* AMD "Zen" thread (inc AM4/APU discussion) *

ubersonic · 22 Dec 2016 at 09:40

humbug said:
+1... Intel would only lower the price by anything even close to that amount if AMD gained a significant sales share from them, that is only possible if a significant number of Intel enthusiasts switched to AMD.

So its not happening.

Intel remember the Athlon XP/64/II/etc, they will want to react ASAP this time to avoid AMD gaining a significant sale share form them.

CAT-THE-FIFTH said:
Its more the case AMD CPUs seem to be better at spanning multiple sockets than their sockets actually lasting a decent amout of time.

Intel were the same prior to 115x, seems they decided that by limiting 1-2 CPU generations per board they could really hose the consumer

Bonjour · 22 Dec 2016 at 10:09

I think Intel will probably wait for confirmation AMD don't have another 'dozer on their hands before adjusting any pricing.

Orangey · 22 Dec 2016 at 11:18

Disco_P said:
I am not sure I really see the need for the 115x sockets by intel. They can't be that different can they?

Well they did play the old hokey cokey with the FIVR.

eddyr · 22 Dec 2016 at 11:21

Aside from the executives watching the balance sheet, I am sure the engineers also like it. Bit more freedom to tinker and innovate. Not constrained by old design choices and also not having to compromise in order to anticipate future requirements when making those design choices.

Mauller · 22 Dec 2016 at 12:03

eddyr said:
Aside from the executives watching the balance sheet, I am sure the engineers also like it. Bit more freedom to tinker and innovate. Not constrained by old design choices and also not having to compromise in order to anticipate future requirements when making those design choices.

All the Girls and Guys working on the Zen architecture must have had lots of fun coming up with new ideas for improvements and feature designs etc.

CAT-THE-FIFTH · 22 Dec 2016 at 15:18

I saw this posted on AT forums:

news.zol.com.cn/620/6203956.html

Apparently a competition to win a Ryzen CPU which ends on the 18th of January.

SiDeards73 · 22 Dec 2016 at 15:34

Apparently AMD official RyZen Store opens on the 18th of Jan or something according to that link?

Uriel · 22 Dec 2016 at 16:40

Other than people speculating, has anyone seen a reliable source indicate that there will be any Ryzens that don't feature SMT?

jigger · 22 Dec 2016 at 19:11

Uriel said:
Other than people speculating, has anyone seen a reliable source indicate that there will be any Ryzens that don't feature SMT?

I've not seen anything, but I think it's safe to say we will see a Athlon X4 RyZen.

Bristol Ridge has a range of non SMT 4 cores.

CAT-THE-FIFTH · 23 Dec 2016 at 00:29

A poster on SA forums has spotted a mention of an Opteron X3000 APU in this document:

http://www.dmtf.org/sites/default/files/standards/documents/DSP0134_3.1.0.pdf

DragonQ · 23 Dec 2016 at 08:47

Something that has just occurred to me is that with Zen being PGA, it's going to be extremely difficult to delid them without bending or breaking the pins. Let's hope they use solder!

Hagar · 23 Dec 2016 at 09:36

DragonQ said:
Something that has just occurred to me is that with Zen being PGA, it's going to be extremely difficult to delid them without bending or breaking the pins. Let's hope they use solder!

Hopefully as you say, there will be no need. But I was a fan of the old bare CPU's. The K5's had a cover/heatspreader over the die but my all timer the K7 was back to naked.

CAT-THE-FIFTH · 23 Dec 2016 at 15:46

Looks like a French computing magazine got hold of an ES,which is running at around 3.15GHZ(according to what is said on Reddit):

https://i.redd.it/tic6i57uob5y.jpg

So in gaming it is 10% slower than a Core i7 6900K and 8% in non-gaming software.

If they can get clockspeeds up and Turbo running properly,it might actually be not bad!!

Harlequin · 23 Dec 2016 at 16:03

10% increase in clock speed puts the numbers on - the live show

CAT-THE-FIFTH · 23 Dec 2016 at 16:05

Harlequin said:
10% increase in clock speed puts the numbers on - the live show

I mean it looks like Ryzen has pretty decent IPC - some of the games tested don't do very well on AMD CPUs either.

If AMD can get 4C/8T CPUs running closer to 4GHZ,its going to be not too bad at all it seems.

mmj_uk · 23 Dec 2016 at 17:02

If you split the Zen result straight down the middle you can deduce that a 4c/8t 3.15ghz Zen will be about 15% slower than a 4c/4t 3.5ghz 6600K (Skylake) on average in productivity.

8c/16t Zen looks like it'll be closer to 6850K performance at stock but consuming almost as much power as 6900K.

All in all it's probably going to come down to how well Zen overclocks and prices.

Harlequin · 23 Dec 2016 at 17:06

mmj_uk said:
If you split the Zen result straight down the middle you can deduce that a 4c/8t 3.15ghz Zen will be about 15% slower than a 4c/4t 3.5ghz 6600K (Skylake) on average in productivity.

8c/16t Zen looks like it'll be closer to 6850K performance at stock but consuming almost as much power as 6900K.

All in all it's probably going to come down to how well Zen overclocks and prices.

erm the 6900k is a 140w part , zen is topping out at 95w

CAT-THE-FIFTH · 23 Dec 2016 at 17:16

mmj_uk said:
If you split the Zen result straight down the middle you can deduce that a 4c/8t 3.15ghz Zen will be about 15% slower than a 4c/4t 3.5ghz 6600K (Skylake) on average in productivity.

8c/16t Zen looks like it'll be closer to 6850K performance at stock but consuming almost as much power as 6900K.

All in all it's probably going to come down to how well Zen overclocks and prices.

No it isn't - the Core i7 6900K is running at 3.2GHZ to 3.7GHZ,so probably around 3.5GHZ,and the Ryzen CPU at 3.15GHZ. The Core i7 6900K is running 10% higher clockspeeds,with production motherboards and quad channel RAM. The Core i7 6900K is 10% faster in games and 14% faster in non-gaming scenarios.

That means at similar clockspeeds,it will be more or less the same performance as a Core i7 6900K whilst using a cheaper base platform. That hints at the Ryzen having a decent memory controller too.

AMD ran their sample at 3.4GHZ/3.45GHZ so I expect with production motherboards,and Turbo working properly,they will be quite close. Remember this is an ES.

Plus the games tested put the Core i7 6850K and Core i7 6900K at similar performance,which means they are more clockspeed dependent. The Core i7 6700K is probably running at 15% to 20% higher clockspeed.

If you consider how much higher the clockspeeds for the Core i7 6700 are over the BW-E CPUs,you can see the IPC difference is basically non-existent.

That means 4C/8T Ryzen CPUs are probably going to be close to the Core i7 6700K/Core i7 7700K as they will be less TDP limited too.

This is FAR more than the SB/IB level IPC many of us thought Ryzen would have and validates the AMD tests in their preview as being accurate(!!).

CAT-THE-FIFTH · 23 Dec 2016 at 17:57

Also this poster mentioned this on SA forums too:

Fottemburg said:
The real game-changer will be the advanced Turbo feature.

He runs one of the Italian tech sites and has been reasonably accurate on rumours in the past. AMD did not have Turbo running during their preview.

Minstadave · 23 Dec 2016 at 18:08

It does seem off we're not seeing turbo in these previews, it's due for release in the next 2-3 months so final silicon must be available.

Competitor rules

*** AMD "Zen" thread (inc AM4/APU discussion) ***

* AMD "Zen" thread (inc AM4/APU discussion) *