Intel Starts Production of Next-Generation Haswell Microprocessors.

[NTMBK]

Don't forget, kids- AMD already broke backwards compatibility, by removing support for 3DNow! from their modern chips.

Of course 3DNow! never had good support in software, especially when SSE did the same thing as it much better and it was never in Intel chips, but it's still a good example.

I think that if MMX went the way of the dodo, no-one would really mourn it- it's a very limited ISA extension with serious limitations (mainly the aliasing onto the x87 registers). But removing SSE would be just crackers. It's been extended and improved over many years, and software is only just getting up to date and utilising SSE4. Plus only a tiny fraction of users have AVX yet.

 

[cytg111]

.. Don't forget, kids- AMD already broke backwards compatibility, by removing support for 3DNow! from their modern chips.

3DNow! was first of its kind IIRC, i still have a .lha somewhere with version of quake2 with this patch .. gave the K6 a slight edge back then as I recall.

 

[ShintaiDK]

Don't forget, kids- AMD already broke backwards compatibility, by removing support for 3DNow! from their modern chips.

Of course 3DNow! never had good support in software, especially when SSE did the same thing as it much better and it was never in Intel chips, but it's still a good example.

I think that if MMX went the way of the dodo, no-one would really mourn it- it's a very limited ISA extension with serious limitations (mainly the aliasing onto the x87 registers). But removing SSE would be just crackers. It's been extended and improved over many years, and software is only just getting up to date and utilising SSE4. Plus only a tiny fraction of users have AVX yet.

Not really, unlike MMX, 3Dnow! wasnt universal. And thats they key point.

 

[Haserath]

Don't most instructions use the same circuits to achieve their performance anyway?

As in, most of the processor's die space is the pipeline for instruction throughput instead of just instruction support.

 

[Idontcare]

- Clearly Intel thought they were going somewhere with IA64, that compilers could/should take on more than they do today. They believed in it enough to try and sink their x86 xeon lines in comparison.
Just saying that, apparantly, somone at Intel thinks/thought it is/was worth breaking backwards compatibility for new hardware/ a new paradigm.
On the other hand, i think we keep reading, here and other places, that the hardware involved with a superscalar design is really not that big of a deal/hindrance for performance onwards.
So call me semiconfused.

The original Itaniums came with x86 hardware support, it could run both ISA's natively. Second gen Itanium dropped the hardware support but still offered emulation support, you could run x86 apps through the emulator.

It wasn't until it became obvious that Itanium was not going to displace x86 that Intel dropped the effort to maintain x86 backwards-compatibility support in Itanium.

 

[Nemesis 1]

I love the Itanics. It amuses me . The fanbois lol at itanic. yet it brings in 4 billion a year. It does my heart good . But the real killer is that faiure of failures larrabbee. I look for this failed project to bring in 5 billion in 2013. Silvermont is late to the party . But it will shine in compute. I look for 2013 to be a record sells year for intel by about 1billion. Haswell will speak well for itself . But when I think of broadwell@ 14nm moving deeper into the low power markets it gives me goosebumps

 

[ginfest]

I love the Itanics. It amuses me . The fanbois lol at itanic. yet it brings in 4 billion a year. It does my heart good . But the real killer is that faiure of failures larrabbee. I look for this failed project to bring in 5 billion in 2013. Silvermont is late to the party . But it will shine in compute. I look for 2013 to be a record sells year for intel by about 1billion. Haswell will speak well for itself . But when I think of broadwell@ 14nm moving deeper into the low power markets it gives me goosebumps

Dude what are you on about? Just because your brother-in-law works in the maintenance dept (janitor) at one of the Intel plants, doesn't mean you have any insight into the company or product lines!

 

[Nemesis 1]

I think I am on about haswell and intels whole line for 2013 . It looks to be lights out products . I especially excited about the heswell 2/core/SoC/SOIX/GT3 for convertiables and tablets. The 15 watt chips for ultras is about right, Intel will sell more chips in 2013 than it ever has and 2013 is just the wake up call.

 

[Piroko]

I love the Itanics. It amuses me . The fanbois lol at itanic. yet it brings in 4 billion a year.

1997:
June: IDC predicts IA-64 systems sales will reach $38bn/yr by 2001
1998:
June: IDC predicts IA-64 systems sales will reach $30bn/yr by 2001
1999:
August: IDC predicts IA-64 systems sales will reach $25bn/yr by 2002
2000:
June: IDC predicts Itanium systems sales will reach $25bn/yr by 2003
2001:
June: IDC predicts Itanium systems sales will reach $15bn/yr by 2004
2002:
March: IDC predicts Itanium systems sales will reach $5bn/yr by end 2004

2004:
December: Itanium system sales for 2004 reach $1.4bn

We 'lol' at Itanic for a reason. It never managed any respectable revenue and is going the way of the Dodos. Itanium wasn't smart marketing at all.

 

[SiliconWars]

I love the Itanics. It amuses me . The fanbois lol at itanic. yet it brings in 4 billion a year. It does my heart good . But the real killer is that faiure of failures larrabbee. I look for this failed project to bring in 5 billion in 2013. Silvermont is late to the party . But it will shine in compute. I look for 2013 to be a record sells year for intel by about 1billion. Haswell will speak well for itself . But when I think of broadwell@ 14nm moving deeper into the low power markets it gives me goosebumps

You expect Phi to bring in more revenue than Nvidia's entire revenue? Don't you think that if the revenue for this kind of chip was there, Nvidia would already be getting it?

 

[BrightCandle]

You expect Phi to bring in more revenue than Nvidia's entire revenue? Don't you think that if the revenue for this kind of chip was there, Nvidia would already be getting it?

While they look like they are the same they really aren't. Cuda/openCL does not run x86 code and hence will never get a large chunk of sales that go into high performance computation. Only a few problems are small enough to be worth rewriting with that particular style of parallel execution end up being converted. For everything else and that there is Phi. I expect Intel to kill Nvidia into legacy with its Phi.

 

[Nemesis 1]

1997:
June: IDC predicts IA-64 systems sales will reach $38bn/yr by 2001
1998:
June: IDC predicts IA-64 systems sales will reach $30bn/yr by 2001
1999:
August: IDC predicts IA-64 systems sales will reach $25bn/yr by 2002
2000:
June: IDC predicts Itanium systems sales will reach $25bn/yr by 2003
2001:
June: IDC predicts Itanium systems sales will reach $15bn/yr by 2004
2002:
March: IDC predicts Itanium systems sales will reach $5bn/yr by end 2004

2004:
December: Itanium system sales for 2004 reach $1.4bn

We 'lol' at Itanic for a reason. It never managed any respectable revenue and is going the way of the Dodos. Itanium wasn't smart marketing at all.

I never said it was smart . I said I find it amusing . Fanbois lol at Itanic yet it bring 4 billion a year in . I bet AMD would like a product that brought in 4 billion dollars a year

 

[grimpr]

I love the Itanics. It amuses me . The fanbois lol at itanic. yet it brings in 4 billion a year. It does my heart good . But the real killer is that faiure of failures larrabbee. I look for this failed project to bring in 5 billion in 2013. Silvermont is late to the party . But it will shine in compute. I look for 2013 to be a record sells year for intel by about 1billion. Haswell will speak well for itself . But when I think of broadwell@ 14nm moving deeper into the low power markets it gives me goosebumps

Well, i wish you good fortune in your Intel stock portofolio but i think that 2013 is gonna be a big flop even for Intel, the economic crisis hit worldwide and it shows, wont stop now.

 

[Nemesis 1]

Well, i wish you good fortune in your Intel stock portofolio but i think that 2013 is gonna be a big flop even for Intel, the economic crisis hit worldwide and it shows, wont stop now.

True but this year we have PHI for a full year we have medfields for a full year we have z2250 for a full year we have silvermont in second half of year. We have haswell up and down the line . Than we have server IVB releasing . Intel will have a record year regaurdless.

 

[tweakboy]

Ivy Bridge E is coming out before the enthusiast Haswell CPUs. If you want Haswell have fun waiting until 2014 lol, just for a faster GPU and lower TDP ,,,,,,,, a heavily overclocked Ivy Bridge E 8 core will dust Haswell any day for now at least.

 

[tweakboy]

the tock is for MUCH enhanced GPU performance,,,,,, it aint the CPU... the CPU is ticking

 

[Piroko]

I never said it was smart . I said I find it amusing . Fanbois lol at Itanic yet it bring 4 billion a year in . I bet AMD would like a product that brought in 4 billion dollars a year

As far as I know it never made 4bn $ revenue for Intel. In its best year 2008 it made 4.4bn $ revenue for HP (who sold 97% of all Itanium Servers in that year). Now substract all the value and components HP added and you'll end up what Intel got out of the deal which was probably below 1/10th of HPs revenue.

 

[NTMBK]

While they look like they are the same they really aren't. Cuda/openCL does not run x86 code and hence will never get a large chunk of sales that go into high performance computation. Only a few problems are small enough to be worth rewriting with that particular style of parallel execution end up being converted. For everything else and that there is Phi. I expect Intel to kill Nvidia into legacy with its Phi.

Phi doesn't support various standard x86 extensions, and uses crazy wide vector units. Getting good performance will require a rewrite. Nowhere near as substantial as the work to run it on a CUDA card, of course, but it's not quite so easy as people think.

 

[Idontcare]

I never said it was smart . I said I find it amusing . Fanbois lol at Itanic yet it bring 4 billion a year in . I bet AMD would like a product that brought in 4 billion dollars a year

$4B? Questionable that.

In 2008, HP agreed to pay Intel $440 million dollars over five years — between 2009 and 2014 — to keep producing the Itanium chips. Of course, HP would also have to pay for the cost of the processors it ordered. Then in 2010, the two companies signed another $250 million deal that would keep Itanium on life support through 2017.

If the TAM for Itanium were such that Intel's piece of the hardware pie amounted to $4B per year then they would not resort to requiring HP to give them $690m for subsidizing the continued development of the microprocessor lineup.

At $4B/year you don't hold discussions with your customer to talk about its imminent EOL unless they open their checkbook and start paying you in advance of deliveries.

No, sorry, the court tore back the veil of secrecy and shined a bright light on the reality of Itanium's precarious market position. If it weren't for HP paying off Intel to keep Itanium alive, it would have been dead back in 2008 or any year since had the "protection" payments not been paid on time

Intel is really just looking out for HP's business here, accidents can happen at any time you know, would be shame if that were to happen...good thing HP saw its way to make its timely contributions to Intel, you know, just to avoid any unfortunate accidents, like say, oh I don't know, maybe the cancellation of Itanium :whiste:

 

[mrmt]

No, sorry, the court tore back the veil of secrecy and shined a bright light on the reality of Itanium's precarious market position. If it weren't for HP paying off Intel to keep Itanium alive, it would have been dead back in 2008 or any year since had the "protection" payments not been paid on time

IDC, what's the nature of the payment? Full payment, cash, directly to R&D or take-or-pay commitments amounting that value?

 

[ShintaiDK]

Sofar Itanium is secure until 2018. And it will share the same platform as x86, that will also take cost and development time down.

In terms of revenue and profit: (Billions)

 

[Nemesis 1]

$4B? Questionable that.



If the TAM for Itanium were such that Intel's piece of the hardware pie amounted to $4B per year then they would not resort to requiring HP to give them $690m for subsidizing the continued development of the microprocessor lineup.

At $4B/year you don't hold discussions with your customer to talk about its imminent EOL unless they open their checkbook and start paying you in advance of deliveries.

No, sorry, the court tore back the veil of secrecy and shined a bright light on the reality of Itanium's precarious market position. If it weren't for HP paying off Intel to keep Itanium alive, it would have been dead back in 2008 or any year since had the "protection" payments not been paid on time

Intel is really just looking out for HP's business here, accidents can happen at any time you know, would be shame if that were to happen...good thing HP saw its way to make its timely contributions to Intel, you know, just to avoid any unfortunate accidents, like say, oh I don't know, maybe the cancellation of Itanium :whiste:

Well thats fine , I never did a google to check . I was going by what another poster had said some time ago about itanic bringing in 4billion in sells for INTEL . You would be surprised by that posters name.

 

[Barfo]

Well thats fine , I never did a google to check . I was going by what another poster had said some time ago about itanic bringing in 4billion in sells for INTEL . You would be surprised by that posters name.

tweakboy?

 

[cytg111]

The original Itaniums came with x86 hardware support, it could run both ISA's natively. Second gen Itanium dropped the hardware support but still offered emulation support, you could run x86 apps through the emulator.

It wasn't until it became obvious that Itanium was not going to displace x86 that Intel dropped the effort to maintain x86 backwards-compatibility support in Itanium.

- Right, support/emulation but at what speed? But that is really a side-point, the primary point i was trying to convey was : What about IA64's architecture makes it, in principle, superior to x86/amd64? Intel must have had a reason to create the damned chip, yet we hear everywhere that everything that makes x86 x86 really has so little impact on performace and die area, that it does not matter. The nature of x86 is not holding x86 back. So, still confused .

 

[Abwx]

Intel must have had a reason to create the damned chip.

Getting a monopoly in server and why not in PCs later if ever
the architecture went successfull , with no more X86 licencees
to compete with.