Zen Engineering Samples, Specs Spotted

[patrickjp93]

Just now, Dabombinable said:

So its more than likely that the programmers in all reality need re training? Because I take it that you came out of University/College a lot more recently.

High Performance Computing courses have been in college curricula in computer science for more than a decade. Part of the problem is these studios hire passionate gaming enthusiasts who will work for dirt cheap and pour their heart and soul into making games, but they aren't the sharpest tools in the shed when it comes to evaluating whether a solution is the best course of action or not.

 

Retraining, and the studios understanding they are going to have to pay good money to get performance engineers who don't quite make the cut to work at supercomputers, are both incredibly necessary to revamp the gaming industry.

[Prysin]

1 hour ago, patrickjp93 said:

I only know about the HPC APU using it for now, but that's also a mid to late 2017 product at the earliest.

server was scheduled for early or mid 2017 according to slides right before the summer.

[Prysin]

17 minutes ago, patrickjp93 said:

Retraining, and the studios understanding they are going to have to pay good money to get performance engineers who don't quite make the cut to work at supercomputers, are both incredibly necessary to revamp the gaming industry.

Thats why i like DICE. The people there truly understand the need for optimizations. Hell, their lead engineers were part of the crew that built Mantle.

[NumLock21]

1 hour ago, patrickjp93 said:

 

Edit: You guys are getting slow or lazy. In 24 hours I'm the one posting all the Intel and AMD news even though the articles were up hours before I found them.

Wat

Posted one 48 hours before yours

Haha!

 

[Doobeedoo]

Nice to see some info on engineering samples, means we're not too far off from release
I know it's all in early stages still, testing and all. Though I expect way more clock for AM4 4-core and 8-core chips, specially cause of those higher core count 24 and 32 being not far in clocks and also tdp taken in account. 

[patrickjp93]

2 minutes ago, NumLock21 said:

Wat

Posted one 48 hours before yours

Haha!

 

Nope. Everyone here knows I am the supreme ruler of all CPU/APU news for Intel and AMD

[DXMember]

I don't get it...

how does this affect Pokemon?

 

on a side note, Intel has more L3 cache on the 8 core

[AlwaysFSX]

28 minutes ago, Prysin said:

Thats why i like DICE. The people there truly understand the need for optimizations. Hell, their lead engineers were part of the crew that built Mantle.

And then they add a gaudy color filter.

[Vode]

Sounds like the Zen 8-core will be around 5960x performance with a lower TPD of 95W.

 

That would actually be pretty sweet.

 

Also:

 

Haswell-E 5960x die size = 352mm2 (22nm)

Zen 8-core die size ~ 250mm2 (14nm) -> It's a guess

 

I guess it will probably cost around 500-600 bucks. That will put some pressure on Intel's current X99 pricing.

[kmanchine]

13 minutes ago, Vode said:

Sounds like the Zen 8-core will be around 5960x performance with a lower TPD of 95W.

 

That would actually be pretty sweet.

 

Also:

 

Haswell-E 5960x die size = 352mm2 (22nm)

Zen 8-core die size ~ 250mm2 (14nm) -> It's a guess

 

I guess it will probably cost around 500-600 bucks. That will put some pressure on Intel's current X99 pricing.

Hold on, it still way to early to talk performance and price. Remember hype only brings disappointment.

[themaniac]

now the question is will they do similar to what intel does and have a quad core with HT, a quad core with no HT, and a quad core with 2 cores disabled or will they have a few different types a quad cores that are all clocked differently and one or two with HT

[Curufinwe_wins]

15 minutes ago, themaniac said:

now the question is will they do similar to what intel does and have a quad core with HT, a quad core with no HT, and a quad core with 2 cores disabled or will they have a few different types a quad cores that are all clocked differently and one or two with HT

They probably won't ship anything less than a quad core on Zen. Remember Bristol Ridge is known to be coming to AM4 with Excavator quad cores, so old FX chips will probably hold the low end for a while.

[BuckGup]

3 hours ago, CUDA_Cores said:

32 cores! I hope they aren't 32 sh*t cores like AMD pulled to bulldozer.

It is 4 with 4 half cores with 4 halfs on the halfs to make 32.

[The Benjamins]

Just now, SamStrecker said:

It is 4 with 4 half cores with 4 halfs on the halfs to make 32.

its 32 cores with 2 threads per core. so 32 + 32

[Vode]

44 minutes ago, kmanchine said:

Hold on, it still way to early to talk performance and price. Remember hype only brings disappointment.

2015 performance in 2017 for less money isn't really hype. It's expected.

 

But since it's an AMD CPU I guess  some degree of scepticisim is justied.

[AlTech]

33 minutes ago, Vode said:

2015 performance in 2017 for less money isn't really hype. It's expected.

 

But since it's an AMD CPU I guess  some degree of scepticisim is justied.

Not for AMD it isn't.

 

Remember the FX CPUs? They didn't deliver on performance with intel's products.

 

This is meant to be within +- 20% of Skylake which is basically Haswell like performance. This itself isn't bad. It's progress. 40% IPC improvement.

[NuclearPeace]

Doesn't cache performance in some way (i forget if its latency or throughput) suffer the larger it gets?

[AlTech]

Just now, NuclearPeace said:

Doesn't cache performance in some way (i forget if its latency or throughput) suffer the larger it gets?

I'm not sure.

 

Level 1 cache is way faster than Level 2 cache and Level 3 cache is way slower than Level 2 cache.

 

It could be due to size. But I highly doubt it.

 

 

[Prokart2000]

when is zen "rumoured" to be released?

[themaniac]

sometime in 2017-2018

[Prokart2000]

Just now, themaniac said:

sometime in 2017-2018

a while yet then

[Tom Hanks]

[VerticalDiscussions]

So, if this turns out to be true, theyll finally get redempted from the "catastrophic" RX-480 hype?

[patrickjp93]

21 minutes ago, NuclearPeace said:

Doesn't cache performance in some way (i forget if its latency or throughput) suffer the larger it gets?

 

18 minutes ago, AluminiumTech said:

I'm not sure.

 

Level 1 cache is way faster than Level 2 cache and Level 3 cache is way slower than Level 2 cache.

 

It could be due to size. But I highly doubt it.

 

 

It suffers on latency the bigger it gets. The algorithms which manage cache searches run at a clock much higher than the cache itself, so suffice it to say that if cache gets big enough, you add a CPU clock cycle to the search time. This can have enormous consequences in performance scaling, which is a big reason why we have seen Intel's and IBM's L1 and L2 cache sizes remaining the same size they have for a long time.

[KillDog]

Very interesting.