Intel engineers calls ryzen competitive

So this would indicate that an 8 core Ryzen is slightly smaller than a 4 core Skylake + IGP. I guess that makes sense since the IGP is about the same size as the 4 cores on a Skylake chip.
 
bc3ce7afd1b4.jpg
 
I expect that Zen will catch up in # of transistors per core in the Zen+ node. They still will be behind in IPC versus the current generation Intel. The good news is this time they are much closer to their competition.
 
It's a feat that they used one less metal layer and note the smaller cache footprint.
 
Whenever you end up mentioning your competition in your own marketing materials, even as a comparison effort, it's a major fail and yes Intel is now doing some quakin' and I don't mean the game series from id Software. :D
 
Whenever you end up mentioning your competition in your own marketing materials, even as a comparison effort, it's a major fail and yes Intel is now doing some quakin' and I don't mean the game series from id Software. :D
That is NOT marketing material. That is an internal training document.
 
Hyperthreading in an i5, new CPU bins on the high end, unlocked i3 CPUs, hyperthreaded pentiums, and Intel saying it's competitive. If AMD hadn't said a word I'd still expect something really good coming.
 
Hyperthreading in an i5, new CPU bins on the high end, unlocked i3 CPUs, hyperthreaded pentiums, and Intel saying it's competitive. If AMD hadn't said a word I'd still expect something really good coming.


Remember that the next generation mainstream Intel processors will have a 6C / 12 T i7. This has been in known for a long time now..
 
It's a feat that they used one less metal layer and note the smaller cache footprint.

Using one less metal layer can be a double edged sword. It saves a bit in production costs, but it can lead to higher operating temps. If you look back at prior AMD CPUs, the Thoroughbred A processors ran very hot. AMD reconfigured the design to add an additional metal layer to the design, which resulting in the Thoroughbred B CPUs that overclocked much better and ran cooler.
 
If you want to read some ACTUAL ARCHITECTURAL ANALYSIS of Zen, David Kanter's usual articles at Real World Tech are now at Microprocessor Report. Because they pay him now, they are usually behind a massive paywall.

LUCKILY, they made this one free to access :D

http://www.linleygroup.com/mpr/article.php?id=11666

Questions are answered on the Real World Tech Forums:

http://www.realworldtech.com/forum/?threadid=164709&curpostid=164717

Sounds very intriguing, might have to look at what they charge for a 6-core to upgrade my aging 2500k.
 
Using one less metal layer can be a double edged sword. It saves a bit in production costs, but it can lead to higher operating temps. If you look back at prior AMD CPUs, the Thoroughbred A processors ran very hot. AMD reconfigured the design to add an additional metal layer to the design, which resulting in the Thoroughbred B CPUs that overclocked much better and ran cooler.

It would be hard for AMD to do any worse than Intel on thermals right now. The new Intel chips and their toothpaste between the heat spreader and CPU die show they just don't give a fuck.
 
If you want to read some ACTUAL ARCHITECTURAL ANALYSIS of Zen, David Kanter's usual articles at Real World Tech are now at Microprocessor Report. Because they pay him now, they are usually behind a massive paywall.

LUCKILY, they made this one free to access :D

http://www.linleygroup.com/mpr/article.php?id=11666

Questions are answered on the Real World Tech Forums:

http://www.realworldtech.com/forum/?threadid=164709&curpostid=164717

Sounds very intriguing, might have to look at what they charge for a 6-core to upgrade my aging 2500k.

Interesting tidbit from that article:

The Linley Group said:
On the basis of our estimates, the 14nm Zen core should offer performance somewhere between that of Intel’s Ivy Bridge and Haswell generations on integer workloads. Although Zen-based processors cannot rival the latest Skylake core in high-end clients, AMD’s eight-core Summit Ridge chip should be a credible contender for midrange desktops.

So based on this article, it sounds like Intel will still hold the performance crown; but AMD has an opportunity to offer some very competitively priced chips with good performance. Since Ryzen is expected to be between Ivy Bridge and Haswell in performance then the 4c/8t will need to be cheaper than the 7700K ($350) and the 6c/12t cheaper than the 6800K ($420) as those two Intel CPUs will likely outperform AMD's offerings by 10% or so. If the 4c/8t Ryzen is $250 and the 6c/12t is $350 then I think AMD is going to have some really great chips on their hands.
 
Dont forget that the article was published last August. I don't know if enough information has been gathered for Kanter to offer a more updated conclusion, but we should aware that this is a little old.
 
Dont forget that the article was published last August. I don't know if enough information has been gathered for Kanter to offer a more updated conclusion, but we should aware that this is a little old.

Yeah, how does that jive with the tech demos of the Ryzen vs stock 6900K? Handbrake and Cinebench?

I know there are some factors that can affect those scores, but it seems odd if Ryzen is stuck between Ivy Bridge and Haswell that it went toe to toe with a similarly clocked Broadwell-E...
 
Dont forget that the article was published last August. I don't know if enough information has been gathered for Kanter to offer a more updated conclusion, but we should aware that this is a little old.

It's an architectural analysis done from their hotchips presentation. I just posted the analysis because they posted it for free recently.

David does a lot better architectural analysis than most other sites, so when he buys into the projected performance, I'm that much more of a believer.
 
Yeah, how does that jive with the tech demos of the Ryzen vs stock 6900K? Handbrake and Cinebench?

I know there are some factors that can affect those scores, but it seems odd if Ryzen is stuck between Ivy Bridge and Haswell that it went toe to toe with a similarly clocked Broadwell-E...

I'll wait for actual performance numbers before I believe results of tech demos presented at a conference from any company; especially AMD after how they presented Polaris.
 
I'll wait for actual performance numbers before I believe results of tech demos presented at a conference from any company; especially AMD after how they presented Polaris.

But you're already basing Ryzen IPC off of articles that aren't real world results? ¯\_(ツ)_/¯

So based on this article, it sounds like Intel will still hold the performance crown; but AMD has an opportunity to offer some very competitively priced chips with good performance. Since Ryzen is expected to be between Ivy Bridge and Haswell in performance

I also hope as you do that AMD offers a good bang for buck competition. But I think you're already writing it off as significantly inferior IPC made up for by core counts, let's just both wait till Kyle can try OCing the shit out of one :)
 
But you're already basing Ryzen IPC off of articles that aren't real world results? ¯\_(ツ)_/¯
The big difference is the Kanter article doesn't carry an obvious bias. A press conference presenting a new product, on the other hand, will obviously only contain positive things things to show.

But as you said, best wait for Kyle and other reviewers to get their hands on these and thoroughly test them before we draw any conclusions.
 
I am just waiting on Kyle's first review to show that they are a winner and I am driving to Microcenter immediately after reading the review. Yes I do not mind being an earliy adopter.

my 3930K had a flaw with VTd and a new revision was released. Intel stood behind it and swapped my chip with a new stepping. If AMD would do the same then I have no quams with being first adopter if there was something wrong with the chip eventually. Just make it right.
 
I'll eat my own hat if a Skylake 4c8t die is 44mm2. Where is this information from?
 
I am just waiting on Kyle's first review to show that they are a winner and I am driving to Microcenter immediately after reading the review. Yes I do not mind being an earliy adopter.

my 3930K had a flaw with VTd and a new revision was released. Intel stood behind it and swapped my chip with a new stepping. If AMD would do the same then I have no quams with being first adopter if there was something wrong with the chip eventually. Just make it right.

If Ryzen turns out to be solid then I may pick up the second generation. This is a completely new architecture for AMD and there is a good chance there will be bugs/hiccups. I just picked up a used i7 6850K for a great price that will hold me off for the next couple years to let AMD sort out any issues.
 
If Ryzen turns out to be solid then I may pick up the second generation. This is a completely new architecture for AMD and there is a good chance there will be bugs/hiccups. I just picked up a used i7 6850K for a great price that will hold me off for the next couple years to let AMD sort out any issues.
Sure you can wait ... and wait ... and wait...

But since you have a really good chip to hold you over you should be good to go.
 
I think that is the core part. Not the interconnects, memory controller, iGPU, L2 cache, L3 cache, etc.
It never really states that in the article.

The actual size of the quad-core skylake + GT2 die is 122.4 mm2.

Divide that into 4 (for four cores) and you get 30.1mm2. This is (falsely) assuming that the ENTIRE die is cores. No interconnects, IMC or graphics.

If you look at 'how much space 4 skylake cores occupy on the die', then we are looking at around an eyeballed 35%; closer to the '44mm2' number, but I can't find this number mentioned anywhere else, even in deep-dive architecture analysis pages.

So essentially, the mysterious '44mm2' measurement is entirely inaccurate and essentially pulled out of someone's ass, and so is this 'news'.

I deem this a FAYKE!!
 
Last edited:
This is coming from a paper only fab that transfers it out to TMSC or GF to produce. I will be shocked and surprised if it's anywhere near as good as intel's chips.
 
this graphic is most interesting:
U126_Zen_F2.png


Arranging these by year of release (2011-2017?) rather than brand I found to be a bit telling of both each company and the market. Not being very familiar with a lot of benchmarks how wrong would it be try to scale this one on clock speed? i.e. where ever that Zen figure has been taken from looks as though a Sandy Bridge at 3.7GHz could come close to it?
 
I would wholeheartedly agree with you except for the fact that it's leaked so, as soon as it hit the Internet it absolutely will be treated as marketing material even if Intel didn't mean for it to happen i.e. all the people really wanting Ryzen to succeed now know implicitly that Intel has something to worry about so they're going to lean towards the AMD purchase even more. ;)
LOL! You are funny.
 
I would wholeheartedly agree with you except for the fact that it's leaked so, as soon as it hit the Internet it absolutely will be treated as marketing material even if Intel didn't mean for it to happen i.e. all the people really wanting Ryzen to succeed now know implicitly that Intel has something to worry about so they're going to lean towards the AMD purchase even more. ;)

Or maybe this is just normal operations for Intel regarding new releases from their competitor, whose to say they didn't do this for Bulldozer when that came out? Best not take things too literally.

Well if AMD Ryzen is a hit at [H], great, more competition in the CPU market.
 
I'd be more concerned about bugs with the first generation motherboards than I would the CPU.

Haven't some of these been used for months with Brystol Ridge CPUs?
 
As of this morning you can't buy an AM4 motherboard off newegg. Maybe I'm missing something but to me that doesn't exactly spell out widespread adoption and testing.

Hoping for the best.
 
Back
Top