Intel Demonstrates 65W Broadwell-K Socketed Processors at GDC 2015

[III-V]

Does it make any difference if there's one or two? The point is that Intel is skipping desktop with Broadwell, apart from that (or those few) rare low TDP SKU(s).

Yeah, there is a difference. Saying there will be only one SKU is almost undoubtedly wrong, and one doesn't have any business participating in a discussion if they're not able to keep their facts straight.

What's interesting is why they are doing that. Heat issues, problems reaching high frequencies, ...? If there were no such problems, don't you think we'd have seen the regular Intel i3/i5/i7 desktop armada by now on 14 nm?

In an ideal world, Intel would have had a full Broadwell lineup out a year ago, instead of Haswell refresh.

I think the main issue however is that on desktop, Broadwell is likely more expensive for a given level of performance compared to Haswell (CPU, that is), thus it would make sense to hold out until that was no longer the case -- i.e., Skylake. Better to soak up the costs of retooling with better yielding, higher margin parts (mobile).

 

[crashtech]

I don't think Intel really expects people to upgrade to Broadwell-K from Haswell-K. The socket compatibility is mainly for OEMs to make their lives easier.

And yet the published material showed that 9-series chipsets would be compatible with Broadwell LGA1150, so there was no reason NOT to believe there would be an upgrade path, should one be desired. Going from a 4570K to an i7 Broadwell-K would not be an irrational decision if the use case merited it.

 

[Fjodor2001]

I don't think Intel really expects people to upgrade to Broadwell-K from Haswell-K. The socket compatibility is mainly for OEMs to make their lives easier.

Yeah, as things are now and with the SKUs we see on the roadmap I agree that's likely what Intel expect. But their customers may have expected otherwise.

 

[Fjodor2001]

Yeah, there is a difference. Saying there will be only one SKU is almost undoubtedly wrong, and one doesn't have any business participating in a discussion if they're not able to keep their facts straight.
In an ideal world, Intel would have had a full Broadwell lineup out a year ago, instead of Haswell refresh.

You're getting stuck on meaningless details. If it makes you happy, I'll admit that it would have been better to not use the SKU terminology until we know if there will only be one or a few SKUs. But the point is that there will only be this Broadwell-K "model" for what we know. Whether there will be only one SKU of that model, or both e.g. a 3.4 and a 3.5 GHz SKU is irrelevant to the point I was making.

 

[erunion]

Does it make any difference if there's one or two? The point is that Intel is skipping desktop with Broadwell, apart from that (or those few) rare low TDP SKU(s).

What's interesting is why they are doing that. Heat issues, problems reaching high frequencies, ...? If there were no such problems, don't you think we'd have seen the regular Intel i3/i5/i7 desktop armada by now on 14 nm?

Why do we keep having to go over this? Your speculation is way off base.

There wont be a full line up for broadwell because of the 14nm delay. Broadwell for desktop was cancelled, and Haswell Refresh was created to fill the void. A year after Haswell Refresh was launched(roughly), Skylake will be launched. Broadwell made sense in 2014, but it missed its window of opportunity. Intel chose to not artificially delay skylake in order to sell a year old product.

The fact that the one broadwell product we will see is 65w doesn't indicate anything as far as process technology. That product (DT Iris Pro) would have been 65w even if the entire product stack had been launched in 2014. Skylake is coming in 2015, its going to have 95W class products, and its using the exact same 14nm process.

Your mistake is thinking that these choices were made due to technology reasons, when really they were simply marketing decisions.

Edit: I should add that there may be other things that do point to technical problems at Intel, I'm just saying this isn't one of them.

 

[Enigmoid]

Yeah, but that's only because Broadwell got so delayed we are now getting into Skylake time frame. Effectively, desktop Broadwell therefore was canceled.

We'd seen desktop Broadwell disappearing from leaked roadmaps, and the reappearing again, only to disappear for good except for this/these rare 65 W TDP SKU(s). Likely Intel never got desktop Broadwell to perform as desired within reasonable time.

All at 65W TDP? CPU performance on par with 4C 84W TDP Haswell too? I highly doubt that.

Easily. The 4770R is 65W itself and only 10-20% behind the 4770k. 14nm brings 30% lower power before any sort of improvements.

 

[Phynaz]

People were made to believe they could upgrade to Broadwell-K with those 1150 boards. And although they may be able to install a 65W TDP Broadwell-K, those SKUs are not to be consider upgrades over 88W TDP Haswell SKUs.

Not considered an upgrade by who, you? Please tell us how high BW will clock and what it's performance will be. Also explain why dropping in a quad BW wouldn't be an upgrade for somebody running a dual HW now.

Your arguments are continuing to fall apart.

 

[Phynaz]

Easily. The 4770R is 65W itself and only 10-20% behind the 4770k. 14nm brings 30% lower power before any sort of improvements.

Don't confuse him with facts.

 

[erunion]

I question the idea that people with Haswell systems actually care about upgrading to broadwell on the same board. I don't.

Intel had no problem selling Ivy Bridge Boards (Z77?) with no future CPUs coming.

 

[SPBHM]

unless it's really bad at overclocking I don't see how one could complain of the 65W TDP or low default clock since this is unlocked anyway...

the IGP is there, you can disable it, and there is a huge l4 cache, improved IPC... unless the price is really high it seems like a win for Z97 boards owners.

better than being limited to just refreshs of the same CPUs for years...

 

[JM Popaleetus]

As far as I'm concerned, Broadwell-K just needs to come out.

 

[crashtech]

I question the idea that people with Haswell systems actually care about upgrading to broadwell on the same board. I don't.

Intel had no problem selling Ivy Bridge Boards (Z77?) with no future CPUs coming.

You expect everyone to come to the same conclusions as you? A recipe for disappointment if ever there was one.

For the record, I have a Haswell system, and I was interested in Broadwell-K as a possible upgrade.

 

[erunion]

You expect everyone to come to the same conclusions as you? A recipe for disappointment if ever there was one.

For the record, I have a Haswell system, and I was interested in Broadwell-K as a possible upgrade.

Was it really necessary to create a straw man?
I obviously never said that.

I am well aware of the limitations of anecdotal evidence.

Anyways, I'm genuinely interested in hearing about why you might want to swap broadwell into your board.

 

[AtenRa]

Also explain why dropping in a quad BW wouldn't be an upgrade for somebody running a dual HW now.

The equation has one more think to consider, PRICE.
The QuadCore GT3e Broadwell will be more expensive than the Quadcore GT2 Haswell. So unless someone was really looking for a GT3e die for the 1150 socket, Broadwell would be irrelevant for an upgrade path.
Why spend more when you can have the same perf with less(Haswell Quad) ??? or why spend more for the GT3e die when you could upgrade to Core i7 4790K ??

 

[Dave2150]

The equation has one more think to consider, PRICE.
The QuadCore GT3e Broadwell will be more expensive than the Quadcore GT2 Haswell. So unless someone was really looking for a GT3e die for the 1150 socket, Broadwell would be irrelevant for an upgrade path.
Why spend more when you can have the same perf with less(Haswell Quad) ??? or why spend more for the GT3e die when you could upgrade to Core i7 4790K ??

Intel are not stupid - if they price Broadwell-K at a significantly higher price than the 4790k, it will not sell well, since people will just wait for Skylake.

I think the price will be reasonable, though we have to see it's performance first before we can make any conclusion.

 

[AtenRa]

Intel are not stupid - if they price Broadwell-K at a significantly higher price than the 4790k, it will not sell well, since people will just wait for Skylake.

I think the price will be reasonable, though we have to see it's performance first before we can make any conclusion.

Intel doesnt want you to buy Broadwell-K, they want you to buy a new socket 1151 motherboard + Skylake CPU

 

[Dave2150]

Intel doesnt want you to buy Broadwell-K, they want you to buy a new socket 1151 motherboard + Skylake CPU

Pretty sure they want people to buy both

 

[Phynaz]

The equation has one more think to consider, PRICE.
The QuadCore GT3e Broadwell will be more expensive than the Quadcore GT2 Haswell. So unless someone was really looking for a GT3e die for the 1150 socket, Broadwell would be irrelevant for an upgrade path.
Why spend more when you can have the same perf with less(Haswell Quad) ??? or why spend more for the GT3e die when you could upgrade to Core i7 4790K ??

Possibly, but I will caution you that you do not know the price of BW.

 

[Fjodor2001]

14nm brings 30% lower power before any sort of improvements.

Across the complete frequency range, from low to high end?

 

[Fjodor2001]

Why do we keep having to go over this? Your speculation is way off base.

There wont be a full line up for broadwell because of the 14nm delay. Broadwell for desktop was cancelled, and Haswell Refresh was created to fill the void. A year after Haswell Refresh was launched(roughly), Skylake will be launched. Broadwell made sense in 2014, but it missed its window of opportunity. Intel chose to not artificially delay skylake in order to sell a year old product.

The fact that the one broadwell product we will see is 65w doesn't indicate anything as far as process technology. That product (DT Iris Pro) would have been 65w even if the entire product stack had been launched in 2014. Skylake is coming in 2015, its going to have 95W class products, and its using the exact same 14nm process.

Your mistake is thinking that these choices were made due to technology reasons, when really they were simply marketing decisions.

Edit: I should add that there may be other things that do point to technical problems at Intel, I'm just saying this isn't one of them.

From what is known, Broadwell was delayed because of yield issues. That effectively means Intel did not manage to get sufficient percentage of produced chips working at sufficiently high frequencies at 14 nm.

As for the Skylake 95 W TDP variants, do we know why they are 95 W? Because of large iGPU, Intel 14 nm requiring high TDP at high frequencies, or something else?

 

[ShintaiDK]

Yield could be about defect rate.

For Skylake the platform is 35, 65 and 95W like previous. Nobody knows yet what the final SKU TDPs will be.

 

[erunion]

As for the Skylake 95 W TDP variants, do we know why they are 95 W? Because of large iGPU, Intel 14 nm requiring high TDP at high frequencies, or something else?

As far as I know the actual TDP isn't known yet. I said 95w class, that includes things like i7-4790k at 88w.

 

[piasabird]

I wonder if Intel really cares about independent OEM's and individual's who want to build their own computers. Only time will tell.

It seems as though Intel is trying to reduce HTPC home builders. Fewer and fewer new parts are available for low power processors. Seems like only OEM's are getting parts suitable for non-gaming HTPC's.

 

[Enigmoid]

From what is known, Broadwell was delayed because of yield issues. That effectively means Intel did not manage to get sufficient percentage of produced chips working at sufficiently high frequencies at 14 nm.

As for the Skylake 95 W TDP variants, do we know why they are 95 W? Because of large iGPU, Intel 14 nm requiring high TDP at high frequencies, or something else?

Or perhaps they accelerated ULV development and dedicated ALL early production into ULV?

 

[ShintaiDK]

I wonder if Intel really cares about independent OEM's and individual's who want to build their own computers. Only time will tell.

It seems as though Intel is trying to reduce HTPC home builders. Fewer and fewer new parts are available for low power processors. Seems like only OEM's are getting parts suitable for non-gaming HTPC's.

Could you give an example?

I assume you think desktop chips only. Skylake gives 35W options, just like Haswell, Ivy and SB(45W).