Why is AMD's Zen more efficient than intel's Core?

[juanrga]

15% behind Haswell? Where? Let alone with SMT. From my perspective it has almost the Skylake IPC and ~10% better SMT, if i compare it to my i5 and some random 6900k from GB4 data base.

~10% behind Broadwell-E clock-for-clock on applications and ~20% behind Broadwell-E on games. Both averages

 

[SonDa5]

Better designed electrical circuitry for flow of electrons to switch 0s and 1s. Better IHS with gold to transfer heat and lower circuit resistance.

I think the R5 series will perform even better than R7 core for core.

 

[NKD]

~10% behind Broadwell-E clock-for-clock on applications and ~20% behind Broadwell-E on games. Both averages

You have some outlier games. Its no where 20% slower across the board.

See for yourself. There are few games that make it look worse than it is. This is against 7700k running at 4.2 that probably has turbo on. It is no where across 20% slower in every game. Average could be skewed because come outlier games. Like fallout 4 clearly there is something up there when you are getting those frame rates may be some optimization can be done.

https://www.techpowerup.com/reviews/AMD/Ryzen_7_1800X/10.html

 

[lolfail9001]

~10% behind Broadwell-E clock-for-clock on applications and ~20% behind Broadwell-E on games. Both averages

Source first claim.

Better designed electrical circuitry for flow of electrons to switch 0s and 1s. Better IHS with gold to transfer heat and lower circuit resistance.

I think the R5 series will perform even better than R7 core for core.

Not sure if serious.

 

[SonDa5]

R5 less impedance, less load, less heat, same IHS thermal transfer design. Should show improved electrical quality/continuity performance..... Even at 1080p gaming.

 

[IdiotInCharge]

I'm not sure why we're talking about efficiency at all in terms of gaming: we've already seen that it's close enough to not give a shit, in terms of gaming.

Process efficiency (workload/watt) is what we need to be getting at there, and that hasn't been touched on from a reviewer's perspective yet, so at best we have educated guesses. And this really matters most in servers, which current Ryzen CPUs will not be populating directly. It matters a little bit in mobile too, but again, these CPUs won't be going in the kind of laptops where people care about battery life (some may find a home in DTRs).

 

[JustReason]

~10% behind Broadwell-E clock-for-clock on applications and ~20% behind Broadwell-E on games. Both averages

Yeah SOURCES would be awesome, although as with any stand I am sure there is some asinine review that will support your claims.

By the by, you can not compare final results as proof of SMT/HT %. You have to show with off against with on. This is shown in a number of other reviews and coverage and it looks a lot like this:

https://forums.anandtech.com/threads/ryzen-strictly-technical.2500572/

Choose the common ones used in most reviews here, there are way more with varying degrees of effectiveness all over the board (oddly Haswell shows better than Skylake in a few).

 

[JustReason]

I'm not sure why we're talking about efficiency at all in terms of gaming: we've already seen that it's close enough to not give a shit, in terms of gaming.

Process efficiency (workload/watt) is what we need to be getting at there, and that hasn't been touched on from a reviewer's perspective yet, so at best we have educated guesses. And this really matters most in servers, which current Ryzen CPUs will not be populating directly. It matters a little bit in mobile too, but again, these CPUs won't be going in the kind of laptops where people care about battery life (some may find a home in DTRs).

Actually the Stilt at the above link I posted comments on Ryzen @ 35 W and its performance is mind blowing. Unfortunately he doesn't comment much more than his surprise.

 

[juanrga]

You have some outlier games. Its no where 20% slower across the board.

See for yourself. There are few games that make it look worse than it is. This is against 7700k running at 4.2 that probably has turbo on. It is no where across 20% slower in every game. Average could be skewed because come outlier games. Like fallout 4 clearly there is something up there when you are getting those frame rates may be some optimization can be done.

https://www.techpowerup.com/reviews/AMD/Ryzen_7_1800X/10.html

Funny, your negation of the gaming performance is accompanied with a link to a review that didn't even test the chip was mentionied.

 

[NKD]

Funny, your negation of the gaming performance is accompanied with a link to a review that didn't even test the chip was mentionied.

WTF? Anything faster at gaming than 7700k? You seriously like to twist shit and go around in circles. So all of sudden broadwell-e is faster at gaming than 7700k? Seriously stop trolling.

 

[juanrga]

Source first claim.

Same source I used above to demonstrate that Broadwell 6900k is more efficient than Ryzen 1800X. Performance clock-for-clock is

"Moyenne applicative" means average of applications. I am taking the best case for RyZen, 235.8 points, which is 10% slower than Broadwell-E

 

[IdiotInCharge]

Actually the Stilt at the above link I posted comments on Ryzen @ 35 W and its performance is mind blowing. Unfortunately he doesn't comment much more than his surprise.

Then- actually- we're still waiting.

I'd like to see an Anandtech or TechReport style writeup here comparing energy usage for different workloads focused on how many watts were used to finish particular jobs. That's efficiency.

 

[juanrga]

WTF? Anything faster at gaming than 7700k? You seriously like to twist shit and go around in circles. So all of sudden broadwell-e is faster at gaming than 7700k? Seriously stop trolling.

I am not trolling. My claims are always backed up with sources. Clock-for-clock Ryzen is about 20% slower than Broadwell on games

Taking the best possible case for RyZen, 171.4 points place it a 22% behind Broadwell on games, clock-for-clock. I rounded down to 20%, because I was being generous...

TechSpot extensive review (16 titles with both SMT-on and off) found the same:

Based on the data above, Ryzen 7 1800X is on average 12% slower than the Core i7-6900K on gaming titles, which seems in line with what we found originally at 1080p.

Correcting for the difference in stock clocks we obtain again the ~20% gap.

 

[JustReason]

Same source I used above to demonstrate that Broadwell 6900k is more efficient than Ryzen 1800X. Performance clock-for-clock is

"Moyenne applicative" means average of applications. I am talking the best case for RyZen, 235.8 points, which is 10% slower than Broadwell-E

You still aren't posting a LINK. Jesus, for the third time where is this article So I Don't Have to search the green hills for it.

Besides I linked solid Findings with a link so you can educate yourself that shows SMT performance. Your insistence to dodge the questions laid before you tempts me to go post some other finding with links so everyone can read and understand the environment used. My guess is that set is pitting against 4 channel memory and you are using that as a boon for Intel to prove your point.

 

[Nobu]

Phoronix did some perf/watt testing, didn't really touch on efficiency, though I guess you could figure it out if you took the data and divided time of test by watts used.

 

[juanrga]

You still aren't posting a LINK. Jesus, for the third time where is this article So I Don't Have to search the green hills for it.

I have given the link dozen of times. I have mentioned this review another dozen of times. And people in this thread was able to find it also. Hint: "Hardware.fr"

 

[JustReason]

I have given the link dozen of times. I have mentioned this review another dozen of times. And people in this thread was able to find it also. Hint: "Hardware.fr"

It isn't that hard to do this:

http://www.pcworld.com/article/3176191/computers/ryzen-review-amd-is-back.html

They show AMD in most of their synthetics and productivity either even or ahead with the 6900k (again 4 channel), but behind in gaming.

https://www.hardocp.com/article/2017/03/02/amd_ryzen_1700x_cpu_review/3

Same here although they used max OCs and the 6900k is actually a 6950X with 2 cores off.

http://www.guru3d.com/articles_pages/amd_ryzen_7_1700_review,24.html

With some indepth testing again a link so one can easily reference any comment I make.

 

[NKD]

I am not trolling. My claims are always backed up with sources. Clock-for-clock Ryzen is about 20% slower than Broadwell on games

Taking the best possible case for RyZen, 171.4 points place it a 22% behind Broadwell on games, clock-for-clock. I rounded down to 20%, because I was being generous...

TechSpot extensive review (16 titles with both SMT-on and off) found the same:



Correcting for the difference in stock clocks we obtain again the ~20% gap.

Techpower up tested a lot of games as well and it seems to better than your claims. Plus I think what you are showing is 720p benches. I get your point.

 

[Formula.350]

Better IHS with gold to transfer heat and lower circuit resistance.

I just wanted to say that they only tinned the IHS with gold. The actual die-to-IHS solder was Indium.
At least, as per this: https://www.pcper.com/news/Processo...onfirms-AMD-Using-Solder-IHS-Ryzen-Processors

"It seems that AMD is using two small pads of Indium solder along with some gold plating on the inside of the IHS to facilitate heat transfer and allow the solder to mate with the IHS."
​

Sorry to correct. Just didn't want anyone thinking it might've been a literal gold-mine to scrap a Ryzen, or that folk start running around shouting "AMD soldered their IHS with gold, muh-fuggas!" lol

 

[lolfail9001]

You still aren't posting a LINK. Jesus, for the third time where is this article So I Don't Have to search the green hills for it.

It is trivial to notice it hardware.fr's review, from the language of labels.

Same source I used above to demonstrate that Broadwell 6900k is more efficient than Ryzen 1800X. Performance clock-for-clock is

See, that's why i hate averages.

We have weirdly disrepant results in 7zip, WinRAR and without investigation of reasons, using them for blind averaging makes little sense.

 

[FrozenSun]

Huh? He wrote "I do think it's incredibly impressive AMD produced a chip whose total performance is good and its efficiency is excellent. FX were turds in both departments." And I just explained in a reply to him why getting RyZen providing efficiency very superior to Piledriver doesn't have any special merit, because Piledriver was a speed-demon microarchitecture on a non-efficient 32PDSOI node.

Literally all I'm saying is that I think Ryzen is a solid overall chip that turned out better than expected. AMD hasn't been relevant in upper tier chips in a decade. As such, my expectations for them producing a good chip were low. They hadn't proven they could in a long, long time. I didn't expect it to compete with the 6900k in *any* workload. The fact that it does is surprising to me. The fact that it is also in the same ballpark of efficiency is also surprising. Your level of defensiveness to such a benign statement is honestly baffling. I'm merely impressed that, against their own track record of duds, they actually released a good chip.

 

[juanrga]

Plus I think what you are showing is 720p benches. I get your point.

Funny, because it is 1080p.

 

[juanrga]

We have weirdly disrepant results in 7zip, WinRAR and without investigation of reasons, using them for blind averaging makes little sense.

I don't find anything weird. RyZen, as any other chip, performs differently on different workloads. RyZen is faster in Komodo and x264, on pair on Stockfish, and slower on 7zip and WinRAR.

 

[juanrga]

Literally all I'm saying is that I think Ryzen is a solid overall chip that turned out better than expected. AMD hasn't been relevant in upper tier chips in a decade. As such, my expectations for them producing a good chip were low. They hadn't proven they could in a long, long time. I didn't expect it to compete with the 6900k in *any* workload. The fact that it does is surprising to me. The fact that it is also in the same ballpark of efficiency is also surprising. Your level of defensiveness to such a benign statement is honestly baffling. I'm merely impressed that, against their own track record of duds, they actually released a good chip.

And as mentioned above, RyZen is better or worse than expected depending if the personal expectations of each one of us. It is very very close to all what I expected and predicted. But it can be different for other people:

• People that expected Skylake IPC will be disappointed because it is well-below Haswell.
• People that expected 60% better efficiency than 6900k will be disappointed because it is 15% behind.
• People that expected at least 4.5GHz on air will be disappointed by the difficulties to break 4GHz.
• People that expected Kabylake gaming performance will be disappointed because it is 15--20% behind Broadwell.

 

[JustReason]

And as mentioned above, RyZen is better or worse than expected depending if the personal expectations of each one of us. It is very very close to all what I expected and predicted. But it can be different for other people:

• People that expected Skylake IPC will be disappointed because it is well-below Haswell.
• People that expected 60% better efficiency than 6900k will be disappointed because it is 15% behind.
• People that expected at least 4.5GHz on air will be disappointed by the difficulties to break 4GHz.
• People that expected Kabylake gaming performance will be disappointed because it is 15--20% behind Broadwell.

HOLY CRAP BATMAN. You sure do make a lot of statements with no back up or links, or hell anything other than blind statements. Many a question has been asked of you but you refuse to answer. Then you state these "expectations" that no rational person expected.

RATIONAL people expected SB IPC. We got Haswell to Broadwell IPC ( Numerous links have already been shown to you and you have yet to acknowledge one).

RATIONAL people expected 3.5Ghz as the ceiling, again this is the one prediction that actually had merit based on the node limitations it had shown. We got 4.0/4.1Ghz, well above expectations.

RATIONAL people really had no gaming expectations really. Seriously in todays games most relative current CPUs including AMD can give adequate performance and if 60Hz gaming superior performance in most games.

Now if you are talking about that small number of individuals that state their wishes and hopes, that then individuals like you then run around parroting as the "expectations" so the results better fit your narrative, then well... You don't really have a point then do you?

 

[RAutrey]

The Ryzen bashing is laughable. Ryzen is like a nice assed truck whereas the 7700K is like a nice assed sports car. That sports car rocks at games but comparatively speaking isn't really practical. That truck is plenty fast enough in games AND you can haul a load and tow to boot.

I'll take the truck if you please.

 

[lolfail9001]

I don't find anything weird. RyZen, as any other chip, performs differently on different workloads. RyZen is faster in Komodo and x264, on pair on Stockfish, and slower on 7zip and WinRAR.

If you do not find anything weird, then you are not looking at all.

 

[TheLAWNoob]

The Ryzen bashing is laughable. Ryzen is like a nice assed truck whereas the 7700K is like a nice assed sports car. That sports car rocks at games but comparatively speaking isn't really practical. That truck is plenty fast enough in games AND you can haul a load and tow to boot.

I'll take the truck if you please.

I want an Audi S6 please. Too bad I'm driving this 2013 Ford Escape. 1.6L on a SUV.

 

[JustReason]

I want an Audi S6 please. Too bad I'm driving this 2013 Ford Escape. 1.6L on a SUV.

I am a Ford man thru and thru, but you have my sympathies. I am not a fan of the new Escapes.

 

[NKD]

Funny, because it is 1080p.

wait so the picture you posted is rocking 242 frames at 1080p in f1 and almost 200 in GTA V, 250+ in civilization VI? Very unrealistic, I highly doubt that, is it on super low settings or what, still pretty unbelievable.

 

[NKD]

And as mentioned above, RyZen is better or worse than expected depending if the personal expectations of each one of us. It is very very close to all what I expected and predicted. But it can be different for other people:

• People that expected Skylake IPC will be disappointed because it is well-below Haswell.
• People that expected 60% better efficiency than 6900k will be disappointed because it is 15% behind.
• People that expected at least 4.5GHz on air will be disappointed by the difficulties to break 4GHz.
• People that expected Kabylake gaming performance will be disappointed because it is 15--20% behind Broadwell.

You know you found a way to list all the negative in there and assumed that all people expected that.

I expected it to 20-25% slower than skylake at IPC, its not. Clock them at same clock and run it its about 5-10% in work loads.

I expected it to be slower in games then intel due to clock speed and how new the architecture is. It did well there and almost destroyed AMDs last chip, ofcourse I didn't expect it to meet intel standards.

I expected it to be 3.2ghz for top model. all the haters here said that, oh you will be lucky to get 3.5ghz after overclock. GF 14nm process is not designed for anything higher bla bla bla bla bla. Guess what we are getting 4.0 to 4.1 in some instances. Still surpassed my expectations and now everyone that wants AMD to fail and they can't put their ego aside are claiming that oh AMD pushed it too hard, ROFL!

It sounds like you are expecting everything for everyone, when most people actually didn't expect even what it delivered along with the clock speeds. So please don't speak for everyone.

 

[DuronBurgerMan]

The Ryzen bashing is laughable. Ryzen is like a nice assed truck whereas the 7700K is like a nice assed sports car. That sports car rocks at games but comparatively speaking isn't really practical. That truck is plenty fast enough in games AND you can haul a load and tow to boot.

I'll take the truck if you please.

Lulz. I bought a Ryzen... And I drive an impractical souped up 15 Mustang GT.

 

[DuronBurgerMan]

I had no expectations for Ryzen. Just hoped for competition, because Intel has kind of stagnated. Probably because they had no competition from AMD anymore.

Ryzen delivered there.

Did the math given my workload and gaming percentages, and Ryzen was a better buy than a 7700k for my use case. The 6900k would be a little better, though... But the price is stupid. So Ryzen won on price there.

I don't really care about this back and forth sh*t. I mean, buy it if it makes sense for you. Don't if it doesn't.

Sure puts my old 2600k to pasture either way. Especially in graphics and video rendering.

 

[NKD]

I had no expectations for Ryzen. Just hoped for competition, because Intel has kind of stagnated. Probably because they had no competition from AMD anymore.

Ryzen delivered there.

Did the math given my workload and gaming percentages, and Ryzen was a better buy than a 7700k for my use case. The 6900k would be a little better, though... But the price is stupid. So Ryzen won on price there.

I don't really care about this back and forth sh*t. I mean, buy it if it makes sense for you. Don't if it doesn't.

Sure puts my old 2600k to pasture either way. Especially in graphics and video rendering.

So true, but its hard to keep everyone at bay. Someone asks a questions you have 2 or 3 people that will come and give you every reason no matter what to not buy ryzen. That usually derails the thread. I think I responded to your thread as well and had my honest opinion, buy 7700k for brute force gaming but I did suggest in your case 1700 and overclocking it would be best bang for buck. But you did get the 1700x I believe. But you won't have these people actually respond in that thread because thats not what they are here for. I wont name people I will probably stop responding to them soon, because they are here for one thing tell you not to buy ryzen and how much it sucks.

 

[Shintai]

You have some outlier games. Its no where 20% slower across the board.

See for yourself. There are few games that make it look worse than it is. This is against 7700k running at 4.2 that probably has turbo on. It is no where across 20% slower in every game. Average could be skewed because come outlier games. Like fallout 4 clearly there is something up there when you are getting those frame rates may be some optimization can be done.

https://www.techpowerup.com/reviews/AMD/Ryzen_7_1800X/10.html

Is it CPU limited there? I think not. Dont get fooled by prescripted and low CPU usage benchmarks that's designed to test GPUs.

Examples:
http://gamegpu.com/action-/-fps-/-tps/rise-of-the-tomb-raider-v-rezhime-directx-12-test-gpu.html
http://gamegpu.com/action-/-fps-/-tps/rainbow-six-siege-test-gpu.html

 

[Shintai]

It is trivial to notice it hardware.fr's review, from the language of labels.


See, that's why i hate averages.

We have weirdly disrepant results in 7zip, WinRAR and without investigation of reasons, using them for blind averaging makes little sense.

You can add any benchmark with a small unrealistic workset to benefit a 512KB L2 or outdated software without AVX support. Or one a bottleneck elsewhere than the CPU.

 

[lolfail9001]

You can add any benchmark with a small unrealistic workset to benefit a 512KB L2 or outdated software without AVX support. Or one a bottleneck elsewhere than the CPU.

Call me when they add SIMD usage to compilers, Mr Outdated Software.

 

[Shintai]

Call me when they add SIMD usage to compilers, Mr Outdated Software.

Its already there. But then you can cheer on benchmark software that's still locked at SSE2 because of compare reasons.

 

[lolfail9001]

Its already there.

Oh, you know a compiler that uses FMA and/or AVX? No, not compiles code with it, but actually uses it?

 

[Pieter3dnow]

Source first claim.

That is so unfair ....