Intel Leaks Core i7-6950X Extreme Edition On Company Website

[matar]

Most of us on x79 are ignoring X99.

Your are so right I think I might do the same and skip broadwell-E because intel skylake will introduce us to a new X motherboard platform.
broadwell-E is the end for X99

 

[galta]

Over the last years, it has been funny to read posts related to new CPUs.
It all starts with someone saying that the extra cores are nice and make all the difference.
Right after that, someone disagrees and says that, unfortunately, modern games are not able to extract all the power of muticore CPU, being limited to 2-4 cores, at most.
Next comes a guy saying how reliable his old Pentium III is, still able to perform greatly in latest titles.
A couple of posts later someone complains about AMD being unable to compete and how that leads to obscene pricing from Intel, adding that he/she is confident that next generation of AMD processors will change the game.

A couple of things we have to accept:

· More’s law seems to be no longer valid, not at least on the per core performance. You are entitled to expect a +30% gain per core every new generation, but be ready for disappointment. It is just not going to happen again, not in the foreseeable future.
· Having said that, Intel has been able to show significant progress on core count. 10 or 12 years ago 2 cores were a must; now you can get 4-6 cores by the same nominal price (or so), plus gains on the single thread performance.
· Although AMDs competition could make Intel sweat, it has not happened at least since 2006. I really wish it were different, but I am not holding my breath.
· It’s been more than 10 years since Intel’s flagship costs $1,000, so I see scarce evidence of them ripping us.
· It is a fact that few titles are able to “extract all the power of multicore CPUs”. The one to blame, however, is not Intel, but software houses. I haven’t seen many complaints about them. Also remember that Intel has not only gamers in their mind when they develop a new CPU.
· No, I am not an “Intel fan boy”.

 

[EarthDog]

You new to forums galta? Normal stuff bub.

 

[galta]

You new to forums galta? Normal stuff bub.

Yeah, new.
But message remains.

 

[PP Mguire]

Over the last years, it has been funny to read posts related to new CPUs.
It all starts with someone saying that the extra cores are nice and make all the difference.
Right after that, someone disagrees and says that, unfortunately, modern games are not able to extract all the power of muticore CPU, being limited to 2-4 cores, at most.
Next comes a guy saying how reliable his old Pentium III is, still able to perform greatly in latest titles.
A couple of posts later someone complains about AMD being unable to compete and how that leads to obscene pricing from Intel, adding that he/she is confident that next generation of AMD processors will change the game.

A couple of things we have to accept:

· More’s law seems to be no longer valid, not at least on the per core performance. You are entitled to expect a +30% gain per core every new generation, but be ready for disappointment. It is just not going to happen again, not in the foreseeable future.
· Having said that, Intel has been able to show significant progress on core count. 10 or 12 years ago 2 cores were a must; now you can get 4-6 cores by the same nominal price (or so), plus gains on the single thread performance.
· Although AMDs competition could make Intel sweat, it has not happened at least since 2006. I really wish it were different, but I am not holding my breath.
· It’s been more than 10 years since Intel’s flagship costs $1,000, so I see scarce evidence of them ripping us.
· It is a fact that few titles are able to “extract all the power of multicore CPUs”. The one to blame, however, is not Intel, but software houses. I haven’t seen many complaints about them. Also remember that Intel has not only gamers in their mind when they develop a new CPU.
· No, I am not an “Intel fan boy”.

Intel has been selling flagship products for 1000 for years at a marginal difference between the rest of the lineup. The first "major" difference was the 5960x that added 2 more cores. This time we're looking at a 500 dollar increase in cost for the flagship and the lineup was simply extended to accommodate current pricing structure. 1500 for a 10 core extreme processor on a platform that will be obsolete by Q4 seems like a rip to me and I don't even have any plans of moving to X99.

 

[remixedcat]

Hell my 3570k seems to do pretty good for its age

 

[galta]

Intel has been selling flagship products for 1000 for years at a marginal difference between the rest of the lineup. The first "major" difference was the 5960x that added 2 more cores. This time we're looking at a 500 dollar increase in cost for the flagship and the lineup was simply extended to accommodate current pricing structure. 1500 for a 10 core extreme processor on a platform that will be obsolete by Q4 seems like a rip to me and I don't even have any plans of moving to X99.

Hummm. Let's see:

• 10 year inflation in the US adds up to 20%
• The number of cores has increased 400% (2 to 10)

Leaving aside per core gains of performance, anything below $4,800 sounds to me like an "anti-rip".
In the end you are saying you have no use for 10 cores. Neither do I, so let's not upgrade. Just understand that it is not a matter of if being expensive (or a rip); it costs a lot of money (different thing), and this lot of money could be actually a bargain if we really needed the performance.

 

[PP Mguire]

Hummm. Let's see:

• 10 year inflation in the US adds up to 20%
• The number of cores has increased 400% (2 to 10)

Leaving aside per core gains of performance, anything below $4,800 sounds to me like an "anti-rip".
In the end you are saying you have no use for 10 cores. Neither do I, so let's not upgrade. Just understand that it is not a matter of if being expensive (or a rip); it costs a lot of money (different thing), and this lot of money could be actually a bargain if we really needed the performance.

You must be new to this, so I'll let it slide.

I'll reply to the end instead, when we need the performance it'll be outdated and still rather costly. Just go look at the pricing for used X58 gear.

 

[galta]

You must be new to this, so I'll let it slide.

I'll reply to the end instead, when we need the performance it'll be outdated and still rather costly. Just go look at the pricing for used X58 gear.

Not quite new. I've been assembling PCs since 486s, so I'll let you slide as well.

When you (and me) actually need 10 cores in 10 years from now, CPUs with 30 cores and 150Mb of cache will be available on the newest X159 chipset for the same $1,000 ($1,500 still to be confirmed) they cost today. If all we need are 10 cores, they we be available for $300, so guess what: complaining that 30 cores cost $1,000 in 2026 will still make no sense.
The price of old and used X99 platforms will be of interest for archeologists only.

Anyway, talk to you again in 2026.

All the best.

 

[cdawall]

Not quite new. I've been assembling PCs since 486s, so I'll let you slide as well.

When you (and me) actually need 10 cores in 10 years from now, CPUs with 30 cores and 150Mb of cache will be available on the newest X159 chipset for the same $1,000 ($1,500 still to be confirmed) they cost today. If all we need are 10 cores, they we be available for $300, so guess what: complaining that 30 cores cost $1,000 in 2026 will still make no sense.
The price of old and used X99 platforms will be of interest for archeologists only.

Anyway, talk to you again in 2026.

All the best.

Prices have dropped substantially since 486 days so bad example. The price jump is just intel being intel and have the market so they can do whatever they want.

 

[galta]

Prices have dropped substantially since 486 days so bad example. The price jump is just intel being intel and have the market so they can do whatever they want.

Dear God...
Once I was guilty of being a newbie, now I am guilty of being old. It never really ends, does it?

Yes, they can do what they want as long as they are willing to face the consequences, and yes, if AMD had good products and some 40% of market share prices would lower.

All things said (or sad), think about it: 10 years ago, when AMD was a real player and Intel's flaship would retail for the same $1,000 it costs today, people would not complain as much as they do today. There is no sense at all.

 

[cdawall]

Dear God...
Once I was guilty of being a newbie, now I am guilty of being old. It never really ends, does it?

Yes, they can do what they want as long as they are willing to face the consequences, and yes, if AMD had good products and some 40% of market share prices would lower.

All things said (or sad), think about it: 10 years ago, when AMD was a real player and Intel's flaship would retail for the same $1,000 it costs today, people would not complain as much as they do today. There is no sense at all.

Prices wouldn't be lower. AMD proved they would do the exact same thing intel does.

Athlon 64 FX 51 (socket 940 aka the server socket) $733
Athlon 64 FX 53 (socket 939 so desktop finally) $733
Athlon 64 FX 55 $827
Athlon 64 FX 57 $1031
Athlon 64 FX 60 $1031
Athlon 64 FX 62 $1031

The next batch of AMD chips were a real value if you didn't include platform costs. AMD's original attempt to compete with the C2Q extremes is a bit of a bust.

Athlon 64 FX 70 $599 pr (2 required for quadfather platform which again is server based)
Athlon 64 FX 72 $799 pr
Athlon 64 FX 74 $999 pr

 

[galta]

Prices wouldn't be lower. AMD proved they would do the exact same thing intel does.

Athlon 64 FX 51 (socket 940 aka the server socket) $733
Athlon 64 FX 53 (socket 939 so desktop finally) $733
Athlon 64 FX 55 $827
Athlon 64 FX 57 $1031
Athlon 64 FX 60 $1031
Athlon 64 FX 62 $1031

The next batch of AMD chips were a real value if you didn't include platform costs. AMD's original attempt to compete with the C2Q extremes is a bit of a bust.

Athlon 64 FX 70 $599 pr (2 required for quadfather platform which again is server based)
Athlon 64 FX 72 $799 pr
Athlon 64 FX 74 $999 pr

Thank you very, very much.

I-hate-Intel-monopoly guys, something else to add?

No? Ok, thanks.
Talk to you guys again when 7xxx is out.

 

[EarthDog]

LOL, this guy is a real gem...

 

[PP Mguire]

Not quite new. I've been assembling PCs since 486s, so I'll let you slide as well.

When you (and me) actually need 10 cores in 10 years from now, CPUs with 30 cores and 150Mb of cache will be available on the newest X159 chipset for the same $1,000 ($1,500 still to be confirmed) they cost today. If all we need are 10 cores, they we be available for $300, so guess what: complaining that 30 cores cost $1,000 in 2026 will still make no sense.
The price of old and used X99 platforms will be of interest for archeologists only.

Anyway, talk to you again in 2026.

All the best.

Then you should know that business major financial math doesn't work in the PC world for a multitude of reasons.

 

[galta]

Then you should know that business major financial math doesn't work in the PC world for a multitude of reasons.

Never said otherwise...

 

[PP Mguire]

Never said otherwise...

*Looks above*....k.

 

[galta]

*Looks above*....k.

Oh, that?

It was not business major financial math. Just arithmetic - which applies everywhere - plus a strong statement susceptible to empirical verification: absent inflationary problems in the US, I expect Intel flagships to remain around $1,000 in the foreseeable future, just like it has over the last 10 years.
Therefore, an i7-6950x for $1,000 today is way cheaper than QX6700 was in 2006, even if gains from previous generation are way smaller.
Again, no finance.
But I believe we have amused our audience enough for now.
Let's talk again in 2026 so we have facts, instead of opinions, to prove who was right.
Right?

 

[johnspack]

I'm waiting for an octo core that can do 5ghz, so I can game as well as transcode and run vms on the same dam machine. How practical would that be? I bet people would pay for that.

 

[Breit]

I'm waiting for an octo core that can do 5ghz, so I can game as well as transcode and run vms on the same dam machine. How practical would that be? I bet people would pay for that.

There were rumors about a quad core Xeon E5-2602 v4 as part of the Broadwell-EP lineup clocking up to 5.1GHz per default a few weeks back. Allthough this chip is not targeted towards the retail market, it shows that the Broadwell-E(P) die can clock very high.
So your dream may eventually become true this generation...

 

[PP Mguire]

I'm waiting for an octo core that can do 5ghz, so I can game as well as transcode and run vms on the same dam machine. How practical would that be? I bet people would pay for that.

Isn't this like the main defense of the AMD FX line?

 

[Vayra86]

I'm waiting for an octo core that can do 5ghz, so I can game as well as transcode and run vms on the same dam machine. How practical would that be? I bet people would pay for that.

They probably will, but the big issue is that not enough people will pay for it.

Not every niche is a worthwhile one to build products for AMD still needs to learn this, FX 9590 and Fury X I'm looking at you.

Also, power consumption will go through the roof. Being able to clock high, and being able to do so for 24/7 use are two different things. You don't go 24/7 on LN2 and you also don't run these high cost chips on the very edge of what's possible in terms of vCore, degradation and all that... And then there is the 8 cores. Broadwell-E 10 cores with a 140w base TDP, I reckon 10x5.1 Ghz will easily go past 350w.