Haswell release date & Intel/industry practices

[Turbonium]

This is a bit of a newb question, but I'm posting it anyway...

I'm no expert on the subject, so I'm wondering: given that Intel is enjoying a lack of competition from AMD in the consumer desktop market, they can now time their product release cycles to maximize profits at the cost of slowing technical progress (arguably). I'm assuming this is a driving force behind why the reported April launch-date of Haswell was pushed back to early June.

I want to build a new system, but am holding out for Haswell.

1) What are the odds that Haswell will be released "as scheduled" in early June?
2) When will we know "for certain" what the launch date of Haswell will be?
3) How soon after "launch" will Haswell actually become available to consumers?

 

[Homeles]

AFAIK, Intel doesn't do paper launches. Not with the CPUs, at least.

 

[sm625]

Intel isnt "enjoying" a lack of competition. They are killing themselves by not even attempting to compete in the mobile space. There isnt going to be a desktop space for much longer. Desktop sales will be cut by 80% in 5 years... only enthusiasts will buy them. Everyone else will buy a mobile device of some sort. And with the way intel is going, almost none of those mobile devices will be powered by intel chips... because there is no way in frozen hell that an intel chip can possibly demand not one hundred, not two hundred, not three hundred, but four frickin hundred dollar premium. NO way. They may as well just close all their fabs now and save themselves the trouble. Piss on a $1000 surface pro. Just forget about it. It will never be viable. Even apple drones arent that stupid.

 

[blackened23]

Intel isn't enjoying a lack of competition, they definitely have fierce competition right now. It should be noted that their focus isn't on desktop so they're more concerned about qualcomm, samsung and nvidia rather than AMD.

I don't think intel has had many issues with missing release dates, I think June is a pretty safe bet for haswell.

 

[ShintaiDK]

Intel isnt "enjoying" a lack of competition. They are killing themselves by not even attempting to compete in the mobile space. There isnt going to be a desktop space for much longer. Desktop sales will be cut by 80% in 5 years... only enthusiasts will buy them. Everyone else will buy a mobile device of some sort. And with the way intel is going, almost none of those mobile devices will be powered by intel chips... because there is no way in frozen hell that an intel chip can possibly demand not one hundred, not two hundred, not three hundred, but four frickin hundred dollar premium. NO way. They may as well just close all their fabs now and save themselves the trouble. Piss on a $1000 surface pro. Just forget about it. It will never be viable. Even apple drones arent that stupid.

Your entire post is so full of flaws. And nag to Microsoft over the Surface Pros price, not Intel. The Surface RT with its gimped specs is already 499$.

 

[Exophase]

AFAIK, Intel doesn't do paper launches. Not with the CPUs, at least.

That may well be the case now and could have been for several years, I don't really know, but I do know that the 1GHz Pentium 3 was quite the paper launch.

 

[moonbogg]

Intel isnt "enjoying" a lack of competition. They are killing themselves by not even attempting to compete in the mobile space. There isnt going to be a desktop space for much longer. Desktop sales will be cut by 80% in 5 years... only enthusiasts will buy them. Everyone else will buy a mobile device of some sort. And with the way intel is going, almost none of those mobile devices will be powered by intel chips... because there is no way in frozen hell that an intel chip can possibly demand not one hundred, not two hundred, not three hundred, but four frickin hundred dollar premium. NO way. They may as well just close all their fabs now and save themselves the trouble. Piss on a $1000 surface pro. Just forget about it. It will never be viable. Even apple drones arent that stupid.

I doubt that. How does a tablet make sense for doing work, even for word processing? I guess you could hook up a keyboard to a tablet, but thats just stupid. I think all of this mobile junk is more of a fad than anything else.
What about needing a large monitor for work and games? I guess you could hook one up to your ipad, but really? Desktops aren't going anywhere and they won't just be for enthusiasts. The only thing going on is that people have other options to spend their money on, but most of us still need desktops.

 

[Torn Mind]

I doubt that. How does a tablet make sense for doing work, even for word processing? I guess you could hook up a keyboard to a tablet, but thats just stupid. I think all of this mobile junk is more of a fad than anything else.
What about needing a large monitor for work and games? I guess you could hook one up to your ipad, but really? Desktops aren't going anywhere and they won't just be for enthusiasts. The only thing going on is that people have other options to spend their money on, but most of us still need desktops.

You don't need a desktop to do many of those things, just a laptop with a decent OS will do.

Where desktops are required will be at the very high end, i.e gaming on extremely high resolutions, professional multimedia editing, etc.

 

[Idontcare]

This is a bit of a newb question, but I'm posting it anyway...

I'm no expert on the subject, so I'm wondering: given that Intel is enjoying a lack of competition from AMD in the consumer desktop market, they can now time their product release cycles to maximize profits at the cost of slowing technical progress (arguably). I'm assuming this is a driving force behind why the reported April launch-date of Haswell was pushed back to early June.

I want to build a new system, but am holding out for Haswell.

1) What are the odds that Haswell will be released "as scheduled" in early June?
2) When will we know "for certain" what the launch date of Haswell will be?
3) How soon after "launch" will Haswell actually become available to consumers?

Intel has always timed their product releases to maximize profits, but only AMD can lay claim to having committed corporate suicide by intentionally slowing down their product development pipeline in the name of boosting short term profits for a quarter or two.

Intel can't slow down unless they want to stall out the entire upgrade pipeline that enables them to hit their revenue numbers today.

To do what you are suggesting would result in Intel shrinking in revenue, shrinking by a lot, which is one of the least effective ways to raise profits.

They could do it, nothing prevents them from being that stupid and taking a page out of AMD's playbook, but it is far from being an inevitable outcome.

Intel only gets to continue being a $50B/year company if they keep developing new products that obsolete the one's they sold the year before and continues to offer them at pricepoints that the typical consumer considers to be acceptable to pay (sub $300).

That may well be the case now and could have been for several years, I don't really know, but I do know that the 1GHz Pentium 3 was quite the paper launch.

How many of Intel's current employees do you think were involved in that decision to paper-launch the 1GHz P3 (or the fabled 1.13GHz variant)?

The company may still have the same name, but that doesn't mean it has the same people or culture.

 

[Fjodor2001]

Intel can't slow down unless they want to stall out the entire upgrade pipeline that enables them to hit their revenue numbers today.

So Intel is primarily competing with itself in the desktop space...

In that case one can't help but wondering if some external competition wouldn't help increasing the rate of innovation?

 

[Lonyo]

It wouldn't.
Intel may only be competing with themselves in the desktop space, but look at their product lineup...

Xeon
Core
Mobile Core

Same core. Same chips often.
If they stop "innovating" in the desktop because of lack of competition, that means they are also no longer innovating in servers or mobile for the same class (Atom excluded)... so no, competition in the desktop wouldn't make any difference. (More cores != innovation)

Competition in mobile and servers already exists in various forms. That's more than enough to keep them moving, even if it's not always "obvious" competition.

 

[Idontcare]

So Intel is primarily competing with itself in the desktop space...

In that case one can't help but wondering if some external competition wouldn't help increasing the rate of innovation?

It wouldn't.
Intel may only be competing with themselves in the desktop space, but look at their product lineup...

Xeon
Core
Mobile Core

Same core. Same chips often.
If they stop "innovating" in the desktop because of lack of competition, that means they are also no longer innovating in servers or mobile for the same class (Atom excluded)... so no, competition in the desktop wouldn't make any difference. (More cores != innovation)

Competition in mobile and servers already exists in various forms. That's more than enough to keep them moving, even if it's not always "obvious" competition.

I think we are all on the same page here.

In terms of "attempting" to innovate, Intel sure is throwing a ton of money towards that effort.

They may fail to innovate despite the billions upon billions being spent, but it sure the heck won't be for lack of trying if that does turn out to be the outcome.

 

[Fjodor2001]

It wouldn't.
Intel may only be competing with themselves in the desktop space, but look at their product lineup...

Xeon
Core
Mobile Core

Same core. Same chips often.
If they stop "innovating" in the desktop because of lack of competition, that means they are also no longer innovating in servers or mobile for the same class (Atom excluded)... so no, competition in the desktop wouldn't make any difference. (More cores != innovation)

Competition in mobile and servers already exists in various forms. That's more than enough to keep them moving, even if it's not always "obvious" competition.

The cores may be the same, but the focus of innovation may not.

What you write above probably explains why Intel is targeting most of their CPU innovations to areas benefiting the mobile space (including laptops). See e.g. with IB and Haswell, where lowering power consumption and improving iGPU is their primary focus.

If Intel had some external competition in the mainstream desktop space, then perhaps we'd see more innovation targeting that market (higher performance, moar cores, ...). So then external competition would help increasing the rate of innovation in the desktop space...

 

[fixbsod]

back on topic...

I would wait until Haswell to at least see what it's like. Haswell using a difft package (LGA 1150) and is the tick, meaning the next CPU (tock) will be the die shrink and use the same LGA 1150 package.

 

[frozentundra123456]

The cores may be the same, but the focus of innovation may not.

What you write above probably explains why Intel is targeting most of their CPU innovations to areas benefiting the mobile space (including laptops). See e.g. with IB and Haswell, where lowering power consumption and improving iGPU is their primary focus.

If Intel had some external competition in the mainstream desktop space, then perhaps we'd see more innovation targeting that market (higher performance, moar cores, ...). So then external competition would help increasing the rate of innovation in the desktop space...

I know this sounds like sitting on the fence, but I agree with you to a certain extent, but also with the other posters. Intel does have to compete with itself and give a reason to upgrade. In the laptop/tablet/phone market, they are competing with themselves at the top end and with apple, samsung, et al at the low end (ARM).

However, I do feel that the upper end of the cpu spectrum has stalled for Intel. I know it was only a tic, but look at the SB-IVB move. A minimal ipc increase and no clockspeed increases except maybe 100 mhz in a few isolated instances. If BD/Vishera was more competitive in overall performance and in power usage, Intel could have easily increased clocks or brought out a mainstream six core IVB. Initially I had high hopes for Haswell, but the more I hear, the less impressed I become for the desktop market as well, unless you are using software that utilizes the new instruction set.

 

[frozentundra123456]

You don't need a desktop to do many of those things, just a laptop with a decent OS will do.

Where desktops are required will be at the very high end, i.e gaming on extremely high resolutions, professional multimedia editing, etc.

I can see laptops replacing desktops for most consumer users. Initially I thought tablets were a fad like the netbook craze of a few years ago, but maybe not. Personally I can see a tablet as a secondary device, but the price/performance is still too high in my book to be able to replace a real computer. Not to mention that Android is still very limited unless it can somehow be made to run real x86 apps. Personally, I absolutely hate android, but I am biased by the tablet I have which is riddled with bugs and seems woefully underpowered.

In the enterprise market, (businesses, scientific research, things like that) a desktop is still by far the most cost effective solution if you need to handle lots of data or run cpu intensive apps.

 

[IntelUser2000]

If Intel had some external competition in the mainstream desktop space, then perhaps we'd see more innovation targeting that market (higher performance, moar cores, ...). So then external competition would help increasing the rate of innovation in the desktop space...

That's no innovation, that's more of the same thing. 70% of their customers don't care of such a thing. How many of you have decreased your upgrade rate in the recent years? It's because you don't need it.

If cores don't bring additional performance because the software is not ready or whatever, you guys wouldn't be buying it anyway.

Desktops are the hardest to improve and also brings the smallest benefits:
-Cores: Servers
-Interconnect: Servers
-Power use: Servers, Laptops
-Graphics: Laptops

None of the above are deal breakers to desktop buyers.

What about needing a large monitor for work and games? I guess you could hook one up to your ipad, but really? Desktops aren't going anywhere and they won't just be for enthusiasts. The only thing going on is that people have other options to spend their money on, but most of us still need desktops.

That has nothing to do with their earnings, because Intel's earnings rely on people BUYING them, not keeping the systems for longer and longer without upgrading.

 

[Fjodor2001]

That's no innovation, that's more of the same thing.

Then lower power consumption and better iGPU is also more of the same thing, right? So it doesn't constitute innovation either?

If cores don't bring additional performance because the software is not ready or whatever, you guys wouldn't be buying it anyway.

Partly a chicken and egg situation. Software may not be redesigned to use more than 4 cores, since the mainstream Intel CPUs (which constitute the bulk of the market) does not have more cores than that, so it's not with the cost of doing it.

With that being said, there's of course always the law of diminishing returns of more cores anyway.

Desktops are the hardest to improve and also brings the smallest benefits:

As the situation is today - yes, probably correct. The question then is what would be needed to actually improve desktop PCs in a substantial way for the bulk of the users?

 

[IntelUser2000]

Then lower power consumption and better iGPU is also more of the same thing, right? So it doesn't constitute innovation either?

It's different, because performance is at a point where less and less users need them, while mobile devices, whether Laptops, Tablets, Smartphones could use faster GPUs and lower power.

Partly a chicken and egg situation. Software may not be redesigned to use more than 4 cores, since the mainstream Intel CPUs (which constitute the bulk of the market) does not have more cores than that, so it's not with the cost of doing it.

This is related to general users not needing that much more performance. There are segments that have insatiable need, and that's in servers, where core matters. But we are moving into era of Apps, and Fixed Function units for specific acceleration and lower power.

Software developers already have a way of taking advantage of more threads, because nearly every Intel chip has Hyperthreading. Benchmarks based on dual core Intel chips with Hyperthreading show that it behaves mostly like having extra cores.

As the situation is today - yes, probably correct. The question then is what would be needed to actually improve desktop PCs in a substantial way for the bulk of the users?

I don't think there is. Those that need it will continue to use desktop setups while the rest will go into more portable setups. As long as Moore's Law continues, its inevitable.

 

[VirtualLarry]

You know, if 2500K was $200, and 2600K was $300, I would pay good money for a six-core with hyperthreading, for $400.

On a mainstread socket, not an enthusiast socket.

I have no idea if memory bandwidth constraints have anything to do with Intel not releasing six-core CPUs into mainstream sockets or not. I haven't read anything about socket 1155 being bandwidth-constrained (the CPU cores at least).

 

[moonbogg]

You know, if 2500K was $200, and 2600K was $300, I would pay good money for a six-core with hyperthreading, for $400.

On a mainstread socket, not an enthusiast socket.

I have no idea if memory bandwidth constraints have anything to do with Intel not releasing six-core CPUs into mainstream sockets or not. I haven't read anything about socket 1155 being bandwidth-constrained (the CPU cores at least).

You can buy a 3930k for $100 more than you said, and you can get an X79 board for the same price as many 1155 boards. Its not that expensive to go 6 cores right now but people seem to think it is for some reason.

 

[frozentundra123456]

You know, if 2500K was $200, and 2600K was $300, I would pay good money for a six-core with hyperthreading, for $400.

On a mainstread socket, not an enthusiast socket.

I have no idea if memory bandwidth constraints have anything to do with Intel not releasing six-core CPUs into mainstream sockets or not. I haven't read anything about socket 1155 being bandwidth-constrained (the CPU cores at least).

This is the way I feel too. We have gone from what, 65nm Q6600 all the way down to 22nm Ivy and not had any increase in the number of cores. I really am not knowledgeable enough to know whether six cores would be worth it though. For me I am sure not, because all I do that is CPU intensive is gaming. But increasing numbers of posters seem to be doing transcoding which I think could use more cores than 4. It seems Intel is actually becoming somewhat complacent again, counting on higher IPC and hyperthreading to make up for having only 4 cores, not to mention charging a pretty premium for HT quads when it actually costs next to nothing to enable. Actually if Intel continues to focus on igp and keeps refusing to bring out mainstream hex cores, AMD could regain the top end market if Steamroller come out on time and meets the performance increases they are claiming. Of course those are 2 very big ifs. Power consumption also is too high, but I could accept that if the performance was there to match.

 

[Charles Kozierok]

Intel isnt "enjoying" a lack of competition. They are killing themselves by not even attempting to compete in the mobile space. There isnt going to be a desktop space for much longer. Desktop sales will be cut by 80% in 5 years... only enthusiasts will buy them. Everyone else will buy a mobile device of some sort.

I'll make a point here again that I've made before: demands for processing power tend to be cyclical, not linear. Technologies are developed and people get "comfortable" with the amount of power they have, and that lasts until something new requiring the next big step up in power is introduced. And then performance matters again. The cycle then repeats.

For the last several years, we've been in the bottom part of one of these cycles. Most people are "comfortable" with the amount of performance they get from their machines, and there's an inordinate amount of attention being paid to miniaturization and power consumption. And so, as always happens in these lulls, people conclude that the need for performance is "over" and this or that technology is going to die.

It could be that this is a real, legitimate sea change in the market. But my on-the-limb prediction is that this situation will last only until the Next Big Thing comes along that demands more processing power. And then, suddenly, people will care more about component performance again.

In the meanwhile, yes, everyone will buy their mobile toys, and for some people, those toys will be sufficient to get the job done. But there will always be a market for people who need real machines. The desktop isn't going anywhere.

 

[tweakboy]

This is a bit of a newb question, but I'm posting it anyway...

I'm no expert on the subject, so I'm wondering: given that Intel is enjoying a lack of competition from AMD in the consumer desktop market, they can now time their product release cycles to maximize profits at the cost of slowing technical progress (arguably). I'm assuming this is a driving force behind why the reported April launch-date of Haswell was pushed back to early June.

I want to build a new system, but am holding out for Haswell.

1) What are the odds that Haswell will be released "as scheduled" in early June?
2) When will we know "for certain" what the launch date of Haswell will be?
3) How soon after "launch" will Haswell actually become available to consumers?

Hello my friend, good post. Haswell in June is only for mobile laptops,, no desktop enthusiast chip. In 2014 Haswell will be buit for enthusiasts.

Sorry to say but in June you have no Haswell unless you wanna upgrade your laptop. its all M chips , mobile,,,,,, dont expect something thats gonna be faster then Ivy Bridge.... it just has super fast GPU and some tweaks to reduce wattage use.

Im waiting for Ivy Bridge E 6 to 12 core monster due Q3 2013. Don't wait for Haswell or youll be waiting til 2014.

 

[poohbear]

You don't need a desktop to do many of those things, just a laptop with a decent OS will do.

Where desktops are required will be at the very high end, i.e gaming on extremely high resolutions, professional multimedia editing, etc.

Uhm, i'm guessing ure in ur 20s & never worked in an office. Do u know how bad it is ergonomically to work on a laptop??? It would destroy peoples necks, lower backs, and wrists to work for hours on a laptop. U have to look down & slouch ur neck, hunch ur back, and type awkardly @ a laptop, there's no way u can expect workers to sit for hours everyday @ a laptop. A 22-27" monitor @ eye level with a normal keyboard is the only possibklle way to get any office/programming work day in & day out.