SOI node...and clocks on GPUs are typically much lower than CPUs; however...that does bode well for the prospect of clock speed across the board.

If you're making a parallel between the GPU process and CPU process, then I think it's not entirely valid. The 16nm FinFET process TSMC and GloFo are using is aimed to sub-2Ghz speeds. Zen should be way above that speed. Saying it in another way, using the 1080 as example, it can go up to 2.2Ghz easily, but if you think about that speed for Zen, it doesn't sound as good 😛

Cheers!

True...however...28nm SHP was designed to promote faster clocks, and yet worked out poorly. My point is, there are preconceived notions about it being a poor process for clockspeed...we can toss preconceived notions aside.

 

[-Fran-]

True...however...28nm SHP was designed to promote faster clocks, and yet worked out poorly. My point is, there are preconceived notions about it being a poor process for clockspeed...we can toss preconceived notions aside.

Now there's an idea!

You have the characteristics of the 14nm node from GloFo and once Polaris gets the official spec and power information, using the transistor count and density, we could make a closer parallel to Zen's top clocks by then.

Having a less dense process for a given die area, would allow faster clocks and everything else equals, it should give us a close estimation on where Zen might land, right?

Now we just need to find the formula, haha.

Cheers! 😛

 

[jimmysmitty]

If Zen is clocked at 2.4GHz it is not the end of the world unless they are on par with Intel per clock with too old of a generation of CPU, especially since Intels 14nm is hitting base clocks of 4GHz.

I think it is better for this discussion if we use the clocks of the 8-core Intel cpu against 8-core Zen, which is 3.2GHz base and 3.7GHz turbo (i7-6900k). We can't mix 8-cores from AMD against 4-cores from Intel.

And why not? We currently do, FX 8350 vs an i7 or even an i5.

I agree though as that will be the market they will try to compete in but even so Intel has a clear clock speed advantage if the 2.4GHz is a top end clock speed. We wont know until closer to release.

Of course the real fun is in the fact that this has done nothing but fuel more rumors and it will eventually fuel more until release causing more crap than needs be.

 

[jdwii]

"http://www.extremetech.com/extreme/229725-report-claims-intel-amd-will-both-delay-next-gen-processors-to-early-2017"

"Report claims Intel, AMD will both delay next-gen processors to early 2017"

 

[8350rocks]

"http://www.extremetech.com/extreme/229725-report-claims-intel-amd-will-both-delay-next-gen-processors-to-early-2017"

"Report claims Intel, AMD will both delay next-gen processors to early 2017"

That link would not work unless I copied it from the quote window and pasted...not sure why...but...I had to get it to work that way.

As for what they are saying, it makes some sense. It would not behoove AMD to launch at the end of november. Nobody competes well with Santa Claus, and their new product needs to splash pretty big to drive momentum. Intel, well...who knows their motivations?

Either way, I have been hearing early Q4 2016/early Q1 2017 for some time...so it is not a big surprise.

 

[jdwii]

"http://www.extremetech.com/extreme/229725-report-claims-intel-amd-will-both-delay-next-gen-processors-to-early-2017"

"Report claims Intel, AMD will both delay next-gen processors to early 2017"

That link would not work unless I copied it from the quote window and pasted...not sure why...but...I had to get it to work that way.

As for what they are saying, it makes some sense. It would not behoove AMD to launch at the end of november. Nobody competes well with Santa Claus, and their new product needs to splash pretty big to drive momentum. Intel, well...who knows their motivations?

Either way, I have been hearing early Q4 2016/early Q1 2017 for some time...so it is not a big surprise.

I'm sorry about that i'm using chrome and i link the site anyways yeah i expected a delay myself anyways, maybe not on the Intel end but then again its not like we are really missing something big from Intel in terms of CPU performance

 

[Cazalan]

True...however...28nm SHP was designed to promote faster clocks, and yet worked out poorly. My point is, there are preconceived notions about it being a poor process for clockspeed...we can toss preconceived notions aside.

Now there's an idea!

You have the characteristics of the 14nm node from GloFo and once Polaris gets the official spec and power information, using the transistor count and density, we could make a closer parallel to Zen's top clocks by then.

Having a less dense process for a given die area, would allow faster clocks and everything else equals, it should give us a close estimation on where Zen might land, right?

Now we just need to find the formula, haha.

Cheers! 😛

One should also take into consideration that AMD hasn't made a pure X86 CPU on a new process since Bulldozer. Everything done at the 28nm node was either a GPU or an APU. They had to make compromises to meet in the middle for standard cell libraries that would be good for both but not ideal for either. With Zen they can choose cells optimal for a pure CPU.

 

[cdrkf]

If Zen is clocked at 2.4GHz it is not the end of the world unless they are on par with Intel per clock with too old of a generation of CPU, especially since Intels 14nm is hitting base clocks of 4GHz.

I think it is better for this discussion if we use the clocks of the 8-core Intel cpu against 8-core Zen, which is 3.2GHz base and 3.7GHz turbo (i7-6900k). We can't mix 8-cores from AMD against 4-cores from Intel.

And why not? We currently do, FX 8350 vs an i7 or even an i5.

I agree though as that will be the market they will try to compete in but even so Intel has a clear clock speed advantage if the 2.4GHz is a top end clock speed. We wont know until closer to release.

Of course the real fun is in the fact that this has done nothing but fuel more rumors and it will eventually fuel more until release causing more crap than needs be.

If I can make 1 prediction- Zen 'top' clock speed isn't going to be 2.4ghz. Looking at this logically, there'd be 0 point in AMD developing and launching an 'almost as high IPC' as Intel with a serious clock speed deficit. That wouldn't gain them anything from where they are now.

Now, I can accept that base clocks could possibly be sub 3ghz (at least on the 8 core parts), however in that case I expect aggressive boost frequencies. I'd also think we'd see higher tdp parts than 95w if the clocks are that low (I mean again AMD have proven they'll push power in order to be performance competitive repeatedly in the past, typically going up to 140W if we ignore the 220w 9590).

AMD are expecting to be 'competitive' with equivalent Intel parts- that to me suggests we should see something close to Intels 8 core. That would suggest base clock in the low 3ghz ballpark. I mean the *current* engineering samples are at 2.4ghz already, and engineering samples are always pretty low clocked compared to a consumer product.

Edit: To put this another way doing a little bit of *very rough math*

The FX 8350 scores 1.1 in the Cinebench 11.5 single thread test. Assuming (generously) that Zen IPC is 1.6x that of Vishera (as the 40% is above Excavator), then at the same speed (4ghz) Zen would score a respectable 1.76 points.

Now if we state that the maximum frequency for Zen is just 2.4 ghz, then the Zen ST score would be just 1.06 points, i.e. slower in real terms than the current FX cpu. That wouldn't be a worthwhile product for them to launch.

 

[juanrga]

"http://www.extremetech.com/extreme/229725-report-claims-intel-amd-will-both-delay-next-gen-processors-to-early-2017"

"Report claims Intel, AMD will both delay next-gen processors to early 2017"

From the source used by extremetech:

both CPU makers have recently postponed their launch schedules to January 2017 at CES 2017

Does this mean that Zen will be not competing with Broadwell-E but with Skylake-E?