Broadwell to Skylake, will be similar to the transition from Prescott to Conroe

[csbin]

http://www.bitsandchips.it/9-hardwa...sara-simile-al-passaggio-da-prescott-a-conroe


Google Translation

Because the development of Skylake then it should be problematic? Due to the high complexity of the project, which is why the team of Haifa decided to minimize the variables to be considered, such as deleting the FIVR integrated for the first time in Haswell.

Skylake seems to have a completely new architecture than the Intel hitherto seen, and this would be taking away considerable time, and because the house of Santa Clara wants a product mostly Bug Free, and because the compiler of the house (the ICC - Intel C ++ Compiler) should be revised to take advantage of the new features that will be integrated.

At the time Intel showed no sample functioning of Skylake not for problems with 14nm FinFET, but because the entire ecosystem (CPU + Software) is still incomplete. Finally, it seems that the frequency of operation of Skylake are not excessively high. The increase in CPI compared with Broadwell, could be really massive.

Perhaps the owners of CPU Sandy Bridge and Ivy Bridge have finally found a replacement of their quasi-immortal processors.

Prescott to Conroe :biggrin:

http://techreport.com/review/10508/cool-and-cheap-screamers-from-intel-and-amd/11

 

[ShintaiDK]

Sounds like rubbish.

 

[AtenRa]

If 4+2 Skylake comes at 95W TDP i may say the rumor may come true.

 

[Vesku]

So quad core Pentium, quad+ht i3, hex i5 and hex+ht i7 and talk up the MT improvements?

My hats off to Intel if a Skylake Quad is Conroe to a Haswell Quad as Prescott.

 

[frozentundra123456]

Very unlikely IMO.

 

[schmuckley]

Like Prescott to Conroe?
I'd be OK with that.

 

[DrMrLordX]

Historically, that was one of the biggest jumps in performance Intel has ever seen, though really the fair comparison was Presler to Conroe (Presler was the prevailing Intel desktop dual-core at the beginning of 2006, even if not very many people bought them). And of course you can see them in the above-posted chart as the PentiumD 950 and 960.

Just looking at the 960 vs the E6700 (3.6 ghz vs 2.66 ghz), the increase in IPC is massive just in Quake4 alone. I would be shocked if Skylake could show up at, oh I don't know 3 ghz and put up gains of nearly +55% vs a hypothetical 4 ghz Broadwell in a video game. At that point, the so-called "14 nm clockspeed problem" that Intel is allegedly having wouldn't matter a whit.

 

[csbin]

Historically, that was one of the biggest jumps in performance Intel has ever seen, though really the fair comparison was Presler to Conroe (Presler was the prevailing Intel desktop dual-core at the beginning of 2006, even if not very many people bought them). And of course you can see them in the above-posted chart as the PentiumD 950 and 960.

Just looking at the 960 vs the E6700 (3.6 ghz vs 2.66 ghz), the increase in IPC is massive just in Quake4 alone. I would be shocked if Skylake could show up at, oh I don't know 3 ghz and put up gains of nearly +55% vs a hypothetical 4 ghz Broadwell in a video game. At that point, the so-called "14 nm clockspeed problem" that Intel is allegedly having wouldn't matter a whit.

Cinebench 2003

 

[DrMrLordX]

Yup, there you've got a ~46% gain for the E6700 over the D960 in multithreaded. And boy did those Core2s have some OC headroom, too.

 

[ElFenix]

For whatever reason I remember conroe decimating the athlons worse than those charts show.

 

[Smoblikat]

Yup, there you've got a ~46% gain for the E6700 over the D960 in multithreaded. And boy did those Core2s have some OC headroom, too.

Used to run my 2.5ghz E5200 at 3.8 no problem.

I would be impressed and surprised if Intel delivered a worthwhile product to us. They havent improved anything since SB.

 

[BigDaveX]

For whatever reason I remember conroe decimating the athlons worse than those charts show.

It was price-performance and performance-per-watt where the Athlon 64 X2s really got stung; the E6300 delivered 4200+ performance for around two-thirds the price, and the E6600 outperformed AMD's entire line-up for a price that was about equivalent to that of the 4800+.

That said, the worse damage didn't come until later that year, with the release of the Core 2 Quad. AMD meanwhile spent a full year flailing around and trying to get the Phenom out (and releasing Quad FX as a stopgap solution), while Intel moved onto Penryn and released a range of cheap quad-cores.

 

[escrow4]

I was talking with somebody who is still using an X2 4200+ in 2015 recently. I nearly collapsed. :awe:

Until benches are out speculation is just that. Skylake could be 5% faster or 55% faster. Only Intel knows.

 

[Dave2150]

I doubt we'll see Skylake for desktop before 2016, judging by the speed at which they are bringing Broadwell to desktop.

 

[tynopik]

Until benches are out speculation is just that. Skylake could be 5% faster or 55% faster. Only Intel knows.

Skylake will not be 55% faster in the general case

we can know that with full confidence now

 

[erunion]

I doubt we'll see Skylake for desktop before 2016, judging by the speed at which they are bringing Broadwell to desktop.

Broadwell for Desktop is cancelled. We've been talking about it for awhile.

Skylake leaks show it coming for back to school 2015.

 

[crashtech]

Broadwell for Desktop is cancelled. We've been talking about it for awhile.

Skylake leaks show it coming for back to school 2015.

I didn't get the memo. Can you link to proof that no LGA 1150 Broadwell CPUs will be released?

 

[MongGrel]

Still using a O/C'd Westmere-EP X5680 @ 4.5 on the main atm and will be until something like that shows up I imagine.

I didn't get the memo. Can you link to proof that no LGA 1150 Broadwell CPUs will be released?

Still wondering about that one myself, but not too hard right now.

 

[Abwx]

For whatever reason I remember conroe decimating the athlons worse than those charts show.

Guess that you gave too much attention to Sysmark...

It was price-performance and performance-per-watt where the Athlon 64 X2s really got stung;

That said, the worse damage didn't come until later that year, with the release of the Core 2 Quad. AMD meanwhile spent a full year flailing around and trying to get the Phenom out (and releasing Quad FX as a stopgap solution), while Intel moved onto Penryn and released a range of cheap quad-cores.

Despite being a node late, and half the transistor budget of the Core 2, the A64x2 were competitive perf/Watt wise, when shrinked a few month later they got better perf/Watt than the Intel counterparts...

 

[erunion]

I didn't get the memo. Can you link to proof that no LGA 1150 Broadwell CPUs will be released?

Several recent broadwell threads have lots of info.

There will actually be two Broadwell SKUs on LGA 1150, both Iris Pro. The rest are cancelled because Skylake is coming in Q3 2015.

http://hexus.net/tech/news/cpu/81883-first-broadwell-desktop-cpus-intel-core-i7-5775c-i5-5675c/


Notice that BDW-S is not on the roadmap. Cancelled. The only LGA BDW are actually BDW-H (Gt3e)

 

[Fjodor2001]

"Broadwell to Skylake, will be similar to the transition from Prescott to Conroe"

Is there any evidence of that, or is it just pure guessing?

 

[III-V]

I've done a lot of research on Skylake -- I've written a ~2500 word article that I've yet to even finish and publish, and it'll probably be around 4000 when I'm done.

As far as I can tell, Skylake is not some completely new architecture. The most important piece of evidence against the idea is the Skylake Geekbench results that have been posted around here a few times. With a new architecture, one would expect its performance to fluctuate quite a bit compared to an old one -- some benchmarks the new arch would excel in, some it would regress in. But in the Geekbench results, Skylake performs pretty similarly in Haswell across the board, just faster. There are a few exceptions, like SHA, but it is already known that SHA cryptography is getting a big boost with Skylake.

Skylake does however have a noticeable boost in multithreaded performance. I am currently attributing this to a couple of changes that are likely to make in into Skylake (pulling from RWT's David Kanter here), the first being a shift to a "tiled" cache hierarchy, where L2 is shared between cores in groups of two, much like Bulldozer (arguably), and Apple's Cyclone. Sharing L2 between two cores is something that Intel has not chosen to do since they bolted an L3 cache on. It's not uncommon to see the L2 shared in L3-less chips (in fact, it's really all you'll ever see), however Intel chose a different route on chips with L3 included. They might be going back to a shared L2, and I think that's a smart move.

The other change that Kanter alluded to was a change in the interconnect fabric: from the current ring bus, to a 2D mesh. I don't really know many of the details surrounding this, although I'm fairly certain it'd be a good improvement to inter-core communication, as I recall ring buses being a pretty outdated concept.

I don't expect a lot from Skylake. I am having issues finding the numbers I compiled, comparing performance per clock against Haswell, but I remember it being fairly comparable to Haswell's gain over Ivy Bridge for single thread, while being a bit more substantial in the MT department. I guess we'll know more as time goes on, but I am fairly certain that Skylake is not an entirely new uarch.

 

[crashtech]

Several recent broadwell threads have lots of info.

There will actually be two Broadwell SKUs on LGA 1150, both Iris Pro. The rest are cancelled because Skylake is coming in Q2 2015.

http://hexus.net/tech/news/cpu/81883-first-broadwell-desktop-cpus-intel-core-i7-5775c-i5-5675c/

Then desktop Broadwell isn't exactly cancelled. Curtailed, perhaps, but LGA 1150 is considered desktop afaik.

 

[witeken]

Skylake will not be 55% faster in the general case

we can know that with full confidence now

It'll be 555% faster, duh.

Edit: Didn't see the "not" in your post, kinda ruins the sarcasm.

Broadwell for Desktop is cancelled. We've been talking about it for awhile.

Skylake leaks show it coming for back to school 2015.

Yes, which is why IDF is, unconventionally, a few weeks earlier than usually.

I've done a lot of research on Skylake -- I've written a ~2500 word article that I've yet to even finish and publish, and it'll probably be around 4000 when I'm done.

So much for an architecture yet to be disclosed?

All I know is that Skylake for sure will make for a few very nice reads on AnandTech, if at least the level of depth and information is as we've known from Anand.

 

[erunion]

Then desktop Broadwell isn't exactly cancelled. Curtailed, perhaps, but LGA 1150 is considered desktop afaik.

See my edit. BDW-S is cancelled.