Ryzen+ 2600x and 2300x cinebench/Cpu-Z results leak (probably fake)

[sof006]

I need an excuse to change from my 4790K! Come on AMD, give me that excuse! Lol

[Trixanity]

13 hours ago, TheCherryKing said:

This is not about me. I'm not going to argue with you if you continue to make this argument personal. 

You argue the point there is no new technology but you can't answer what new technology means. Only say that every new piece of tech in Ryzen is not new. By that logic we haven't had new CPU tech in several decades because everything has pretty much been a reiteration since then. Unless we count instruction sets as new tech which would then mean we've had new tech on pretty much every new release from Intel and AMD but those are often just reiterations on existing work.

 

This is not personal at all. Just refuting misinformation and poking fun at the fact that you haven't said anything but "it's not new!!1111". You could at least have said what you consider a new technology after several posts of me answering your question to no avail. That's good etiquette.

 

13 hours ago, AresKrieger said:

It also is the main reason Ryzen has so many issues with memory speeds, I don't see amds methods to be a better option they are simply what they managed to make decent enough to compete with, now could they fix the issues with the design as launched, absolutely but you need to keep in mind that performance and reliability are all that matters everything else is just marketing bs

 

That has nothing to do with intel not having the capability to do so more so it wasn't a profitable endeavor for them to do so as it would cut into xeon sales to serve a market that has been niche (video editing and streaming market), however it is obvious that if the market changes due to AMD pushing new products in Intel will adapt out of necessity.

 

This is why competitive markets are generally a good thing.

Or maybe AMD's IMC is just so awful like it's been before IF?

 

So you're saying Intel couldn't repurpose Xeon CPUs without cutting deep into profits but AMD can? Sounds an awful lot like AMD's approach have given them an advantage yet you say Intel has the same capability? I mean yes they could in theory if they want to ruin their own business. But that's the entire point isn't it? Bringing core count down in price and AMD succeeded. 

[jagdtigger]

7 hours ago, Drak3 said:

Except it is true. Not once in history has an x86 MCM design performed at the same level as a monolithic design where the core and subsystems are effectively the same. Talk between the dies incurs latency, and there are bandwidth limitations, more so than the bus that allows CPU cores to communicate. Some tasks don't take a meaningful hit. Some tasks, like gaming, do. Ryzen comes close, but that's it.

 

It's awesome, but not new. Infinity Fabric is nothing more than a Mesh interconnect, and functions not dissimilarly than Intel's QPI or AMD's older HyperTransport when either were used for multiple NUMA systems.

 

Nothing on Ryzen is actually new. SenseMI, Infinity Fabric, XFR, MCM, are all AMD's latest iteration of existing technologies. And IME is not a present factor on all Intel CPUs, just like AMD's counterpart.

 

It's the equivalent of Dodge throwing a larger turbo and an after market super on a Cummins engine, and calling it the Cerberus. It's not a new engine.

I suggest to you read up on that one, its in every intel CPU since many years now regardless of which market the CPU is targeted for. Onto the limits of IF, with an updated kernel cross CCX talk can be reduced dramatically.

[Drak3]

35 minutes ago, jagdtigger said:

I suggest to you read up on that one, its in every intel CPU since many years now regardless of which market the CPU is targeted for.

I have. Intel ME is precent enough to say it's in virtually every mainstream CPU since '08 (whereas every x86 CPU AMD has made since '13 includes PSP/AMD Secure Technology), however, low power systems and server platforms have counterparts called TXE and SPS (respectively).

[TheCherryKing]

3 hours ago, Trixanity said:

You argue the point there is no new technology but you can't answer what new technology means. Only say that every new piece of tech in Ryzen is not new. By that logic we haven't had new CPU tech in several decades because everything has pretty much been a reiteration since then. Unless we count instruction sets as new tech which would then mean we've had new tech on pretty much every new release from Intel and AMD but those are often just reiterations on existing work.

 

This is not personal at all. Just refuting misinformation and poking fun at the fact that you haven't said anything but "it's not new!!1111". You could at least have said what you consider a new technology after several posts of me answering your question to no avail. That's good etiquette.

 

Or maybe AMD's IMC is just so awful like it's been before IF?

 

So you're saying Intel couldn't repurpose Xeon CPUs without cutting deep into profits but AMD can? Sounds an awful lot like AMD's approach have given them an advantage yet you say Intel has the same capability? I mean yes they could in theory if they want to ruin their own business. But that's the entire point isn't it? Bringing core count down in price and AMD succeeded. 

This is not about you or me, It is about Intel or AMD.

[gbergeron]

You guys are funny if you think AMD will not raise their price if they make a better CPU for gaming. They will like any company raise the profit

 

It would be nice for competition tho

[Trixanity]

Just now, gbergeron said:

You guys are funny if you think AMD will not raise their price if they make a better CPU for gaming. They will like any company raise the profit

 

It would be nice for competition tho

Naturally. However they have to provide the better value than Intel. They're the 'risky' choice so the incentive has to be more than 'equal or slightly higher performance'. Price is very important. I imagine AMD will slowly raise prices over time depending on their competitiveness and their position in the market. If AMD could, they'd sell it in the same ballpark as Intel.

[gbergeron]

1 minute ago, Trixanity said:

Naturally. However they have to provide the better value than Intel. They're the 'risky' choice so the incentive has to be more than 'equal or slightly higher performance'. Price is very important. I imagine AMD will slowly raise prices over time depending on their competitiveness and their position in the market. If AMD could, they'd sell it in the same ballpark as Intel.

true

 

But dont expect this incentive to be like suuuuper good price

[gbergeron]

Just now, The Benjamins said:

I think they will after they make back a good chunk of market share. but if they are both one up each other each generation prices will not get that high.

np i will make sure to get higher than the prices

[TheCherryKing]

7 minutes ago, gbergeron said:

You guys are funny if you think AMD will not raise their price if they make a better CPU for gaming. They will like any company raise the profit

 

It would be nice for competition tho

The probably will. Everything that is designed for gaming has a price premium.

[WkdPaul]

Thread cleaned.

 

Any replies that are simply made to antagonize others will be removed without notifications, same goes with replies quoting those.

 

Please follow the Community Standards at all times.

Edited December 12, 2017 by wkdpaul

[AresKrieger]

9 hours ago, Trixanity said:

So you're saying Intel couldn't repurpose Xeon CPUs without cutting deep into profits but AMD can? Sounds an awful lot like AMD's approach have given them an advantage yet you say Intel has the same capability?

AMD needed to bring core count up in the consumer market, intel did not the reason has nothing to do with amd's vs intels process and everything to do with their market shares, amd needed to get back into the market this approach achieves that, intel did not hence they haven't bothered to do so until amd made sitting back not a viable option.

 

Also selling a chip for less than you were previously is what cuts into the profit, you are simply twisting my words to try and make amd look like they are some how at an advantage which currently they are not.

[cj09beira]

2 minutes ago, AresKrieger said:

AMD needed to bring core count up in the consumer market, intel did not the reason has nothing to do with amd's vs intels process and everything to do with their market shares, amd needed to get back into the market this approach achieves that, intel did not hence they haven't bothered to do so until amd made sitting back not a viable option.

 

Also selling a chip for less than you were previously is what cuts into the profit, you are simply twisting my words to try and make amd look like they are some how at an advantage which currently they are not.

8 cores really is the sweet spot, its small enough to were the die is cheap and so selling it as a 4 core is viable and large enough to were the same die can be sold for 300+dollars (single one), so amd will probably keep the die size around the same for some time, so expect the next core increase to be at 7nm, as they will have much more space then to increase core count while keeping the die affordable 

[AresKrieger]

5 minutes ago, cj09beira said:

8 cores really is the sweet spot, its small enough to were the die is cheap and so selling it as a 4 core is viable and large enough to were the same die can be sold for 300+dollars (single one), so amd will probably keep the die size around the same for some time, so expect the next core increase to be at 7nm, as they will have much more space then to increase core count while keeping the die affordable 

They may increase or they may not hard to say, but 8 is good for streaming and competitive in scalable applications, realistically 8 cores is more than the average user needs but if people want to take a crack at streaming and feel that a higher core cpu is what they want then someone will sell them it.

[Trixanity]

9 minutes ago, AresKrieger said:

AMD needed to bring core count up in the consumer market, intel did not the reason has nothing to do with amd's vs intels process and everything to do with their market shares, amd needed to get back into the market this approach achieves that, intel did not hence they haven't bothered to do so until amd made sitting back not a viable option.

 

Also selling a chip for less than you were previously is what cuts into the profit, you are simply twisting my words to try and make amd look like they are some how at an advantage which currently they are not.

I'm not twisting anything. Just remarking how what you're saying sounds off and it still does sound off.

AMD didn't need to bring up anything necessarily. They did do it though because it gave them an edge in core count and their design allowed them to scale up very easily and cheaply. Intel only just made 6 core monolithic mainstream designs and they're practically sold out everywhere which points to supply issues. AMD has none. Of course Intel probably moves a lot more volume than AMD however Intel's fab business is more reliable than GloFo and they have better facilities to get chips to market so I'm not so sure the disparity is as great as it might otherwise seem.

 

Fact is their design lends itself to scaling up at no extra charge. Intel's does not. Which is part of the reason why they're moving to a mesh design. Still a monolithic design but it scales better to keep up with AMD. 

 

Having an advantage in one area does not mean superior across the board which it sounds like you interpret it as.

[AresKrieger]

9 minutes ago, Trixanity said:

 

Having an advantage in one area does not mean superior across the board which it sounds like you interpret it as.

It is mostly because it sounded like a blanket statement, but since it wasn't intended no need to continue this back and forth

[jasonc_01]

On 12/10/2017 at 2:58 PM, jagdtigger said:

Then you can continue to pay a hefty premium for intel...

I will until they hey something more compelling. I don't want competitive, I want competing. Hopefully all in due time.

[Taf the Ghost]

1 hour ago, cj09beira said:

8 cores really is the sweet spot, its small enough to were the die is cheap and so selling it as a 4 core is viable and large enough to were the same die can be sold for 300+dollars (single one), so amd will probably keep the die size around the same for some time, so expect the next core increase to be at 7nm, as they will have much more space then to increase core count while keeping the die affordable 

Yup, dies around 200 mm² are in a really good spot for yields, even on a "bad" process. That's exactly why AMD has gone that direction. We'll see what the next uArch changes come with "Rome" and a die-shrink. We'll probably get split Server & Client dies, it seems like.

[cj09beira]

51 minutes ago, Taf the Ghost said:

Yup, dies around 200 mm² are in a really good spot for yields, even on a "bad" process. That's exactly why AMD has gone that direction. We'll see what the next uArch changes come with "Rome" and a die-shrink. We'll probably get split Server & Client dies, it seems like.

i kind of hope not and hope it does at the same time, from one side we would get 12 cores on am4 but from the other it wouldn't be able to be as frequency happy as it could be 

[Taf the Ghost]

13 minutes ago, cj09beira said:

i kind of hope not and hope it does at the same time, from one side we would get 12 cores on am4 but from the other it wouldn't be able to be as frequency happy as it could be 

After the leak about 16c parts, my assumption is we're getting 2 designs on the higher density process. "Zen2" is the core design itself for the new CCX. At the moment, I would assume we're getting an 8c (2 CCX) mainstream design and a 16c (4 CCX) Server + HEDT part. The Mainstream part will still show up in up to 32c parts for Epyc 2, but they'll go up to 64c parts as a result of the second set of dies.

 

Epyc 2 ("Rome") will be on a different package design. Too many extra cores, but that approach would let them work in both versions of the die. Depending on how they arrange the dies, the Epyc 2 dies would probably be under 300 mm². Since they'll be server parts, the Clocks of the yield are easier to bin.

 

But, I do want to hedge a little, we could see a 12c Mainstream design. Given the area shrink that AMD will be getting on 7nm, there's a lot of room for more cores at the same rough 200 mm² size.

[cj09beira]

35 minutes ago, Taf the Ghost said:

After the leak about 16c parts, my assumption is we're getting 2 designs on the higher density process. "Zen2" is the core design itself for the new CCX. At the moment, I would assume we're getting an 8c (2 CCX) mainstream design and a 16c (4 CCX) Server + HEDT part. The Mainstream part will still show up in up to 32c parts for Epyc 2, but they'll go up to 64c parts as a result of the second set of dies.

 

Epyc 2 ("Rome") will be on a different package design. Too many extra cores, but that approach would let them work in both versions of the die. Depending on how they arrange the dies, the Epyc 2 dies would probably be under 300 mm². Since they'll be server parts, the Clocks of the yield are easier to bin.

 

But, I do want to hedge a little, we could see a 12c Mainstream design. Given the area shrink that AMD will be getting on 7nm, there's a lot of room for more cores at the same rough 200 mm² size.

wasn't rome supposed to be up to 48 cores though, i haven't seen the rumors you have it seems, toss them my way if you can  

[hconverse02]

Honestly I'm only gonna be stoked about ryzen+ if it has much higher clock speed

[Taf the Ghost]

2 hours ago, cj09beira said:

wasn't rome supposed to be up to 48 cores though, i haven't seen the rumors you have it seems, toss them my way if you can  

It was 48c/96t back in 2015 as "Starship". There's been one leak from a credible source with 16c design that would be right at the Tapeout range for 7nm. (It was the French site, and the much larger 16mb L3 Cache is the give away for where AMD would need to go for this type of design anyway.)

 

Zeppelin is an 8c design that's been used in Mainstream, HEDT and Server. I expect we're getting 2 designs for 7nm. "Rome" will be the Eypc 2 design for the highest Core Count, and then we'll have another design for Mainstream. Do they cross them over? How many cores are in the Mainstream design? Given that the area shrink is going to be fairly massive (14nm to 7nm is about a 1.5 node jump for AMD), they can fit a lot in there. So, we'll see.

[jagdtigger]

7 hours ago, jasonc_01 said:

I will until they hey something more compelling. I don't want competitive, I want competing. Hopefully all in due time.

For marginally better performance? Well good for you that you have that kind of money to waste...

[jasonc_01]

35 minutes ago, jagdtigger said:

For marginally better performance? Well good for you that you have that kind of money to waste...

I do...