Intel Broadwell Thread

[superstition]

Surprisingly Broadwell-C is in stock at Newegg. It's $299 for the i5 and $371 for the i7.

i5 is $275 shipped from NextWarehouse. 35 in stock. Not a bad choice for gaming-specific builds.

i7 from B&H is $370 shipped. Newegg wants a shipping fee, so B&H wins.

 

[superstition]

I simply want to buy the best CPU I can this year, for the fastest build I can make this year, and sit on that for a while before I repeat the process a few years down the line.

Given the improvements to the Skylake core I don't understand why anyone would choose to buy a Broadwell without EDRAM (especially if they're not planning to upgrade as soon as Skylake E is released).

Something like the 5775C makes sense for gaming builds but what's the sense in getting 6-10 core Broadwell for a high price since it doesn't have the latest cores, or the EDRAM, and Haswell E is still around for less money?

Broadwell E, frankly, makes zero sense to me as a product. Outdated cores. No EDRAM. Higher price.

 

[Arachnotronic]

With no EDRAM, too, I bet.

Let's all get an inferior core for a high price that doesn't have the one thing that made desktop Broadwell attractive for gamers.

Broadwell-E won't be bandwidth starved, AFAIK.

 

[MrTeal]

From a very reliable source it has been confirmed 10 Core Broadwell-E at $1500

Sadly, I am hearing the same.

That's brutal. Sadly, it probably means the octo-core will probably slot in around the same price as the 5960X. No mid-platform upgrade here if this is true, onward and upward to SKL-E.

 

[Arachnotronic]

That's brutal. Sadly, it probably means the octo-core will probably slot in around the same price as the 5960X. No mid-platform upgrade here if this is true, onward and upward to SKL-E.

Yeah...

 

[DrMrLordX]

Broadwell-E won't be bandwidth starved, AFAIK.

That's probably true. Broadwell-E has no iGPU to feed, after all.

 

[RussianSensation]

Broadwell E, frankly, makes zero sense to me as a product. Outdated cores. No EDRAM. Higher price.

BW-E makes a lot of sense.

> 6-core BW-E will have IPC barely 6-7% slower (that's being generous to Skylake) compared to lowest SKU will have 2 more cores. That means, the CPU is going to be much better well-rounded for gaming + productivity and is much more future-proof to keep for 4-5 years.

> 8- and 10-core offerings will be awesome for productivity.

> X99 platform will have future upgrade paths to Xeon CPUs dirt cheap, something we've seen with X58 and X79.

If anything, once BW-E launches, it's Skylake mainstream platform i7 6700K that makes no sense. I'd much rather have 2 extra cores over the next 4-5 years over mostly 'useless' 6-7% difference in IPC. At least the extra cores have serious potential down the line while the tiny difference in IPC will hardly matter.

If BW-E 6-core managed to overclock to 5.0-5.1Ghz because it has solder vs. $2 TIM Intel uses with i7 6700K, then it's going to slaughter 6700K @ 4.8Ghz as a hands down much better well-rounded CPU.

Also, some of us don't want to pay for Intel's iGPU and for me personally I would never pay $400+ USD for a quad-core i7 6700K out of principle since I think glorified i5 with HT should not cost that much in 2016.

 

[frozentundra123456]

Given the improvements to the Skylake core I don't understand why anyone would choose to buy a Broadwell without EDRAM (especially if they're not planning to upgrade as soon as Skylake E is released).

Something like the 5775C makes sense for gaming builds but what's the sense in getting 6-10 core Broadwell for a high price since it doesn't have the latest cores, or the EDRAM, and Haswell E is still around for less money?

Broadwell E, frankly, makes zero sense to me as a product. Outdated cores. No EDRAM. Higher price.

This post makes no sense. BW-E will be available before Skylake E. When it comes out it will be the best HEDT chip, and I would assume unchanged in price from HW-E. So it would be the obvious choice. Yes, skylake is the obvious choice for a quad core, but that is a separate issue.

Edit, I assume BW-E will be soldered as well, so considering your other thread, that should finally make you happy.

 

[superstition]

BW-E will be available before Skylake E.

Artificially available before Skylake E. Obviously the Skylake cores have already been made. They're being sold right now.

BW-E will be soldered as well, so considering your other thread, that should finally make you happy.

Solder won't make outdated cores current nor will it make 6-10 core chips quads with hyperthreading.

6-core BW-E will have IPC barely 6-7% slower (that's being generous to Skylake)

That makes sense? Pay a lot for cores that are inferior to cheaper quads?

Skylake should have been postponed in favor of Broadwell and Broadwell E or Broadwell E should have been canceled.

 

[jcwagers]

There is no "artificially" available. As I recall, Intel has been on this schedule for several years. The "E" version doesn't come along until after the next refresh has been released.
Such as...... Ivy Bridge......Haswell.....Ivy Bridge E.....Broadwell.....Haswell E.....Skylake.....Broadwell E...etc.

What exactly is it that makes the Broadwell cores "outdated"? The fact that they aren't Skylake? If you look at a lot of the performance numbers, Haswell E performs VERY close to Skylake when both are overclocked and outperforms it depending on the core applications. So.....Broadwell E should be very competitive with Skylake or beat it depending on what applications are being used. What they are trying to tell you is that 2 FULL cores is going to be more beneficial in the long run than a few few % difference in IPC + hyperthreading. Hyperthreading is nice.....and useful but it's not like actually having 8 fully active cores. That means that the Broadwell E won't be "outdated". It'll compete just fine with Skylake and it will fulfill the place in the lineup that it was designed to occupy. It may be more expensive than what a lot of consumers want to pay. If that's the case, they have Skylake. If they are willing to pony up the cash for more physical cores, they have Broadwell E. They still have options and they aren't forced to buy anything that they don't want. Just my .02...

 

[ZGR]

Will Broadwell C become even more expensive as stock runs out? The i7-5775C looks like an awesome chip, but I don't feel like it is worth $400.

 

[ShintaiDK]

Will Broadwell C become even more expensive as stock runs out? The i7-5775C looks like an awesome chip, but I don't feel like it is worth $400.

Its being sold to 370$ in Europe.

Its more a retailer issue than anything else.

And the chip is still in production.

 

[moonbogg]

So it looks like Broadwell 8 core will simply replace the 5960x @ $1,000.00. So its not really an upgrade, its just a replacement in the same way Ivy-E replaced Sandy-E. Of course you could get an upgrade for $1,500.00 with the ten core version. If CPU's scaled in games like GPU's did, then a CPU price like that would work for a lot more of us, but 10 cores will only help those doing productivity stuff so I could never hand over a whole $1,500.00 cold hard cash for a CPU unless it bangs out the FPS hard core and will continue to do so for a long time.
So where's the upgrade for Haswell-E people? Looks like it will cost them a few extra bucks this time around, but it can be expected since 10 cores is well beyond the curve at this point.

 

[ozzy702]

So it looks like Broadwell 8 core will simply replace the 5960x @ $1,000.00. So its not really an upgrade, its just a replacement in the same way Ivy-E replaced Sandy-E. Of course you could get an upgrade for $1,500.00 with the ten core version. If CPU's scaled in games like GPU's did, then a CPU price like that would work for a lot more of us, but 10 cores will only help those doing productivity stuff so I could never hand over a whole $1,500.00 cold hard cash for a CPU unless it bangs out the FPS hard core and will continue to do so for a long time.
So where's the upgrade for Haswell-E people? Looks like it will cost them a few extra bucks this time around, but it can be expected since 10 cores is well beyond the curve at this point.

I'm curious how Broadwell-E will overclock compared with Haswell-E. What's the average OC on Haswell-E? I know that Broadwell-C wasn't a great overclocker.

 

[frozentundra123456]

So it looks like Broadwell 8 core will simply replace the 5960x @ $1,000.00. So its not really an upgrade, its just a replacement in the same way Ivy-E replaced Sandy-E. Of course you could get an upgrade for $1,500.00 with the ten core version. If CPU's scaled in games like GPU's did, then a CPU price like that would work for a lot more of us, but 10 cores will only help those doing productivity stuff so I could never hand over a whole $1,500.00 cold hard cash for a CPU unless it bangs out the FPS hard core and will continue to do so for a long time.
So where's the upgrade for Haswell-E people? Looks like it will cost them a few extra bucks this time around, but it can be expected since 10 cores is well beyond the curve at this point.

Yea, it is not the price so much as the lack of applications that can use 10 cores. Even though it is 50% more cost for 25% more cores, I am sure there are quite a few people that would be willing pay it if it could scale linearly in games. Problem is for that 50% additional cost, I doubt you will see any improvement in gaming, and maybe even worse if it overclocks less than the hex or octa core.

But lets see what DX12 ,14/16 nm dgpus and VR bring to the table. Maybe a few games will be able to use 10 cores, but I am very skeptical.

 

[frozentundra123456]

I'm curious how Broadwell-E will overclock compared with Haswell-E. What's the average OC on Haswell-E? I know that Broadwell-C wasn't a great overclocker.

Yea, a very good question. It seems like it would be almost embarrassing for intel to bring out a new 1500.00 chip if the line-up overclocks worse than Haswell.

 

[moonbogg]

OC plays a roll in the value for sure, but I can't say I would have replaced my sandy bridge for an ivy even if ivy OC'd better (which it didn't). You are still losing money to replace a chip and if its only for a couple hundred Mhz, then Broadwell-E isn't for Haswell-E people unless they are going with the 10 core model. If they wanted an 8 core model they would have already had a 5960x, unless they were holding out for a $600.00 8 core broadwell, in which case they will be disappointed it seems.

 

[frozentundra123456]

Well, lets see what happens with Zen. Maybe if it is a killer product, it will force Intel to cut prices, at least on the hex and octa core models. The problem is, there is just not much room for a price cut on the hex core, since the cheapest is very close to the 6700K.
An interesting question is whether Intel has any insider information on what will be the performance and pricing of Zen, and if that influenced the pricing of BW-E. Of course, they can charge a much as they can until Zen comes out (Edit: and gets in good supply), and then cut the price if necessary.

 

[crashtech]

If there was any pressure from AMD at all, the 6700K would be $360 or less, and the hexacore could slot in at around the current street price for the 6700K, about $419.

 

[superstition]

There is no "artificially" available. As I recall, Intel has been on this schedule for several years. The "E" version doesn't come along until after the next refresh has been released.
Such as...... Ivy Bridge......Haswell.....Ivy Bridge E.....Broadwell.....Haswell E.....Skylake.....Broadwell E...etc.

What exactly is it that makes the Broadwell cores "outdated"? The fact that they aren't Skylake? If you look at a lot of the performance numbers, Haswell E performs VERY close to Skylake when both are overclocked and outperforms it depending on the core applications. So.....Broadwell E should be very competitive with Skylake or beat it depending on what applications are being used. What they are trying to tell you is that 2 FULL cores is going to be more beneficial in the long run than a few few % difference in IPC + hyperthreading. Hyperthreading is nice.....and useful but it's not like actually having 8 fully active cores. That means that the Broadwell E won't be "outdated". It'll compete just fine with Skylake and it will fulfill the place in the lineup that it was designed to occupy. It may be more expensive than what a lot of consumers want to pay. If that's the case, they have Skylake. If they are willing to pony up the cash for more physical cores, they have Broadwell E. They still have options and they aren't forced to buy anything that they don't want. Just my .02...

So you're saying Skylake cores aren't improved enough over Broadwell to matter? Why, then, did we need desktop Skylake in its current form at all? Why not just make Broadwell and save everyone the expense of having to replace their boards and buy DDR4? It's not like Intel has to offer every desktop Broadwell SKU with the iGPU enabled, which reduces yields.

That's interesting. So much for Skylake being such a revolutionary product. The one thing Skylake offered of any relevance, improved cores, is so inconsequential that people should be happy about paying $1000-$1500 for old Broadwell cores.

Also, arguing that something must be done because it was planned is fallacious. Definitely artificial since the Skylake cores are already made and being sold, too.

Finally, extra cores (over a quad with hyperthreading) is irrelevant to gaming and equivalent to Haswell E.

The EDRAM makes Broadwell C relevant. Broadwell E seems to have no reason to ever exist.

 

[frozentundra123456]

You seem determined to mix up HEDT line up and mainstream. Either that or you are being deliberately obtuse. I agree that it would be nice if there was a mainstream hex core. But there is not. Deal with it. All the complaining in the world is not going to change that.

HEDT platform lags usually one generation behind mainstream. Because of the delays on 14 nm, and the rush to put out skylake it is now two generations, but will be one again when BW-E comes out. But this has nothing to do with whether skylake is out on the mainstream or not. The HEDT platform goes from Haswell to Broadwell to Skylake. Broadwell should bring a small IPC improvement, more cores, and possibly reduced power consumption. Then Skylake should bring another slight IPC improvement, and maybe will overclock a bit better. But to say there is no reason for BW-E to exist because of desktop skylake is just a total apples to oranges non-sequitor, and every other logical fallacy one could describe. In actual fact, there is a very sound reason for BW-E to exist and that is the availability of 10 cores if the user wants them. That obviously does not interest you, (or me for that matter), but it is definitely something new that BW-E brings to the table.

 

[superstition]

HEDT platform lags usually one generation behind mainstream.

Appeal to tradition is a fallacy.

the rush to put out skylake

The unnecessary release of Skylake that I mentioned. Intel had the choice between keeping around Broadwell exclusively on the desktop (and in "E" form) and killing off Broadwell entirely in favor of Skylake. Instead, it is doing the worst of the three possible choices, something that only favors them not consumers.

• Poorly supported Broadwell C (boards are becoming harder to find and board makers have no incentive to continue to improve their BIOSes).

• Already obsolete Broadwell E, offering minimal improvement over Haswell E and worse cores than cheaper Skylake chips already on the market.

• Also, no higher wattage enthusiast Broadwell with EDRAM for gaming.

• No Skylake quad with EDRAM for gaming.

All of these decisions are designed to maximize how many pennies Intel can squeeze out of its customers. (The only exception is that it decided to produce more Broadwell C chips, which is a small favor for gamers since their price is not exactly low and board support is already stagnated — but it's better than being stuck just with Skylake and Haswell. Even so, Skylake on the desktop with EDRAM should already be out and it's not, so so much for the favor.)

Broadwell should bring a small IPC improvement, more cores, and possibly reduced power consumption. Then Skylake should bring another slight IPC improvement, and maybe will overclock a bit better. But to say there is no reason for BW-E to exist because of desktop skylake is just a total apples to oranges non-sequitor, and every other logical fallacy one could describe.

Nope. The existence of Skylake cores for sale makes a mockery of Broadwell E. It's absurd to have superior cores available for inferior-tier products — with nothing available in the higher tiers as an option. Broadwell E is too redundant with Haswell E and already obsolete before it has shipped.

Intel is withholding its existing tech, doling out already inferior tech, and telling people they have no choice. This is why Skylake on the desktop is inferior to Broadwell C for gaming and why Broadwell E is being released at all.

 

[jcwagers]

At this point, you're just looking to stick to your own opinion. That's fine. The fact that I didn't explain it very well and someone else DID and you still continue to refuse to see the other point-of-view.......that says it all. There is no mockery or injustice here. It's called BUSINESS. Intel can withhold whatever they like......they OWN it. They release what they want to release and when they want to release it. Is it ideal? Of course not! They have no reason to release huge leaps in technology because they are already in the driver's seat. They can release it as they want and right now they have no competition except on the lower end.

I DO see what you're saying. From a certain perspective, it doesn't make sense to release something that's not equal with what's currently available. But from the other perspective, some of the features on what's currently available may not be as important to certain aspects of the market. They may be concerned with more cores for their particular needs. That doesn't make the product inferior......that makes the product what those people are looking for. It is a step forward from what they may already have. It could be a large step or a very small step. That's for them to decide. People always have a choice. They can upgrade, if they like or keep what they have. Nobody is twisting their arm to upgrade. If they upgrade, it's because they chose to do so.

I'm not here to start a big war with you. You're certainly entitled to your opinion and I'm not belittling that opinion. Just don't get upset when people don't agree with your opinion. They have opinions of their own and while theirs may be different, it doesn't mean that their opinion is anymore important than yours. It also doesn't mean that your opinion is right and that theirs is wrong.....or vice versa. Just means we see things differently. We'll have to agree to disagree. Just my .02...

 

[ozzy702]

I really see quad core CPUs today in the same light that highly clocked dual cores used to be. The slower quads simply weren't utilized in gaming but were much more powerful in productivity software. Fast forward to today and even the fastest clocked dual cores struggle in gaming and productivity compared to their quad core alternatives.

The same story will play out with 6, 8, and 10 core CPUs, they will age very, very well and the HEDT platform is always a better platform for cheap future upgrades. I'm kicking myself for buying 1156 and 1155 setups instead of the HEDT alternative.

DDR4, and M.2. support on 2011 v3 means that it's going to be close to a decade before even a haswell-e based 6 or 8 core is obsolete, let alone broadwell-e (which we know very little about).

Shoot the free 1366 mobo and $80 six core cpu I'm rocking in my extra box is still FAST by today's standards in multithreaded workloads and rarely chokes on single threaded tasks and that platform is almost 8 years old.

IPC and clock speed improvements are hitting walls and moar cores will eventually be the direction everything heads as efficiency in mobile and server platforms is focused upon at the expense of outright per core speed.

Intel's pricing may not be exciting, but I still think the broadwell-e chips are pretty darned exciting, especially with the 10-core that may very well OC well.

 

[III-V]

HEDT platform lags usually one generation behind mainstream.

It didn't use to be that way, I'm fairly certain. They used to be released in relative tandem. The gap really started to grow with Sandy Bridge, though.

Of course, there didn't use to be the diverging of product lines that there is today.

I'd expect that gap to grow over time, not shrink, and I expect prices to go up for HEDT relative to general consumer hardware. Lastly, I expect the (multi-threaded) performance gap between HEDT and mainstream DT to grow over time as well.

I think the single-threaded performance gap will grow as well. What I think will eventually happen is that HEDT chips will have strong cores alongside weak cores, in order to mitigate this (talking on-die, not Xeon Phi).

Someday, this mixed setup would "trickle down" to consumer hardware as well. Or "trickle up," because it's perhaps even more critical in high end mobile devices.

Frankly, I'm a bit baffled that Intel doesn't already do this, although I guess the expansion of the GPU is sort of similar. I wonder why they don't go more of the fixed function route with the GPU, add small cores for to take care of high-bandwidth, general-purpose compute, and then have big cores for latency sensitive applications.