Fjodor2001
Diamond Member
- Feb 6, 2010
- 4,581
- 731
- 126
If it ever exists. Both Haswell IPC (except perhaps in some cherrypicked benchmark) and 4 ghz seem highly unlikely.
Based on what? Speculation?
If it ever exists. Both Haswell IPC (except perhaps in some cherrypicked benchmark) and 4 ghz seem highly unlikely.
Of course not, that's why I qualified it with the assumption that AM4 is a ~1000 pin PGA like every AMD PGA socket since Socket 939. Maybe AMD will switch to a higher count LGA, or changing to a 1mm pitch PGA socket, but that would be a little surprising.
Note that I never said it wouldn't be able to fit more than one x16 interface. Another x16 interface isn't really that many pins. Adding another dual channel DDR4 controller is close to 350 pins though.
If it ever exists. Both Haswell IPC (except perhaps in some cherrypicked benchmark) and 4 ghz seem highly unlikely.
A 6 core 12 threaded haswell IPC 4ghz zen processorBased on what? Speculation?
Kabylake is an APU with up to 4 cores 8 threads,
First ZEN SKUs (with up to 16 threads ??) will be competing against Socket 2011-V3 and Broadwell-E .
A 6 core 12 threaded haswell IPC 4ghz zen processor
Is pure speculation.... I don't see your point
Zen will compete with whatever Intel wants it to compete with.
He made a statement saying it was unlikely. So what is that statement based on?
Fjodor2001 said:You're just bitter because in 2016 you'll be sitting on an expensive and slow 4 core Intel CPU, while others will be using a cheaper and faster 8 core AMD CPU.
So you consider KabyLake a candidate for being a Zen competitor, if Intel desires?
Shintai, am I missing something? That slide doesn't seem to say anything about the exact number of PCIe 3.0 lanes on Zen. All it shows are some lanes dedicated to storage, extra USB ports, and Promontory.
It specifies x16 for Summit Ridge. The fabled 8 core Zen.
They want to attack the Server (Data Center and Cloud) market first, thus the big dies are first and then the APUs.
I thought you were only buying 4-5 year old used Intel systems![]()
Like Zen has any chance in servers. They are doing a CPU-only design because they need to get something out ASAP to keep the charade up to Wall Street. Adding in an IGP would only delay things more even though they would need one to sell in any volume.
Literally the only market Zen has a realistic shot in is gamers, and that's a tiny market. And it's still going to be a tough sell.
IF (with a double capital and bold), ZEN has 40% higher IPC than Excavator then it should be close to Haswell (I dont count Cinebench). 4GHz for a 6C 12T at 95W TDP on the 14nm FF LPP is also reasonable.
I'm not how you can come to that conclusion. Heck, cloud servers haven't seen a decent revolution in a while and it sucks. There needs to be someone to push technology. I'm tired of seeing "atom" crap processors pushed on me when they have a very, very limited reach.
The problem with Zen in general is that it's designed with if it was still 2010 when it's coming out in 2017.
4GHz for a 6C 12T at 95W TDP on the 14nm FF LPP is also reasonable.
Cannonlake ain't coming in first half of 2017, certainly not since the new roadmap shows Kabylake as a stopgap. Now if Intel wants to do a token launch, like with Broadwell followed by Skylake, then sure you could say AMD's 14nm will fight it out with Intel's 10nm but that is in the distinct realms of possibility.I still fail to see how this is a win for AMD though?
With Zen pushed to late 2016 / sometime in 2017, and Cannonlake proposed for release around Q1 / Q2 of 2017, Zen has to be able to compete with a 10 nm die shrink of Skylake.
That's a pretty big challenge for a processor architecture for which we still know very little.
Cannonlake ain't coming in first half of 2017, certainly not since the new roadmap shows Kabylake as a stopgap. Now if Intel wants to do a token launch, like with Broadwell followed by Skylake, then sure you could say AMD's 14nm will fight it out with Intel's 10nm but that is in the distinct realms of possibility.
